Ignorance, Knowledge, and Dummy Categories:
Social and Cognitive Aspects of Expertise


John B. Gatewood

[ Copyright (c) 1996, John B. Gatewood ]

Presented at the 95th Annual Meeting of the American Anthropological Association, San Francisco, CA (November 19-23, 1996), in an invited session organized by David B. Kronenfeld entitled, "Culture as Distributed Cognition."

Compared to other species, human beings have large brains, enabling each of us to learn and retain a great deal of information during our life spans. Private neurological storage, however, has upper limits. Eventually, brain size limitations on cultural development were overcome by a group-strategy built on the principle of reciprocal ignorance, i.e., a division of labor entailing non-redundant distribution of cultural knowledge among individuals. Maintenance of the group's social heritage ceased being a matter of replicating uniformity and rested instead on organizing diversity, superficially similar to the group life of the social insects. Today, any given individual learns only a portion of his or her group's heritage, and for most domains, individuals range along a novice-to-expert gradient. Partial knowledge, along with the ability to access what others know, is sufficient for most purposes for most people. This paper discusses the role of linguistic categories in bridging very asymmetric knowledge boundaries in frequently occurring social contexts. Particular attention is given to the kinds of knowledge (knowing of, knowing about, and knowing how) that distinguish Nacirema experts from novices in the domain of mixed alcoholic drinks.



Given the "retrospective" theme for this year's AAA meetings, let me begin by noting some theorists who are especially relevant to my paper:
    WHO:              WHY:
    Herbert Spencer   evolution of complexity
    Emile Durkheim    division of labor / collective representations
    Alfred Kroeber    subcultures, occupational cultures
    Gregory Bateson   schismogenesis / standardization through
    Donald Campbell   evolution of morals
    Melford Spiro     false dichotomy of culture and personality
    Anthony Wallace   end-linkage / replicating diversity
    John Roberts      information-processing view of society
    Theodore Schwartz individual as distributive locus of culture
    F. K. Lehman      dummy categories
    others ... some of whom are sitting in this room
I'm hoping we share enough common culture that just mentioning these names up front will tell you where I'm coming from ...

My game plan is to sketch a general framework for thinking about culture as distributed cognition, then to report on some research findings that illustrate one piece of the larger puzzle.


The biologist Roland Hotchkiss defined life as "the repetitive reproduction of ordered heterogeneity." It follows from this definition that all life forms can be viewed as varieties of information storage and processing systems (Gatlin 1972). What, then, is special about the information systems of human beings?

Compared to other species, human beings have large brains, enabling each of us to learn and retain a great deal of information during our life spans. Extra-genetic storage of this sort is an essential prerequisite for culture, but private neurological storage has upper limits--the amount one person is capable of learning and remembering.

Over several million years, our ancestors eventually overcame the brain-size limitation on cultural development. They evolved the neurological capacity for symbolic communication, and this enabled an extra-somatic storage and retrieval system in which the differentiated social group rather than the individual became the vehicle of acquired knowledge. This new group-strategy was built on the dual principles of reciprocal ignorance and mutual dependence. The end-result was human societies superficially similar to those of the social insects (Campbell 1983), i.e., societies with a division of labor entailing a largely non-redundant distribution of knowledge among individuals. Accompanying the shift toward extra-somatic storage and retrieval systems, maintenance of the group's social heritage ceased being a matter of replicating uniformity and rested instead on organizing diversity.

Today, any given individual learns only a portion of his or her group's heritage. Partial or incomplete knowledge is sufficient for most purposes for most people. Individuals partake of their socially distributed culture on a "need-to-know" basis, where the trade-off is slower retrieval time in exchange for much greater storage capacity (sort of a RAM versus hard-disk analogy). But for this memory-efficient, culture-as-distributed-cognition system to work, individuals must be able to access what others know (find information on the hard-disk). And this highlights the enabling role of language in the evolution of complex human societies. Simply being able to talk the group's talk, whether or not you know what you're talking about, is adaptive for the individual and the group.

For explicit, verbalizable knowledge, the road from total ignorance to full expertise has three clearly identifiable signposts along the way:

1. Never heard of X.
2. Heard of X.
3. Heard of X, and know about X (something / all there is to know).

When we include learned motor habits in our definition of culture (following in Boas's dance steps), we must distinguish at least three general kinds of cultural knowledge: knowing of, knowing about, and knowing how. Let me expand.

Knowing Of. This is the minimal form of cultural knowledge. To know of a category implies that the category is part of the collective representations of some social group: it is something people talk about among themselves. When a person hears of it (and remembers), then that category comes into being for that person. Even when the category is empty of content for a given person--a "dummy" category--its existence in the cognitive makeup of the individual provides a frame of reference for substantive knowledge distributed elsewhere. Thus, simply knowing of a category is the first step toward linking otherwise ignorant individuals to substantive knowledge distributed across the larger social group. Knowing of things provides the foundation for orchestrating substantive diversity.

Method Note: Inter-individual patterning can be revealed through free-listing tasks and direct questioning (e.g., Ever heard of ___?).
Knowing About. Knowing about knowledge is also purely linguistic, but includes such things as the category's defining characteristics, its connotative and situational associations, its relations vis-a-vis other categories, and what kinds of people are likely to know more about it. How much a given person knows about something is variable through his or her life, and individuals differ greatly among each other.
Method Note: Inter-individual patterning can be revealed through taxonomic interviews, identification tasks, list-properties tasks, consensus scores on judged similarity tasks, consensus scores on appropriateness judgments, and so on.
Knowing How. Knowing how knowledge is knowing how to do something. When speaking of performative skills--such as riding a bicycle, blacksmithing, fishing for salmon, bartending, or medical care--expertise involves sensory-motor abilities and tacit knowledge that may or may not be verbalizable (Polanyi 1958; Dougherty and Keller 1985; Gatewood 1985, 1991; Dreyfus and Dreyfus 1986). Such motor habits and tacit knowledge are acquired slowly through personal immersion in meaningful performative contexts. If the constituent actions and thought processes are topics of conversation and novices learn the skill via coaching, then the three kinds of knowledge--knowing of, about, and how--are likely to be learned together. In general, however, an expert in knowing how to do something may or may not be an expert with respect to associated "talk knowledge," e.g., an accomplished pianist may know little about the history of tonal music.
Method Note: Inter-individual patterning can be revealed through time trials, mistake/success ratio, peer-evaluations of ability, and so on.
When we move to the question of how these different kinds of knowledge are socially distributed, we find that inter-individual differences in partial knowledge can take several forms--random variation, uniformity, expertise gradients, or systematic differences (Boster 1985). Compared to the other two kinds, "knowing of" knowledge tends to be distributed redundantly (more uniformly) across adults in a society, whereas "knowing about" and "knowing how" knowledge tend to be unevenly distributed, either as expertise gradients or as systematic differences.

Now, I'd like to illustrate social distributions for all three kinds of knowledge (see Gatewood 1994), but time is short. So for the rest of my paper I'll concentrate on one extended-play example--Americans' knowledge about liquor and mixed drinks.

I like this example for several reasons. First, booze is a culturally "thick" domain for the Nacirema. Second, Nacirema are highly variable in how much they know about it. Third, liquors and liquor knowledge are generally considered inappropriate for children, such that most Nacirema learn what they know from multiple and diffuse sources, usually without much explicit instruction, and over many years. And fourth, perhaps because of the age-grading of such knowledge, Nacirema regard drink orders as highly expressive and use them to define social situations and identities.


John Lowe and I did some commercial research for United Distillers and Schieffelin & Somerset in New York City (Lowe and Gatewood 1995). I have their permission to talk about "generic liquors"--gin, vodka, scotch, etc.--but not brand names or other aspects of the project.

One of our research questions was: "Do Nacirema have a cultural model(s) regarding the mixability of generic liquors?" That is, we wanted to examine a particular kind of "knowing about" knowledge: the combinatorial possibilities, or grammars, of different liquors when making mixed drinks. What ingredients plausibly go with what?

[Note: Some of you may feel mixing rules just conform to whatever tastes good. Let me assure you that a scotch margarita tastes pretty good, it just sounds wrong.]

We already knew Americans differ considerably in how much they know about liquor and mixed drinks, and we hypothesized that inter-individual differences in this domain would be patterned in the fashion of "single cultural models with expertise gradients." With this in mind, our sample of paid informants was composed of two a priori groupings--Group1: young/inexperienced college students (n=21), and Group2: older/experienced lounge lizards (n=20), with roughly equal numbers of men and women in each.

We collected data of two sorts. (1) To see whether Nacirema have similar conceptual models of liquors as a semantic domain, we did triadic comparisons among seven generic liquors--gin, vodka, scotch, bourbon, brandy, tequila, and rum. (2) To explore the combinatorics of liquors, we asked informants to rate the plausibility of an ungodly number of two-ingredient mixed drinks on a Yes/No/Maybe scale. The portion of these mixability data I'll be talking about can be thought of as an 8x32 "booze by other-boozes + mixers" matrix, where each cell is a judged pair-wise combination.

Six Specific Questions and Findings

I. Do Nacirema share a single cultural model with respect to the overall similarity of generic liquors? Thus, the basic categorical relations among gin, vodka, scotch, bourbon, tequila, brandy, and rum are, for the most part, distributed redundantly (uniformly) across our 41 informants. A two-dimensional image of these relations is:

*** MDS of 7 generic liquors ***


II. Overall, do young & inexperienced drinkers (underage college students) differ significantly from their older & experienced counterparts?

Thus, based on the both perceived conceptual similarities among generic liquors (triadic data) and the mixability matrices taken as a whole, there is no basis for segmenting the sample, at least when considering generic liquors collectively.

III. Are there cultural models governing the mixability of generic liquors considered one at a time, and if so, for which liquors?

           Ratio of      Mean
Liquor     Fac1/Fac2     competence    Decision

Vodka         6.309        .657       YES, single cultural model
Gin           4.785        .615       YES, single cultural model
Scotch        4.221        .571       YES, single cultural model
Bourbon       3.342        .499       YES, single cultural model
Light Rum     3.130        .480       YES, single cultural model
Tequila       2.282        .415       No.
Dark Rum      2.199        .469       No.
Brandy        1.672        .331       No.

Thus, understandings of plausible mixed drinks take the form of expertise gradients for some liquors, but not for others.

Vodka / Gin / Scotch / Bourbon / Light Rum: Variability is a matter of idiosyncratic deviations from a core of shared beliefs, i.e., "partial grammars" deviate randomly from the more expert norms.

Brandy / Tequila / Dark Rum: These liquors are not well understood: they show systematically different ways of thinking about their mixability, but for different reasons:

-- BRANDY: Lack of consensus is likely due to a semantic ambiguity
(a) "brandy" as cognac --> drink straight in snifters, don't mix
(b) "brandy" as cheap, flavored booze --> mixable

-- TEQUILA / DARK RUM: Both are "exotic" liquors that have diffused into Nacirema culture differently for different people. Individuals subject to direct diffusion use the mixability grammar of Mexico and the Caribbean, respectively. Nacirema who know about Tequila and Dark Rum only through stimulus diffusion have a different view of their plausible mixings.

IV. Do individuals acquire liquor competencies across the board or one liquor at a time?

Vodka Gin Scotch Bourbon L.Rum
Vodka -----  0.29  0.48  0.29  0.30 
Gin 0.29  -----  0.52  0.45  0.32 
Scotch 0.48  0.52  -----  0.49  0.18 
Bourbon 0.29  0.45  0.49  -----  0.26 
L.Rum 0.30  0.32  0.18  0.26  ----- 

Thus, partial knowledge seems to be filled in one liquor at a time, and each liquor has its own characteristic "profile" vis-a-vis the grid of twenty-four mixers, as illustrated here:

*** PROFILES: Gin, Vodka, Scotch, Bourbon, L.Rum ***


V. For the five liquors where expertise gradients exist, do age/experience groups differ in their grammatical competence, their grasp of the cultural rules?

Preliminary. Comparing the MEANS and VARIANCES of consensus scores for the two groupings may shed some light on the age at which booze grammars are acquired. Given that expertise gradients exist and holding aside sampling issues, from within a general cultural acquisition framework there are three plausible/interpretable patterns:

A. The younger group (college students) knows the cultural model as well as their elders, implying that a liquor's grammar is acquired either before going off to college or very rapidly upon arriving.

MEANyoung = MEANold ; VARyoung = VARold
B. Acquisition of the liquor's grammar occurs as a function of age, generalized social experience, and individual (e.g., familial) factors.
MEANyoung =< MEANold ; VARyoung >= VARold
C. Nacirema booze culture is undergoing generationally-based change, i.e., a "new" cultural model is afoot.
MEANyoung >= MEANold ; VARyoung =< VARold
Results of t-tests and F-ratios were as follows:
   PATTERN   LIQUOR            MEANS           VARIANCES

        A    Vodka              n.s.              n.s.

        B    Scotch             n.s.        Younger > Older
        B    Bourbon            n.s.        Younger > Older
        B    Gin          Younger < Older   Younger > Older
        B    Light Rum    Younger < Older         n.s.

        C    ---                ---               ---
Thus, the implications concerning age of acquisition are: VI. Does overall similarity among liquors at a conceptual level predict mixability with specific ingredients? Thus, categorical similarity among generic liquors (triads task) does predict mixability to some extent. As Berlin or Hunn might say, general-purpose classifications are related to specific uses. On the other hand, categorical similarity explains only 31% of the variance in the generic liquors' mixability profiles.


To relate my extended example back to the bigger framework I began with, let me note that several of our novices are non-drinkers. The puzzling thing, then, is why they've bothered to learn what they have about liquors?

Collective representations, in the minimal sense of knowing of, tend to be distributed widely and redundantly among members of the society. Even when collective representations are "dummy" categories for a given person, they are a strong social glue. They enable the person to acquire more knowledge about the category as motivations and opportunities allow. Thus, it is not surprising that non-drinkers might pick up liquor knowledge in bits and pieces through life. Indeed, we seem to be culture sponges for all kinds of "talk about" knowledge--we pick up and remember a great deal of apparently trivial details.

The real pay-off for having shifted to an extra-somatic storage and retrieval system comes with respect to "knowing how" knowledge. We can be wonderfully oblivious to all the motor skills and tacit knowledge required by other people's labor. Butchers, bakers, candlestick-makers ... they do what they do, AND I DON"T HAVE TO BE BOTHERED WITH HOW THEY DO IT! All I need to know is that such people exist, how to find them when I want to, and how to communicate with them (talk, money) to get the fruits of their mysterious expertise.

In the case of booze grammars, even non-drinkers seem to know of the generic liquors and a fair amount about them. Granted, the cultruralogic for Gin seems to require post-graduate education, but Vodka and, to a lesser extent, Scotch and Bourbon are understood pretty well by college students. Perhaps the most interesting finding was that grammars for Tequila and Dark Rum have not yet standardized--there are systematic differences regarding their mixing potentials. But for the more familiar liquors, Nacirema do seem to converge on the unspoken mixing rules, and they do this despite informal and diffuse learning contexts. Armed with a rudimentary grasp of the culturally correct, novices can take advantage of bartenders' skilled labor without making fools of themselves and they can interpret other's savior faire. Thus, knowing something about liquors is socially useful, even for non-drinkers.


Boster, James S. 1985. "Requiem for the Omniscient Informant": There's Life in the Old Girl Yet. In Janet Dougherty, ed. New Directions in Cognitive Anthropology, pp. 177-197. Urbana: University of Illinois Press.

Campbell, Donald T. 1983. The Two Distinct Routes beyond Kin Selection to Ultrasociality: Implications for the Humanities and Social Sciences. In Diane Bridgeman, ed., The Nature of Prosocial Development: Theories and Strategies, pp. 11-41. New York: Academic Press.

Dougherty, Janet W. D. and Keller, Charles M. 1985. Taskonomy: A Practical Approach to Knowledge Structures. In Janet Dougherty, ed., New Directions in Cognitive Anthropology, pp. 161-174. Urbana: University of Illinois Press.

Dreyfus, Hubert L. and Dreyfus, Stuart E. 1986. Mind over Machine: The Power of Human Intuition and Expertise in the Era of the Computer. New York: The Free Press.

Gatewood, John B. 1985. Actions Speak Louder than Words. In Janet Dougherty, ed., New Directions in Cognitive Anthropology, pp. 199-219. Urbana: University of Illinois Press.

Gatewood, John B. 1991. Taking Orders, Making Drinks: Active Cognition in Bartending. Paper presented at the 50th meeting of the Society for Applied Anthropology, Charleston, S.C.

Gatewood, John B. 1994. Personal Knowledge and Collective Representations. Paper presented at the 93rd meeting of the American Anthropological Association, Atlanta, GA.

Gatlin, Lila L. 1972. Information Theory and the Living System. New York: Columbia University Press.

Lowe, John W.G. and Gatewood, John B. 1995. The Culturalogic of Mixed Drinks. Report for Schieffelin & Somerset and United Distillers, New York, NY. 82 pp.

Polanyi, Michael. 1958. Personal Knowledge: Towards a Post-Critical Philosophy. Chicago: University of Chicago Press.

Number of visitors to this page: