COGNITIVE PSYCHOLOGY 7, 573-605 (1975)
'Family Resemblances: Studies in the Internal Structure of Categories'
ELEANOR ROSCH AND CAROLYN B. MERVIS
University of California, Berkeley
Six experiments explored the hypothesis that the members of categories which are considered most prototypical are those with most attributes in common with other members of the category and least attributes in common with other categories.
In probabilistic terms, the hypothesis is that prototypicality is a function of the total cue validity of the attributes of items.
...
As speakers of our language and members of our culture, we know that a chair is a more reasonable exemplar of the category furniture than a radio, and that some chairs fit our idea or image of a chair better than others.
However, when describing categories analytically, most traditions of thought have treated category membership as a digital, all-or-none phenomenon.
That is, much work in philosophy, psychology, linguistics, and anthropology assumes that categories are logical bounded entities, membership in which is defined by an item’s posses-
[574]
sion of a simple set of criterial features, in which all instances possessing the criterial attributes have a full and equal degree of membership.
In contrast to such a view, it has been recently argued that some natural categories are analog and must be represented logically in a manner which reflects their analog structure.
Rosch (1973, 1975b) has further characterized some natural analog categories as internally structured into a prototype (clearest cases, best examples of the category) and nonprototype members, with nonprototype members tending toward an order from better to poorer examples.
While the domain for which such a claim has been demonstrated most unequivocally is that of color there is also considerable evidence that natural superordinate semantic categories have a prototype structure. Subjects can reliably rate the extent to which a member of a category fits their idea or image of the meaning of the category name , and such ratings predict performance in a number of tasks.
However, there has, as yet, been little attention given to the problem of how internal structure arises.
That is, what principles govern the formation of category prototypes and gradients of category membership?
For some categories which probably have a physiological basis, such as colors, forms, and facial expressions of basic human emotions, prototypes may be stimuli which are salient prior to formation of the category, whose salience, at the outset, determines the categorical structuring of those domains.
For the artificial categories which have been used in prototype research—such as families of dot patterns and artificial faces—the categories have been intentionally structured and/or the prototypes have been defined so that the prototypes were central tendencies of the categories.
For most domains, however, prototypes do not appear to precede the category and must be formed through principles of learning and information processing from the items given in the category.
The present research was not intended to provide a processing model of the learning of categories or formation of prototypes;
rather, our intention was to examine the stimulus relations which underlie such learning.
That is, the purpose of the present research was to explore one of the major structural principles which, we believe, may govern the formation of the prototype structure of semantic categories.
This principle was first suggested in philosophy;
Wittgenstein (1953) argued that the referents of a word need not have common elements in order for the word to be understood and used in the normal functioning
[575]
of language.
He suggested that, rather, a family resemblance might be what linked the various referents of a word.
A family resemblance relationship consists of a set of items of the form AB, BC, CD, DE.
That is, each item has at least one, and probably several, elements in common with one or more other items, but no, or few, elements are common to all items.
The existence of such relationships in actual natural language categories has not previously been investigated empirically.
In the present research, we viewed natural semantic categories as networks of overlapping attributes;
the basic hypothesis was that members of a category come to be viewed as prototypical of the category as a whole in proportion to the extent to which they bear a family resemblance to (have attributes which overlap those of) other members of the category.
Conversely, items viewed as most prototypical of one category will be those with least family resemblance to or membership in other categories.
The latter part of this
hypothesis
tends to get lost in the shuffle, but it is just as important as the rest. Carroll has emphasized the
truism of logic
that
everything resembles everything else in some respect.
This is crucial, but it goes less than halfway toward painting the empirical picture he (and we) are after. If some resemblance
can always
be found, then how and why is one
ever not
found?
There seem to be at least two parts to the answer, one of which I have already said more about than I am qualified to say: This is the difference between 'seeking' and 'finding', implicit and explicit learning, perception and apperception, etc. The role of
stimuli which are salient prior to formation of the category
has already been touched on above. On this, Carroll's "truism" is highly misleading unless we add:
' . . . the salience of these resemblances varies widely, from the unignorable to the imperceptible . . . '
With this corollary in place, the disregard of
certain manifest or unexhibited properties
in favor of
perceptible resemblances
can no longer reasonably be called a
misuse of the notion of family resemblance.
Even spurious "resemblance" must have a target object. If offspring "descend" from an "unexhibited" parent, there is no resemblance because there is nothing for the offspring to resemble.
The other thing to consider here is
nonresemblance
as itself a robust and operative empirical concept; that is,
not
as the proverbial 'other side of the coin', as it necessarily is
rhetorically,
but rather as a 'concept'
sui generis,
one which has 'a use in the language' that extends beyond this elementary 'rhetorical' role.
If the above
hypothesis
is correct, then we can anticipate that no profusion of shared properties among a set of objects can outweigh a
lack
of differentiation from other sets of objects. Indeed, per the "truism," every new similarity noted of a target-object
just is
a (likely) vector of
de-differentiation from every other object: other objects may
share
the adduced property, but few (perhaps none) will share the full
combination
of adduced properties.
Here then is another aspect of 'family resemblance' theory underplayed by Carroll:
prototypicality
is borne not of brute force but
minumum force;
it is not a simply matter of piling up similarities until the preponderance of
positive
evidence-for-similarity becomes overwhelming.
If we keep 'seeking' for similarities, we will keep 'finding' them; and then, very quickly, everything will come to resemble everything else, not just in 'some' respect, but in
many
or even
most
respects.
That
is where 'seeking' leads. But needless seeking is not much of an empirical concern. (Principle of Least Effort!)
The real concern, I think, is that different people are differently acute observers in different domains for different reasons. 'Salience' is undefeated, of course; still, I can't help but think that boilerplate cognitivist levels of agreement in the 70-80% range hardly preclude some quite extreme variation in our perceptions of similarity and difference.
Perhaps the
nonresemblance
concept is acknowledged obliquely in the art-definition search, i.e., whenever a candidate definition is deemed to be 'overinclusive'; but this issue really is deeper and more granular than those sorts of wild goose chases can accommodate.
Now metaphysically and existentially:
What kind of world do YOU want to live in?
In what and how many 'respects' would you prefer for 'everything to resemble everything else'? If you think you don't care, think harder about some likely consequences. Consider that words can be at odds with actions.
When prompted explicitly to 'seek' an answer, we say:
'More is more!'
'It's all connected, man!'
'Gimme one with everything!'
But this is not at all what the preponderance of unprompted human action suggests we need. What we need, I think, is to live in a society and culture which enshrines our unsought classifications as First Principles while requiring everything we seek after to prove itself scientifically and pragmatically, in some essential human domain, before being accorded the least respect or importance. We need to find and hold an optimal proportion of
resemblance
to
nonresemblance.
(I was going to say 'a balance of opposites', but neither the optimality of 'balance' nor the mere fact of 'opposition' can be assumed a priori.)
Total compartmentalization and total dedifferentiation are both madness-inducing. Once we stop thinking so hard about it and simply go about our day, we find ourselves constructing well-balanced gestalts without trying. We don't seem to be very good at doing this wilfully, however; and so, if our quite wilfully-constructed institutions predictably also struggle with this, it would be better to refer them and their endless 'seeking' to everything we have already 'found' elsewhere.
When
must
we 'seek' for solutions to category problems? Carroll first coughs up a couple of examples of
practical and political questions:
Whether or not something is art might determine whether or not it is eligible for an award from a government arts agency or whether its sale or import should be taxed.
He then purports to show that
Identifying whether something should be classified as art or not is crucial to ascertaining how we should respond to it.
Presumably, for art-philosophers, this latter question is where the action is. If so, I beg to differ. The category problem,
once it has been made explicit,
is among the most recalcitrant philosophical problems we face. It would be better
not
to have to 'face' it in 'practical and political' life any more than is absolutely unavoidable. Otherwise, the preliminary definition of terms threatens to go on forever and to swallow practical politics whole.
Could we
just not have
government arts agencies and import taxes? I can imagine several arguments, some of which I could accept, others not; but that's not the point. I am not arguing for or against Fiscal Conservatism, nor Boomer Institutionalism, nor Democratic Socialism. What if
some
form of 'import tax' arises under all three regimes? Then all three regimes inherit Carroll's category problem, at least in its throwaway version. If these are the contenders, then the only way to win is not to play the game.
I couldn't even guess which
-isms entail the total abolition of import taxes, and for all I know that could be the best or the worst idea. The point is: Here we come face-to-face, finally, with the
genesis
of
unintended consequences.
Now, I can't say I've ever heard this version:
In order to tax imports by category, you have to be able categorize your imports.
But this is indeed true. Carroll's example is apt in this respect, but otherwise it is an archetypically Boomerblind example. The reason I think this is because I think that the need to transport artworks across national borders is more fundamental than the need to levy taxes on them. If I'm wrong about this, then I am a backwards person in at least this respect. But this is what I think. And I think I
know
that a category problem is
created
by the decision the levy taxes on imports. This is not much of an argument against those taxes, and again, it is not intended to be. You could say that it is the transnational transporter and not the bureaucrat or the state who 'creates' the problem; but what kind of -ist are you if you think that? I merely hope to suggest that Carroll relies here upon a highly contigent example, something that very well could be otherwise, and which very many people are very concerned to show
indeed ought to be
otherwise. I am generally ignorant of these arguments. What
do
I believe in? As Dr. Grayson said of composition, the best way out of a tough spot is not to get into one in the first place. I am saying: if certain forms (and instrumentations and occasions) put even the great composers in tough spots, then the problem is with the forms and not with the composers.
In natural categories of concrete objects, the two aspects of family resemblance should coincide rather than conflict since it is reasonable that categories tend to become organized in such a way that they reflect the correlational structure of the environment in a manner which renders them maximally discriminable from each other.
The present structural hypothesis is closely related to a cue validity processing model of classification
in which the validity of a cue is defined in terms of its total frequency within a category and its proportional frequency in that category relative to contrasting categories.
Mathematically, cue validity has been defined as a conditional probability—specifically, the frequency of a cue being associated with the category in question divided by the total frequency of that cue over all relevant categories.
Unfortunately, cue validity has been treated as a model in conflict with a prototype model of category processing where prototypes are operationally defined solely as attribute means.
If prototypes are defined more broadly—for example, as the abstract representation of a category, or as those category members to which subjects compare items when judging category membership, or as the internal structure of the category defined by subjects’ judgments of the degree to which members fit their “idea or image” of the category—then prototypes should coincide rather than conflict with cue validity.
That is, if natural categories of concrete objects tend to become organized so as to render the categories maximally discriminable from each other, it follows that the maximum possible cue validity of items within each category will be attained.
The principle of family resemblance relationships can be restated in terms of cue validity since the attributes most distributed among members of a category and least distributed among members of con-
[576]
trasting categories are, by definition, the most valid cues to membership in the category in question.
We use the term family resemblance rather than cue validity primarily to emphasize that we are dealing with a description of structural principles and not with a processing model.
We believe that the principle of family resemblance relationships is a very general one and is applicable to categories regardless of whether or not they have features common to members of the category or formal criteria for category membership.
In all of the studies of the present research, family resemblances were defined in terms of discrete attributes such as has legs, you drive it, or the letter B is a member.
These are the kinds of features of natural semantic categories which can be most readily reported and the features normally used in definitions of categories by means of lists of formal criteria.
How we'd just love to know about everything that is perceived but is not readily reported or normally used in definitions!
Insofar as the context in which an attribute occurs as part of a stimulus may always affect perception and understanding of the attribute, discrete attributes of this type may be an analytic myth.
However, in one sense, the purpose of the present research was to show that it is not necessary to invoke attribute interactions or higher order gestalt properties of stimuli in order to analyze the prototype structure of categories.
That is, even at the level of analysis of the type of discrete attributes normally used in definitions of categories by means of criterial features, we believe there is a principle of the structure of stimulus sets, family resemblances, which can be shown to underlie category prototype structure.
The present paper reports studies using three different types of category;
superordinate semantic categories such as furniture and vehicle,
basic level semantic categories such as chair and car,
and artificial categories formed from sets of letter strings.
For each type of stimulus, both aspects of the family resemblance hypothesis (that the most prototypical members of categories are those with most attributes in common with other members of that category and are those with least attributes in common with other categories) were tested.
Superordinate semantic categories are of particular interest because they are sufficiently abstract that they have few, if any, attributes common to all members.
Thus, such categories may consist almost entirely of items related to each other by means of family resemblances of overlapping attributes.
In addition, superordinate categories have the advantage that their membership consists of a finite number of names of basic level categories which can be adequately sampled.
Art certainly would seem be of this type, no?
Superordinate categories have the disadvantage that they do not have contrasting categories (operationally defined below);
thus, the second half of the family resemblance hypothesis (that prototypical members of categories have least resemblance to other categories) had
[577]
to be tested indirectly by measuring membership in, rather than attributes in common with, other superordinate categories.
Basic level semantic categories are of great interest because they are the level of abstraction at which the basic category cuts in the world may be made.
However, basic level categories present a sampling problem since their membership consists of an infinite number of objects.
On the positive side, basic level categories do form contrast sets, thus, making possible a direct test of the second part of the family resemblance hypothesis.
Artificial categories were needed because they made possible the study of prototype formation with adequate controls.
In natural language domains of any type, categories have long since evolved in culture and been learned by subjects.
Both prototypes and the attribute structure of categories are independent variables; we can only measure their correlations.
Artificial categories are of use because attribute structures can be varied in a controlled manner and the development of prototypes studied as a dependent variable.
PART I: SUPERORDINATE SEMANTIC CATEGORIES
Experiment 1
...
[584]
...
Experiment 2
...
[586]
...
PART II: BASIC LEVEL CATEGORIES
It has been previously argued (Rosch, in press-a; Rosch et al., in press) that there is a basic level of abstraction at which the concrete objects of the world are most naturally divided into categories.
A working assumption has been that, in the domains of both man-made and biological objects, there occur information-rich bundles of attributes that form natural discontinuities.
These bundles are both perceptual and functional.
It is proposed that basic cuts are made at this level.
The dreaded Form-Content distinction looms enormous here.
It is precisely some workable concept of
natural discontinuities
that is needed to lend the concepts of
Form and Content some 'analytic' coherence. Otherwise, as Carroll shows, they simply collapse into each other. He is wishy-washy here even by his standards. He seems to prefer the notion of 'form' as
relations between parts of a work.
If we could all agree upon that definition and move on, I would be all for it. This would at least help us to
understand
what critics mean by 'form' and 'content', which I confess I usually do not. Analytically, however, this merely passes the buck to the terms 'relations' and 'parts'. (Incidentally, see Dilman,
previously,
for why we had better not get too hung up on 'between'.)
Can't we always name as many or as few 'relations' as we want to?
Can't division into 'parts' go on infinitely? Aren't both of these possibilities equal parts unprincipled and unbounded? Only if empirical inquiry indicates as much. Here perhaps I myself have underplayed the prospect of
most natural divisions
of
concrete objects into categories.
This would seem to open the possibility of principled designations of "objects" as 'content' if and when (1) some
salient stimulus
triggers
(2) recognition of some
bundle of attributes
at
(3)
the appropriate
basic level of abstraction.
It would then become impossible
not
to experience
some
meaningful subjective distinction between 'parts' and wholes. I think it would nonetheless be
very easy
to avoid positing 'relations' among parts without significant further prompting; but of course it can always be argued that this is going on implicitly, and I presume it's very difficult to settle that question empirically. Failing that, critics can continue to prompt us in this direction, and we can choose to ensnare ourselves in their web of intrigue if for some reason we find this prospect appealing. (I never have.)
It's crucial to realize, in this slapdash theory of mine, that the "attributes" are NOT the 'parts'. Rather, the "objects" are the 'parts'. And that is to say,
there may be no
"concrete objects" in R&M's sense above; not unless the artwork in question is, say, an array (or a twisted pile) of them. It seems to me that it is unproblematic to borrow the laws of "concrete objects" on behalf of the proposed laws of 'parts' of artworks, even for 'parts' of novels and symphonies; if I'm wrong about that, then of course the theory is dead in the water.
I resist, as always, any suggestion that "salience" is monolithic. Exceptions which are small and occasional enough to deter scientists not in the least in claiming universality for a behavior seem to me, nonetheless, more than adequate (especially at present population scale, connectivity, and, where applicable, density) to engender far wider aesthetic divergence than will permit acceptance of Carroll's Aesthetic Realism without significant qualifications. There is plenty of room for divergence even under the strictures laid out by R&M here. It is of course quite laborious to try to say why . . .
Basic objects (for example, chair, car) are the most inclusive level of abstraction at which categories can mirror the correlational structure of the environment and the most inclusive level at which there can be many attributes common to all or most members of the categories.
The more abstract combinations of basic level objects (e.g., categories such as furniture and vehicle used in Experiments 1 and 2) are superordinates which share only a few attributes; the common attributes are rather abstract ones.
Categories below the basic level are subordinates (e.g., kitchen chair, sports car).
Subordinates are also bundles of predictable attributes and functions, but contain little more information than the basic level object to which they are subordinate.
Basic categories are, thus, the categories for which the cue validity of attributes within categories is maximized:
Superordinate categories have lower cue validity than basic because they have fewer common attributes within the category;
subordinate categories have lower cue validity than basic because they share
[587]
attributes with contrasting subordinate categories (e.g., kitchen chair shares most of its attributes with living room chair).
In a converging series of experiments, it was confirmed that basic objects are the most inclusive categories in which clusters of attributes occur which subjects agree are possessed by members of the category;
sets of common motor movements are made when using or interacting with objects of that type;
commonalities in the shape, and, thus, the overall look, of objects occur;
it is possible to recognize an averaged shape of an object of that class;
and it is possible to form a representation of a typical member of the class which is sufficiently concrete to aid in detection of the object in visual noise.
In addition, basic objects were shown to be the first categorizations made by young children, and basic object names the level of abstraction at which objects are first named by children and usually named by adults.
The present research concerned the question of whether the family resemblances of items in basic level categories were related to prototypicality in the way in which it had proved to be in the superordinate categories studied in Experiments 1 and 2.
Do subjects agree concerning which members of basic object categories are the more prototypical—do they agree, for example, about which cars more closely fit their idea or image of the meaning of car?
And, if agreement in prototypicality ratings is obtained, does it hold, as it did in the case of superordinate categories, that the more prototypical category members are those with most resemblance to members of that category and least resemblance to other categories?
...
Experiment 3
...
[589]
...
Experiment 4
...
[591]
...
PART III. ARTIFICIAL CATEGORIES
In the four preceding experiments, it was shown for a sample of naturally occurring categories that items rated more prototypical of the category were more closely related to other members of the category and less closely related to members of other categories than were items rated less prototypical of a category.
Categories designated by the words of natural languages have the advantage for study that they have evolved and occur in actual human usage;
however, they have the disadvantage that the variables of interest occur in uncontrolled and, thus, unanalyzable conjunction with each other and with other extraneous factors.
In the previous experiments, the object was to determine the structure of preexisting categories.
In the following two experiments, artificial categories were constructed in which items differed only in the degree of family resemblance within categories or amount of overlap of attributes between categories.
In these experiments, the structure was provided as an independent variable; our hypothesis was that this structure would affect rate of learning of category items; reaction time in judging category membership once the categories were learned; and ratings of prototypicality of items.
[592]
...
Experiment 5
...
[596]
...
Experiment 6
...
[598]
...
GENERAL DISCUSSION
The results of the present study confirmed the hypothesis that the most prototypical members of common superordinate, basic level, and artificial categories are those which bear the greatest family resemblance
[599]
to other members of their own category and have the least overlap with other categories.
In probabilistic language, prototypicality was shown to be a function of the cue validity of the attributes of items.
In the particular studies in this paper, we defined and measured family resemblance in terms of discrete attributes;
however, previous studies indicate that the principle can be applied, to some extent, to other types of categories,such as dot patterns distorted around a prototype and categories consisting of items composed of continuous attributes which have a metric.
In such categories, the prototype dot pattern and the pattern with attributes at mean values have more in common with (are more like) the other items in the category than are items further from the prototype or the mean.
Family resemblances (even broadly defined) are undoubtedly not the only principle of prototype formation—for example, the frequency of items and the salience of particular attributes or particular members of the categories (perceptual, social, or memorial salience) as well as the as yet undefined gestalt properties of stimuli and stimulus combinations, undoubtedly contribute to prototype formation—
however, the results of the present study indicate that family resemblance is a major factor.
Such a finding is important in six ways:
(a) It suggests a structural basis for the formation of prototypes of categories,
(b) It argues that in modeling natural categories, prototypes and cue validity are not conflicting accounts, but, rather, must be incorporated into a single model,
(c) It indicates a structural rationale for the use of proximity scaling in the study of categories, even in the absence of definable category dimensionality,
(d) It offers a principle by which prototype formation can be understood as part of the general processes through which categories themselves may be formed,
(e) It provides a new link between adult and children’s modes of categorization, and
(f) It offers a concrete alternative to criterial attributes in understanding the logic of categorical structure.
Family resemblance as a structural basis for prototype formation.
The origin of prototypes of categories is an issue because, as outlined in the introduction, there is now considerable evidence that the extent to which members are conceived typical of a category appears to be an important variable in the cognitive processing of categories.
From that previous work alone, it could be argued that ratings of prototypicality are only measures of the associative linkage between an item and the category name and that it is such associative strength which determines the effects of typicality on processing tasks such as those used in semantic memory.
While in a processing model, associative strength may, by definition, be directly related to typicality effects, associative strength need not be conceived only as the
[600]
result of the frequency of (arbitrary or accidental) pairings of the item with the category name.
The present experiments have attempted to provide a structural principle for the formation of prototypes;
family resemblance relationships are not in contradiction to, but, rather, themselves offer a possible structural reason behind associative strength.
The principle of family resemblance is similar but not identical to two recent accounts of prototype effects:
the attribute frequency model (Neumann, 1974) and
an element tag model (Reitman & Bower, 1973).
Both of these models were designed to account, without recourse to an “abstraction process,” for the findings of several specific previous experiments ...
Both models predict memory (particularly the mistaken memory for prototype items which were not actually presented) from the frequency with which elements appear in a learning set.
A family resemblance account of prototypes is of greater generality than these models.
In the first place, it accounts for prototypes in terms of distributions of attributes rather than in terms of the simple frequency of attributes (a factor which also distinguishes family resemblances from a narrow definition of cue validity).
In the second place, it includes an account of the distribution of attributes over contrasting categories rather than focusing only on the category in question.
That it is distribution rather than simple frequency of attributes which is most relevant to prototypes in natural categories is argued by two facts:
(a) The measure of distribution used in the present study was highly correlated with ratings of prototypicality for superordinate categories,
whereas, a measure of the frequency of items (which is necessarily correlated with frequency of attributes) in the category is not correlated with prototypicality, and
(b) The overlap of attributes with contrasting categories is itself a distributional property not a property of simple frequency.
(In the artificial categories of Experiment 5 of the present paper, distributional and simple frequency were equivalent; however, in the other experiments, they were not—clarification of the relations between distribution and frequency of attributes is an issue which requires further research.) That the distribution of attributes over contrasting categories is as important a principle of prototype formation as distribution of attributes within a category is argued by the results of Experiments 2, 4, and 6.
At this point, it should be reiterated that the principle of family resemblance, as defined in the present research, is a descriptive, not a processing principle.
Family resemblances are related to process models in two ways:
(a)
Any account of the processes by which humans convert stimulus attributes into mental or behavioral prototypes (such as an attribute tag model) should be able to account for the family resemblance attribute structure of categories outlined by the present research,
and
(b)
[601]
Classification by computation of cue validity and classification by matching to a prototype have been treated as alternative process models which are in conflict;
however, the principle of family resemblance suggests that, for natural categories, both should be aspects of the same processing model.
Family resemblance as an argument for the compatibility of cue validity and prototype models.
Probability models, such as cue validity, and distance models, such as matching to a prototype, have been treated as two fundamentally different forms of categorization model whose conflicting validities must be tested by empirical research.
However, the present study has shown that empirically defined prototypes of natural categories are just those items with highest cue validity.
Such a structure of categories would, in fact, appear to provide the means for maximally efficient processing of categories.
Computation and summation of the validities of individual cues is a laborious cognitive process.
However, since cue validity appears to be the basis of categories, it is ecologically essential that cue validities be taken into account, in some manner, in categorization.
If prototypes function cognitively as representatives of the category and if prototypes are items with the highest cue validities, humans can use the efficient processing mechanism of matching to a prototype without sacrificing attention to the validity of cues.
...
In short, humans probably incorporate probabilistic analysis of cues and computation of distance from a representation of the category into the same process of categorization;
future research on categorization would do well to attempt to model the ways in which that incorporation can occur rather than to treat cue validity and prototypes as conflicting models.
Family resemblance as a basis for proximity scaling.
Just as it has been customary to treat categories in terms of logical defining features which were assumed to be common to all members of the category, it is also not uncommon to treat proximity scaling of items in categories only as a means of determining the general dimensions along which items of the category are seen to differ.
However, the results of the multidimensional scaling of the items of the superordinate categories in Experiment 1 (performed with Smith, Shoben, and Rips) indicated that family resemblance was predictive of centrality of items in the derived similarity space regardless of interpretability of dimensions or of item clusters.
It should, in general, be the case that the more that items have in common with other items in a class (the closer the items are to all other items irrespective of the basis of closeness), the more central those items will
[602]
be in a space derived from proximity measures.
The demonstration of the importance of family resemblances (and of prototypicality) in classification provided by the present research suggests that the dimension of centrality may itself be an important aspect of and deserve to be a focus of attention in the analysis of proximity spaces.
Family resemblance as a part of the general process of category formation.
The concept of family resemblances is also of general use because it characterizes prototype formation as part of the general process by which categories themselves are formed.
It has been argued by Rosch et al., (in press) that division of the world into categories is not arbitrary.
The basic category cuts in the world are those which separate the information-rich bundles of attributes which form natural discontinuities.
Basic categories have, in fact, been shown to be the most inclusive categories in which all items in the category possess significant numbers of attributes in common, and, thereby, are used by means of similar sequences of motor movements and are like each other in overall appearance.
Basic categories are the categories for which the cue validity of attributes within categories is maximized since superordinate categories have fewer common attributes within the category than do basic categories and subordinate categories share more attributes with contrasting categories than do basic categories.
Basic categories are, thus, the categories which mirror the correlational structure of the environment.
The present study has shown that formation of prototypes of categories appears to be likewise nonarbitrary.
The more prototypical a category member, the more attributes it has in common with other members of the category and the less attributes in common with contrasting categories.
Thus, prototypes appear to be just those members of the category which most reflect the redundancy structure of the category as a whole.
That is, categories form to maximize the information-rich clusters of attributes in the environment and, thus, the cue validity of the attributes of categories;
when prototypes of categories form by means of the principle of family resemblance, they maximize such clusters and such cue validity still further within categories.
Family resemblance as a link with children’s classifications.
The principle of family resemblances in adult categories casts a new perspective on children’s classifications.
Young children have been shown to classify objects or pictures by means of complexive classes, that is, classes in which items are related to each other by attributes not shared by all members of the class.
For example, Vygotsky (1962) speaks of the child in the “phase of thinking in complexes” starting with a small yellow triangle, putting with it a red triangle, then a red circle—in each case matching the new item to
[603]
one attribute of the old.
Bruner et al. describe the young child’s tendency to classify by means of “complexive structures,” for example, “banana and peach are yellow, peach and potato round. . . .”
Such complexive classes have been considered logically more primitive than the adult preferred method of grouping taxonomically by “what a thing is”—that is, grouping by superordinate classes and justifying groups by their superordinate names.
However, the present research has shown that family resemblances, a form of complexive grouping, appears to be one of the structural principles in the composition of the superordinate classes themselves, and, thus, one of the structural principles in adult classification.
Since adult taxonomic classes such as furniture or chair themselves consist of complexive groupings of attributes, it would appear appropriate to study the development of the integration of complexive into taxonomic categories rather than the replacement of the former by the latter.
Family resemblance as a logical alternative to criterial attributes.
There is a tenacious tradition of thought in philosophy and psychology which assumes that items can bear a categorical relationship to each other only by means of the possession of common criterial attributes.
The present study is an empirical confirmation of Wittgenstein’s (1953) argument that formal criteria are neither a logical nor psychological necessity;
the categorical relationship in categories which do not appear to possess criterial attributes, such as those used in the present study, can be understood in terms of the principle of family resemblance.
REFERENCES
...
[605]
REFERENCE NOTES
...
(Accepted June 3, 1975)
GEORGE LAKOFF
HEDGES: A STUDY IN MEANING CRITERIA AND THE
LOGIC OF FUZZY CONCEPTS
(1973)
[458]
1. DEGREES OF TRUTH
Logicians have, by and large, engaged in the convenient fiction that sentences of natural languages (at least declarative sentences) are either true or false or, at worst, lack a truth value, or have a third value often interpreted as 'nonsense'.
And most contemporary linguists who have thought seriously about semantics, especially formal semantics, have largely shared this fiction, primarily for lack of a sensible alternative.
Yet students of language, especially psychologists and linguistic philosophers, have long been attuned to the fact that natural language concepts have vague boundaries and fuzzy edges and that, consequently, natural language sentences will very often be neither true, nor false, nor nonsensical, but rather true to a certain extent and false to a certain extent, true in certain respects and false in other respects.
It is common for logicians to give truth conditions for predicates in terms of classical set theory. 'John is tall' (or 'TALL(j)') is defined to be true just in case the individual denoted by 'John' (or 'j') is in the set of tall men.
Putting aside the problem that tallness is really a relative concept ... , suppose we fix a population relative to which we want to define tallness.
... how tall do you have to be to be tall? ... How old do you have to be to be middle-aged? ... the concept is fuzzy.
Clearly any attempt to limit truth conditions for natural language sentences to true, false and 'nonsense' will distort the natural language concepts by portraying them as having sharply defined rather than fuzzily defined boundaries.
... Eleanor Rosch Heider took up the question of whether people perceive category membership as a clearcut issue or a matter of degree. ...
[459]
... She asked subjects to rank birds as to the degree of their birdiness, that is, the degree to which they matched the ideal of a bird. ... a fairly well-defined hierarchy of 'birdiness' emerged.
...
Further experiments by Heider showed a distinction between central members of a category and peripheral members. ...
[460]
...
I think Heider's work shows clearly that category membership is not simply a yes-or-no matter, but rather a matter of degree.
Different people may have different category rankings depending on their experience or their knowledge or their beliefs, but the fact of hierarchical ranking seems to me to be indisputable.
...
Suppose now that instead of asking about category membership we ask instead about the truth of sentences that assert category membership.
If an X is a member of a category Y only to a certain degree, then the sentence 'An X is a Y' should be true only to that degree, rather than being clearly true or false.
My feeling is that this is correct, as (4) indicates.
(4) Degree of truth (corresponding to degree of category membership)
| a. A robin is a bird. | (true) |
| b. A chicken is a bird. | (less true than a) |
| c. A penguin is a bird. | (less true than b) |
| d. A bat is a bird. | (false, or at least very far from true) |
| e. A cow is a bird. | (absolutely false) |
Most speakers I have checked with bear out this judgement, though some seem to collapse the cases in (4a-c), and don't distinguish among them.
My guess is that they in general judge the truth of sentences like those in (4) according to the truth of corresponding sentences like those in (5).
|
(5)
a. A robin is more of a bird
than anything else. |
(True) |
|
b. A chicken is more of a bird
than anything else. |
(True) |
[461]
|
c. A penguin is more of a bird
than anything else. |
(True) |
|
d. A bat is more of a bird
than anything else. |
(False) |
|
e. A cow is more of a bird
than anything else. |
(False) |
That is, some speakers seem to turn relative judgments of category membership into absolute judgments by assigning the member in question to the category in which it has the highest degree of membership. ...
2. FUZZY LOGIC
Although the phenomena discussed above are beyond the bounds of classical set theory and the logics based on it, there is a well-developed set theory capable of dealing with degrees of set membership, namely, fuzzy set theory as developed by Zadeh.
The central idea is basically simple:
Instead of just being in the set or not, an individual is in the set to a certain degree, say some real number between zero and one.
...
[462]
... As a subjective approximation we might say
[Figs. 1 & 2]
that if someone is smaller than 5'3", then he is not tall to any degree.
If he is 5'7", we might say that he is tall to, say, degree 0.3.
If he is 5'11", we might say that he is tall to, say, degree 0.8.
And if he is over 6'3", then he is tall, period.
...
Undoubtedly the function which maps height into tallness is itself fuzzy.
However, I do think that the curve in Figure 2 is not a bad approximation to my own intuitions ...
We should also ask how seriously we should take the fact that the function for tallness given in Figure 2 is continuous, assigning an infinite
[463]
number of values, in fact filling the uncountable infinity of values in the real interval between zero and one.
After all, human beings cannot perceive that many distinctions.
Perhaps it would be psychologically more real not to have an infinity of degrees of set membership, but rather some relatively small number of degrees, say the usual 7±2.
From Google AI Overview:
"Miller's Law, from cognitive psychology, states the average person can hold about 7 (plus or minus 2) "chunks" of information in their working memory at once, meaning overloading people with too much data leads to confusion . . .
"Chunking:
The key isn't just the number but what you're remembering; bigger, meaningful chunks (like words) are easier to handle than individual bits (like letters)."
Named after
George A. Miller (1956).
On the other hand, one might consider the interesting possibility that the finiteness of human perceptual distinctions is what might be called a surface phenomenon.
It might be the case that the perception of degrees of tallness is based on an underlying continuous assignment of values like that given by the curve in Figure 2.
The finite number of perceived distinctions would then result from 'low level' perceptual factors, though perhaps the number of perceived distinctions and their distribution would depend on the shape of the underlying curve ... and various contextual
[Fig. 3]
factors.
I think that the latter proposal has a high degree of plausibility, and I think that some of the facts discussed below will make it even more plausible. For this reason, I will stick to continuous assignments of values.
...
[464]
...
... let P='John is tall' and Q='Bill is rich'.
Let P=0.7 and Q=0.4,
that is, suppose John is tall to degree 0.7 and Bill is rich to degree 0.4.
'John is not tall' will be true to degree 0.3, while 'Bill is not rich' will be true to degree 0.6.
'John is tall and Bill is rich' will be true to degree 0.4, which is the minimum of 0.7 and 0.4.
'Either John is tall or Bill is rich' will be true to degree 0.7, the maximum of 0.7 and 0.4.
[Figs. 4 & 5]
[465]
For an example of how semantic entailment works, let P='John is very tall' and Q='John is tall'.
Clearly P semantically entails Q.
Consider an assignment of values to 'very tall', as in Figure 4 and Figure 5.
If you compare the tables in Figures 1 and 4, you will find that in each case, given a height, the value for 'VERY TALL' is LESS THAN OR EQUAL TO the value of 'TALL'.
For example, at the height of 5'11", the value for 'VERY TALL' is 0.3, while the value for 'TALL' is 0.8.
...
Thus, no matter what John's height is, the value of 'John is very tall' will be less than or equal to the value of 'John is tall', and by the above definition, the former will semantically entail the latter.
...
[471]
...
3. HEDGES
Let us begin with a hedge that looks superficially to be simple: sort of.
Just as very is an intenisfier in that it shifts values to the right and steepens the curve ... , so sort of is, in part at least, a deintensifier in that it shifts the curve to the left and makes it less steep.
However it also drops sharply to zero on the right.
Consider the notion 'sort of tall' ...
The values for 'sort of tall' are greatest when you are of intermediate height.
If you are of less than intermediate height, then the values for 'sort of tall' are greater than those for 'tall'.
But above intermediate height the values for 'sort of tall' drop off sharply.
If you're really tall, you're not sort of tall.
The same thing is true in the case of birdiness.
| (1) | a. A robin is sort of a bird. | (False - it is a bird, no question, about it) |
| b. A chicken is sort of a bird. | (True, or very close to true) | |
| c. A penguin is sort of a bird. | (True, or close to true) | |
| d. A bat is sort of a bird. | (Still pretty close to false) | |
| e. A cow is sort of a bird. | (False) |
Sort of is a predicate modifier, but one of a type that has not been previously studied in formal semantics in that its effect can only be described in terms of membership functions for fuzzy sets.
It takes values that are true or close to true and makes them false while uniformly raising values in the low to mid range of the scale, leaving the very low range of the scale constant.
The effect of sort of cannot even be described in a two-valued system, where sentences are either true or false and individuals are either set members or not.
Consider again example (4) of section I, where
[472]
we saw that there were speakers who did not distinguish between the (a), (b), and (c), sentences, but rather lumped them together as all being true, as in the corresponding sentences of example (5) of section I.
However, even such speakers distinguish the (a) sentence in example (1) of this section from the (b) and (c) sentences.
In order for them to do this, they must have been able to make an underlying distinction in degree of birdiness between robins on the one hand and chickens and penguins on the other.
The effect of the predicate modifier sort of depends upon just such a dis-
[473]
tinction.
There are other types of predicate modifiers that reveal such distinctions.
| (2) | a. A robin is a bird par excellence. | (true) |
| b. A chicken is a bird par excellence. | (false) | |
| c. A penguin is a bird par excellence. | (false) | |
| (3) | a. A chicken is a typical bird. | (false) |
| b. In essence, a chicken is a bird. | (true) | |
| (4) | a. In a manner of speaking, a bat is a bird. | (true or close to true) |
| b. In a manner of speaking, a cow is a bird. | (false) | |
| c. In a manner of speaking, a chicken is a bird. | (nonsense - (c) presupposes that chickens are not really birds, which is false). |
As (2) reveals, par excellence requires the highest degree of category membership.
Robins fit, chickens and penguins don't.
Typical, in (3), also requires a high degree of membership, which is why chickens don't fit.
But a high degree of membership isn't sufficient for typical, as (5) shows.
| (5) | a. A robin is a typical bird. | (true) |
| b. An eagle is a typical bird. | (false, or at least far from true) |
Even though eagles seem to rank high in birdiness, the fact that they are predators makes them atypical of birds.
What examples (2) - (5) seem to show is that people do make the full range of distinctions in the birdiness hierarchy.
Though these distinctions may be subtle, they can be thrown into clear relief by hedges.
But hedges do not merely reveal distinctions of degree of category membership.
They can also reveal a great deal more about meaning.
Consider (6).
| (6) | a. Esther Williams is a fish. |
| b. Esther Williams is a regular fish. |
(6a) is false, since Esther Williams is a human being, not a fish.
(6b), on the other hand, would seem to be true, since it says that Esther Williams
[474]
swims well and is at home in water.
Note that (6b) does not assert that Esther Williams has gills, scales, fins, a tail, etc.
In fact, (6b) presupposes that Esther Williams is not literally a fish and asserts that she has certain other characteristic properties of a fish.
Bolinger (1972) has suggested that regular picks out certain 'metaphorical" properties.
We can see what this means in an example like (7).
| (7) | a. John is bachelor. |
| b. John is a regular bachelor. |
(7b) would not be said of a bachelor.
It might be said of a married man who acts like a bachelor — dates a lot, feels unbound by marital responsibilities, etc.
In short, regular seems to assert the connotations of "bachelor', while presupposing the negation of the literal meaning.
(7) reveals the same fact, though perhaps more clearly.
| (8) | a. Sarah is a spinster. |
| b. Sarah is a regular spinster. |
(8b) asserts that Sarah has certain characteristic properties of spinsters — presumably that she is prissy and disdains sexual activity.
(8b) would not be said of someone who was literally a spinster, but might be said either of a married woman or a girl who was not yet past marriageable age who acted like a spinster.
What (8b) asserts is the connotation of 'spinster' — prissiness and lack of sexual activity, while presupposing the negation of the literal meaning.
If this account of the meaning of regular is essentially correct, a rather important conclusion follows.
It is usually assumed that the connotations of words are part of pragmatics — the wastebasket of the study of meaning.
Certainly most philosophers seem to take it for granted that connotations and other pragmatic aspects of meaning are irrelevant to the assignment of truth values (leaving aside sentences containing indexical expressions).
Truth is usually taken to involve literal or denotative meaning alone.
Yet in sentences with regular, such as (6b), (7b) and (8b), the truth value of the sentences as a whole depends not upon the literal meaning of the predicates involved, but strictly upon their connotations!
What this indicates, I think, is that semantics cannot be taken to be independent of pragmatics, but that the two are inextricably tied together.
In the above discussion I used the terms 'literal meaning' and 'connota-
[475]
tion' as though they were adequate to describe at least informally the types of meaning components affected by hedges and related words.
But as might be expected the situation is more complex.
We can see this if we try to find some hedges that are opposites of regular, ones which pick out literal meaning alone.
Two promising candidates are strictly speaking and technically.
| (9) | a. A whale is technically a mammal. |
| b. Strictly speaking a whale is a mammal. |
Technically and strictly speaking seem to have the same effect in (9a) and (b).
However, in other sentences they produce radically different results.
| (10) | a. Richard Nixon is technically a Quaker. | (true) |
| b. Strictly speaking, Richard Nixon is a Quaker. | (false) | |
| (11) | a. Ronald Reagan is technically a cattle rancher. | (true) |
| b. Strictly speaking, Ronald Reagan is a cattle rancher. | (false) | |
| (12) | a. Strictly speaking, George Wallace is a racist. | |
| b. Technically, George Wallace is a racist. |
As (10) and (11) show, technically picks out some definitional criterion, while strictly speaking requires both the definitional criterion and other important criteria as well.
Richard Nixon may be a Quaker in some definitional sense, but he does not have the religious and ethical views characteristic of Quakers.
...
Ronald Reagan meets the definitional criterion for being a cattle rancher since he seems to have bought cattle stocks as a tax dodge ...
However, he does not meet all of the primary criteria for being a cattle rancher.
Note that, as (12) shows, technically seems to mean only technically, that is
it asserts that the definitional criteria are met but that some important criterion for category membership is not met.
Hence the strangeness of (12b).
Strictly speaking contrasts interestingly with loosely speaking.
| (13) | a. Strictly speaking, a whale is a mammal. |
| b. Loosely speaking, a whale is a fish. |
(13) shows the need for distinguishing between important or primary properties on the one hand and secondary properties on the other hand.
[476]
(13a) says that whales classify as mammals if we take into account important criteria for distinguishing mammals from fish.
For example, they give live birth and breathe air.
(13b) seems to say that we can classify whales as fish if we ignore the primary properties and take into account certain secondary properties, for example, their general appearance and the fact that they live in water.
Thus, we need to distinguish between primary and secondary criteria for category membership.
However, loosely speaking still differs sharply from regular, as the following examples show:
| (14) | a. Harry is a regular fish. |
| b. Loosely speaking, Harry is a fish. | |
| (15) | a. Loosely speaking, a whale is a fish. |
| b. A whale is a regular fish. |
What is strange about (14b) is that it asserts that Harry is a member of the category fish to some degree by virtue of having some secondary property of fish.
(14a) simply says that he swims well and at is home in water, while it presupposes that he is not a member of the category fish to any degree whatever.
The distinction between (14a) and (14b) indicates that we must distinguish between those properties capable of conferring some degree of category membership and those properties which happen to be characteristic of category members, but do not confer category membership to any degree at all.
No matter how well you swim, that won't make you a fish to any degree at all.
But if you are a living being, live in the water, are shaped like a fish, and your only limbs are flippers and a tail, it would seem that, like the whale, you are loosely speaking, that is by virtue of secondary criteria, a member of the category fish to some extent.
Perhaps Carroll's 'nonmanifest' or 'unexhibited' properties ought to give way to this
primary/secondary
distiction.
'Strictly speaking,
Fountain
is a urinal.'
'Loosely speaking,
Fountain
is an artwork.'
But a decades-tenured Duchampian curator at a major museum might put it the other way around, and it would be pointless to disagree with them. Or, a Philosopher of Art may claim actually to have
proven
rationalistically that
Fountain
is,
in the
strictest possible
sense,
in fact
both
urinal
and
artwork.
Here indeed,
semantics cannot be taken to be independent of pragmatics.
From here, this problem could be built out with reams of formal excursus on Philosophy of Mind, Organizational Sociology, and laboratory studies of 'depragmaticized' perception; but there seems to be a simpler answer: When it comes to categorization, I just don't understand why or how we could expect to hold anyone accountable to things not present, things neither 'manifest' nor 'exhibited'. "Pragmatics" cannot be entirely a matter of slant, of mere 'bias' in both the colloquial and the McLuhanist senses; It also encompasses the most basic distinction, we might say in Carroll's language, between the 'exhibited' and the 'unexhibited'.
The
misuse
of terms here is the Institutional Theorist's misuse: the equivocation between an
empirical
and an
ontological
sense of 'resemblance' is all theirs.
I haven't yet turned up any suggestion that the Wittgensteinian diaspora was concerned with anything but the empirical question.
Incidentally, it is no shock that the search for a definition would veer into the social/sociological: The lowest hanging fruit vis-a-vis the transfiguration of empirical into ontological claims is that of simple
aggregation. This
is
quantitative sociology. But then,
is it even Philosophy?
Again, it seems that such a definition could succeed perfectly in drawing the boundaries of art and yet remain empirically useless, since no single observer could possibly have empirical
(i.e. "pragmatic") access to the necessary lines of evidence in any given instance. (The next paper below wrestles with just this sort of issue.)
Note that (15b) is odd in that it presupposes that the whale is not a member of the category fish to any extent.
An adequate account of the functioning of characteristic-though-incidental properties should provide an understanding of at least one type of metaphor.
Suppose I say 'John is a fish'.
I am using a metaphor to indicate either that he swims well or that he is slimy (in the nonliteral sense).
The mechanism for this is, I think, something like the following.
Since it is presupposed that the subject, John, is not literally a member of the category fish, one cannot be asserting membership in that category if the sentence is to make sense.
Instead, the sentence is understood in essen-
[477]
tially the same way as 'John is a regular fish',
that is, the contextually most important incidental-though-characteristic properties are asserted.
...
By looking at just four hedges — technically, strictly speaking, loosely speaking and regular — we have seen that we must distinguish at least four types of criteria for category membership:
| (16) | TYPES OF CRITERIA | |
| 1. Definitional | ||
| 2. Primary | — capable of conferring category membership to a certain degree depending on various factors | |
| 3. Secondary | ||
| 4. Characteristic though incidental | — not capable of conferring category membership to any degree, but contributes to degree of category membership if some degree of membership is otherwise established. |
These distinctions are necessary for even a primitive account of how such hedges function.
Such a primitive account is given in (17).
| (17) | An Informal and Inadequate Approximation to an Understanding of Some Hedges |
| TECHNICALLY | — Truth value depends upon values of definitional criteria alone. Implies that at least one primary criterion is below the threshold value for simple category membership. | |
| STRICTLY SPEAKING | — Truth value depends on value of definitional and primary criteria. Values for each criterion must be above certain threshold values. | |
| LOOSELY SPEAKING | — Truth value depends primarily on secondary criteria. Implies that threshold values for definitional and primary criteria are insuf- |
[478]
| ficient to confer category membership. | ||
| REGULAR | — Truth value depends upon characteristic-though-incidental criteria. It is presupposed that the values of other criteria are insufficient to establish any degree of category membership. |
The facts in (17) cannot be handled within the framework of fuzzy logic as developed above, since they require a distinction between types of criteria for category membership.
Nor can they, so far as I know, be handled by any logic developed to date.
Let us consider what type of logic would be needed to handle such cases.
4. FUZZY LOGIC WITH HEDGES
...
[483]
...
5. SOME INADEQUACIES OF THE TREATMENT OF HEDGES IN SECTION 4
...
[484]
5.1. Dependence upon Context
The valuations for hedges given in Section 4 were independent of context.
However, it is fairly easy to show that any adequate treatment will have to take context into account.
Consider (1).
| (1) | Technically, this TV set is a piece of furniture. |
As Eleanor Heider (personal communication) has observed,
there is no generally recognized technical definition accepted throughout American culture (or any other) that will tell you whether a particular TV set is or is not a piece of furniture.
The range of TV sets goes from small portable ones that can easily be carried (perhaps in one's pocket) to large consoles with fancy wooden cabinets.
But whether a given TV set is technically a piece of furniture will vary with the situation.
For example, insurancе companies or movers may set different rates for furniture, appliances, and other personal property.
technical =
'technique'
?
i.e.,
= 'for a purpose'
?
and then,
the situation =
'the purpose'
?
Sure enough, Carroll gives two very similar examples of purpose-driven technique-as-context for art-categorization. Seeing that this is inadequate to his argument, he gives the purportedly intrinsic example of
ascertaining how we should respond
(208).
It might as well be countered:
Only if it really matters 'how we respond', and if there is known danger of failure to respond in just such an appropriate manner.
That something is art signals how and even whether we are to respond to it interpretively, aesthetically, and appreciatively.
(208)
But
why
must we respond in such strict accordance with 'some'-'thing's art status?
Better, what if 'how we respond'
just is
the 'signal that something is or is not art?
In such situations, technical standards have to be set and it is doubtful that there will be much uniformity.
Yet, the truth or falsity or even the appropriateness of (1) in a given context will depend on what those standards are, if there are any.
Moreover, different cultures, subcultures, or even individuals may differ as to which criteria for a given predicate are primary and which are secondary.
In fact, it would not be surprising to find that which criteria were considered primary and which secondary depended on context. Consider (2).
| (2) | a. Strictly speaking, Christine Jorgenson is a woman. |
| b. Strictly speaking, Christine Jorgenson is a man. |
One can imagine contexts in which either (2a) or (2b) would be true or very close to true.
Take contexts where current sex is what matters, ...
In such situations, (2a) would be true and (2b) would not.
Take, on the other hand, situations where former sex might matter, ...
In such situations, one could imagine that (2b) might be true and (2a) not.
That is, current sex might be primary for determining manhood vs. womanhood in some contexts and former sex primary in others.
[485]
5.2. Modifiers that Affect the Number of Criteria Considered
Under Zadeh's proposals for the definition of words like VERY and SORT OF, such modifiers affect only the absolute values of the predicates modified.
However, consider cases like VERY SIMILAR and SORT OF SIMILAR.
Things are similar or dissimilar not just to degrees, but also in various respects.
In judging similarity one picks out a certain number of contextually important criteria, and determines degree of similarity on the basis of how closely the values match for the criteria chosen.
In determining the values for VERY SIMILAR, there are two possibilities.
First, one can, for the fixed number of criteria considered in judging mere similarity, require that the values assigned to the various criteria be closer.
Secondly, one can require that more criteria be taken into account
For example, consider (1).
| (1) | a. Richard Nixon and Warren G. Harding are similar. |
| b. Richard Nixon and Warren G. Harding are very similar. |
In judging (1a) to be true to a certain degree, one might take into account merely their records as president.
One might then want to go on to assert (1b) by taking into account other criteria, for instance, the personal lives, moral values, etc.
SORT OF has the opposite effects when applied to SIMILAR.
| (2) | a. George Wallace and Adolf Hitler are similar. |
| b. George Wallace and Adolf Hitler are sort of similar. |
(2b) can be a hedge on (2a) in two different respects.
First, on the given criteria considered, one may require less closeness of values for (2b) than for (2a).
Secondly, in judging the degree of truth of (2b) versus (2a), one may take fewer criteria into account.
These considerations show that an adequate account of the meanings affect of VERY and SORT OF cannot be given simply in terms of how they the absolute values of the predicates they modify;
one must take into account the way they change the consideration of vector values.
In the case of similarity that includes both the closeness of selected vector values and the number of them.
5.3. Some Hedges Must Be Assigned Vector Values
In the treatment given in Section 4, all of the hedges were assigned only
[486]
absolute values. That this is inadequate can be seen by considering an expression like VERY STRICTLY SPEAKING, as in (1).
| (1) | a. Strictly speaking, Sam is not the kind of person we want to hire. |
| b. Very strictly speaking, Sam is just the kind of person we do not want to hire. |
One can imagine a situation in which one might say (1a) and then follow it up with (1b).
Suppose one were running a business and had certain criteria for filling a certain job — objective qualifications, honesty, personality traits, etc., with some criteria being more important than others.
Given that Sam did not measure up according to the primary criteria, one might accurately say (1a), though perhaps nothing stronger.
Suppose that, one then isolated the most important ofthe primary criteria and looked at how Sam ranked with respect to those.
With respect to those, he might not merely be unqualified but might actually be injurious to the business.
One might then be in a position to make the stronger statement (1b).
One of the things that VERY does, when applied to STRICTLY SPEAKING, is further restrict the number of categories considered most important:
this can be viewed as changing the weights assigned to various criteria at the upper end of the spectrum.
This is, incidentally, the opposite of what it does when applied to SIMILAR — and I have no idea why.
Be that as it may, VERY seems to operate on the vector value of STRICTLY SPEAKING, not just on the absolute value.
This means that we must find a way of assigning vector values to hedges like STRICTLY SPEAKING.
The same is true of LOOSELY SPEAКING, as expressions like VERY LOOSELY SPEAKING show.
In this case, however, one of the effects' of VERY is to increase the number of criteria considered — or at least increase the weights assigned to the lower end of the spectrum — the opposite of what happened in the case of VERY STRICTLY SPEAKING.
Any adequate description of the meaning of VERY will have to take such considerations into account.
Another thing suggested by these facts is that there may not be a strict division between primary and secondary criteria;
rather there may be a continuum of weighted criteria, with different hedges picking out different cut-off points in different situations.
[487]
5.4. Perhaps Values Should Not Be Linearly Ordered, But only Partially Ordered
So far we have discussed the concept 'true to a certain degree';
we have paid hardly any attention to the concept 'true in a certain respect'.
Any serious study of hedges like IN SOME RESPECTS, IN A SENSE, IN A REAL SENSE, etc. requires it.
What these hedges seem to do is say there are certain criteria which, if given great weight, would make the statement true.
Consider (1).
| (1) | a. In some respects, Nixon has helped the country. |
| b. In a sense, J. Edgar Hoover was a great man. | |
| c. In a real sense, Nixon is a murderer. |
But very often, sentences without such hedges are meant to be taken in the same way.
| (2) | Nixon is a murderer and he's not a murderer. |
The usual sense of (2) is not either a statement of a contradiction nor a statement that Nixon is a murderer to a degree.
Rather it would usually be understood as saying that if you take into account certain criteria for being a murderer, Nixon qualifies, while if you give prominence to other criteria, he doesn't qualify.
On a reading such as this, sentence (2) could be true.
But one of the inadequacies of fuzzy logic as we have set it up is that we have no way of assigning values in such a way that (2) comes out to be true.
...
When a member of the New Left says:
| (3) | Nixon is a murderer. |
[488]
and the local Republican spokesman replies
| (4) | Nixon is not a murderer. |
the disagreement is not over the facts of the world.
They may agree completely on just what Nixon has and hasn't done.
The disagreement is one of values.
What criteria should be considered important in conferring membership in the category of murderers?
The issue is by no means trivial.
Similar cases arise every day in most people's speech.
Any serious account of human reasoning will require an understanding of such cases.
5.5. More Problems With VERY
We saw above ... that the meaning of VERY cannot be adequately represented simply by taking a function of the absolute value of the predicate modified;
vector values must be taken into consideration.
There are other considerations that seem to me to indicate this.
According to Zadeh's treatment of VERY, ... the curve for VERY TALL hits the values 0 and 1 at exactly the same places as the curve for TALL.
... [But] It seems to me that it can be absolutely true that someone is tall without it being absolutely true that he is very tall.
The situation, of course, gets worse with VERY VERY TALL, VERY VERY VERY TALL, etc., ...
...
[Fig. 10]
[489]
[Fig. 11]
...
... there is no way to get the effect of such a shift simply by having VERY operate on the absolute value of TALL.
Rather any function giving the values for VERY TALL in Figure 11 would have to range over heights, which would be included in the vector value for TALL.
...
My feeling at present is that a complete understanding of VERY is very far from our grasp.
5.6. Restrictions on the Occurrence of Modifiers
Some modifiers can apply to other modifiers, but the combinations are quite limited.
We get VERY STRICTLY SPEAKING, but not VERY RATHER.
Moreover, there are firm restrictions on what modifiers can
[490]
modify what predicates.
We get NEARLY EQUAL TO 5, but not VERY EQUAL TO 5, though we get VERY CLOSE TO 5.
A few of these restrictions follow automatically from certain of the above proposals.
...
But most other restrictions on the occurrence of hedges seem not to follow automatically from what has been said above.
Such restrictions should follow automatically from any adequate account.
Hedges raise some interesting questions:
| A. |
How Do Hedges Interact With Performatives?
Take a sentence like (1). |
| (1) | Тechnically, I said that Harry was a bastard. |
What (1) would generally be taken to mean is that I said it but I didn't mean it.
That is, TECHNICALLY in (1) seems to be cancelling the implicature that if you say something, you mean it.
Or suppose a sergeant says (2).
| (2) | You might want to close that window, Private Snurg. |
I think it would be appropriate to describe such a situation by the sentences in (3).
| (3) | a. Strictly speaking, the sergeant didn't order the private to close the window. | |
| b. Essentially, the sergeant did order the private to close the window. |
Obviously hedges interact with felicity conditions for utterances and with rules of conversation.
...
In addition, Robin Lakoff (personal communication) has observed that certain verbs and syntactic constructions convey hedged performatives.
| (4) | a. I suppose (guess/think) that Harry is coming. | |
| b. Won't you open the door? |
[491]
(4a) is a hedged assertion. (4b) is a 'softened' request. An investigation of these would also be revealing.
| B. | Are There Hedges in Lexical Items? |
Robin Lakoff has suggested that one might want to describe a word like 'pink' as a hedge between red and white.
This is also suggested by the metaphorical term 'pinko', which is a hedge on 'red'.
| C. | What are the Primitive Fuzzy Concepts in Natural Language? |
We will say that a fuzzy set is primitive if its membership function cannot be decomposed, ...
The question as to what such primitives are in natural language is a fundamental question about the nature of the human mind.
The question has, of course, been raised innumerable times before, but to my knowledge the possibility that the primitives themselves might be fuzzy has not been discussed.
Fundamental
indeed.
"“psychologism” for Husserl is the error of collapsing the normative or regulative discipline of logic down onto the merely descriptive discipline of psychology. It would make mental operations (such as combination) the source of their own regulation. The “should” of logic, that utter necessity inhering in logical inference, would become no more than the “is” or facticity of our customary thinking processes, empirically described."
(IEP,
Edmund Husserl (1859—1938))
There must be some danger that the advent of fuzziness, by expanding the purview of all that 'logic' is able to deal with, concurrently expands the temptation to 'psychologism'? Just a thought.
(I don't really know what 'logic'
is, which is why I've excised everything here of that kind that is the least bit technical. This becomes harder to justify below.)
| D. | What are the Possible Types of Membership Functions? |
...
6. CONCLUSIONS
6.1. The Logic of Fuzzy Concepts Can Be Studied Seriously
Fuzzy concepts have had a bad press among logicians,
Perhaps also among certain Analytic Philosophers of Art? Or has there been such an audience at all?
especially in this century when the formal analysis of axiomatic and semantic systems reached a high degree of sophistication.
It has been generally assumed that such concepts were not amenable to serious formal study.
I believe that the development of fuzzy set theory by Zadeh and the placement of it by Scott ... within the general context of recent work in modal and many-valued logics makes such serious study possible.
6.2. In Natural Language, Truth is a Matter of Degree, Not an Absolute
Heider (1971) has shown that category membership is a matter of dègree.
Sentences asserting category membership of an individual or object correspondingly display a degree of truth.
This is made clear by the study of
[492]
modifiers like SORT OF, PAR EXCELLENCE, TYPICAL, IN ESSENCE, and IN A MANNER OF SPEAKING (see Section 3 above), whose effect on truth conditions can only be made sense of if the corresponding sentences without those modifiers admit of degrees of truth.
6.3. Fuzzy Concepts Have Internal Structure
The study of hedges like TECHNİCALLY, STRICTLY SPEAKING, LOOSELY SPEAKING, and REGULAR requires the assignment of vector values to the predicates they modify.
Each component is a meaning criterion, itself a membership function for a fuzzy set.
There are at least four types of meaning criteria, three of which are capable of conferring category membership to some degree, one of which is not.
6.4. Semantics is Not Independent of Pragmatics
The study of the hedge REGULAR by Bolinger 1972 reveals that sentences with REGULAR assert connotations, not any aspect of literal meaning.
Connotations are considered to be part of pragmatics and, as such, to have nothing to do with truth conditions, since semantics has been assumed to be independent of pragmatics.
However, since the truth conditions of sentences with REGULAR depend only on connotations, it follows that if connotations are part of pragmatics, then semantics is not independent of pragmatics.
Since connotations are closely tied to the real-world situation, it seems reasonable to maintain the traditional view that connotations are part of pragmatic information.
6.5. Algebraic Functions Play a Role in the Semantics of Certain Hedges
Hedges like SORT OF, RATHER, PRETTY, and VERY change distribution curves in a regular way.
Zadeh has proposed that such changes can be described by simple combinations of a small number of algebraic functions.
Whether or not Zadeh's proposals are correct in all detail, it seems like something of the sort is necessary. ...
6.6. Perceptual Finiteness Depends on an Underlying Continuum of Values
Since people can perceive, for each category, only a finite number of gradations in any given context, one might be tempted to suggest that fuzzy logic be limited to a relatively small finite number of values.
But the study
[493]
of hedges like SORT OF, VERY, PRETTY, and RATHER, whose effect seems to be characterizable at least in part by algebraic functions, dicates that the number and distribution of perceived values is a surface matter, determined by the shape of underlying continuous functions.
For this reason, it seems best not to restrict fuzzy logic to any fixed finite number of values.
Instead, it seems preferable to attempt to account for the perceptual phenomena by trying to figure out how, in a perceptual number model, the shape of underlying continuous functions determinės the and distribution of perceived values.
6.7. The Logic of Hedges Requires Serious Semantic Analysis for All Predicates
In a fuzzy predicate logic with hedges, (FPrLH) the notion of a valuation is fundamentally more complex than the corresponding notion in other logics developed to date.
The reason is that each predicate must be assigned a vector value as well as an absolute value and the models for each FPrLH must contain functions mapping vector values into absolute values, as well as the functions prim, sec, def, and char.
Google is no help with the latter four, which is to say: I'm so clueless that I can't seem to coax out the answer I'm looking for.
What this amounts to is that the assignment of truth values in an FPrLH requires a much deeper analysis of meaning than in a classical predicate logic.
In fact, by comparison the assignment of values to predicates in a classical predicate logic is a triviality.
For each n-place predicate we set up in a model a corresponding (classical) set of n-tuples of individuals.
Thus, an expression like 'BIRD(x)' is true on an assignment of individuals to variables just in case the individual denoted by x is in the set of birds.
Nothing is said about whether it has to have wings or a beak, whether it typically flies, ... etc.
Nor is anything said in classical predicate logic about what type of criteria these are and how they contribute to degree of category membership
In a fuzzy predicate logic with hedges, all these matters must be taken into account in every valuation for the predicate BIRD.
The reason is that all of these matters enter into the assignment of truth values when BIRD is modified by one or another of the set of hedges.
Simply saying that an individual is or is not in the set of birds will tell you next to nothing about how to evaluate sentences where BIRD is modified by a hedge.
In short, fuzzy predicate logic with hedges requires serious semantic analysis for all predicates.
[494]
6.8. Claim: Hedges Show That Formal Semantics is the Right Approach to the Logic of Natural Language and That Axiomatic Theories Will Be Inadequate
Considering the cleverness of logicians in devising axiomatizations, this claim should be hedged considerably.
However, I think it will turn out to be correct.
Suppose Zadeh is right in suggesting that hedges like SORT OF, PRETTY, VERY, etc. require algebraic functions such as those discussed above to account for their meaning, at least in part.
It seems to me unlikely that one is going to be able to get complete axiomatizations for fuzzy predicate logics containing such hedges.
At least, the question should be raised as a challenge to logicians.
...
6.9. In Addition to Degrees of Truth, Degrees of Nonsense are Needed to Account for Certain Hedges
Suppose P presupposes Q and Q has some intermediate degree of truth.
Does P make sense?
Is it complete nonsense?
Or does it have an intermediate degree of nonsense?
A study of the hedge TO THE EXTENT THAT IT MAKES SENSE TO SAY THAT . . . indicates that intermediate degrees of nonsense are necessary.
...
...
[515]
Current Anthropology Volume 42, Number 4, August–October 2001
Are Ethnic Groups Biological "Species" to the Human Brain?
Essentialism in Our Cognition of Some Social Categories
Francisco J. Gil-White
[Abstract]
If ethnic actors represent ethnic groups as essentialized “natural” groups despite the fact that ethnic essences do not exist, we must understand why.
This article presents a hypothesis and evidence that humans process ethnic groups (and a few other related social categories) as if they were “species” because their surface similarities to species make them inputs to the “living-kinds” mental module that initially evolved to process species-level categories.
The main similarities responsible are (1) category-based endogamy and (2) descent-based membership.
Evolution encouraged this because processing ethnic groups as species—at least in the ancestral environment— solved adaptive problems having to do with interactional discriminations and behavioral prediction.
Coethnics (like conspecifics) share many strongly intercorrelated “properties” that are not obvious on first inspection.
Since interaction with out-group members is costly because of coordination failure due to different norms between ethnic groups, thinking of ethnic groups as species adaptively promotes interactional discriminations towards the in-group (including endogamy).
It also promotes inductive generalizations, which allow acquisition of reliable knowledge for behavioral prediction without too much costly interaction with out-group members.
Sure, it's
reliable
'on the whole', i.e., at the very highest level of
adaptive
outcome. I would think it becomes much less reliable as we move down from the highest level of aggregation of cases, towards the single case. In capsule, that's the reason we (post)moderns find ideas like this so off-putting. But this isn't even the
really
off-putting part. The
very
worst part is that our social adaptations entrained us into 'dis-adaptation' by way of transforming
the ancestral environment
into something unrecognizable. Here is ground zero for the Hegelian seeds-of-destruction theory.
But yes, let's continue . . .
...
Constructivist studies in recent anthropology have made an intellectual and moral contribution by repeatedly demonstrating that neither supposed “races” nor ethnic groups—or “ethnies”²—are natural kinds in any biological sense.
Biologists have reached the same conclusion...
However, establishing the ontological fact may have clouded our understanding of local epistemologies.
...
Even though ethnies and races do not have essences, we still need to investigate why ordinary people often believe that they do and how this affects their behavior.
...
2. I prefer the term “ethnie” to the more common “ethnic group” because these are “groups” only in certain special cases. Generally speaking they are categories rather than groups of people, and the usage of the word “group” has served only to mislead theorists, who often appear to confuse, for example, processes such as ethnic mobilization with ethnogenesis.
[516]
...
These days “good” anthropologists do not essentialize groups, and therefore no self-proclaimed essentialists are found in anthropology journals.
But ordinary folk are not good anthropologists or sophisticated constructivist scholars.
Quite to the contrary, they are naive essentialists, and I will try to explain why.
...
What Is A "Group"?
...
In psychology, although some distinctions have been made between so-called primary (face-to-face) and other kinds of groups, the fields of stereotyping and group perception are strongly biased by the explicit or implicit assumption that the term “group” represents a scientific category...
... a large literature has focused on groups with no history or content, formed in the lab on the slimmest and least ecologically valid of categorical distinctions...
...
However, recent experiments with ethnic groups, more realistic stakes, and a procedure that makes the allocators’ costs and benefits contingent on choices made by allocators show that allocators may favor the out-group...
And the in-group bias in the original minimum groups is easily made to disappear with what appear to be minor framing variations...
In sum, a straightforward application of the results of the minimum-group literature to ethnic groups and other similar moves are valid only if “group” is a useful scientific category, which is likely to be true only if the astonishing variety of social distinctions that qualify for the label “group” are meaningfully unified in human perception and experience.
Are they?
The only theoretical effort to confront this issue head-on ... originates with Campbell
[517]
(1958).
Campbell, D. T. 1958. Common fate, similarity, and other indices of the status of aggregates of persons as social entities. Behavioral Science 3:14–25
Campbell argued that a collection of individuals is perceived as a group to the degree that this collection has the characteristics of an entity.
...
[arguments following from this]
...
...
The concept of entitativity needs conceptual clarification, and the empirical tests typically pit not a highly against a less entitative social category but rather a social category against an aggregate ...
...
At an intuitive level, it seems implausible that entitativity is closely related to essentialism:
Not to me. But, continue.
firms, for example, are highly entitative, but they are not essentialized—and they are characterized by much weaker stereotypes than much less entitative social categories such as ethnic groups.
Hmm. When I tell people the
firm
I've worked for for the past decade, their responses do suggest
essentialization
and
stereotyping
of that "firm."
Do I only
think
that because I am privy to the granular details?
What is it to say that these are
weaker stereotypes
than the
ethnic
kind? Is that saying anything at all? Perhaps only that ethnic stereotypes are the 'strongest' we've got. That much seems uncontroversial.
...
Finally, the use of the word “essence” in the above literature is self-consciously derived from the categorization literature and its recent focus on natural-kind categories;
however, the claim of essentialism among the above authors appears implicitly to have become a stand-in for “stereotyping” or “causal reasoning” even though these are not the same thing.
The usage does a disservice to the explicit pedigree of the term “essentialism.”
After all, one can use a predictive stereotype for a social category without seeing it as an essentialized natural kind.
What about the alternative to the gestalt approach?
Perhaps there is specialized mental machinery for processing particular social categories in distinct and relatively discontinuous ways.
This hypothesis flows naturally from the categorization literature, which has gotten much mileage from focusing on the structure of categories as a window into the way in which the brain parses the world.
...
Particularly interesting here is the distinction between “natural” and “artifact” categories, which are processed in different ways ...
Some have taken this as an inspiration to argue that certain social categories are processed more like “natural kinds” and other social categories more like “artifacts”...
My own argument is in this vein.
I will argue... that domains important to our survival and reproduction in the past have probably selected for machinery specifically dedicated to processing the domain-relevant inputs.
Each of these dedicated “mental organs” or “modules” is described in a cognitive sense (not necessarily implying physical brain modularity) as a set of processing biases and assumptions activated by the domain-relevant inputs.
...
The "Ugly Duckling" Hypothesis
...
[519]
...
If it is true that we naively and intuitively process ethnies as species, this is likely to improve our understanding of the behavior of ethnic actors in different contexts.
This is an increasingly urgent concern of anthropologists, sociologists, and political scientists, for obvious reasons.
In these literatures, the prejudice of “circumstantialists” (a.k.a. “instrumentalists”) is that ethnies are socially “constructed” as people rationally follow their associative interests.
But if ethnies recruit their members, mate, behave, and perceive each other as natural living kinds, then they are not constructed from the individual political decisions of rational actors in relatively short time-scales.
They are constructed—though not for that any less real, mind you—with ideologies of descent-based membership that constrain the constructive process.
The present argument thus supports some of the prejudices of “primordialists” ...
Natural Kinds in Context: A Brief History of Categorization Theory
...
[520]
...
... though people may make fuzzy guesses when assigning people to ethnic categories (if they have imperfect background information about them), this is fuzziness and ambiguity in identification procedures.
It does not imply ambiguity in concept representation (category structure) if the fuzziness disappears when people are given the missing information.
Claims by constructivists that ethnic categories are fuzzy (ambiguous, malleable, etc.) may result from confusing identification with representation...
We've travelled quite far from art and aesthetics here, but here is a deceptively simple point that never quite seems to get its due in those arenas:
We
categorize
using
the
information
that is found,
not that which is
missing.
I don't know what more to say about this beyond what I have already written in this thread. Here is yet another flail.
Our practices
cannot reflect any "information" that the practitioners do not possess; If we can explain those practices only in that manner, then we do well to reopen our search for candidate explanations (or, if circumstances permit, to stop trying so hard to explain).
Any
analysis
of practical
concepts
is subject to a pernicious contradiction: The more thorough the analysis, the noisier it becomes.
i.e.
The destiny of conceptual analysis seems to be: to describe a 'concept' which no practitioner possesses, because indeed
no practitioner can
possess such an aggregation as good 'analysis' is bound to turn up.
I wonder if
any
purported conceptual
ambiguity
might not be
entirely a product of inquiry-into-practice?
After all, we may ask: Whose
representation
of any
concept
at all is truly "ambiguous" at any given point in time?
This seems, at minimum, to be a misleading description; and maybe it is actual nonsense. I can't tell just how domain-dependent FGW above takes this problem to be. Is all variation here not entirely
between
subjects rather than
within
them?
That question broached, it seems clear enough that there is no such ambiguity
in
practice, nor
among
practitioners, since as soon as we think we have located 'practice' or 'practitioners' we are, yes, presented with all the "information" we could ever desire; It then remains only to decide
(1)
what shall we do
with whatever we've already got?
and (2)
how hard should we dig
for everything that has not yet presented itself?
One of the papers Bambrough singles out for priase is John Wisdom's 'Metaphysics and Verification'. A first go suggests that this paper could shed much light on the above problem. I defer that project for now, hopefully only to the next post.
Now, . . . whither Art? We are incorrigibly
fuzzy
in our
identification procedures.
How goes our
concept representation?
FGW suggests that these two "procedures" cannot simply borrow each other's evidence. So, if we ask people to sort ("identify") the artworks from the nonartworks, what happens? Doesn't everyone have all of the same information? Aren't we interested in controlling such an experiment along precisely this axis? Then, everyone just tells us, 'just by looking', which category of object it is that they 'see'; these results are unconfounded by information asymmetry; subjects' "identifications" therefore
do
reveal differences in "concept formation", at least relative to each other. This seems bound to fail, not because it is unreasonable or outlandish to think that people
mostly
judge objects in just such an unreflective manner, but because the variable: "posession of information" is
retrospectively leaky,
no matter how well we succeed in stanching the prospective flow. "Identification" as well as "concept formation"
just is
a product of
all life's information,
and we cannot control for that. As always, there are infinite gradations among all the ways we might be
given
some information, but one overarching distinction is:
implicit-explicit.
To eliminate "fuzziness" via
explicit
information-giving is not possible in every domain; and that, if you care to read the paper unabridged, is the power of FGW's argument above. I doubt that any evidence quite that strong can be found for art-identifications.
===========
I would like to ask another question about art-categorization, one I find more interesting and actually important. It is oblique to the present frame, but we require this frame in order to ask it productively.
Is a category which includes both Michelangelo and Duchamp
necessarily
any 'fuzzier' or 'ambiguous' than one which includes only Michelangelo and his close contemporaries?
Problems of "information" abound here, but that is not my concern this time. Rather, I'm concerned with what Rosch above and FGW here call "discontinuity". e.g., Rosch writes of
information-rich bundles of attributes that form natural discontinuities.
I presume 'natural' to suggest the limits of any relativism or subjectivism; but
(1)
what are
those limits?
and (2)
what exactly is it that they are they limits
upon?
What is 'naturally discontinuous' between the
Sistine
and
Fountain
as artworks?
Surely we can
find
'discontinuities' among close contemporaries too, if we are willing to look hard enough; but are these 'natural'? If so, why should we have to look so hard?
Consider the prospect of applying FGW's coinage
ideologies of descent-based membership
to art-categorization.
Is that not precisely what
the
Danto of 1964
assumes to be a durable feature of his 'artworld'?
Ditto the phenomenon of mere
missing information
being sufficient on its own to clear up any residual
fuzziness
in "concept representation".
That is precisely the precondition to theory-boundedness that the early Danto should have the most trouble proving.
FGW has already made the distinction between
artifacts
and
natural kinds.
There is much more to come below on that distinction, and what leaps off those pages is the near identity of this account of 'natural kinds' with Carroll's account of the Institutional Theory. But if artworks themselves are artifacts, what are we to make of this?
I gather this issue has been amply broached in more recent Aesthetics, so I defer the nuts and bolts to any future exploration of that work.
For now:
It takes a whole lot of
constructivism
before 'society', e.g., begins to resemble an 'organism', or before 'art' begins to appear as an 'institutional' phenomenon.
I fear that revelations of this sort bear not in the least upon either the "identifications" or the "representations" of anyone who has spent less than half of their waking hours buried in books. I think this is, among other things, one reason why relativism and subjectivism can be variously 'refuted' but can't quite be killed: It's because each 'constructivist' revelation made is an
un-natural' standpoint created. Of course there are infinite ways for standpoints to diverge, but this seems to be among the more pernicious: Constructivists
believe
in their standpoint, the more zealous the more difficult it has been to achieve it; and anyway, failing that, it confers power, perhaps directly through technique, or perhaps contingently through valuation, prestige and any resulting 'soft power'. i.e. One way to create asymmetry of information is to
grow your own.
It's tremendously inefficient but supremely effective. So, I don't see how we can just rule out 'subjectivism'
tout court
if we're entirely surrounded by 'natural' demonstrations of it. In Carroll's case, the issue seems to be that he would not consider everything I have described here to constitute 'subjectivism', on which point I fear that we're just bickering over terminology, and I (you can forgive me, right?) simply can't keep up with capital-P Philosophy's semantic treadmill.
Now that we have
too much information,
what shall become of it? If 'categorization' is anything like every other perceptual-cognitive nexus, what will happen is that we will let a tiny amount of this information in while filtering out all the rest. So, present the
Sistine
alongside
Guernica
and
Yellow Painting;
or, present three Haydn symphonies. Present all of the contextual and genetic "information" that has yet been gathered; or print single-sheet programs and present only what can fit
au verso. The presentational possibilities are vast and amenable to fancy; not so the cognitive-perceptual possibilities. On what basis, then, can we simply assume that the boundaries of the category:
art
have become wider, or fuzzier, or more permeable? Already by the advent of Kristeller's 'modern system',
what more could we possibly cram in there without getting rid of something else?
It seems more that the Constructivists among us can, with great benefit of hindsight and leisure (and inclination!), more readily unspool a much longer litany of everything the program annotator's allotted space would not permit. But it is the notion of 'allotted space' and
not
the notion of free annotation which is the better metaphor for the 'perceptual-cognitive nexus'.
Now, say there is ample white space remaining
au verso.
Must we fill it up? Does info rush in to fill such voids as inevitably as it is filtered out in the event of a surplus? I would think not, or at least not with the same consistency and inevitability; and if not, this sinks my three-symphony argument decisively. Perhaps we
can
enumerate endless complexities in and distinctions among, e.g., three Haydn symphonies, but that doesn't mean we
will.
Dare I give the more pointed version: Surely this
has
been done by someone; just as surely, it simply
is not
done by almost anyone. I think the who's-who contingencies of this are far less interesting than the mere fact of just how far our Great Enumerators have been able to take this; doubly so just how far they've been able to get this way. No such sociological or psychological explanation of given cases in terms of conditioning, experiences or genetics is quite up to dealing with the sheer
material carnage
that such a proliferation of irreconcilable standpoints begets. Foremost, if too-obviously: The principals cannot possibly have understood this really to be a consequence of their project. Explicitly, at least, many settled on the conceit that they were in fact doing exactly the opposite. But the tell, as always, is in the actual deployment of the mission, which necessarily involves the extreme reduction of all this expert Haydn scholarship, e.g., down to whatever would fit in whatever white space remained
after
the minimum procedural info had been given. Peculiar, eh?
...
from classical categories to fuzzy ones
From the time of Aristotle until recently, all categories were thought to have necessary and sufficient conditions ...
An “object” is always either “A” or “not A.”
...
Ludwig Wittgenstein recently developed the concepts of “family resemblances,” “centrality,” and “gradience.”
The members of some categories (e.g., game ), he noted, did not all have certain properties in common, and for them no set of necessary and sufficient conditions could establish category membership.
...
Beyond membership, some members may be more typical of the category than others.
...
Finally, some categories are characterized by “gradual” membership.
For example, crisply separating poor from rich is impossible,...
To age is, as it were, gradually to lose membership in the category young and gain it in old .⁵ ...
Eleanor Rosch’s empirical and theoretical work, with that of others, revolutionized the way we think of categories and led to the current understanding that not all categories have the structure “A or not A.”
She introduced the term “prototype” to stand for the “cognitive reference point” or most central member (or subcategory of members) of a category and demonstrated the corollary “prototype effects”:
some members of a category are learned, recognized, etc., more easily, which is consistent with their being more representative or typical of the category.
At first category structure was inferred directly from these prototype effects, but Rosch
5. A similar criterion should operate for results that are “statistically significant,” by the way, but the classical view has strangely prevailed over what should obviously be a fuzzy category.
[521]
(1978) eventually recognized that demonstrating prototype effects did not, in fact, reveal category structure,
as different concept representations may be consistent with the same prototype effect.
For example, if we establish that some tokens of a category are more typical because people tell us they are better examples of the category or because these tokens are identified more quickly, does this show that the category in question has a probabilistic structure such as young people?
That is, will some category members have higher percentual membership (e.g., be younger) than others?
Not necessarily. Armstrong, Gleitman, and Gleitman (1983) ... found that subjects indeed felt that some numbers, such as 3 or 7, were more representative of odd number than others, such as 109 and 2,003.
However, “if subjects are asked directly whether typical odd numbers are or can be more odd than atypical ones, they will flatly deny it” (Keil 1989:30, my emphasis).
This result shows that even classical categories with necessary and sufficient conditions, such as odd number , may show typicality effects.
Rosch cogently argued that the demonstration of prototype effects constrained the range of possible category structures to those which could account for them but in the same breath maintained that necessary and sufficient conditions could not.
As we see above with the case of odd number, however, it seems that they can, and so classical categories may not be as rare as was initially believed ...
i.e. If you cannot be more or less
x,
then you either are
x
or you are not.
Typicality
of
x
is oblique to membership in
x.
Typicality effects are not incompatible with newer understandings of classical categories, for they may result from processing during identification of members rather than concept representation.
This is tantalizing and deeply
Wittgensteinian:
identification and representation are different pathways, at least somewhat.
Perhaps each act of identification is
sui generis
and 'mechanical' whereas representation is always ongoing and thus has the chance to become both highly refined and highly inertial.
The optimist: Look what jewels representation can make from the unrefined ore of identification.
The pessimist: Look what a hash is made of experience by the grinding gears of identification. It takes a lifetime to make a Realist, and by then he is only good for dying.
Thus, though some token members may be considered more typical and therefore be processed better and recognized more quickly, it is still possible that for some categories membership is a yes/no distinction.
Natural kinds, despite demonstrable prototype effects, appear to be classical categories with necessary and sufficient conditions of membership.
natural kinds as classical categories
Locke’s (1964[1690]) distinction between nominal and real essences has led to one between nominal and natural kinds (Keil 1989:36–37).
Nominal kinds reflect more or less arbitrary conventions of usage or else have definitions based on their purposes for human use (e.g., artifacts such as pencil ),
whereas natural kinds (e.g., fish) at least carry the intuition that they carve out important causal domains and processes in the natural world.
When natural kinds are revised—
or rather, when we revise our ideas about which objects out there in the world should be excluded from, or included in, a natural kind
—we feel this less as a definitional shift than as a gain in understanding of the category itself , whose definition we discover rather than arbitrate⁶.
In natural kinds typical appearances defer to our intuitions of “deep,” nonobvious explanatory properties causally responsible for category inclusion.
Consider that humans seem ready to accept, if the information is conveyed by an expert source (e.g., a biology teacher), that a dolphin is not a fish, even though a cursory inspection suggests that it is.
Similarly, we happily leave fool’s gold to fools, even though we ourselves need an expert to decide which is which.
Artifact categories, however, are a different matter;
try to imagine someone accepting the statement by a librarian that what looks like a book is really a magazine.
" . . . or that a snow shovel is an artwork . . . "
Malt (1989) has systematically investigated these intuitions and found that for borderline cases on the basis of appearance, people feel that they need experts to sort natural kinds, but with artifacts such cases are felt to be a matter of opinion
(e.g., the statement “According to experts, this is a shirt” is silly, but “According to experts, this is a fish” is not).
Criteria of identification (i.e., appearances) supply—for most instances (and in particular those typical of the category)—a rough-and-ready way of guessing that an object is inside the category (if it looks like a duck, we’ll guess “duck”).
The guesses
(1) are probabilistic,
(2) are graded by the number of category-relevant characteristics that are readily apparent in a given object,
(3) imply judgments of typicality, and
(4) will generate prototype effects.
However, just as in odd number, these effects do not imply fuzzy or ambiguous definitions, for appearances do not determine criteria of inclusion in natural kinds (Gelman and Markman 1987:1532):
Natural kinds are categories of objects and substances that are found in nature (e.g., tiger, water, cactus). . . . natural kind terms capture regularities in nature that go beyond intuitive similarity . . . . Natural kinds have a deep, nonobvious basis; perceptual features, though useful for identifying members of a category, do not always serve to define the category. ... Because natural kinds capture theory-based properties rather than superficial features, some of the properties that were originally used to pick out category members can be violated, but we will still agree the object is a member of the kind if there is reason to believe that “deeper,” more explanatory properties still hold.
Because in natural kinds we seem to privilege something deeper than mere appearances, some have proposed that we intuitively conceive of inner “essences”:
“Almost everyone has had the intuition that things are not always what they seem and that there is something deeper and more basic to a kind than what is immediately apparent.
One way to capture this intuition is to
6. Even if by “arbitrate” we mean the emergent “Saussurian” consensus which is the result not of any act of legislation but of more or less mutually reinforcing patterns of practice.
[522]
argue that things have essences that are often difficult to discern immediately” (Keil 1989:36).
Thus, inclusion in a natural kind follows from a necessary and sufficient condition—possession of the essence (or, by extension, from evidence of meeting the conditions for having the essence).
Surface characteristics are responsible for typicality effects but not category inclusion, for we treat the former as the typical manifestations or consequences of the essence rather than direct evidence of essence possession.
What this implies is that the categories are not fuzzy even if our appearance-based guesses may sometimes be.
For species categories, Keil (1989) has demonstrated that young children faced with gradual pictorial transformations of one species into another (e.g., lion into tiger) are very reluctant to designate an animal as simultaneously belonging to both.
They will consider the animal to have been transformed, but this happens abruptly, for the children “saw the animals as fully changing in kind across the critical pair of adjacent pictures”.
This suggests that even fuzzy transformations cannot overwhelm a strong intuition that the categories are not fuzzy.
All of the preceding arguments apply to both natural-kind substances such as gold and natural living kinds such as mouse.
What is specific to the cognition of “species” categories is our intuitions about (1) how the category essence is acquired/transmitted, which has implications for (2) how it can be established that something has one essence rather than another.
In species categories, the possession of an essence of X, I argue, promotes a strongly held intuition that one mates with X and produces Xs in reproduction.
Thus, whether a token has an essence of X answers to the questions (1) Does it produce Xs in reproduction? (2) Is it descended from Xs? and (3) Does it mate with X?
Identification Versus Categorization in Mongolia
My study population consists in the main of Torguud seminomadic pastoralists (a small Mongol ethnic group in western Mongolia).
...
Apart from Torguuds, there are other Mongol ethnic groups in the area, as well as a large population of Kazakhs (perhaps 30% of the local population), with the greatest local ethnic contrast being that between Mongols and Kazakhs.
...
This site was chosen because it offers desirable controls for a number of variables.
...
The questionnaire I constructed for the study was as follows:
1. If the father is Kazakh and the mother Mongol, what is the ethnicity of the child?
2. The father is Kazakh, the mother Mongol, but everybody around the family is Mongol and the child has never even seen a Kazakh outside of the father.
... ?
3. A Kazakh couple has a child that they don’t want. They give it in adoption to a Mongol couple when the child is under a year old.
... ?
...
[523]
...
The responses to question 1 show that Mongols (like other pastoralists) are patrilineal.
However, this usually refers to clan and subclan ascription and material inheritance;
here we see that fathers also transmit ethnic ascription.
The question was open in that they were not forced to choose among predetermined options, but in another sense it was forced by presuming that children have an ethnic status at birth.
However, if respondents considered membership a matter of one’s absorbed culture, perhaps they should have objected that “it depends” and explained on what.
No such answer was ever given.
...
The answers to questions 2 and 3 show that my respondents categorize the child according to descent.
In question 3 this implies that one can be an X and not know it ...
The dark side of social construction:
We'll TELL you what you are, tyvm.
Why 'the dark side'?
Dilman
has said:
"When I judge this to be similar to that in such-and-such a respect, my judgement is objective if it is responsible to criteria that are independent of my thoughts."
Generally it's good (necessary, actually) that we have SOME grounding in
"objectivity", SOMETHING
"independent"
of us ourselves with which to reckon. Dilman's objectivity accommodates ideas as well as materials, and those ideas could indeed be whatever unholy consensus-category prevails amongst our cohort.
I say 'dark' and 'unholy', though, because
Hell is other people. We're all "independent of each other's thoughts", yet we can, through structure or coercion, be held
"responsible to criteria"
of another, FOR NO GOOD (objective) REASON at all. Any such chain of relational objectivity is bound to be missing a link or three. Hence relational objectivity does not 'scale up'. It's an introspective tonic and an intersubjective toxin.
"No human relationship can bear the burden of godhood, and the attempt has to take its toll in some way on both parties. The reasons are not far to seek. The thing that makes God the perfect spiritual object is precisely that he is abstract—as Hegel saw. He is not a concrete individuality, and so He does not limit our development by His own personal will and needs."
(Becker, DoD, p. 166)
Short of a God or a godhood, we can simply give our consent to abide by some shared criterion, so long as it is an
"abstract"
criterion. That's the easy part. The hard part is establishing institutions and norms which facilitate this kind of consent rather than just imposing it or neglecting it.
This is especially clear in 1998, where the randomized questions do not imply to subjects (as the 1997 sequence “1, 2, 3” does) that I expect and want enculturation-based answers.
The prevalence of descent in question 3 rules out a “rearing” patrilineal model.
Of particular importance here is that biological descent is for most a necessary and sufficient criterion of inclusion, even though members are typically identified by recourse to prominently displayed cultural traits.
I gather that the so-called Historical Definitions of Art are similarly
descent-based.
But there remains the thorny question: What kind of kind is art?
Further, what are
our “deep” intuitions
worth if this is not how
members are typically
identified?
Will Our Man address any of this in his section of historical definitions?
Are we too far from home here to be entitled to borrow anything substantive?
When the adopted child is called a “Mongol” (1998 sample) this is despite the fact that my informants agree that the child will not look or behave anything like a Mongol...
If she [the child] existed, a naive local traveler would doubtless classify her as Kazakh.
But learning that the biological parents were Mongol would lead to a revision of the guess and a recognition that surface features had misled one as to the “‘deeper’, more explanatory properties”...
...
... the ethnographic literature suggests that all over the world—no matter how culturally marked ethnic actors may be—the “rule” for making ethnic ascriptions is based on blood much more than on enculturation.
Some may object that perhaps my informants are rea-
[Fig. 1]
soning not about ethnic groups but about what they believe are “races.”
Hirschfeld (1996) has shown essentialism in the cognition of “race.”
There are some problems of comparison because of differences in the details, but one of Hirschfeld’s manipulations was an adoption situation very similar to mine, and his results amount to finding that children reason that “race” will be unmodified by early adoption into the out-group family ...
I think that my respondents do “racialize” the ethnic groups, but we must then explain why this occurs even when there are no sharp phenotypic differences ...
...
...
[524]
...
The Species “Essence” in Human Cognition
One could argue that so far there is no demonstration of Torguuds’ processing their local ethnic categories as “species.”
What the data show is that one assumes the ethnic ascription of one’s biological parents.
It could be that, for them, an ethnic group is simply a descent group ...
To show that ethnic groups are processed as natural kinds one must go beyond the above evidence to show that putative essences are attached to the labels.
...
Most of the cognitive psychological work on human reasoning about living kinds has been conducted in Western settings with Westerners.
More cross-cultural work is needed, but what little we have is suggestive.
...
This work is so far consistent with the hypothesis that many important aspects of living-kind cognition constitute human universals.
...
psychological essentialism
Relative “similarity” has no objective basis because an infinite number of things can be predicated of any object and therefore unconstrained comparisons between objects A, B, and C cannot tell us which two are more similar.
Thus, the similarity we see between any two things depends entirely on what our brains consider material and immaterial about them—
what matters is not the shared predicates but the represented shared predicates.
I would think also, re: what exactly becomes material and immaterial, that exposure effects and the 'availability heuristic' must have a part to play.
“For example, both tennis balls and shoes share the predicate ‘not having ears’, but it is unlikely that this predicate is part of our representation of either tennis balls or shoes”.
Medin and Ortony speak of “psychological essentialism” as the stance that places two “objects” in the same category if they are believed to share the same constitutive and inalienable hidden essence (even when they are superficially different).
They believe that this is what governs category inclusion in natural kinds.
The argument does not commit us to the belief that essences actually exist.
For some natural kinds, such as “substances” and “elements,” chemical or atomic constitutions may qualify as legitimate essences,
but the Darwinian revolution made it clear that biological natural kinds cannot be characterized as having real essences.
In any given population, for any given genetic locus, either (1) there is variation now or (2) there will be variation at some point in the future,
and such variation does not ipso facto force an unusual variant or mutant out of the category which names the population
(in fact, typically it does not).
However, we may nevertheless think in essentialist terms.
Psychological essentialism is a claim about people’s categorization processes, not about the world these processes organize;
it is “not the view that things have essences, but rather the view that people’s representations of things might reflect such a belief (erroneous as it may be)”.
Medin and Ortony argue further that the description of an essence may be a complete mystery to us, but this is no detriment to essentializing a category, for we will simply assume that the essence—whatever it is—is there.
They call this assumption the “essence placeholder”, and they note that it forces a way of conceptualizing the category:
whatever category members look like and do, it is because of that unseen underlying “essence” or “nature” (whatever it is).
essences and development in living kinds
...
[525]
...
Gelman and Wellman presented children with examples of animals (e.g., tiger) that were raised with other kinds of animal (e.g., horses) in the complete absence of their own kind.
... children relied more on category than on environment.
For example, they answered that a tiger raised with horses (and never having seen another tiger after the early transfer) would display tiger traits and behaviors as an adult rather than horse traits and behaviors.
Children seem aware that the nature of an animal is relatively impervious to the environment of development and that adult traits and behaviors not present in the earlier stages of ontogeny are the product of the animal’s intrinsic developmental program ...
Pet owners nota bene.
This is the central point of the The Ugly Duckling.
It is not merely that the ugly duckling is really a swan because the parents are swans
but also that being a swan it will develop into a particular kind of adult regardless of varying environments of rearing.
essences and reproduction
How is a living-kind essence acquired?
To what extent do we see the process of reproduction as uniquely responsible for transmitting a species essence?
...
Rips (1989) had a sample of adults listen to a story about a “sorp,” a flying animal ... typical of a bird.
This sorp falls into some toxic waste and is transformed:
it ... acquires transparent, membranous wings ... —everything typical of a flying insect.
The transformed sorp eventually meets a normal female sorp and mates, and this leads to the laying of eggs which produce normal sorps.
... most thought that the transformed sorp was more likely to be a bird than an insect ...
...
This shows, first of all, the independence of similarity and categorization.
If participants in this study were merely comparing the features of the transformed sorp with those of the concept “bird” ... they would have concluded that the transformed sorp was not a bird.
Second, it shows the importance of what I believe is the necessary and sufficient condition for categorization in a natural living kind:
biological descent.
...
Whereas Rips tested adults,
Keil (1989:chaps. 8, 9) conducted experiments pitting descent relations against appearances with children and found that by age 10 they had strong intuitions that membership in a species category is a matter of descent, regardless of current appearance or transformations.
Yoruba adults in Nigeria, with no exposure to Western-style schooling in biology, reasoned similarly (Jeyifous 1986).
...
essences and insides
Gelman and Wellman (1991:215) carefully distinguish between insides and essences as follows:
The insides of an item are the matter residing physically behind or under its outer layer ... Insides are concrete and ultimately observable, yet typically remain unobserved. An essence is the unique, typically hidden property of an object that makes it what it is. . . . Essences generally are never observed, and in fact may remain unknown. . . .
...
The distinction is salutary because humans probably do not equate the insides with the essence as if they were coextensive.
However, it seems very likely that we think of essences as somehow located “inside.” Whatever it is that makes a thing what it is
—its essence—
we imagine it not as lying on the surface of the thing but as somehow inside it,
even though we may deny that the essence amounts to nothing more and nothing less than the thing’s insides.
Simons and Keil (1995) have shown that young children know very little about the insides of living things ...
Moreover, children’s understanding of how the insides are related to biological function is completely unspecified, as was the same understanding for adults before the development of modern biology.
And yet, Gel-
[526]
man and Wellman (1991) found that with insides-relevant items, including species categories,
children reason that the function changes when the insides rather than the outsides are removed and that identity is more likely to change when the insides rather than the outsides are removed.
This is consistent with the idea that they believe the essence is inside and that removing the insides will take the essence with them, thus altering the nature of the thing.
...
Are Torguuds Essentialists?
...
torguuds on development
A sample of “hard” primordialists in the 1998 study ... were asked an additional question.
... ,
“This adopted child, will he become exactly like the Kazakhs, or will he be somewhat different?”
When respondents answered that the child would be somewhat different ... , I asked, “How will the child be different?”
Almost invariably, the response to this was that the child’s features would be like those of a Mongol, revealing that subjects had understood the question as it was intended, in terms of similarity to Kazakhs as a group and not to the adoptive parents in particular.
Then I asked,
“And how will the child behave? Will the child behave exactly like the Kazakhs, or will he behave somewhat differently?”
A majority of respondents (17 of 23) replied that the child would behave not quite like the Kazakhs but somewhat like a Mongol.
This suggests that, in addition to believing ethnic ascriptions to result from descent, Torguuds consider this to be more than mere labeling of one’s ascent.
Apparently, being a Mongol or being descended biologically from Mongols implies resisting enculturation into an out-group to a certain degree, ...
I also asked a small sample—all primordialists in question 3—after how many generations of intermarrying the adopted child’s descendants in the male line would become Kazakhs.
... the samples of essentialists ...and nonessentialists ... are here much too small to draw conclusions.
However, one observation appears significant:
fully half of the essentialists responded that the child would never become a Kazakh, and this was the modal response for this group.
In contrast, not one of the nonessentialists gave such an answer.
...
conscious theories and intuitive “theories”
As ... I experienced people’s beliefs outside the rigid context of the interview format, I came to believe that my questionnaires were good tools for revealing people’s conscious but not their intutitive “theories.”
I use the term “intuitive theory” to substitute for “theory” or “naive theory” as the term is used in cognitive psychology, where it refers to the organizing and constraining (but subjectively unnoticed) content that underlies human concepts and categories.
By “conscious theory” I mean an elaborated belief that organizes knowledge and that the individual is aware of having.
The responses my questions elicited, I submit, result from an interview context that forces people to use their conscious theories ...
...
That my questionnaire investigates such models and not intuitive theories is no cause for despair.
...
However, if the design of the human brain is what we are after, conscious theories can be misleading, ...
For example, it has been demonstrated that, when thinking intuitively, people who understand a law of statistics which is in conflict
[527]
with their innate biases for interpreting probabilities and distributions will use the innate bias,
although they will recognize that they have made a mistake when the problem they were asked to consider is rephrased in terms of the statistical law.
These mistakes are “errors of application.”
...
...
[529]
...
“ethnie” as a privileged category
Do other descent categories show similar essentialism?
If all do, then there is nothing particular about ethnies, as distinct from other descent categories.
Some have argued that ethnicity is just kinship writ large, ...
But if ethnicity is not primarily processed as a kinship category (merely being occasionally rendered as one for the purposes of mobilization), then there should be differences in the processing of ethnic and kinship categories.
The Torguuds of Bulgan Sum divide themselves into five khoshuun, which have most of the properties usually associated with clans.
...
To investigate whether clans are processed as are ethnic groups, I presented 14 individuals with the setup for question 3 but substituted clan names for the ethnic groups.
...
I had earlier pretested, with a different sample, to see whether most Bangyakhan thought that there were cultural differences among the clans.
Almost everybody in that sample did, even though they considered them very minor and were absolutely incapable of listing them ...
A very interesting thing happened:
question 3 suddenly became very difficult to understand.
["A [y] couple has a child that they don’t want. They give it in adoption to a [z] couple when the child is under a year old. ... "]
Every time I asked the question
“Will this child be exactly like the Beelinkhen, or will she be somewhat like the Bangyakhan?”
people began talking about individual differences.
...
It was very hard to make them see that the question had to do with being exactly like the Beelinkhen (versus somewhat like the Bangyakhan) as a group, rather than exactly like the adoptive Beelinkhen parents versus somewhat like the biological Bangyakhan parents.
This is a misunderstanding that virtually never occurred ... when the categories used were Mongol and Kazakh.
The novel difficulty is in itself telling.
...
...
[530]
...
I conclude that ethnies are not processed as large kin groups
(though they are often popularly rendered as such, and this is interesting).
Kinship seems to prime schemas that deal with individually varying differences such as one will find in the personalities of different individuals within an ethnie.
Ethnicity seems to prime essentialism about qualities general to all members of the kind.
Why Should Ethnies Be, Cognitively, Species?
I have made an empirical case that Torguuds essentialize ethnies in the manner of species.
If this is a panhuman proclivity of our psychological design, as I claim, then natural selection must have favored it.
Per a few commenters (below), this claim is controversial.
I will argue that there are good reasons to believe this.
First of all, species essentialism motivates inductive generalizations about hidden properties, so that we will assume that anything nonobvious learned about one beaver, say, will be true of all beavers.
This is adaptive because beavers in fact do share many nonobvious properties and such generalizations reduce the costs of learning.
In ethnies, I will argue, inductive generalizations yield similar benefits.
inductive inferences in species
Let us assume that shared appearances are typical of members of a species but not determinative of membership ...
Let us further assume that what is determinative of membership is the essence, whatever it is.
Finally, let us assume that essences are cognized as “hidden,” “more than meets the eye,” ...
It seems reasonable, then, that
(1) knowledge that an item is a member of natural kind A will lead to the automatic assumption that it has an A essence;
(2) learning that this item has hidden property P should lead to the assumption that P is either caused by or a part of the essence and thus to the generalization that P is true of other A’s;
Per p. 521 above, presumably beliefs about
nominal kinds
are not corrigible in this way.
What
kind
of sicko goes around
hiding
properties
in
artifacts
and playing gotcha?
and
(3) As that do not look like the target item will be thought of as having property P anyway, and non-A’s bearing a strong similarity to the target will nevertheless be thought of as lacking P ...
Studies involving 4- and 3-year-olds have confirmed these propositions and Gelman and Markman’s explicit hypothesis that natural kinds will favor categorical rather than appearance-based inductive inferences.
For example, children did not make inductive inferences from a target cat with a white stripe down its back to a skunk even though they looked almost identical, but they did generalize to another cat that looked quite different from the target.
Why do natural kinds promote inductive generalizations of nonobvious properties?
Gelman and Markman observe that natural-kind categories have rich, correlated structures and that many of the properties that correlate strongly are nonobvious at first ...
...
A Darwinian unpacking of this statement might go like this:¹³
Any animal that relies heavily on learning will benefit by reducing the costs of the learning process.
If we can reliably learn about whole suites of objects merely by examining one of them, then evolution would have failed us if it had not provided mechanisms for doing so.
Of course this argument applies to artifact categories (which are also characterized by rich, correlated structures) as much as to natural kinds.
What makes induction in natural kinds special is that inductions are easily made for nonobvious (i.e., “hidden”) properties,
and this is because members of natural kinds in fact do share many nonobvious properties.
Now consider artifacts:
If I showed you a ceramic pitcher and told you that it breaks easily (a hidden property),
you would be wrong to infer that this is true of all pitchers, since they can be made of wood, stone, coconut shell, metal, and, in modern times, unbreakable plastic
13. Quine (1977:166) first observed the usefulness of natural-kind induction and its likely Darwinian explanation.
[531]
(cf. Gelman 1988). The things that members of an artifact category typically share are perceptually obvious:
their parts and their interconnections are constrained to be similar because they must fulfill the same function;
Do all
artifacts
have
a
function?
Is that
what it is
to be one?
the hidden properties (such as the properties of the materials they are made of) can vary widely so long as the artifact itself fulfills the same function.
Clearly, art-artifacts do not all fulfill the same function.
Thus, if I show you that a ceramic pitcher breaks easily, you would generalize that to ceramic anythings (because ceramic is a “substance”), but you will not think of this as a “pitcher property” ...
Recently, Atran et al. (1997) tested to see whether people made inductive generalizations for “hidden” or nonobvious properties ... equally at all levels of the biological taxonomic tree.
They found that people have a strong preference for making such generalizations at the species level
even though, as shown by Rosch et al. (1976), this is not the “basic level” (established on perceptual grounds) for biological taxonomies.
Although a domain-general mechanism does indeed seem responsible for the rank-similarity of the basic level in all sorts of taxonomical domains,
the biological domain has the idiosyncrasy that nonobvious properties maximally cluster not at its basic level (the “life-form” level, e.g., bird, fish) but at the species level.
We should say:
" . . .
*relevant*
nonobvious
properties
. . . "
Right?
Relative “similarity” has no objective basis because an infinite number of things can be predicated of any object
...
Thus, the similarity we see between any two things depends entirely on what our brains consider material and immaterial about them—what matters is not the shared predicates but the represented shared predicates.
(524)
Our cognition thus apparently favors inductive generalizations in species because this is where it makes most sense,
and the cognitive idiosyncrasy suggests that we indeed have a privileged biological domain of cognition.
...
inductive generalizations in ethnies
Processing an ethnie as a species ... is adaptive because members of an ethnie—like members of a species—share many important “properties” (norms) specific to the ethnie.
This claim of norm clustering may be counterintuitive for anthropologists and sociologists, post-Barth (1969), who have decided that ethnic groups and “cultures” are not coextensive,
and therefore I elaborate.
Barth’s (1969) views ...
He is often presented as debunking the idea that ethnic boundaries organize culture.
Following his critique, anthropologists are now very skeptical that ascriptive boundaries closely correspond to “culture” boundaries ...
Such interpretations of Barth are, however, dead wrong.
What Barth debunked was the idea that ethnic boundaries organize the totality of culture in a holistic way—
...
But practically in the same breath he insisted that what ethnic boundaries do enclose is “ethnic organization” (Barth 1969:12)
and that the cultural content most relevant to ethnic groups is
(1) the diacritical features that signal membership and
(2) “basic value orientations: standards of morality and excellence by which performance is judged”.
So long as we agree that standards of morality and performance are “culture,”
Barth is not really subtracting all culture from ethnicity, as if ethnic ascriptions were truly arbitrary,
but rather insisting on a kind of culture as most relevant to ethnicity: ...
...
... concepts of personhood, as such phenomena are realized through the observation of performative norms with interactional consequences, are nonobvious— ...
If Barth is right, then, fully socialized members of an ethnic out-group (and even of the in-group!) will have many strongly correlated hidden properties.
Indeed,
even of the in-group.
In brackets here, it's interesting to consider the various cross-cutting hypotheticals:
out-group
shares obvious properties with
in-group;
in-group's
shared properties are
nonobvious.
etc.
All humans share
some
obvious properties and
many
criteria of saliency, but perhaps the crucial (and Wittgensteinian) term above is
strongly correlated.
These hidden properties are extremely important, for they determine how easy or difficult it will be to coordinate with another human being.
Barth argues that this is precisely why ethnic groups form:
they demarcate “ways of being” ...
According to Barth, we must define ethnic groups in terms of the actors’ own ascriptions because these imply commitments to certain interactional norms:
“ ...
Coethnics therefore understand each other to be “playing the same game.”
But why do norms cluster in intercorrelated clumps?
Barth explains the maintenance of ethnic boundaries in terms of the punishments that accrue to those who fail to fulfill the normative expectations of their coethnics.
...
[532]
...
Although Barth says nothing about conformism resulting from coordination costs, this suffices for boundary maintenance and is probably originally responsible, without punishment, for the emergence of intercorrelated clustering of interactional norms.
Norm conformism allows humans to maximize the number of potential interactants in their local community with whom to engage in mutually beneficial cooperative and coordinated endeavors.
Ain't that the truth!
Much evidence from psychology suggests that we are indeed norm conformists.
...
... the proverb “When in Rome, do as the Romans do.” ...
In an ancestral environment composed of wise conformists, immigrants would absorb the norms of the host community, and this would prevent the blending-of-norms effect that immigrants would otherwise have on cultural variation,
ensuring in this way that at least some important interactional norms would regularly clump discontinuously across the human landscape.
Thus, it seems plausible that after the emergence of the conformist adaptation, the ancestral environment became populated with more or less well-defined, discontinuous norm clumps.
There are two reasons that thinking of such incipient ethnic groups as “species” is advantageous in these circumstances:
1. Interactions with those socialized into different interactional patterns and expectations will not as often be felicitous, and the costs incurred—in energy and time wasted—will concede the evolutionary advantage to any mutants who manage somehow to discriminate.
...
2. To the extent that out-group members are not irrelevant, adaptation to a norm-clumped world presents the dual challenge of avoidance and prediction.
...
... a plausible story of how ethnic groups came to look like species.
... in simple societies marrying off one’s children begins a long-term alliance with affines.
Parents were bound to learn through trial and error that coordination in long-term reciprocity was more difficult and costly with people outside the norm boundary.
Incipient ethnic groups thereby become normatively endogamous in part because of parental enforcement.
Next, because much of culture is acquired at a young age and is developmentally stable, expert practitioners of a group’s norms—when groups are endogamous—would tend to be those born of two member parents, creating the illusion that their cultural mastery was biologically inherited.
Finally, such incipient groups began labeling and conspicuously marking themselves with cultural “phenotypes” to improve the accuracy of the interactional discriminations.
...
Is the Ugly Duckling Hypothesis Robust?
the ugly duckling compared with other proposals
The Ugly Duckling hypothesis predicts that categories that look like a species (i.e., meet the brain’s “input criteria” of a species) will tend to be essentialized, especially when the perceptions of descent-based membership and category-based endogamy, in particular, are strong.
A corollary is that inductive generalizations of nonobvious properties—which essentialist intuitions motivate—will be more easily made in categories that look like species.
An alternative hypothesis posits a smooth entitativity continuum.
The more a social group looks like the prototypical “entity”—a human individual—the more it is essentialized, because we essentialize human individuals.
The constituents of a human—its organs or
[533]
cells—are all closely bunched together (proximity), they all die together if major organs cease to function (common fate), and they are divided into specialized, hierarchically organized functions in order jointly to contribute to the human’s survival and reproduction (structure).
The more a social category has these properties, this argument goes, the more we will essentialize it.
My work in Bulgan Sum does not support this hypothesis.
My respondents appeared to process clans and ethnies in different ways.
Despite some similarities between clans and ethnic groups, entitativity seems much higher for clans.
The Bangyakhan clan members all know and interact closely with each other, migrate together, and live in a relatively small territory ...
... these clans ... are composed of different exogamous patrilineages and are closely coextensive with the smallest administrative unit, ...
... [they] rank higher than ethnies in entitativity if one accepts the privileged variable of “structure”...
Despite all this, however, essentialist thinking seems absent at the level of the clan but quite evident for the ethnie.
It seems more plausible that essentialized categories are perceived as entities than that meeting the objective criteria of entitativity causes essentialism.
...
Many groups that we do not essentialize (e.g., firms) are much better examples of entities than is the average ethnic group.
And “races”, which we do essentialize, are vast collections of individuals with virtually nothing in common.
...
I argue that the ultimate cause for essentializing ethnies but not clans is that norms cluster at ethnic but not at clan boundaries, so selection has favored essentialism of ethnies.
But in terms of proximate causes (priming inputs), what is it that makes ethnic groups “good to essentialize”?
Perhaps in Bulgan Sum endogamy is one of the key differences.
Even though the Bangyakhan are highly endogamous as a clan, this is not a rule but a preference expressed at the individual level, and parents never interfere when children wish to marry someone in another clan or, indeed, in another Mongol “tribe.”
The rule, for most, is that their children may marry any Mongol they please, but they may not marry a Kazakh.
...
...
Rothbart and Taylor (1992:21) argue that
inalterability (the fact that membership is neither initially chosen nor changeable)
and
inductive potential (the idea that one member of the category is informative about all members)
are what suggest a social category as an essentialized natural kind to human cognition.
In species categories, I have argued, possession of the essence is necessary and sufficient to category membership, and the only way to acquire it is through biological descent.
... inalterability is, from this perspective, closely linked to reproduction and is primed by (1) descent-derived membership and (2) category-based endogamy.
But inductive potential as such seems to play no promixate role in priming (i.e., activating) the living-kinds module.
...
The Weimar Jews [e.g.] were the product of several generations of acculturation to Germans.
If inductive potential drives categorization, then it should have been difficult and counterintuitive to group the Weimar Jews as a naturalized category apart from Germans ...
This is not an isolated case.
Across the world, a medieval Chinese dynast likewise sought to eradicate an ethnic group that had become completely acculturated to and indistinguishable from the Chinese.
...
These and other examples suggest that inductive potential does not really drive categorization.
Rather,
social categories with normative endogamy and descent-based membership will be naturalized as having different, not-fully-knowable essences,
which assumption then motivates our inductive generalizations.
What this means is that inductive potential is an ultimate cause (a selection pressure shaping our brains over the eons) rather than a proximate cause (an input stimulus that activates a schema in the brain) of social-category essentialism.
Novel environments can undermine the original utility or function of a cognitive adaptation ...
This seems to occur regularly in the case of state structures with institutionally absorbed and culturally assimilated—but persistently endogamous and self-ascriptive—groups.
...
Also absent in the ancestral environment were sharp
[534]
“racial” differences between neighboring groups
(these are the product of modern migrations ... ).
✨ ✨ 👆 👆 ✨ ✨
This is a fundamental Left-Conservative point. Left-lefts prefer to not to think that our species might be so basically unfit for their project that the project eventually has to be abandoned; Right-rights prefer to think that migrants seek 'economic opportunity' voluntarily rather than that they are merely the ants who suffer most when the imperial elephants wrestle.
Norm conformism in the ancestral environment would have led to a landscape peppered with the self-ascriptive norm clusters that we now like to call ethnic groups.
The fact of their norm differences explains why we essentialize them,
but since neighbors were very similar in biological phenotype, why then are we apparently so prepared to essentialize “races”?
As I have argued, members of different norm communities have resorted to marking themselves ... in order to broadcast membership.
...
It is plausible... that natural selection acted on our psychology so that we interpret sharp differences in cultural phenotypic markers as signaling different essences.
It is also plausible that dramatic differences in skin color, hair type, etc.... are interpreted by our brains as ethnic diacritics.
In other words, we essentialize races because we mistakenly “think” they are ethnic groups.
We thus process “races” as ethnies even though not by the longest stretch of the imagination can they be characterized as representing norm or behavioral boundaries of any kind, which is the original reason for exapting the living-kinds module.
(This point also undermines the idea that inductive potential promotes the naturalizing of a social category,
for their is no inductive potential whatsoever to a racial category qua “race.”)
Darwinians should recognize this as a “big-mistake” hypothesis for racialist thinking (the resulting pun is not really intended but entirely appropriate).
...
the ugly duckling cross-culturally
...
In a previous paper (Gil-White 1999) I examined the most important cases put forth to support the claim that an ethnic group is an instrumental association with chosen membership (which would deny all of the basic points argued here).
...
... despite the (now famous) claims of their authors about people freely choosing to switch ethnies, the ethnographic data that they present show that no such transition takes place in the first generation ...
The earliest any such transition can be completed is in the second generation, and this almost always requires a tie of blood through intermarriage.
...
In the most thoroughgoing examination of such transitions, Levine (1987) documents in Nepal a great deal of instrumental accommodation by different groups of people to the caste and ethnic categories that best serve their interests in the sociopolitical environment in which they find themselves.
But even in such a fluid place, Levine documents the following:
(1) Many political manipulations and accommodations involve using labels that are only for outsiders or government officials;
at the local level the ethnic distinctions remain intact, as do the patterns of normative endogamy that support them.
(2) When genuine transitions from one ethnic category to another take place, these are typically not completed in political time-scales but usually require several decades and—most important—intermarriage.
Furthermore, changes in caste or ethnic status very often lag considerably behind the political/residential/economic accommodations.
...
...
[535]
...
One might object that this neglects the United States—perhaps a major exception ...
I readily admit that ethnic affiliation, for many U.S. citizens, is less salient than a national identity as Americans, ...
But the history of the United States is not inconsistent with the hypothesis presented here.
The relevant factors are, here,
(1) the rampant intermarriage among the European ethnic groups,
(2) the failure to keep alive ethnic norms to a significant degree ... , and
(3) the absence of a clear, uniparental descent-rule for ethnic identity in cases of intermarriage ...
These three facts, in the light of the Ugly Duckling hypothesis, will be expected to produce just such a weakening of ethnic identity and a reformulation of a new ethnicity coincident with the new endogamy boundary (i.e., “whites”).
Despite all this, the intuitions—such as they are—that American whites have about their European ethnicity nevertheless rely on “blood.”
Americans who say that they are half Irish, one-quarter Italian, and one-quarter Mexican are not explaining the degrees to which they have mastered these different cultures or choose to participate or affiliate with them but making a calculation based on the ethnicities of their grandparents.
Quite the
punning reprieve here, re:
calculation.
Intended
or not, this too is
entirely appropriate.
a better explanation for ethnic conflict?
The most prominent psychological perspective on intergroup conflict has been social identity theory, which began by finding that people will discriminate against any out-group, however defined ...
This has been interpreted as a general answer for the root of ethnic hatred and other forms of conflict.
However, even supposing that this demonstrates a general tendency to discriminate against outgroups rather than an idiosyncrasy of the experimental setup, it still leaves us with the question:
why ethnic hatred and warfare but, for example, no architect/lawyer riots?
Well, don't rule out a trad-bebop melee.
Perhaps architects and lawyers have already been selected for a sub-baseline inclination to
riot,
whereas (I presume) there are few ethnic groups worthy of the name (Quakers??) for whom this is true.
(I would guess that 'professional' musicians are similarly selected, but 'great' musicians are not.)
(Where I work, fighting was not unheard of into the 1990s, or so I'm led to believe. The change, if there has been one, correlates with an enormous increase in the granting of academic degrees with a 'music performance' specialization. Most people I work with have one or more. Probably a much bigger factor is the also-enormous expansion of formal policy, supervision and discipline, which I presume also to be the handiwork of those with much schooling and little education.)
(If I was
a fighter, there are double-digit people I work with who would be in the crosshairs. The collective musical talent and accomplishment of this group is tremendously high.)
(I should stop before I say something I'll regret.)
...
...
Let us provisionally accept that
(1) ethnies are interactional-norm groups,
(2) people essentialize ethnies, and
(3) ethnic membership is neither chosen nor alterable without subterfuge...
Now let us suppose further that people, inside their own group, are good at learning whom to trust.
...
Now turn to the out-group:
Members of the out-group are always violating my in-group’s norms.
From my perspective, their behavior is “wrong,” and therefore they are “bad” people because I will judge them with the same mechanism that judges norm breakers in the in-group.
Being tolerantly cosmopolitan rather than ethnocentric confers no advantage on me because the costs of miscoordination with out-group ethnics are very similar to the costs of being cheated by coethnics who “should know better” ...
Thus natural selection has favored my interpreting the fact that they break my norms as a moral lapse rather than a cultural difference.
This means that my baseline attitude towards outgroup ethnics, in the best of times, is one of at least mild distrust.
Since I essentialize them, I think that their “moral lapses” result from their corrupt nature, which cannot be helped.
...
This account can also explain why the contact hypothesis has failed.
This hypothesis states that contact between members of different racial, ethnic, or cultural groups fosters positive intergroup attitudes.
However, empirical tests have found that contact achieves this end only when a long laundry list of joint conditions are met:
“the contact must involve cooperative activities;
the participants must have equal status;
they must also be similar on nonstatus dimensions;
they must hold no negative views of each other at the outset;
the outcome of the interaction must be positive; and so on” ...
In other words, experimental manipulations that meet all the above conditions improve intergroup attitudes—contact doesn’t.
...
[536]
Conclusion
...
Comments
rita astuti
London School of Economics,...
Gil-White’s article raises some very important and challenging issues.
...
While entirely sympathetic in principle with this endeavour, I don’t think that the evidence presented takes us much closer to understanding why actors—unlike “constructivist” scholars—persist in essentializing the differences between human groups.
But this is what they persist in doing?
Gil-White’s hypothesis is motivated by the straightforward evolutionary assumption that “domains important to our survival and reproduction in the past have probably selected for machinery specifically dedicated to processing the domain-relevant inputs.”
...
...
... such a story only generates the hypothesis that humans are endowed with a certain “mental machinery”;
...
... what evidence does Gil-White provide ... ?
The evidence consists in the results of a questionnaire ...
Its formulation raises several methodological problems.
First, the respondents were asked to reason about the ethnic identity of a child born of an interethnic marriage.
This is puzzling, because if endogamy is what primes actors to perceive ethnic groups as animal species, intermarriage should be inconceivable.
However, if intermarriage is conceivable ... , it follows that actors do not define ethnic groups as animal species.
A strange objection. No ethnic group has a monopoly on tacit illogic, and the resulting force of contradiction is limited only by the force of circumstances.
Second, respondents were asked about the ethnic identity of a child born into one group but adopted and raised by another;
but this is problematic, since the participants were told that, rather than asked whether, the adopted baby would learn the customs and language of the adoptive group.
But how would you answer, Professor?
...
[The] results are very interesting, but they do no more than provide an ethnographic example of essentialist reasoning about ethnic groups.
... Gil-White is ... certainly right in claiming that his results are not idiosyncratic.
However, his claim that “all over the world” people base ethnic ascriptions on blood much more than enculturation is more contentious and is disproved by his own data:
the same questionnaire applied to neighbouring Kazakhs shows a majority of respondents reasoning that ethnic identity is acquired performatively through enculturation.
My own ethnographic and experimental work ... with the Vezo of Madagascar (Astuti 2001a) offers further robust evidence that ethnic and racial groups are not universally essentialized (for converging ethnographic evidence see, e.g., Fox 1987, Linnekin and Poyer 1990, Gow 1991, Bloch 1993, Carsten 1995).
Astuti (2001). Are we all “natural dualists”? A cognitive developmental approach. (The Malinowski Memorial Lecture.) Journal of the Royal Anthropological Institute 7:429–47.
Bloch (1993). Zafimaniry birth and kinship theory. Social Anthropology 1:119–32.
Carsten (1995). The politics of forgetting: Migration, kinship, and memory on the periphery of the Southeast Asian state. Journal of the Royal Anthropological Institute 1:317–35.
Fox (1987). “The house as a type of social organization on the island of Roti,” in
De la hutte au palais: Socie´te´s ‘a` maison’ en Asie du Sud-Est insulaire.
Edited by C. Macdonald.
Paris: CNRS.
Gow (1991).
Of mixed blood: Kinship and history in Peruvian Amazonia.
Oxford: Oxford University Press.
Linnekin and Power, Eds. (1990).
Cultural identity and ethnicity in the Pacific.
Honolulu: University of Hawaii Press
[537]
Even incontrovertible evidence of cross-cultural universality would in any case be insufficient to prove Gil-White’s claim that human cognition is innately designed to process ethnies like animal species.
... the existence of cross-cultural universality in adults is not conclusive evidence for the existence of an innate domain-specific cognitive module,
since universal convergence in adult knowledge could result from theory-building capacities applied to a world that, across cultures, provides consistent and convergent evidence;
As a novice here, I am at a loss to understand where
theory-building capacities
might come from if not from some
cognitive module(s).
It does seem daunting to try to say
which
"module" might be responsible for any given "theory."
in other words, universal patterns among adults could quite plausibly be “constructed.”
But how
un-sophisticatedly!
construct-ed
≠
construct-ivist!
It follows that evidence of innate cognitive modules should be based on empirical studies of infant knowledge that emerge well before infants can conceivably have acquired it from the environment and appear to constrain the way infants learn.
...
Although Gil-White does not report work specifically with children, he does mention that his younger Kazakh informants (no age specified) claimed that ethnic identity is fixed at birth, by contrast with a majority of (adult) responses that were nurture-based.
This raises the possibility that he has captured a significant moment in the construction of the adult theory out of early childhood intuitions ...
scott atran
Institute for Social Research, University of Michigan,
...
Cognitive anthropology’s demise as a major player in current anthropology owes, in part, to the superficiality of the psychological mechanisms employed to explain the cross-cultural patterns revealed ...
Analyses followed widespread views of minds as vast, amorphous neural networks with little content-specific, innate structure. ...
This afforded little insight into how minds form representations of the world or how individual minds causally link up ...
Cognitive psychology today focuses more on domain-specific mechanisms than on general-purpose “intelligence.”
These mechanisms presumably evolved over millions of years of biological and cognitive evolution to deal with specific recurrent problems in ancestral environments (“task demands”) ...
This “computational mind” consists of a variety of distinct information-processing devices.
Each has a particular “content bias” that targets some particular domain of stimuli in the world (input sets): ...
The particular inferential structure of each domain-specific processor takes exemplars of the stimulus set episodically encountered in a person’s life and spontaneously projects these relatively fragmentary samples onto richly structured categories (output classes) ...
Hirschfeld (1996) has proposed a domain-specific “folksociology” ...
Gil-White amplifies and challenges Hirschfeld’s insights with preliminary cross-cultural data.
Both researchers focus on a puzzle and problem of human existence:
people everywhere aggregate into essentialized groups, such as racial and ethnic groups, whose apparent regularities are thought to be caused by some deep underlying nature unique to the group and not transparent from observation (perhaps even unknowable).
...
The puzzle is that people willfully ignore massive evidence of biological or cognitive incoherence.
...
Mere belief in the group’s causal unity creates a “looping effect” (Hacking 1995) whereby people strive (or compel others) to conform to group norms and stereotypes. This brings causal coherence to a group where initially there was none.
Hacking (1995). “The looping effect of human kinds,” in Causal cognition. Edited by D. Sperber, D. Premack, and A. Premack. Oxford: Clarendon Press.
...
[538]
...
Gil-White attributes the cognitive bias toward essentializing human kinds to an “ethnicity module” that he characterizes as an exaptation of what I have dubbed “the living-kind module” for species detection (Atran et al. 2001).
Atran, Medin, Lynch, Vapnarsky, Ucan Ek’, and Sousa (2001). Folkbiology doesn’t come from folkpsychology: Evidence from Yukatek Maya in crosscultural perspective. Journal of Cognition and Culture 1:3–43.
Other essentialized human kinds—race, caste, class—are supposedly derivative.
Once domain-specific mechanisms for discriminating and establishing normative group behavior were in place, any number of perceptible groupwide features could serve as input to the ethnicity module for marking ethnic boundaries.
I find Gil-White’s evolutionary perspective compelling but doubt that humans evolved an ethnicity module that targets some real-world domain of “norms.”
I suspect that essentialization of human groups by race is cognitively privileged—
not because of an evolved “race module” or “cultural construction” but because
(1) putative racial (and perhaps linguistic) characteristics readily meet minimum perceptual triggering conditions for firing the living-kind module ... and
(2) species essentialization of human groups creates (through a looping effect) human kinds together with inferential means for making reliable predictions about them (if only as self-fulfilling prophecy).
This (2) seems to suggest that the other half of essentialization's adaptive value is that it actually creates in-groups, not just that it streamlines interaction with out-groups.
Norm markers for ethnic boundaries are culturally created and manipulated to trigger the living-kind module in certain contexts ...
Derivative norm markers may be even more critical—and more “real”—for group differentiation than the normative behaviors they are supposed to signal.
For some important markers there may be no underlying normative behavior or consensus to speak of ...
michael banton
The Court House, Llanvair Discoed,
...
...
Gil-White shows that Mongols have a folk concept of ündesten ...
"the Mongolian word used to describe groups at the level of contrast between Mongols and Kazakhs"
"Ündesten
is a good translation for “ethnic group.” Not only is the word used at the same level of contrast but it has the same polysemic ambiguities, shading into “race” at one end and “tribe” at the other.
(p. 524)
He contends that members of the Mongol group cognize ündesten in an essentialist manner and that such cognition could serve an evolutionary function.
It is difficult ... to keep such an argument separate from the assumption that ethnic alignment must be something “out there in the world of social relations,” something with an objective reality that can be grasped by an analytical concept.
At the heart of the problem is the multidimensionality of social groups.
A given group may be distinguished by its members’ outward appearance, language, religion, ethnic origin, and other characteristics, ...
Groups have to be maintained by human activity.
Some of us have held that similar processes underlie the maintenance of different kinds of group, but that is not to deny the possible importance of differences in the material from which groups are constructed.
Multidimensionality is one of the factors associated with the historical continuity of groups.
...
...
[539]
...
pascal boyer
College of Arts and Sciences, Washington University,
...
...
... the cognitive underpinnings of essentialist inferences should be made more specific.
Human minds can be reasonably described as comprising a number of functionally distinct, domain-specific inference engines, each of which focuses on particular kinds of information picked up from an environment and produces specific kinds of inferences.
Although the delivered inferences may be consciously accessible, the inferential processes are generally not.
A single inference engine can be activated by two ontologically distinct domains ... and a single object can activate different inference engines ...
Different inference engines may be activated in the interpretation of a single input: ...
It is probable that ethnic groups too trigger several parallel inference engines ...
As Gil-White suggests, a causal-essence inference engine (probably evolved to afford quick induction about living kinds) is activated whenever people represent a group as descent-based and endogamous.
...
But the causal-essence inference engine also delivers potential inferences that do not match actual features of ethnic groups.
For instance, it suggests that intercategory reproduction is impossible when it is clear that it is perfectly possible to mate across ethnic lines— ...
Conversely, many common intuitions about ethnic groups are delivered not by the causal-essence inference engine but by others.
Representing “impure” and despised castes or groups as different by virtue of birth requires some essentializing,
but then people also have the intuitions that (a) members of these groups carry some dangerous, invisible substance (“pollution”), (b) contact with them can transmit that substance, and (c) the amount or frequency of contact is irrelevant.
These inferences are typical of what could be called the contagion-contamination system, an inference engine that produces strong feelings of aversion ...
As Rozin, Haidt, and McCauley (1993) have shown, easy acquisition of such disgust reactions is vital to generalists like rats and humans.
Rozin et al (1993). “Disgust,” in Handbook of emotions. Edited by M. Lewis and J. M. Haviland. New York: Guilford Press.
More generally, pathogen avoidance is made very efficient by the three intuitions listed above.
In some contexts, some groups can be described in a way that activates this system, and people spontaneously draw such inferences about them.
But this is not (or not only) causal essentialism.
Finally, humans are extremely good at using coalitional affiliation to carry out collaborative endeavours by efficiently allocating trust among cooperators (Kurzban 1999).
Kurzban (1999). The social psychophysics of cooperation in groups. Ph.D. diss., University of California, Santa Barbara, Calif.
...
... after being attacked by one member of the group one can retaliate by attacking another member.
...
Such intuitions are not intrinsically about essence-based groups, since they work very well in the maintenance of small coalitions in any group.
Different historical circumstances may make particular kinds of inferences more efficient in the explanation of a particular interaction.
Thus people may well entertain several not necessarily congruent potential representations of the ethnic landscape at once, though Gil-White is right to suggest that a causal-essence version is usually most salient.
susan a. gelman
Department of Psychology, University of Michigan,
...
...
[540]
...
Gil-White insightfully notes that not all social categories are essentialized ...
His proposed explanation is that we have evolved in such a way that we process ethnic groups as if they were species
when two priming conditions are met: that the groups are endogamous and that they are descent-based.
I suspect that endogamy cannot be the whole story.
For one thing, endogamy cannot be a priming condition for essentialism unless it is perceived and understood.
In this regard, I suspect that children may essentialize social categories even before they acquire knowledge of endogamy.
...
A second point is that, among the social groups that Gil-White examines, endogamy is confounded with several other factors, ...
... we have a cluster of properties nearly all of which intercorrelate and any one of which could be priming essentialism.
Third, there would appear to be counterexamples to Gil-White’s prediction.
Gender is marked by neither endogamy nor descent-based membership, yet clearly it is essentialized.
Similarly, some people essentialize traits such as intelligence or shyness although they are not endogamous.
Another problematic example is caste essentialism in India.
Using a switched-at-birth task, Mahalingam (1998) finds that upper-caste adults essentialize caste differences but low-caste adults do not.
They tend instead toward social constructivist sorts of accounts, despite the presumed descent-based membership and endogamy of caste.
In these cases, we need to look for an explanation outside of normative endogamy or descent-based membership.
And if we need to look elsewhere to explain gender, traits, and caste, then it becomes plausible that we need to look elsewhere to explain essentializing of ethnic groups as well.
The evolutionary/biological component to the model, though artfully presented, seems unnecessary to account for the data.
Although all cognition is surely “in the brain,” the worry is that this approach may preclude close examination of relevant proximate factors, including language.
This is akin to Astuti's theory-building capacities argument.
...
Gil-White has numerous insightful things to say about issues of measurement, two of which are particularly crucial for studying essentialism:
(1) Identification procedures are distinct from categorization.
(2) Implicit measures may yield different results from explicit, metacognitive judgments.
This latter point raises a host of vexing questions:
How can we know when we are studying conscious versus intuitive theories?
Which sorts of theories reflect more stable, more powerfully predictive, or more representative beliefs?
How coherent or consistent are people’s belief systems,?
One of the great strengths of Gil-White’s article is that it generates a fertile set of predictions and conceptual contrasts.
Further tests of the theory are certain to yield rich insights into the nature of folk essentialism.
david l. hamilton, steven j. sherman, and jeremy d. sack
Department of Psychology, University of California, Santa Barbara,
...
(Hamilton and Sack)/Department of Psychology, Indiana University,...
(Sherman)
...
Gil-White begins by asking, “What is a group?”
Our work has also pursued an answer to that question, and, as with Gil-White, the actual concern is “What is a group as perceived by observers?”
... however ...
Gil-White’s primary concern is why ethnic groups are perceived as species.
Our research has focused on when an aggregate is perceived as a group, how, and with what consequences,
and our analysis has emphasized perceived entitativity.
Entitativity is the unity, coherence, bondedness among individuals that causes them to be perceived as a group.
Our research ...
[541]
... has investigated both
the antecedents of perceiving entitativity (the cues used to infer that an aggregate is a group)
and the consequences of perceiving entitativity (greater integrative processing of information about the group, effects on memory and judgment, perceptions of collective responsibility among group members).
We have referred to groups’ falling along an “entitativity continuum” ... Lickel et al. (2000) have obtained ratings of 40 groups documenting that their “groupness” varies widely.
In addition, Lickel et al.’s analyses have identified five distinct types of groups:
intimacy groups (family, friends, street gang),
task groups (committee, jury, co-workers),
social categories (women, blacks, Jews),
loose associations (people living in the same neighborhood, people who like classical music), and
transitory groups (people waiting at a bus stop).
These types differ not only in degree of perceived entitativity (decreasing across types as listed) but also in the properties (interaction among members, common fate, similarity, etc.) that contribute to that perception.
...
We too believe that the mental representation of group types serves important functions,
but we focus on social-motivational functions
and have suggested that the three main group types (intimacy, task, and social category) serve the needs for affiliation, achievement, and social identity, respectively.
Of particular relevance here is the fact that social categories are only one of several types of groups about which perceivers have well-developed cognitive representations.
Moreover, whereas essentialist thinking is most likely to be applied to social categories (or at least some of them; we don’t think about the “essence” of, for example, a social club or a committee),
they are lower than both intimacy and task groups in degree of perceived entitativity.
Thus, the two concepts are distinct.
...
Some researchers have proposed that stereotyping often includes endowing a social category with some essence that justifies its differential perception.
... the viability of this approach requires specification of the conditions under which the idea of an underlying essence will guide people’s perceptions of a group.
Another complexity ... :
A family is a prime example of an intimacy group as perceived by our subjects and is very high in perceived entitativity.
It would also be high in essentialism and seemingly meets Gil-White’s criteria for being perceived as a natural kind.
Yet sweeping (stereotypic?) generalizations about members of a family seem intuitively less likely than broad characterizations of social categories.
People seem more comfortable making generalizations about what blacks are like than about what members of the Filan family are like.
I wonder if expectations play a role here? Is a
family
just the kind of
group
that we expect (perhaps wish!) to be
unified?
Are violations of such an expectation (especially if it is a wish!) a sure route to salience?
By the same token, if we're 'white' and wish that
blacks
were more like us (or, uh . . . us more like them!), then doesn't this magnify the salience of seemingly small differences?
Likelihood
is one thing, desire is something else.
The inductive potential afforded by a group that is high in both entitativity and essentialism can, as in this example, be rather limited.
... Gil-White’s application [of analyses of essentialism] to the perception of ethnies as species is laudable and thought-provoking.
However, this is a narrower and more focused analysis than is required to answer the more general question of a how a group is perceived to be a group.
So, we have been un-parsimonious in accounting for the data (Gelman above), but over-parsimonious in explaining how a group is perceived to be a group.
Entitativity is a more general concept than essentialism, and recent research indicates important distinctions between them.
...
tim ingold
Department of Sociology and Anthropology, University of Aberdeen,
...
Gil-White’s argument is so manifestly absurd that I cannot believe it is meant to be taken seriously. ...
[542]
...
...
There is, indeed, abundant ethnographic evidence that in many societies persons and organisms are not identified by fixed, inner attributes.
They are identified, rather, by their positions vis-à-vis one another in unfolding fields of relationships.
...
...
david d. laitin
Department of Political Science, Stanford University,
...
Gil-White here addresses a fundamental problem ...
The phenomenon to be explained might be called “everyday primordialism”.
...
...
... Gil-White asks a hypothetical about the offspring of a Kazakh father and a Mongol mother.
It seems that every respondent accepted the terms of the question—that is to say, none refused to
[543]
accept the premise of the question by saying that it was against nature even to imagine that a Kazakh and a Mongol would have a common offspring.
Surely some would have if asked about the offspring of a giraffe father and an elephant mother.
This yet again recapitulates Astuti's comment, re: the mere conceiveability of intermarriage (p. 536). Again, I fail to see anything remarkable or improbable in the divergence of explicitly accepted terms, on the one hand, and evinced popular beliefs, on the other.
Astuti's consistent and convergent evidence argument cuts both ways here: presuming that we never encounter giraffe-elephant offspring but do sometimes encounter cross-ethnic offspring, it is then easy to explain why the particular 'bias' of a living-kind module, e.g., would permit even an unreconstructed primordialist to "accept the terms" of the first question but not those of the second.
Vis-a-vis acceptance of terms, the implicit-explicit distinction is very nearly isomorphic with the (far simpler) known-unknown distinction. I would think. i.e. We don't have to guess whether any given pastoralist defines 'species' in the the now-standard scientific way. Rather, if we think that definition is scientifically correct, then we also think that any purported outlandish exceptions suggest a problem with the observer and not with the definition.¹
Now, shouldn't an 'explicit' question such as this one be sufficient to jolt the interviewee into correspondingly explicit reasoning? I'm not sure. I would guess that we're dealing here with a counterintuitive (to some) asymmetry between the 'known' and the 'unknown'.
|
contradiction of
positive belief |
contradiction of
negative belief |
|
affirmation of
positive belief |
affirmation of
negative belief |
I think that the logical oppositions between any adjacent boxes are only logical; otherwise, each box has a slightly different 'flavor' phenomenally, pragmatically and socially.
If we do know that bi-ethnic offspring exist, then there is no shock in the mere implication that they exist (bottom left above).
We equally well think we know (though really we are just assuming) that a giraffe and an elephant cannot mate; if someone implies that they have, I can only agree that we should reject their premises (bottom right above).
The problem with arguing from a counterfactual here (giraffe-elephant) is that this relies on a purported 'symmetry' with the given fact (biethnic child); but there is only an 'asymmetry' here: the counterfactualist creates it himself, right along with his counterfactual, only so that he can later find it right where he left it.
Dare I venture that there is something profoundly Wittgensteinian afoot here in the possibility that our "accepting the terms of a question" has little to do with classification and everything to do with identification. And there is something Leibnizian afoot when we accept terms less based on beliefs than on whatever it is that we have actually observed.
Perhaps this hones in on the engines of 'asymmetry' between logical opposites.
|
uncorroborated
positive belief |
uncorroborated
negative belief |
|
corroborated
positive belief |
corroborated
negative belief |
Laitin thinks he catches FGW ascribing an 'uncorroborated negative belief' to his informants and then contradicting it with his question; hence if this is their belief they should have resisted, and if they did not resist it cannot be their belief. But nothing of the sort has happened. Neither categorization nor identification maps so directly onto belief.
|
negative
categorization |
positive
identification |
|
positive
categorization |
negative
identification |
"Many species can produce offspring with other closely related species, creating hybrids like mules (horse/donkey), ligers (lion/tiger), wolfdogs, and Savannah cats (domestic/serval), though most hybrids, like mules, are sterile; successful fertile hybrids often occur in plants and some animals (like certain cichlid fish, canids, or lynx/bobcats) where genetic differences are smaller or speciation is recent."
(Google AI Overview, 17 Dec 2025, topline)
Why not go for the lion-tiger counterfactual instead of giraffe-elephant? If someone knows that lions and tigers
are not
the same species but
are
able to produce offspring, then here we at least have "terms" which ought to be "accepted", and therefore a closer analog to Gil-White's actual method; but then,
are
lions and tigers species
'to the human brain'?
Are wolves and dogs? Are there
natural discontinuities
in our parsing of various cichlids?
Presumably there are, or at least there would/could be if we had the opportunity and the incentive; but for present purposes this matters not in the least, not unless these latter 'discontinuities' are very nearly isomorphic with the ethnic 'discontinuities' to which they are set in analogy.
Laitin wants to
assure
'discontinuity' among his 'species' examples
without bothering to look for it himself,
because he thinks that this is precisely Gil-White's mistake; but the same dismorphology which generates the counterfactual also stretches and breaks the analogy.
In this vein, we can construct infinite counterfactuals from among the available materials, running the gamut from ant-hippo to lynx-bobcat (axis: 'surface similarity'), and from horse-donkey to lynx-bobcat (axis: 'fertility of offspring'). None can fare any better than the others. Why? Because Gil-White has put quotation marks around "species", whereas the counterfactualist has not.
¹ A humorous and mundane example of 'purported outlandish exceptions' in music: In college I once watched a student composer angrily insist that he had often heard trombone slide glissandi covering over an octave. The maximum possible is just barely a tritone, and even that is only possible by starting in first position, which is to say, starting only from certain notes. Almost every 'pure' slide gliss you've ever heard has spanned a fourth or less. Is this a giraffe-elephant or a dog-wolf situation? Was our man fully insane or just a victim of the mind's silent attenuation? I would think the latter, but of course it's not out of the question that distortions on the level of giraffe-elephant could ascend to become shared beliefs without anyone ever observing them. That sort of thing happens all the time.)
... not all respondents were essentialists; essentialism existed in different degrees among them.
Gil-White’s most trusted informant turned out to be a closet primordialist ... , but even he thought that many cultural characteristics were learned, ...
Again the adaptation of species-thinking for ethnies is not quite perfect.
...
Granted some power to the species analogy ... is there another way to make sense of Gil-White’s data?
I would suggest that he underestimates the gains from trade (in goods and in wives) between ethnic groups.
The transaction costs of such long-distance trade are high, ...
... empiricist strategies ... lend themselves to generalizations that are resolutely primordialist.
While some ... will see these group characteristics as species-like, it is not necessary for all to do this or for anyone to accept the full consequences of the analogy ...
...
One way to reduce transaction costs is to engage in a form of species-analogizing.
Species analogizing, because it is useful, has a powerful grip on the human imagination,
but it is unevenly distributed and breaks down in key areas ...
It need not be transmitted across generations as a Darwinian cultural trait;
rather, it is a rational response to the problem of asymmetric information when there are gains to trade.
Cannot rational responses and living-kind modules coexist?
One test ... would be to examine whether everyday primordialism is constant over time.
My surmise ... is that the rigid rules of ethnic categorization that seem so obvious in today’s world are fairly recent inventions.
In the era when religious membership was a hegemonic social category, transactions across the religious divide allowed for different modes of addressing asymmetric information.
Is this a joke?
Certainly if you already have a
religious divide
to work with, then you have already marked yourself off from someone/something, and maybe that is all the marking-off you require.
...
...
If this proclivity for essentializing ethnic difference has been selected for, why do ethnic entrepreneurs need to expend such effort policing the boundaries ... ?
Everyday primordialism requires work.
Giraffes need no policing to restrain them from seeking to pass for or mate with elephants.
...
ma rong
Institute of Sociology and Anthropology, Peking
University,...
...
The argument sounds plausible, but there are several points that need to be clarified.
... it can be accepted that the two “most diagnostic features of ‘species’ [are] group-based endogamy and descent-based membership.”
But ... human groups are different from “natural kinds” such as animals, birds, or fish, which, in general, cannot mate and have offspring ...
Endogamy is a regulation established by human norms and not restricted by natural essences.
...
This is yet again the Astuti/Laitin argument.
My field research in Inner Mongolia and Tibet (where there is much intermarriage) indicates that recognizing “ethnies” as “species” is not precisely the
[544]
way local residents think.
...
... the classification of species in the natural world is processed not on a single plane (as different columns in a table) but as a “tree” with species, subspecies, etc., as branches ...
Classifying human groups is also complicated, with races, ethnic groups, subgroups, clans, and families, ...
Considering “species” and “ethnies” similar categories while ignoring this complexity is simplifying the issue.
Classification?
Or
processing?
Which will it be?
Perhaps the
sophisticated constructivist scholar-ordinary folk
duality is too triggering. One part of this is to clearly distinguish scientific classification from naive category formation. These are not the same.
Gil-White devotes as much space to post-Wittgensteinian preliminaries as to his own fieldwork. He does not seem to believe in
single plane
theories of
either
scientific or naive classification.
Ingold goes as far as to question Gil-White's sincerity:
"We
know that the people among whom we work are, invariably, far from naïve and that we have to struggle to match the sophistication with which they understand the world around them."
(542)
But this is just a motivated equivocation of "naïve", and of "sophistication", so as to turn one into an insult and the other into a compliment.
What the folk are naïve of, very specifically, is constructivism itself. What's wrong with that? Ingold seems to have more against it than does Gil-White. Ingold reads it as an insult, presumably because, if he had said it himself, that's what it would be. There is more justice than hyperbole in calling this 'dissociative'
... although we are still improving our knowledge of species in the natural world ... , there is usually expert consensus on categorizing, and the experts’ conclusions are accepted by society.
Nice try.
Categorizing human ethnic groups is complicated by their artificial aspects.
Now that is either a libel against the folk or evidence of a ghastly ideoligical inversion!
There are many stages in the continuum of group identity from family to humankind, and ethnic groups may be placed at different points on this continuum under different conditions.
" . . .
constructivists may, at great expense of time and effort, converge upon ethnicity's location in the broadest possible empirical schema of 'group identity', though no single person's 'schema' could ever be nearly this 'broad', not even if they are a constructivist themselves
. . . "
The rest of this paragraph is about government "recognition" of ethnic groups in China. Gil-White's treatment of
instrumental accommodation
(534)
and
the contact hypothesis
(535)
seem relevant here.
Surely
different conditions
yield different outcomes. The point seems to be that "instrumental" actions are utterly uninformative about people's 'real' inner representations and explicit beliefs. As for "contact", Gil-White basically argues that the circumstances have to be perfect; positive relations are the exception, not the rule.
Put these points together and we're left with a very old paradox: the "conditions" we seek can only be engineered, but we cannot be allowed to
know
that they have been engineered; the more we want something, the harder it is to attain it; etc.
I think the trick is to turn away from vast, messy engineering projects like 'ending racism' and focus instead on, say, fixing the sidewalk or sharing appliances.
...
Fourth, ethnic groups change much more easily and more rapidly over time than species.
Some groups have disappeared and some have emerged in the space of several generations, ...
Besides, members of a species cannot change their membership from one species to another, but members of ethnic groups can and do.
And, if natural-kind knowledge is indeed corrigible, then we can expect people to actually heed changes of ethnic affiliation.
Still, there is a vast difference between personal and
generational
time-scales, even if it is not quite as vast as that between the generational and the epochal. How much ethnic relaignment does any one of us witness?
There have been many cases in contemporary China in which people changed their “ethnic status,” ... that the Chinese government offers to ethnic minorities.
I smell instrumentality. This was very explicitly addressed in the paper.
Gil-White presents data from his field studies ...
Several points need further discussion:
(1) Religion plays a very important role in group identification.
This point has been made repeatedly in the literature on ethnicity ...
The distance between Mongols and Kazakhs ...
Religion is more important and should be used to explain the ethnic boundary.
There are many intermarriages between ethnic groups that are Muslim ... but few between Mongolian and Muslim groups in Xinjiang.
(2) No explanation is given for the reported difference in ethnic categorizing between older and younger generations.
If we consider the political/ideological environments in which they grew up, the picture becomes clearer.
The older generation grew up with the communist ideology ...
The younger generation grew up under a new nationalism ...
This is the strongest counterpoint yet, I think, at least concerning these local data.
From a broader view, however, perhaps this is a textbook example of a populist reaction against a top-down anti-essentialist
ideology.
i.e.,
This is ground zero for the
failure
of antiessentialism
even instrumentally,
not to say
cognitively,
or yes, even
spiritually.
Anti-essentialism could thrive only under
conditions
that indeed proved transient; they proved transient because they were intolerable on the whole, in spite of any tolerable successes.
For those who say that the "evolved cognitive module" is an answer in search of a question,
here is your question.
(3) “Something inside” is the term used to describe the essence of an ethnic group by the respondents ... , but what is this “inside” exactly?
The capacity for casting curses is not a general example for examining “essences,” because [some groups] believe in curses but other[s] may not.
Blood ancestry can be traced, but the “essential difference” of Weimar Jews was only the opinion of the Nazis, not a general or universal opinion, and cannot be used to support the “Ugly Duckling” hypothesis.
And we're back to throwing our cucumber slices.
I find this objection doubly bizarre. The paper is unequivocal that the typical belief in the ethnic
"inside"
is not rational, that it could be (has been) scientifically disproven. The question is: Why have
experts' conclusions
not
been
accepted
as readily here as elsewhere?
(Present company seem to agree
that there is
such a nonacceptance, though perhaps the present commenter is an exception.)
Further, nowhere is it suggested that there is anything
general or universal
about the
content
of such beliefs; rather, it is argued that their
form
is universal to the extent that it is a product of
natural selection
in
the ancestral environment.
So, of course a Southern farmer would not essentialize Weimar Jews quite like a Nazi, if he ever met one at all; but the form of Southern racism is not unrecognizable here, even though the content diverges. I thought that was the whole point?
(Lookie what happened to the American racist once he
was
able to 'meet' everyone in the entire world onscreen: a symphony of content, a metronome of form.)
Finally, there are differences in norms between ethnic groups, but they also have many norms in common.
...
The spirit of Confucianism is that all groups can be taught norms and everyone is born good.
The great success of the Chinese dynasties was the acceptance of the Confucianist norms by frontier “barbarians” who were never considered essentially “different.”
Worth considering, except for the part about
teaching
norms
to
groups.
We have seen above that norms
make
groups, so "teaching" a group new norms is just a polite way of saying:
un-make this group.
Insofar as group affiliation is itself a human need, to that same extent is a baseline
difference
in norms
something to be fiercely conserved, not something to be eradicated.
No differences, no groups!
The world is diverse, and a general conclusion may be reached only after careful study.
Playing the uncertainty card. Usually a distress signal.
...
myron rothbart and marjorie taylor
Department of Psychology, University of Oregon,
...
Gil-White has given us a wide-ranging, thought-provoking paper.
...
However, we disagree with Gil-White’s view that humans have evolved a mental module dedicated to processing information about species-level categories.
We have problems with
(1) the characterization of this type of processing as modular,
(2) the rationale that it would be adaptive for ethnies to be processed by such a module, and
(3) the view that essentialist thinking is determined by the activation of the module rather than by social beliefs and social context.
Okay, I don't really know what module and modular mean here, so let's rustle up some quick-and-dirty highlights.
Key Concepts
Modules as Networks: ... brain regions (nodes) are highly connected within their own module but sparsely connected to nodes in other modules.
Functional Specialization: Each module is thought to perform a discrete function ... the module's primary function is relatively autonomous or "informationally encapsulated," meaning it doesn't need to access all the brain's information to do its job.
Integration and Coordination: ... the brain also needs to integrate information across different functions ... This is achieved by "connector nodes" (or hubs) that have a high number of long-distance connections to various modules ...
Hierarchical Organization: ... modules can contain sub-modules, which in turn contain sub-sub-modules, and so on, at multiple anatomical and functional scales.
...
Theoretic Debate
Strong Modularity (Fodorian): ... input systems (like perception and language) are highly modular, fast, mandatory, and strictly informationally encapsulated. ... higher cognitive functions (like reasoning) are not modular.
Massive Modularity (Evolutionary Psychology): ... the entire mind, including higher-level processes, is composed of a large number of modules, each evolved to solve a specific ancestral adaptive problem. ... a looser definition of "module" that does not require strict encapsulation.
Integrated View: Modern neuroscientific evidence generally supports a view of the brain as a balance of specialized, modular networks that work together through global integration, ...
(Google AI Overview, 22 Dec 2025, query: "in neuroscience, what is modular processing?")
Jerry Fodor's nine criteria for modularity in his 1983 book The Modularity of Mind ...
1. Domain Specificity: Modules process only a specific type of input (e.g., faces, colors, sounds).
2. Mandatory Operation: They run automatically and cannot be voluntarily turned off once triggered.
3. Fast Processing: Modules deliver their outputs quickly, crucial for rapid responses.
4. Informational Encapsulation: They operate without access to information from other modules or central beliefs (e.g., you can't unsee something).
5. Limited Central Accessibility: Only the outputs of modules are available to higher cognitive processes, not their internal workings.
6. Shallow Outputs: Outputs are simple, concrete representations rather than complex judgments or beliefs.
7. Fixed Neural Architecture: Modules are associated with specific, dedicated brain structures.
8. Characteristic & Specific Breakdown Patterns: Damage to a module results in predictable, specific deficits (e.g., prosopagnosia for face modules).
9. Characteristic Ontogenetic Pace & Sequencing: Modules develop and mature in a predictable order and timeframe.
('what are fodor's nine criteria of modularity?')
First, although a module can be posited without requiring that all nine of Fodor’s (1983) original criteria for modularity be met,
at a minimum “module” suggests that the processing of the input cannot be influenced by the person’s knowledge or beliefs (i.e., informational encapsulation) and that it is rapid,
[545]
automatic, and unconscious. Does Gil-White really want to make these claims about the processing of ethnies?
Based on my own self-awarenress and, uh . . . recent life experience, I would think that
these claims
can indeed be defended.
Without learning vastly more about this than I currently know, the
knowledge or beliefs
clause seems very tricky to parse. e.g. I am startled by a mouse although I neither "know" nor "believe" mice to be dangerous; I react
as if I do
know/believe this, but in fact
I neither know nor believe it;
ergo, there has been no
influence
of "knowledge or beliefs" upon
the processing of the input.
Yet it strains credulity, to me at least, to think that this "influence" could ever quite be narrowed down to zero.
For one thing, this is not (I think?) a 'mouse' module but rather a 'scurrying animals' module, or a 'disgust' module, or a 'startle' module. The form-content distinction is again worth making. So, of course my beliefs about mice won't influence my response if the stimulus is, for all intents and purposes,
not
specifically a mouse at all but rather something vaguer. It is when I
do
realize it's 'just a mouse' that the startle dissipates. In this scenario, encapsulation could as easily be a function of the stimulus as of the brain.
Now, can I
learn
not to be startled by mice?
Isn't protracted exposure to scurrying mice in shadowy rooms all but guaranteed to attenuate the
automaticity?
Does that show that I have not chosen the right example here? Or does it show that
there is no example
of a "module" which is impervious to "influence"?
Do we only ever get one iteration of actionable evidence from each subject, since the first iteration itself "influences" all subsequent ones? Is no mouse ever so frightening as your first one? That does not sound like a "module" that
is not
influenced by knowledge, rather like one that
has not
been. So, if some knowledge
simply cannot
be gathered empirically, then there would remain modules which
simply do not
become influenced by that knowledge-which-cannot-be-had; really it is the inaccessibility of the knowledge that is decisive, not the pristineness of the module; but because the two necessarily appear in tandem, they cannot be empirically differentiated. The pristine module then becomes robustly isolable empirically simply because the 'influence profile' of certain modules is not accessible to empirical investigation;
wherever it is
so accessible, candidate modules are eliminated from the strict Fodorian scheme per the "knowledge and belief" clause.
This issue of attenuation through experience is variously hinted at throughout this document, though not at all in these terms; sometimes via theoretical admonitions to consider "context", other times via actual countervailing empirical evidence. The possibility can never be ruled out that "naive essentialists" are simply people who manage to avoid receiving certain countervailing empirical feedback; the better question, I think, is whether or not it's reasonable to expect otherwise. This includes the question
du jour
of
what is lost in order to 'gain'
an environment wherein this feedback cannot be passively or actively avoided. Better to abandon Gil-White's module theory, I think, than to abandon honest answers to that question, as a few of these "good anthropologists" seem to have done.
Second, the justification for the adaptive value of processing ethnies as species is weak.
Gil-White focuses on the costs of coordinating between groups ...
However, it is possible to argue that ... seeking out such interactions, could be highly adaptive.
Out-group members offer new ideas and contrasting perspectives—exactly the type of input that reduces the likelihood of bad decisions associated with insular processing (e.g., groupthink) and that can promote technological and cultural innovation.
Historical analysis indicates that creative productivity tends to cluster in periods and places in which there is sufficient ethnic diversity to encourage cross-fertilization of ideas and beliefs ...
...
Third, Gil-White’s focus on ethnicity as the basis for “species”-like essentialism is misguided for several reasons:
(1) it focuses on the reality—rather than the perception—of group boundaries and inductive potential;
Hmm. Really?
Isn't this precisely the difference (or one of them) between "naïve essentialism" and "sophisticated constructivism"?
Didn't we devote almost as much space to prelimiaries as to experiments? And weren't those preliminaries mostly about
perception?
(2) it views essentialism as all-or-none rather than as continuous (even though his Mongolian data seem to suggest the latter);
I think this criticism evaporates if we read the paper for broad thrust rather than for granular details.
(3) it fails to recognize the dynamic nature of social perception, wherein the essentialist view of a given group can increase or decrease in strength depending on changes in social context;
Basically the "informational encapsulation" problem again. The module does not appear to be "encapsulated" from those "central beliefs" arising from social context.
and
(4) by placing so much weight on endogamy and descent as determinants of essentialism, it fails to recognize that the strength of essentialist perception for a given target group can vary both across different groups of perceivers and from individual to individual.
...
... Gil-White realizes that race is often strongly essentialized and therefore concludes that race is mistaken for ethnicity.
Clans, on the basis of his Mongolian data, are not considered ethnies and not essentialized.
On the other hand, caste and class can be essentialized, again presumably because of their mistaken similarity to ethnies.
We find this logic tortured.
Moreover, we view this attempt to identify which groups will and will not activate the “species module” as overly narrow and static.
...
If severe and extended conflict developed between the clans over, say, grazing territory, would not the changing intergroup conflict lead to deep, hostile attributions that would also be essential in their character?
I would think yes, probably, but this would indeed have to be
severe and extended.
I finally swallowed my pride and chatted with Gemini (17 Dec 2025) about this paper. This turned up an interesting morsel in a summary of Astuti's work in Madagascar:
"The Vezo are a coastal fishing people who define themselves by praxis (what they do) rather than descent (who their parents are).
"The Living (Performative): If you fish, use a pirogue, and live on the coast, you are Vezo. If you move inland and start farming, you become Masikoro (the neighboring group). There is no "hidden essence" preventing this change.
"The Dead (Essentialized): Crucially, Astuti found that when the Vezo talk about the dead, they switch to a descent-based model. Your ancestors are fixed; you belong to their tomb because of blood."
No partisan of Rank and Becker could be surprised by this. For
"Living (Performative)"
read: 'creaturely'; and for
"Dead (Essentialized)"
read: 'symbolic'. There is much to be said about this, far too much for this occasion, but here is a little bit.
No American will fail to notice here the faint promise of salvation from our lingering collective 'symbolic' conflict of vicious race-essentialism. This could be as simple as having more meaningful work to do, but we aren't willing to live that life, one so simple that
fisherman-farmer
leaps out as a highly salient and very nearly 'essential' distinction among people. If we're going to keep on watching TV, writing poetry and waxing our crotches, all with the irresistable force which those practices presently exert, then the countervailing immovable objects must be proportionally
severe and extended.
(Of course for these good "constructivist scholars",
grazing territory,
e.g., is an entree not to material conditions but to
social conditions,
i.e., to
intergroup conflict.
C'mon guys, it's not anthropology without a
little
bit of vulgar Marxism!)
Essentialism can take place at the dyadic, family, clan, class, gender, ethnic, or national level, and we view it as a mistake to restrict it to one or a few types of social categories.
We continue to believe that two factors strongly related to essentialism are perceived unalterability and perceived inductive potential, both of which are powerfully influenced by social beliefs and context.
... descent-based categorization, ceteris paribus, is more likely to be essentialized than others, but again the perceived homogeneity/inductive potential of the group varies as a function of individual and group beliefs.
...
In our view, inductive potential is itself a complex function of social reality and social perception ...
The reality of Western European Jewry was extremely diverse ...
It was Nazi ideology that imposed unity on this diversity ...
In our view, perceived homogeneity and intergroup boundaries are established by social belief and social context, and it is these factors ... that contribute most ...
takeyuki tsuda
Center for Comparative Immigration Studies, University of California, San Diego,
...
...
As Gil-White notes, few scholars have in fact advocated an essentialist approach that naturalizes ethnicity as primordial, biologically inherited, and ontologically predetermined,
suggesting that the constructivists are constructing a straw man to highlight what is new about their perspective.
More important, however, ... constructivists do not fully realize that most of our informants perceive ethnicity (emically) in a highly essentialist manner.
They often seem to assume that the people they write about are just as theoretically sophisticated as they are ...
...
[546]
One wonders whether the evolution of this type of ethnic cognition follows the law of natural selection,
since human thought in general does not always aim to maximize cognitive and interactional efficiency.
It is doubtful that cognitive patterns that are less instrumentally beneficial and adaptationally advantageous than others will always become extinct, especially where they do not threaten the social survival of a group.
Ultimately, such social evolutionary theories tend to be based less on empirical, historical evidence than on purely logical reasoning.
This does seem to be a problem with evolutionary theory. We can assume that many dogs are not barking, but we cannot say which ones.
...
Gil-White claims that because ethnic groups are perceived to possess a relatively immutable “nature” and essence, most individuals around the world do not and cannot freely choose and switch their ethnicities in response to sociopolitical benefits (contrary to what the constructivists claim) until they reach the second generation and begin intermarrying.
However, ... I doubt that even radical constructivists would argue that ethnicity can be completely manipulated and invented.
Right.
Not all
construction
is not
invention
per se. In other words, to say something is "socially constructed" is
not
to say that it could have been other than it is, and to say it is "essentialized" is not to say that it could have been only exactly as it is.
The
constructivist-essentialism
polemic is not (strictly speaking) a mere proxy for
free will-determinism.
Tbh, I personally lack the jolly and sporting collegial 'nature' to actually suspend judgment in this way. Constuctivists mostly seem convinced that things
ought
to have been otherwise than they in fact are, while essentialists would like to keep things as close as possible to the way they have always been; the valences with our present-day ideological moieties are obvious here, and I conspicuously lack the generosity to fully suspend my awareness of that.
...
I also wonder about the cross-cultural applicability of the analysis.
... Gil-White ... himself notes that his theory does not really apply to the United States, where ethnic affiliation is becoming less salient than a national American identity (many scholars argue that precisely the opposite is happening).
🤔 🤔 🤔
It seems that the real reason many individuals in the United States (and, by extension, other multiethnic complex societies) no longer essentialize ethnic affiliation is that
they have developed multiple and hybrid ethnic affiliations (only very briefly mentioned in the article) that defy easy classification, allowing them to present themselves somewhat differently in different situations.
Yep. And the way we
present
ourselves
doxastically often has nothing to do with what we actually believe.
i.e.
Essentialism Without Essentialists.
Here is an insidious way to
manipulate
and invent
ethnicity. The 'melting pot' is fleshly, ethnic, social, cultural, . . . and so ultimately
semiotic.
The melting pot cooks up precisely the raw material of self-caused self-styling that human beings have always dreamed of. We quite accidentally cracked the code of
self-construction:
diversity + connectivity + dematerialization. The Other no longer defines you if your self-presentation speaks to them in their own language; but that language also must meet minimum threshholds of diffusion and complexity in order to gain traction.
We have always dreamed of this, but it is not good for us.
Gil-White’s evolutionary model of ethnic consciousness ...
... it would apply only to a limited number of simple societies in which ethnic groups are still relatively isolated ...
However, there are not many simple societies of this type today.
This is the problem with an analysis that is built upon a dichotomy of ethnic constructivism versus essentialism, since most people’s ethnic behavior and consciousness fall somewhere in between.
What about their beliefs?
Although individuals do construct and manipulate their ethnic identity to a certain extent, they do so under the essentialist constraints of socialized and “inherited” cultural and racial characteristics which prevent the free crossing of ethnic group boundaries.
Well, if you get rid of the boundaries . . .
Reply
francisco j. gil-white
Los Angeles, Calif.,
...
My commentators in general consider that the empirical results are important, demand an explanation, and point out deficiencies in previous theorizing.
With one exception, their criticisms engage the ideas seriously.
Methodological issues and hypothesis testing.
Laitin observes that my respondents easily granted “that many cultural characteristics were learned, ... ”
So, he intimates, they are not as essentialist as all that.
...
... despite granting this extreme enculturation premise, subjects resisted the idea that the child would be quite like the Kazakhs ...
To assume that there is a Mongol “essence” is to think that the child has the potential and tendency naturally to behave like a Mongol, but people know that humans learn.
The burden of demonstrating essentialism should not be that of showing people to be fools who deny learning.
Astuti finds problematic (but does not say why) that in question 3 “participants were told that, rather than asked whether, the adopted baby would learn the customs and language of the adoptive group.”
The point was to see whether enculturation or descent was more important for ethnic ascription.
By telling participants that the adopted child was thoroughly socialized into the adopting ethnie’s culture, I set the contrast that would test that question.
...
She finds the interest of the result mitigated, in any case, by the fact that it is no more than one ethnographic case ...
But scattered exceptions ...
[547]
... will hardly disprove a prediction about an expected pattern.
... I am ready to accept that my theory may find its predictions unsupported by cross-cultural data.
But we must be clear on what my main claim is:
it is not quite that most ethnies will be essentialized.
...
I have used the term “ethnie” merely to communicate a rough understanding that is in the overlap of all the definitions and extensions used,
but my claims and predictions do not use or need the term, and they are quite specific.
The prediction is that
those social categories which show normative endogamy and descent-based membership will be the ones more likely to be essentialized and essentialized more strongly.
The prediction is about such categories, not narrowly about what any particular scholar may insist is an “ethnie,” and is thus applicable to castes and any other category that has the stated properties.
...
Rothbart and Taylor seem to think that I have argued just the opposite, and they complain that different kinds of categories, not just ethnies, can be essentialized.
But we are in complete agreement on that point.
Laitin, for his part, presents the variability in essentialist beliefs over space and time as weakening my hypothesis, but it is the nature of the variability that will undermine my claims.
If essentialist beliefs weaken when the cues that make a social category resemble a species are attenuated, this variability, rather than disconfirming my hypothesis, will follow its predictions rather precisely.
It is obvious that I failed to be clear ... , given that Banton focuses exclusively on the fact that scholars don’t agree on the use of a particular lexeme—“ethnic”¹— ... rather than on my arguments ...
Ma, similarly, interprets my argument as being about ethnies as in fact immutable entities ...
Rothbart and Taylor focus on a definition of “module” that they favor rather than on my claims, which do not need the word “module” (which, again, was used to communicate a rough idea
1. ... Scholars disagree when they write down explicit definitions of “ethnic group” that they hope will advance a priori theoretical commitments. This naturally leads to a proliferation of definitions, for here ideology holds more sway than science and one makes a point not by showing something but by redefining the terms everybody else is using. ...
👆 👆 👆
This is just the kind of "naïve" Wittgensteinianism I can get with.
Less said the better.
Don't tell me,
show
me.
It's fun to spool and unspool "sophisticated" antitheoretical arguments, but if basic dialogic continence and hygiene carry the day then all of that becomes unnecessary.
rather than a fundamental theoretical point).
And Tsuda thinks that my claims overreach because not all ethnic groups are ... easily demarcable or impermeable ...
I hope I have now made clear why these objections miss the point,
but I accept the blame if my presentation was suboptimal.
" . . . I'm sorry YOU FEEL that way . . . "
Gotta love the academic bloodsport.
Astuti defends Hirschfeld’s (1996) attention to child cognition ...
Evidence from adults, she claims, ... would not constitute support for my theory ...
I cannot agree with her position.
If we find that, all over the world, adults intuitively essentialize normatively endogamous categories with descent-based membership even when their explicit cultural models tell them not to,
an innatist explanation for this pattern becomes more plausible than the alternative.
Pity the layperson who seeks to 'do their own research' about what we find all over the world. Anthropologists must uniquely appreciate the moiety organization of scholarship itself!
...
What could falsify my hypothesis?
Perhaps the fact that we essentialize gender, Gelman ventures.
She observes that endogamy and descent-based membership cannot explain why gender is essentialized.
I would make the list longer, ...
But how are these counterexamples to my hypothesis?
I have claimed that endogamy and descent-based membership will increase the likelihood and the extent to which a social category will be essentialized,
not that nothing else can cause essentialism or that living-kind essentialism is the only kind possible.
Perhaps invoking "modules" was a big mistake, then, since module-theory does seem to be absolute in precisely these ways.
How about some in-depth forensic psychological interviews with people who are outlier exceptions to their prevailing local
likelihoods
and
extents?
i.e. Studies of 'the single case' in Stephenson's vein? Surely that would at least be interesting.
...
The Vezo, Astuti claims, constitute a counterexample.
But to know this we must not merely show them to be nonessentialists;
in addition, we must determine whether for the Vezo
(1) membership is a matter of descent and
(2) category endogamy is normative.
If this is the case, we have a genuine counterexample, ...
Hmm. So, here is your 'social construction'? Right here?
If any ethnography of
these
variables has been presented here, I have overlooked it. What else do
these
correlate with? If the answer is: 'Ethnic essentialism, of course!' then we would be going in a circle; this would be
quite
informative all the same!
...
Failure to find caste essentialism would be a real counterexample, for castes are normatively endogamous and descent-based.
The Mahalingam (1998) results that Gelman refers me to are equivocal, however.
...
Meritocracy is taking hold in India, ...
... denying [traditional caste] assumptions serves
[548]
the legitimizing of lower-caste upward mobility ...
...
In any case, the proper analysis must try to see whether essentialism, normative endogamy, and descent-based membership travel together.
The strongest refutation of my hypothesis will come if those explicitly committed to the descent-based acquisition of caste status and to the undesirability of intercaste unions show explicit and intuitive nonessentialist reasoning about attributes.
Gelman sees important confounds in my data ...
... [but] the items in Gelman’s list are highly redundant.
...
... the only thing in the Mongol/Kazakh contrast in addition to endogamy and descent that might cause essentialism is “cultural differences.”
What, then, differentiates the ethnic contrast from the clan contrast?
Endogamy practices?
Only in a minor quantitative sense; ...
Cultural differences?
Again, only in a quantitative sense, ...
...
The only dramatic and qualitative difference between the two contrasts is the rule of endogamy.
...
Theoretical issues.
In the summary of history that Atran favors, mutual predation by human groups is what drives psychology, and essentialism reinforces both ingroup similarity, cooperation, and cohesion and intergroup demarcations.
I am skeptical that this process can really be the engine of ethnic essentialism given that for most of history the antagonistic political units have typically been either smaller ... or larger ... than ethnies.
Political units are not strongly essentialized unless they coincide with ethnic boundaries.
My argument is not about cooperative coalitions locked in combat but about the emergence of clustered variation in interactional norms,
which results if people are unevenly distributed in the physical landscape and adapted to conform to locally common interactional norms.
My account treats the imperative of navigating adaptively a world composed of norm clusters as causing ethnic-category essentialism.
Atran does not comment on this argument, however, except to observe that he doubts (without saying why) that there is a domain of “norms” that could have selected for elements of human psychology.
Atran believes that social-category essentialism results not from specialized derived psychology but purely from the “big mistake” misfiring of an ancestral cognitive trait ...
This process results in a “looping effect” ...
I see two problems with this argument. First, phenotypic differences hardly appear necessary for essentialism.
...
Closer to home, it was not long ago that the English considered the Irish a nonwhite “race.”
...
Not that I would know, but this point strikes me as being on the right track, and the gamut of neglect spanned in the comments seems to be just as clearly off-track.
One problem seems to be that
phenotypic differences
are sufficient,
even if they are not
necessary.
In the abstract that is a simple enough notion, but it means that the phenomena it purports to describe are epistemically booby-trapped:
sufficiens
and
necessitiens
become counterexamples to each other,
qua each other; but the capacity in which they contradict each other is not the capacity that matters for the present argument. That's because (I think?)
normative endogamy
can't guarantee intergroup "differences"
tout court
but it can restrict the range of variation within each group; therefore, perceptible phenotypic differences which already exist will strongly tend to be maintained, but new ones will very rarely be generated.
The other problem is that this essentialism in absence of obviously perceptible differences
does,
after all, seem to assume that the essential categories are, in these cases, learned
explicitly, or at least
not
purely implicitly; but then, as with all explicit learning, there must also (later) be a time of implicit 'sinking-in' which so to speak 'covers the tracks' of any explicit process. This is just a guess, and an unparsimonious one at that.
We will not know for sure until developmental data are collected ...
Nevertheless, Hirschfeld (1996:102–7) already has a result supporting the argument I defend ...
He found—to his surprise—that three-year-olds essentialize people of the same phenotype but dressed in different uniforms (e.g., police officer) virtually as much as they essentialize people with dramatically different phenotypes.
Why this result?
Because differences in clothing should prime the living-kind module?
A more plausible alternative is that children are specifically looking out for uniforms as a shortcut for guessing which categories ought to be essentialized.
Clustering of norms in the ancestral environment, after all, would have made it sensible to signal which norm-group one came from, ...
... young children presumably have yet to learn that policemen do not typically marry policewomen ...
When children become adults they no longer essentialize police
[549]
officers as a natural kind because by then they have learned about the importance of descent and marriage for “living kind-ness” ...
Well, do modules learn or not? If they do not learn, have we explicitly ruled them out with the above interpretation?
...
Hirschfeld finds that by age seven dramatic phenotype contrasts are almost unanimously essentialized,
which may follow from increasing awareness of normative prescriptions against interracial marriage.²
The second problem is that ...
There is simply nothing to infer from membership in a putative race qua “race,” ...
It is in the subgroups—neighboring ethnies with strongly overlapping phenotypic distributions—that normative differences cluster, rather than at the boundaries of “races.”³
Rothbart and Taylor note that interaction with ethnic out-group members could be adaptive rather than maladaptive ...
I agree that acquiring superior technology from others is sensible, but I disagree that interaction across the ethnic boundary is thereby rendered generally adaptive.
My argument is about interactional norms, ... [e.g.] making promises and contracts, hosting a guest, ... etc.
It is certainly adaptive for me to obtain superior knowledge ... but it will be maladaptive to adopt a set of interactional norms that the people in my own community neither expect nor endorse, and it will be costly to attempt many interactions with people in other communities ...
A current study of knowledge transfer between two different ethnic communities in Guatemala shows that very little and very superficial contact ... is sufficient for the transmission of adaptive knowledge ...
Knowledge flows from the biologically expert Itza’ Maya to the recently arrived Ladino through only one or two human vectors, prestigious Ladino individuals who learn from prestigious Itza’ individuals.
Other Ladinos then acquire this knowledge from the prestigious Ladinos.
...
2. Perhaps Rothbart and Taylor will find this logic tortured, but I find no value to parsimonious explanations that fail to account for the facts. In evolution, where selection pressures are always changing direction and magnitude and adaptive lags are common, the facts often call for explanations that fall short of optimal parsimony.
3. Marked phenotypic discontinuities occur in given localities only because of modern migrations anyway. The supposedly “racial” categories themselves all blend insensibly into each other.
...
The point is that very little actual interaction—and certainly not very intimate interaction—is sufficient for transfers ... ; one interethnic interactant will suffice.
All others will be faced with either cheaply obtaining the information from this coethnic or doing so by means of costly interactions with out-group ethnics.
The adaptive path is clear.
Excellently argued once again.
I wonder how Marx's much-belabored harangue about trade originating between communities rather than within them has fared in the intervening century-and-a-half of anthropological data collection? If it is a fact, then it is a fact
about trade
and not just about 'communities'.
Gil-White returns often to such observations as:
it will be maladaptive to adopt a set of interactional norms that the people in my own community neither expect nor endorse, and it will be costly to attempt many interactions with people in other communities who play the games of life in different ways.
Jane Jacobs titled one of her books
Systems of Survival,
and that phrase is as good as any for naming the nexus of these two arguments:
technological innovation
is lovely, but 'survival' comes first. Subsistence
tout court
is prior to susbsistence
in style.
The pacifying effects of trade have to be
discovered
before they can be assumed known. Intergroup trade must have seemed dangerous at first, and presumably it really was dangerous by contemporary standards.
(Gemini mentions "'silent trade,' where different groups would leave goods at a border, retreat, and wait for the other group to leave an equivalent offering.")
Leaving aside the extent to which danger itself may actually become enticing to certain people, aren't we entitled to assume that the 'incentive' (whatever it was) to trade with scantly-known outsiders must have been not just strong but ultimate?
Rothbart and Taylor assert that the Nazis ideologically imposed homogeneity on the diverse Weimar Jews, ...
Even if correct, this observation does not undermine my argument that actual inductive potential is not the proximate priming input that causes essentialism.
...
... by correcting my characterization ... , Rothbart and Taylor make my point for me in a different way:
Weimar Jews were probably essentialized ... because descent-based membership and category-based endogamy were perceived as coextensive with the label “Jew.”
The Universalist says:
Don't want to be essentialized?
Then don't practice endogamy!
and
just to be safe
don't signal your membership
quite so lustily!
The Particularist says:
But people NEED groupish identities!
We need to HAVE them
and also
to TRANSACT IN them
just to know and feel that we are alive.
Here is a vicious circle.
I think that there are simple and obvious solutions, and that these solutions are very unpopular with Universalists but are recently gaining popularity with Particularists.
Hamilton, Sherman, and Sack take on the issue of the domain which my theory is supposed to explain and construe my effort as that of finding out what a “group” is from the point of view of observers.
But, in fact, I am skeptical that the question “What is a group?” is a useful one.
The extension of “group,” as used by social psychologists and other social scientists, includes a staggering diversity of things ...
“Group” is a rather strange yet mainstream abstraction among scholars ...
... properties that are most strongly associated with the term and that emerged from “small-group research” [get projected] onto massive categories that don’t have those properties.
As Deutsch notes, the usage of “group”
“is consonant with the intuitive notion that a group is an entity that consists of interacting people who are aware of being psychologically bound together in terms of mutually linked interests.
A group is thus to be distinguished from [a] . . . category . . . which consists of people who are classified together because of some common characteristic.”
My question is:
How do observers decide that they are
[550]
looking at one kind of social category and not another?
... it is only recently, with the advent of ethnonationalism, that some ethnies have converged somewhat on those properties [associated with "groups"].
The term “ethnic group” is therefore mostly a misnomer, ...
...
When Laitin ... observes that the boundaries of the ethnic group need to be policed by ethnic entrepreneurs,
he is thinking more about ethnic-wide cooperation—a prisoner’s dilemma payoff matrix with incentives for cheaters—than about coordination—a payoff matrix without incentives to cheat but rather costs to those who fail to match their standards of behavior and signaling systems
(the latter, not the former, is the substance of my evolutionary argument ... ).
This focus on ethnic cooperation ... confer[s] on [ethnies] a degree of “groupness” that they do not have.
...
Coming back to the question of perception,
Hamilton and colleagues say that “entitativity is the unity, coherence, bondedness among individuals that causes them to be perceived as a group.”
But this communicates little unless it is self-evident that “group” is supposed to tell us something about cognitive processing.
...
Let us find out what certain inputs do or do not prime in social reasoning, without the awkward term “group” and its baggage to muddy our understanding.
...
... it is probably best to drop “entitativity” as it is currently employed.
Hamilton and colleagues’ claim that ethnies are low in perceived entitativity appears to confuse the degree of (objective) entitativity that a scientist can measure with a person’s (subjective) intuition that a given category is like an entity.
Are ethnies low in the latter?
I doubt it.
It seems more likely that laypeople (and scholars of ethnicity!) process ethnies as coherent entities because they are essentialized and despite the fact that their objective entitativity is low.
...
My points concerning endogamy have met with considerable resistance.
...
Laitin chooses to mate a giraffe with an elephant.
That is indeed a strange mating.
The point would be made more forcefully with the example of a bear that mates with a fly or a whale that mates with a protozoan.
But if we consider interethnic marriage like mating a dog with a wolf or a donkey with a horse the force of this objection seems to evaporate.
Such matings are not only behaviorally and anatomically plausible but occur and produce viable offspring.
Importantly, these examples concern matings between species the close relation of which is apparent in the similarities they show in morphology and behavior.
...
The intuitions which make us think that the mating between a wolf and a dog is possible but unnatural could easily be responsible for similar reasoning about interethnic marriages.
The claim I have made is that normative endogamy helps trigger the intuition that one sees a “species”—
that is, the trigger is not the perception of the impossibility of an interethnic mating but the perception that such things are immoral and therefore also unnatural (or vice versa),
which may be underlain to a greater or lesser degree by the actual rate of intermarriage.
The objection just addressed seems to flow from the perception that priming inputs have to be either “on” or “off.”
But in this as in so many other domains, the brain is probably primed on a continuum:
the stronger the public normative prohibition against intermarriage in a time and place, the more closely such a prohibition is observed, and the more rigid the requirement of membership by descent, the closer the input match will be to that of a species-like category and therefore the more likely humans will be to activate the essentialism exapted from the living-kind module.
That is my claim.
Gelman points out that endogamy cannot work as a priming input unless it is perceived and understood and suspects that children may essentialize social categories even before they acquire knowledge of endogamy.
However, this does not undermine my claims unless the point at which humans make the link between “living kindness” and endogamy is material to the substance of my argument.
Suppose that at first children ... initially have [only] an “essence placeholder” ...
The thing which initially primes a child to essentialize may be nothing more than certain key surface appearances.
Audrey Lorde's 'baby maid' anecdote seems to be an (unfortunate) example of this.
... the assumptions
[551]
that Markman (1991) describes as responsible for object-word acquisition may automatically include, for objects that appear to be animate, the additional assumption that there is a causal essence common to all members of the class.
But any specific content about what the essence is ... remains mostly to be filled in ...
If this picture is reasonable, then Gelman’s suspicion that children will essentialize social categories before they learn about endogamy is likely to be correct,
for endogamy is merely something that shared living-kind essence is supposed to imply, and this supposition is part of what is waiting to be filled in.
...
To think of members of a living kind mating and reproducing is practically to see the essence in the process of being transmitted ...
...
References Cited
...
No comments:
Post a Comment