Richard A. Shweder
Department of Comparative Human Development
University of Chicago
Richard A. Shweder is the William Claude Reavis Distinguished Service Professor of Human Development at the University of Chicago and a Carnegie Scholar. This essay expands on remarks delivered on June 11, 2005 at Columbia University at a Festschrift in honor of Walter Mischel titled “Toward a Science of the Person: Paradigm Change in Psychological Models of Human Nature” (organized by Yuichi Shoda and Daniel Cervone)
For those who have a taste for the conceptual foundations of personality theory 1968 and 1973 were vintage years. All the fun began, especially for those who imbibe explanatory theories of individual differences in behavior, when Walter Mischel published his monumental book-length critique of trait psychology titled Personality and Assessment and then followed it up with a (now classic) Psychological Review publication in which he aimed to “reconceptualize personality.” In my brief essay I want to say a few words about what I take to be the pre-eminent take home message of Mischel’s 1973 re-conceptualization.
Thus in a moment I will explicate and react to a few quotations from his classic essay. Mischel’s main message, as I interpret it, is this: Explanatory (and predictive) success in both personality psychology and social psychology requires that both disciplines revise their most basic (and shared) explanatory model, in particular, the “person/situation” model; that explanatory model is based on the assumption that all human behavior is a joint function of two mutually exclusive (separable, independent) and exhaustive causal factors, namely, inside the person vectors (such as global personality traits) X outside the person situational demands. As an alternative both disciplines should seriously consider adopting a “preference/constraint” model of explanation, which is based on the assumption that all variations in human behavior (including individual differences in behavior) are a joint function of variations in individual preferences (ends or goals) X variations in constraints (means of various sorts) as mediated by human understanding and human agency. Special attention should be given to purposeful strategic actions motivated by our capacity for instrumental or means/ends reasoning and other aspects of human rationality.
However, before I develop that theme – which is a story about Walter Mischel’s invitation to substitute a “preference/constraint” model of explanation for a “person/situation” model - permit me to make a comment about one aspect of his famous book. To do so, even minimally, requires that I initially hazard a definition of the concept of an abstract mental state, because the abstract mental states of personality trait psychology are the central objects of his critical concerns.
Mischel’s Critique of Abstract Mental State Concepts
By a mental state concept (such as “the feeling of anxiety” or “the desire to be dependent” or “valuing honesty”) in contrast to a physical state concept (such as “mass”, “weight” or “gravity”) or a mathematical state concept (such as “zero”) I mean a concept that refers to or is about something a person knows, wants, thinks, feels or values as good or bad (for example, that they feel afraid, or that they want others to support and take care of them rather than to do things for or by themselves). By an abstract mental state concept I mean a mental state concept that refers to something a person might know, want, think, feel or value (for example, feeling afraid) apart from any particular or concrete instance of it (for example, feeling afraid to take financial risks on the stock market or feeling afraid of frogs).
In Personality and Assessment (1968) Mischel challenged his readers to explain why, when ecologically valid behavior observational evidence is available for examination, particular or concrete instances or manifestations of the same abstract mental state concept do not typically co-occur to even a moderate degree across individual differences in mental functioning and behavior. He brought to everyone’s attention a provocative pattern of empirical findings, which might be illustrated as follows: The person who is more likely than others to get angry when contradicted in an argument is not typically the person who is more likely than others to get angry when cut in front of in a line. The person who feels more anxious than others while sitting in a dentist chair is not typically the person who is more likely than others to feel anxious when standing in a crowded elevator. The child who is more likely than others to seek help from his or her teacher is not typically the child who is more likely than others to seek help from his or her peers. The financial risk-taker is not typically a social risk taker; the intellectual risk-taker is not typically a physical risk taker, and vice versa, all the way around.
Personality theory, when viewed as a theory of individual differences, is concerned to describe and explain those mental states or processes that are both relatively stable in, and distinctive of, an individual person. By design Mischel’s provocative book invited us to have doubts about the role and value of abstract mental state concepts (e.g., the need for affiliation) in predicting and accounting for variations in behavioral outcomes (e.g., the likelihood that someone is going to seek or take advantage of this or that particular opportunity for social contact). In effect one finished the book pondering the following question: if abstract mental states do not organize and motivate individual behavior then precisely what types of mental states do?
Some Critical Responses to the Critique
Mischel’s original (1968) critique of abstract mental states and global personality trait theory provoked many responses, including objections and calls for clarification. In this essay, I plan to focus entirely on what I take to be the major point of his Psychological Review (1973) clarification. Thus, this is not an occasion on which I will address the objections, some of which I consider sound, logical and legitimate. For example, I agree that Mischel’s doubts about the role and value of abstract mental state concepts as explanatory concepts do not necessarily follow from the empirical finding that particular case-based and context-embedded instances of an abstract mental state (for example, taking financial risks in poker, taking intellectual risks in a debate, taking social risks in dating, taking physical risks on a ski slope) do not typically correlate in an impressively positive direction or hang together as a behavioral package deal across individual differences in (for example) risk-taking behavior.
The philosopher William Alston (1975, p. 22) surely has a point when he writes (with regard to mental state concepts such as needs, desires, expectations, interests, values, attitudes and abilities) that “…an ability is typically manifest in its exercise, a need in efforts to satisfy it, a favorable attitude in actions directed to promoting or benefiting its object. Nevertheless it is not part of what we mean in attributing (even a strong degree) of some ability, need, or attitude that such manifestations will frequently occur. A person may have abilities he rarely exercises. Thus a man may be a crack pistol shot, but because he doesn’t have a pistol or because of lack of interest, rarely exercises his ability. A person may have a strong need for close relationships, but because of fear of rejection rarely or never seeks to satisfy it.”
Or, taking this type of critical response one step further, consider a personality descriptive sentence of the form “So and so is disposed to act [here name a dispositional quality associated with an abstract mental state concept: for example, aggressively] in social situations”. One agrees with Alston that an accurate translation of that sentence might state that he or she is more likely than not to do something or other that is aggressive, and that the sentence does not imply that he or she will not be selective in the type of way that they aggress.
Yet Alston’s particulars seem particularly apt to me, in part because they exemplify some potential explanatory features of a preference/constraint model of precisely the sort that Walter Mischel himself endorses (or so I shall argue below). ). In general, a preference/ constraint model of explanation allows that variations in behavior might be caused either by variations in constraints (holding preferences constant) or variations in preferences (holding constraints constant), or both. It is noteworthy that the field of micro-economics is defined in part by a specialized (and rather bold) version of a preference/constraint model of explanation according to which one tries ones best to explain all variations in behavior as instrumental (rational means-ends) responses to variations in constraints (available material and symbolic resources, social capital, personal skills, causal beliefs) based on the assumption that (at some very high order of abstraction) the ordering of preferences (the desired ends or goals) for members of a population can be viewed as shared and stable and that any supposed “end” can always be re-interpreted as a means (to some other end) ad infinitum (thus making it possible for all behavior to be explained in instrumental terms).
Perhaps it is debatable whether that specialized economic application of a preference/constraint model is always justified, based as it is on that assumption that all differences in behavior are adaptive responses to differences in constraints and thus should be interpreted as by-products of the application of instrumental reasoning in the fulfillment of fixed preferences or ends. Mischel himself appears to embrace the preference/constraint model of explanation in its more general or inclusive sense, whereby changes in either preferences or constraints might be invoked to explain changes in behavior. Nevertheless, it is worth noting that at least one powerful way to explain inter-individual and intra-individual variations in behavior is by reference to instrumental means-ends thinking and by focusing theoretical attention on variations in “constraints” (the means that are or are not available for attaining some given end). Alston’s examples, the marksman who lacks the physical resource (the pistol) to make manifest his skill, or the person whose need for close relationships is rarely exhibited in behavior because he or she subscribes to the substantive causal belief (which is itself a “constraint” on that person’s behavior) that social initiative results in rejection, focuses our attention in precisely that way.
Dispositional and Non-Dispositional Representations of the “Person”
Alston raises many other points with regard to Mischel’s critique of personality trait theory. For example, he is uncertain whether the critique is merely an objection to the generality or breadth of the proposed abstract mental state concepts in the personality literature (“risk-taker” rather than “financial risk taker in a game of poker”; “extroverted” rather than “likes to talk before groups of wealthy patrons”) or whether it is also an objection to the tendency to characterize individual differences (whether general or specific, broad or narrow) exclusively in “dispositional terms”. Alston notes that to say that a person has a certain “disposition” (e.g., conscientiousness) is to assert that if he or she is in a certain type of situation he or she will emit a certain type of response (plus the minimal assumption of all personality theory that there must be some mental state or other that is either recurrent or relatively stable in that person’s constitution causing the emission of that type of response). Quite significantly, Alston notes that a “personality” can be conceptualized in either dispositional or non-dispositional terms.
In clarifying the distinction between dispositional versus non-dispositional representations Alston uses the example of the “fragility” of a physical object. Thus, he states, the quality of “fragility is construed dispositionally when we think of x’s being fragile simply as the truth of the hypothetical proposition, ‘If x is struck sharply x will break.’ But if we can discover what micro-structure of x is responsible for the truth of that hypothetical proposition, we can think of the same attribute in non-dispositional terms.” Applying that distinction to theories of personality one might suggest that a “person” can, in theory, always be represented “dispositionally”: as a structured list of response tendencies to behave in certain ways in certain types of situations. Alternatively, a “person” can, in theory, always be represented non-dispositionally: as an “agent” with a latent preference/constraint cognitive microstructure – a mental being motivated by his or her desires/goals, beliefs, feelings, values, skills and other resources, whose behavior is the outcome of choices made in the pursuit of valued ends in the face of some limited set of means for (or constraints on) the reaching of those ends.
Construed in those terms, one way to formulate my view of the main take home message of Mischel’s reconceptualization of personality theory is as follows. Namely, that once we are able to produce a valid non-dispositional representation of the latent cognitive micro-structure of human behavior (for example, in terms of goals, causal beliefs, expectations of success) it will become obvious why dispositional representations of individual differences in observable behavior in particular situations can’t rely on abstract mental state concepts (such as an abstract ability to resist temptation/exercise self-control) if they are to be useful from a predictive point of view. Why that should be obvious will become more obvious once I have a chance to be a bit more explicit about Mischel’s re-conceptualization of personality theory, but it has something to do with the explanatory importance of what Clifford Geertz (1973) (following Gilbert Ryle) has called “thick description.” The more detailed the representation of a person’s preferences and constraints the more apparent it becomes why (e.g.) two occasions to take risks either did or did not result in the same level of risk taking behavior.
A Final Preliminary Remark: The Systematic Distortion Hypothesis
One final preliminary remark is also probably in order. My original interest in Walter Mischel’s critique of global personality trait theory arose in the context of language and thought studies in anthropology, where Roy D’Andrade had pointed to personality trait theory as an example of the confusion of propositions about language with propositions about the world. In this essay, however, I do not plan to address (what came to be called) the “systematic distortion hypothesis” or discuss in any detail its connection to debates about the evidence (or lack of it) supporting a global trait theory of individual differences.
Roy D’Andrade first proposed “the systematic distortion hypothesis” in 1965 and both he and I were working on the topic between 1968 and 1973 when Mischel’s critique of global personality traits burst forth on the academic scene (e.g., D’Andrade 1965, 1974; Shweder 1975, 1981; Shweder and D’Andrade 1981, 1987). In effect, the “systematic distortion hypothesis” complemented Mischel’s critique by providing a “method effect” explanation for much of the existing (spurious or pseudo?) evidence in the personality psychology literature that purports to support global trait theories.
In the personality psychology literature the inter-item covariance clusters, factors, syndromes, dimensions or traits (for example “extroversion”: active, outgoing, assertive, gives parties a lot, likes speaking before groups) so frequently discovered by means of correlation analyses of subject response patterns on recall based personality rating tasks are typically interpreted as evidence of the existence of broad or global personality dispositions in the organization and motivation of individual behavior. The “systematic distortion hypothesis” proposes an alternative interpretation. Namely that (a) error in judgment, inference and memory on personality rating tasks is not only sufficiently frequent to influence the structure of inter-item co-variations but also systematically biased in the direction of pre-existing highly schematic and oversimplified understandings about how language and reality are organized (for example, the folk understanding that “friendliness” and “aggression” do not “go together” while “dominating” and “aggressive” do), and (b) those schematic understandings (many of which are derived from lay notions about the components of meanings of words such as “friendly”, “aggressive”, “dominating”) are not very good maps of actual inter-item co-variations across individual differences in human mental states and behavior (for example, friendliness and aggressiveness are often positively correlated in actual behavior; and dominance hierarchies may actually reduce the frequency of aggressive behavior).
The following claim is common to both the “systematic distortion hypothesis” and Mischel’s critique of personality trait theory in his 1968 book: the closer one gets to ecologically valid on-line evidence of mental functioning and behavior the less it looks broadly trait-like in its organization. Mischel critique brought to our attention the rather low or insignificant inter-correlations in actual behavior among different instances of the same abstract mental state, while the “systematic distortion hypothesis” treated as dubious and problematic (“halo effect” like) the rather high positive inter-correlations among specific abstract mental state indicators on memory based personality rating forms. So much for preliminary remarks (which at this rather advanced point in this essay are not so preliminary)!
“This Is It”: From Persons and Situations to Preferences and Constraints
I have recently located in an ancient filing cabinet an old and very marked up personal copy of Mischel’s 1973 classic. Some decades ago I had scribbled the phrase “This is it” at several places in the margins of the text. In identifying the central message of Mischel’s re-conceptualization of personality theory I am going to focus on a few of those places in his essay where I had written “This is it”.
In his famous Psychological Review article Mischel (1973: 255) wrote: “Is information about individuals more important than information about situations? The author has persistently refrained from posing this question because phrased that way it is unanswerable and can only serve to stimulate futile polemics. Moreover, in current debates on this topic, ‘situations’ are often erroneously invoked as entities that supposedly exert either major or only minor control over behavior, without specifying what, psychologically, they are or how they function.”
I have already previewed Walter Mischel’s main “reconceptualization” message. It is not primarily that abstract mental states are weak (rather than strong) causal factors in the explanation of individual differences in behavior. His message is much more profound because it calls for a shift in meta-theoretical assumptions about the nature of the “person” in both personality and social psychology (see Shweder and Sullivan 1990). His pre-eminent message is that there are good reasons for moving from a model of explanation in which behavior is assumed to be the by-product of “person” variables and “situational” variables to a model of explanation in which behavior is assumed to be a by-product of “preferences” and “constraints.’ The first explanatory model (behavior is a function of person X situation) conceptualizes the person as a vessel for autonomous mental states or response dispositions. The second explanatory model (behavior is a function of preferences X constraints) conceptualizes the person as a purposeful agent whose behavior in any context can best be explained by reference to aspects of human rationality, including instrumental means- ends thinking.
There in the above quotation Mischel expresses some forebodings about the person/situation distinction. Not only does he avoid trying to answer an unanswerable question (about the relative important of person versus situation in the control and regulation of behavior); he also suggests why the question is unanswerable. Namely that the very distinction between person/situation (which of course pre-supposes that all relevant causal forces are either person forces or situational forces but not both) makes it impossible to coherently pose the really key question, which he tries to formulate by asking: “how do situations function psychologically?”
Notice that by posing the question that way the very fabric of the contrast between a person and a situation begins to unravel, because the “situation” is now at least partly a mentally mediated fact (inside the person) and “the person” (including the “meaning of things” to that person) is not fully separable from one’s description of the situation (which must now be represented or described in “person-centered” mental terms and no longer can be said to be “external”). In asking how situations function psychologically Mischel reveals that he is reaching for new units for causal analysis that circumvent the person/situation divide. Those units of analysis (describing the non-dispositional cognitive micro-structure of personality) are indeed different (and paradigm shattering) because they derive from a preference/constraint perspective on the explanation of behavior. The irony is that, within the terms of the person/situation meta-language, the call to study how situations function psychologically may appear to incoherently require the dissolution of the very definitional boundary between person and situation that is pre-supposed by the person/situation model of explanation. Indeed it is that irony in Mischel’s formulation which suggests the need for a new meta-language.
2. This Is It: Preferences and Constraints, Thick Description and the Native Point of View.
Here are three more quotations from Mischel’s re-conceptualization of personality essay (1973: 259, 259, 270).
“Rather than argue about the existence of ‘consistency,” it would be more constructive to analyze and study the cognitive and social learning conditions that seem to foster – and to undermine – its occurrence.”
“When the probable reinforcing consequences to the person for cheating, waiting, or working differ widely across situations depending on the particular task or circumstances, the behavior of others, the likelihood of detection, the probable consequences of being caught, the frustration induced, the value of success, etc., impressive generality will not be found. Conversely, when similar behaviors are expected and supported in numerous situations, consistency will be obtained. Because most social behaviors produce positive consequences in some situations but negative ones in other contexts, the relatively low associations found among an individual’s response patterns even in seemingly similar situations should not be surprising.”
“One type of expectancy concerns behavior-outcome expectancies under particular conditions. These behavior-outcome expectancies (hypotheses, contingency rules) represent the ‘if___; then ___’ relations between behavioral alternatives and probable outcomes anticipated with regard to particular behavioral possibilities in particular situations. In any given situation, the person will generate the response pattern which he expects is most likely to lead to the most subjectively valuable outcomes.”
And let me add to this list one other quotation from a Mischel and Mischel essay on morality and self-regulation (1976): “Even the noblest altruism still depends on expected consequences, although the consequences are often temporally distant, are not in the immediate external environment, and are not easily identified, and reside in the actor himself rather than in social agents.”
I can now conclude this essay by merely explicating the obvious. In Mischel’s 1973 reconceptualization of personality the “reconceptualization” is based on a preference/constraint meta-language and model of explanation. He moves away from the received approach in personality and social psychology that tries to predict and explain behavioral outputs from a combination of external situational forces and internal personality traits. He opts instead for a preference/constraint meta-theory of human motivation and tries to predict and explain behavior from a combination of ends or goals of the agent and available cognized means and resources, etc. All personality "traits", dispositions, or response mode summaries, whether broad or narrow (from a cross-situational or cross response mode point of view) become grist for a non-dispositional exegesis. You don’t describe a person as “clumsy”; instead you give an account of his or her preferences (what if the goal on that occasion was to look like a clown?) and constraints (for example, the skills he or she lacks or possesses to navigate smoothly through that particular environment). In effect, observed behavior is rendered intelligible through an interpretative analysis which makes reference to what the person knows, thinks, feels, wants and values (as good or bad) as a meaning-maker –the “semiotic subject” of personality theory (Shweder and Sullivan 1990). Once the agent of behavior (and subject of personality theory) is thus conceptualized in terms of his or her preferences and constraints, the instrumental rationality and the means-ends thinking of both the actor and the personality theorist come to play a central role in explanations of behavior. As a personality theorist one is encouraged to construct an accurate non-dispositional representation of the micro-structure of human personality, which, as noted earlier, is very likely to be a “thick description” of preferences and constraints (including “behavior-outcome expectancies”, available skills, causal beliefs, etc) from a “native point of view”. Speaking as a cultural anthropologist and cultural psychologist who regularly attends the economist’s ‘rational choice” workshop at the University of Chicago it seems to me that all this makes very good sense!
Alston, W.P. (1975). Traits, Consistency and Conceptual Alternatives for Personality
Theory. Journal for a Theory of Social Behavior, 5:17-48.
D’Andrade, R.G. (1965). Trait Psychology and Componential Analysis. American
D’Andrade, R.G. (1974). Memory and the Assessment of Behavior. In T. Blalock
(Ed.), Social Measurement. Pp. 159-186. Chicago, Illinois: Aldine-Atherton.
Geertz, C. (1973). Thick Description: Toward an Interpretive Theory of Culture. In C.
Geertz, The Interpretation of Cultures. Pps. 3-32. New York: Basic Books.
Mischel, W. (1968). Personality and Assessment. New York: Wiley.
Mischel, W. (1973). Towards a Cognitive Social Learning Theory Reconceptualization
of Personality. Psychological Review, 80:252-283.
Mischel, W. and Mischel, H.N. (1976). A Cognitive Social Learning Approach to