Home Introduction Cognitive Psychology Cognitive Perspective Social Perception Social Memory Social Categorization Social Judgment Language Automaticity Self Social Neuropsychology Personality Social Intelligence Development Sociology of Knowledge Social Construction Conclusion Lecture Illustrations Exam Information

The Cognitive Perspective on Social Interaction

Link to an integrated overview of personality and social psychology.

Link to an overview of the traditional psychology of personality.

The cognitive perspective on social interaction begins with the assumption -- actually, more like an axiom -- that humans are intelligent creatures.  We do not behave merely by reflex, taxis, instinct, and conditioned response.  Rather, our behavior is a response to the meaning of the stimulus, and reflects active cognitive processes of perceiving, learning, remembering, thinking, and communicating through language.  But humans are also social creatures. Our experiences, thoughts, and actions take place in an explicitly social context of cooperation, competition, and exchange, family and group memberships, institutional, social, and cultural structures.  For that last reason, psychologists need to understand the relations between psychological processes within the individual and social processes that take place in the world outside. 

This cognitive perspective is echoed in a famous line from Harper Lee's To Kill a Mockingbird (1960).  The novel involves a widowed Southern lawyer, Atticus Finch, and his two young children, Scout and Jem.  After a particularly difficult first day back in school, Atticus tells Scout:  "You never really understand a person until you consider things from his point of view -- until you climb into his skin and walk around in it". 

We'll see a little later how understanding the subject's point of view can shed new light on a certain classic psychological experiment.

While much of cognitive psychology is concerned with how the individual acquires, represents, and uses knowledge in perception, memory, thought, and language, social psychology is concerned with the role that cognitive (as well as emotional and motivational) processes play in social interactions -- between individuals, and between individuals and groups.  As such, social psychology is not just the study of mind in social relations -- it is also very much the study of mind in action.


A Capsule History of Social Psychology

Social psychology is part of psychology, and the history of psychology begins with declarations of its impossibility as a science.  In the 17th century, Descartes had drawn a sharp distinction between humans and animals -- humans had minds, whereas animals operated solely by physical (physiological) mechanisms like the reflex.  But according to Descartes' doctrine of substance dualism, mind was made of an immaterial substance.  In the later 18th century, Kant drew the inescapable conclusion:

Kant may have been the smartest man of his time, but his assertions about psychology were quickly disproved, in the 19th century, by studies of sensation and perception. 

So, by 1867-1868, a full-fledged science of psychology was off and running.

Wundt had held that an experimental, quantitative psychology had to be limited to "immediate" experience, by which he meant sensation and perception.  But very quickly, the domain of experimental research in psychology expanded:

Many of these experiments involved only a few subjects.  However, researchers began to recognize important individual differences in mental function:

While Kant had declared psychology an impossible science, Wundt doubted that topics in social psychology could be studied with the new science.  Wundt distinguished between Naturwissenschaft, including studies of sensation and perception, and Geisteswissenschaft including everything else.  He also distinguished between experimental psychology, which could be accomplished in the laboratory, and Volkerpsycholgie, which had to be based on uncontrolled field studies.  Just as the psychophysicists had proved Kant wrong, and just as Ebbinghaus and Hull had proved Wundt wrong about higher cognitive functions, Wundt was to be on the losing side again.

The arc begun by the psychophysicists' studies of sensation and perception was completed by Sherif's studies of social influence on the autokinetic effect (1935), and Asch's (1956) studies of conformity in perception.

As experimental research in personality and social psychology began, specialized societies were formed to encourage this work, specialized journals were established to publish it:

Following McGuire (1986), we can distinguish four major periods in the history of social psychology:

Throughout all this history, we can identify several milestone crises and challenges:

Since the mid-1980s, social psychology has been characterized by a series of reactions to the "Cognitive Revolution":


The Domain of Social Psychology

Over the years, social psychology has been defined in various ways:

The definition I favor is broader than any of these:  

Social psychology is the study of the relation between
the individual's mental structures and processes and
structures and processes in the social world outside the individual.

Following Gleitman (1980), we can further identify four distinct domains of social interaction:

There are, in fact, two rather different versions of social psychology -- one as practiced by psychologists, the other as practiced by sociologists.

Sociological social psychology tends to be more popular in Europe, while psychological social psychology tends to be more popular in America.  In fact, with its emphasis on the individual, psychological social psychology has a distinctly American flavor to it.


Symbolic Interactionism

The cognitive perspective in social psychology has its origins in symbolic interactionism, a term coined by Herbert Blumer (1937, 1989), a student of George Herbert Mead who  founded the sociology department at UC Berkeley.  In Blumer's view, symbolic interactionism rests on three premises: 

In symbolic interactionism, interactions are "symbolic" because they occur in the person's head, symbolically, before they occur in reality, in the world outside the person.

As Blumer makes clear, symbolic interactionism itself is rooted in the work of Blumer's mentor, George Herbert Mead, author of the seminal treatise Mind, Self, and Society (1934).  A number of Mead's concepts, as described by Blumer (1989), will illustrate the connection:

Mead's contribution to the cognitive perspective on social interaction is clear in his assertion that "If a thing is not recognized as true, then it does not function as true in the community (1936, p. 29).  What matters, then, is not so much what is true, or what exists; what matters is what is believed to be true, and what is believed to exist.


Other Cognitive Forebears

But the cognitive perspective has roots that go back even further than Blumer and Mead, to what R.K. Merton (1976) has dubbed The Thomas Theorem, which appeared in a book on adolescence by William Isaac. Thomas and Dorothy Swain Thomas (1928, p. 529):

"If men define situations as real, they are real in their consequences."

Merton (1976, p. 174) has called this quote "Probably the single most consequential sentence ever put in print by an American sociologist".

Although The Thomas Theorem is properly attributed to Thomas and Thomas writing together, its essence had been articulated by W.I. Thomas, writing alone, some five years earlier (Thomas, 1923, p. 42-43):

"Preliminary to any self-determined act of behavior there is always a stage of examination and deliberation which we may call the definition of the situation.  And actually not only concrete acts are dependent on the definition of the situation, but gradually a whole life-policy and the personality of the individual himself follow from a series of such definitions."

Similarly, Theodore Newcomb, discussing the findings of his pioneering study of the consistency of social behavior across situations, which actually found precious little consistency, attributed the individual's behavior in a particular situation to his beliefs about that situation (1929):

"There are always slight differences in both internal and external stimuli which are important in determining behavior, yet are not recordable....  situations are necessarily so different that large measurable consistency is not to be expected" (pp. 77).

"To cite an obvious example, whether or not Johnny engages in a fight may depend on whether or not he thinks he can 'lick' his opponent" (p. 39, emphasis added).

The cognitive perspective in psychology was neatly summed up by a British psychologist, Sir Frederick C. Bartlett, in his critique of classical psychophysics and of Ebbinghaus' research on memory:

"The psychologist, of all people, must not stand in awe of the stimulus."

For Bartlett, perception was not merely the analysis of a stimulus object or event; rather, perception involved the construction of a mental representation of the stimulus.  And memory was not merely the reproduction of some past event; rather, remembering involved the reconstruction of that event.  Both construction and reconstruction involved "higher" cognitive activities such as reasoning, inference, judgment, and problem-solving -- what Bartlett called "effort after meaning".  


The Behaviorist Revolution

Watson.JPG (99908 bytes)But that was then.  Just as social psychology was beginning to get going, the behaviorist revolution initiated by John B. Watson (1913, 1919) took hold in psychology in general, and in social psychology in particular.  Whereas William James had defined psychology as the science of mental life, Watson saw a conflict between the private and subjective nature of mental life, and the requirement of science for objective, publicly observable facts.  Accordingly, he redefined psychology as a science of behavior, and restricted a scientific analysis to publicly observable environmental stimuli and publicly observable behavioral responses to them. 


For Watson:

Skinner.JPG (104992 bytes)The behaviorist perspective on human behavior, including human social behavior, can be summarized by a Doctrine of Situationism expressed most vigorously by B.F. Skinner. 

For Skinner, as for Watson:



            (128810 bytes)The behaviorist perspective was quickly embraced by social psychology, particularly in an important textbook by Floyd Allport (1924).

F. Allport's behaviorism is complete: people respond to stimulation by other people, and the behavior of groups is the sum of the behavior of the individuals within those groups.

Very quickly, and especially in the years after World War II, social psychology -- and especially American social psychology -- evolved as a variant on functional behaviorism.  In this way, environmental control was equated to stimulus control.  Cross-situational variability, not cross-situational consistency, was to be expected in behavior, depending on the individual's reinforcement history, and on the conditioned stimuli and discriminative stimuli present in the environment.  Throughout this period, the emphasis of (American social psychology) was on the situational control of individual behavior -- by which social psychologists meant the objective situation, not the situation as defined, as cognitively constructed, by the individual.

            (123778 bytes)This can be seen in the classical definition of social psychology, offered by Floyd Allport's younger brother Gordon (G. Allport, 1954, p. 5):



"With few exceptions, social psychologists regard their discipline as an attempt to understand and explain how the thought, feeling, and behavior of individuals are influenced by the actual, imaged, or implied presence of other human beings.... [S]ocial psychology wishes to know how any given member of a society is affected by all the social stimuli that surround him."

G. Allport was not a strict behaviorist: he was interested in thoughts and feelings as well as in behavior.  But still, the behaviorist view can be seen in his emphasis on thoughts, feelings, and behaviors as responses to stimuli impinging on the individual from the external social environment.


Gestalt Psychology and Field Theory, Balance and Dissonance

Whereas the behaviorist approach conceived of social behavior as a more-or-less mechanical (conditioned or unconditioned) responses to stimuli in the social environment, the first glimmerings of a cognitive approach began to emerge in the 1950s. 

The first reaction to the behaviorist viewpoint came by way of Gestalt psychology, a movement led by Kurt Koffka, Wolfgang Kohler, and Max Wertheimer, and which arose in Europe as a reaction to the atomism of classical 19th-dentury Structuralism, with its emphasis on stimulus determination/  But Gestalt psychology also had appeal in opposition to early 20th century behaviorism, especially of the kind espoused by Watson, with its atomistic description of both stimulus and response.   Gestalt, of course, roughly translates as "whole configuration", and the Gestalt theorists focused on the tendency of the mind to organize individual stimuli into groups or sets -- in broader terms, to fuse individual stimulus elements into a perceptual whole.  From a Gestalt point of view, we cannot analyze perceptual experience into its elementary constituents (as the Structuralists sought to do), because the individual elements itneract and combine with each otehr in such a way that "the whole is different than the sum of its parts".  The Gestalt principles of perception, such as proximity, similarity, and symmetry made it clear that perception was not determined solely by the stimulus, but also by internal processes.  Similar principles were applied to memory, such as the von Restorff effect, that memory is better for stimuli that stand out against their background; and also to thinking, such as Kohler's own studies of insight in problem-solving.  Gestalt psychology, along with (honest!) psychoanalytic object-relations theory, kept interest in cognition alive during the dark days of behaviorist hegemony -- and what it did for "experimental" psychology, it also did for personality and social psychology.

In fact, by the mid-1950s cognitivism was visible enough that Martin Sheerer was commissioned to write a whole chapter on the approach for the first edition of Lindzey's Handbook of Social Psychology (1954).  Scheerer is largely forgotten now, but he and Kurt Goldstein had published an important monograph on Abstract and Concrete Behavior (1941), in which they distinguished between two types of behavior, abstract and concrete, in turn dependent on corresponding abstract and concrete attitudes -- which, in turn they construed as "capacity levels of the entire personality" (p. 1).  People have abstract or concrete attitudes, in differing degrees, and these attitudes determine how they will behave.

Lewin's Field Theory 

The Gestalt viewpoint was initially brought into social psychology by Kurt Lewin, especialy in the papers from the 1930s and 1940s collected as Field Theory in Social Science (1951).  Lewin argued for a dynamic psychology in which behavior was determined by various psychological forces (Lewin was especially interested in conflict.  For our purposes, the most important idea is that of the Life Space, which consists of the "Gestalt" of the person and the psychological environment -- by which Lewin meant not the physical environment, as it might be described by the behaviorists, but rather the perceived environment -- or, better yet, the meaning of the environment, which of course was grist for the cognitive mill. 

Heider's Theory of Interpersonal Relations

A more explicitly cognitive take on social interaction was supplied by Fritz Heider in his Psychology of Interpersonal Relations (1958).  Heider agreed with Lewin about the interdependence of the person and the situation, and that what was important about the situation was how the person viewed it.  He focused on "common-sense psychology, or what we would now call folk psychology -- the views about mind and behavior held by ordinary people on the street, as opposed to the scientific theories developed by professional psychologists.  After all, he argued, it's folk psychology that determines our behavior toward other people; just as important, Heider argued that folk psychology is also, often scientifically correct.

Heider agreed with Brunswik (1934), whom we shall discuss more fully in the lectures on Social Perception, that person  perception was governed by the same principles as the "impersonal" perception of nonsocial objects.  At the same time, he argued that persons have properties that are not possessed by impersonal objects such as tables and chairs: these include abilities, emotions, intentions, wishes, sentiments, purposes, and other aspects of mind.  Person perception is the process by which we perceive these qualities in other people.followed   Moreover, Heider aknowledged that, in the social case, the object of perception, who is a sentient being like the perceiver him- or herself, is perceiving the perceiver in turn.

The Perceiver (P) is perceiving the Other (O), who is perceiving P in turn.

The big concept with Heider is phenomenal causality (Heider, 1944), a topic which we'll discuss further in the lectures on Social Judgment.  In social perception, we are trying to understand another person's social behavior.  Our understanding will determine how we behave toward that person.  It doesn't matter so much what really caused the behavior -- that is a topic for scientific psychology.  What matters is what we perceive, or believe, caused the behavior.  Therefore, it's important to understand causality from the perceiver's point of view.

In social perception, the object of perception is also a sentient being, and P is always aware that O's perception will have an effect on P (by determining O's behavior).  O's perceptions, and the behavior that follows from them, will affect P in three different ways:

Heider is sometimes called the "father of situationism" in social psychology (e.g.  Ross & Nisbett, 1991), but he's not.  As I'll make clear later, Heider, like Lewin,  argued that the person and the situation constituted an interdependent whole.  If "the situation" means the psychological situation, as perceived by the individual (Ross & Nisbett acknowledge this), and if the perceived situation is not determined wholly by the stimulus input, then, at the very least, the person makes just as important a contribution to behavior as the situation does. 

Heider's Balance Theory

Heider also initiated an important theoretical tradition in social psychology known generically as balance theory (1946).  All of the balance theories assume that people attempt to achieve consistency among their cognitions, defined broadly to include beliefs, knowledge, expectations, attitudes, and all sorts of other internal mental states and dispositions of the sort that were thoroughly rejected by the behaviorists. Similarly, any inconsistencies among cognitions are assumed to be affectively aversive, leading to various cognitive maneuvers intended to reduce the discrpancies and the consequent negative feelings.

Heider's own theory was called the p-o-x theory, because it dealt with three elements: the attitude of a person (p) toward another person (o) and an object (x) belonging to o or is related to o in some way..  Heider understood that we have lots of attitudes towards lots of people, and things, but he argued that we try to impose some order on this vast network of attitudes.  In line with Gestalt theory, he argued that p, o, and x constituted a unit or a Gestalt-like whole bound together either by unit relations like family, nationality, or gender; or by sentiment relations such as liking, admiration, and approval.  Both sorts of relations tend toward harmony.

More generally:

When a state of imbalance occurs, there are several things that p can do to set things right (remember, balance or imbalance is always defined with respect to p).

Festinger's Theory of Cognitive Dissonance

Heider's was the first balance theory, but there were others, including Newcomb's (1953) A-B-X System and Osgood and Tannenbaum's (1955) congruity theory.  But the most famous and influential balance theory, by far, was Festinger's (1957) theory of cognitive dissonance.  Of all the balance theories, this is the one that generated the most experimental research.  And it's the only one that's still presented in introductory textbooks (most of them, anyway).  Me, personally, I never understood what all the fuss was about.

Festinger's theory is different from the others, because it purports to be a general cognitive theory, not concerned specifically, or solely, with social interaction.  Again, by "cognitive" Festinger really means to refer to any internal mental state -- not just expressly "cognitive" states of knowledge and belief, but also "emotional" states like attitudes and feelings and "motivational" states like habits and desires. 

Festinger adopted the terms consonance and dissonance to describe the relations between pairs of cognitive elements, A and B.

Some of the most famous evidence supporting dissonance theory comes from studies using the forced compliance paradigm, in which subjects are asked to do something that contradicts their personal attitudes or beliefs.  To take a classic example, Festinger and Carlsmith (1959) engaged subjects in tasks that were extremely boring, such as turning a series of knobs for precisely 1/4 turn.  After completion of their task, the subjects were then asked to tell the next subject, newly arrived form the experiment, that the tasks involved were very interesting (which, decidedly, they were not).  Some subjects were given $20 to lie in this way, while others were given only $1.  The lie was intended to induce cognitive dissonance. 

Here's why this happened, according to cognitive dissonance theory.

These notions of discounting and insufficient justification played a big role in certain theories of causal attribution, as discussed in the lectures on Social Judgment.

Note:  Alternative interpretations of the F&C experiment, and others like it, are always possible.  For example, in a resurgence of behaviorism, Bem's (1967, 1972) self-perception theory denied that subjects in the low-payment condition changed their attitudes, because it denied that they had any attitudes to begin with.  According to self-perception theory, people don't have attitudes stored in memory, as it were, ready to be retrieved when asked their opinion about some topic.  Instead, they compute what their attitude must be, on line, based on their observations of their own behavior.  In the large-payment condition, the behavior could be discounted as relevant to one's attitude, much as Festinger argued: it is sufficient justification to explain the behavior.  However, because the smaller payment was insufficient to justify the behavior, the subjects therefore inferred that they had the corresponding attitude -- i.e., that the tasks really were engaging and exciting.

A similar sort of argument was made by Lepper, Greene, and Nisbett (1973) in a seminal experiment on intrinsic motivation.  This experiment is discussed more fully in the General Psychology lecture on Motivation.

Attitudes, Communication, and Persuasion

The Festinger & Carlsmith study was about belief (concerning the tasks), but it could just as easily be taken as a study of attitude -- which many theorists, following Allport (1935), have argued is the central concept in social psychology, distinguishing it from other subfields (I don't think that's true, but I respect those who do).  That is, the subjects may have come to believe that their tasks were engaging and interesting; or they could have changed their attitude towards the tasks, from negative to positive.  It works either way.  But as a matter of historical interest, the balance theories, and especially Festinger's theory of cognitive dissonance, led to an important shift in the study of attitudes. 

McGuire (1986) has traced three (actually, four) periods of social-psychological research on attitudes.,

  1. The 1920s-1930s were devoted to developing techniques for attitude assessment, by means of interviews and questionnaires, by Thorstone and Chave (1929) and Likert (1932 -- he of the Likert scale), among others.
  2. What were social psychologists doing in the late 1930s-1940s?  McGuire calls this the group dynamics interlude, focused on intro-group and inter-group dynamics.
  3. Research in the 1950s-1960s focused on persuasion and attitude change: this work was greatly influenced by balance theory, and especially by dissonance theory.
  4. From about 1965-1985 interest in attitudes diminished during what McGuire called the "cognitive interlude".
  5. In the 1980s-1990s (MGuire was projecting here), research was dominated by the question of attitude structure, represented by research by Judd and Milburn (1980) and Button et al. (1993).

Attitudes are "cognitive" in the broad sense that they are internal mental states that dispose people to respond in particular ways to particular objects and events.  But strictly speaking, attitudes are emotional, not cognitive constructs, because they are less about knowledge and more about positive and negative feelings toward the attitude objects.  Still, feelings can be based on knowledge and beliefs, and feelings can change when knowledge and belief change.  Accordingly, the process of persuasion and attitude change is relevant to social cognition: how do people come to think, and feel, differently about something?

Classic research on attitude formation and change, characteristic of the 1930s and 1940s, and thus focused on things like the Nazi and Soviet propaganda machines, focused on processes of persuasion and communication -- in the "Lasswellian" formulation of the Yale Communicatiaons Research Program, "Who says what to whom in what channel with what effect" (Lasswell, 1948, p. 117; Hovland et al., 1953, p. 12). Thus, whether an individual comes to hold an attitude, or changes his attitude, depends on a number of different elements:

The 1950s and 1960s saw the influence of balance and dissonance theories.  Dissonance between a pre-existing attitude and a persuasive communication, or between two attitudes, or between attitudes and behavior.  One way to do this is to change the perception of the behavior.  Another way is to engage in selective perception, learning, or memory that that favors attitude-consistent information. 

And finally, beginning in the 1970s, following the cognitive revolution in social psychology, theories of attitude change more expressly focused on information-processing and other aspects of cognition.

For an excellent exposition of Lewin's field theory, see the treatments by Hall and Lindzey in Theories of Personality (1e, 1957); for Heider and the balance theories, see the corresponding presentation by Shaw and Costanzo in Theories of Social Psychology (1e, 1970) -- an important textbook that was expressly modeled on the success of Hall and Lindzey.  Hall and Lindzey, for their part, modeled their text on Hilgard's classic Theories of Learning (1e, 1948). 

The Cognitive Revolution in Social Psychology

Beginning in the 1950s, and especially in the 1960s, psychology became disenchanted with the radical behaviorism of Watson and of its most devoted exponent, B.F. Skinner.  In the mid-1960s, around the time that Neisser's seminal textbook, Cognitive Psychology (1967) was published, the cognitive revolution in experimental psychology washed over into social psychology.  The thrust of the cognitive revolution in social psychology was to reassert The Thomas Theorem that what controls individual experience, thought, and action is not the situation, as it might be objectively described by a third-person, but rather  the individual's mental representation of the situation, which is in turn a product of his or her internal, cognitive processes -- though, frankly, social psychologists did not always make attribution to the Thomases themselves.  

In some respects, the transition between a crypto-behavioristic approach to social psychology and a more cognitive point of view is illustrated by two lines of research separated by less than five years.

The first is Stanley Milgram's classic studies of obedience to authority (1963, 1964).  As is well known, Milgram contrived a situation in which two subjects were brought into the laboratory for a study ostensibly on the effect of punishment on learning.  One subject was ostensibly assigned to the role of teacher, the other to the role of learner.  However, only one of these individuals was an actual subject: the person assigned to the role of learner was, in fact, a confederate of the experimenter, whose behavior in the experiment followed a pre-arranged script.  When the "learner" made mistakes, the "teacher" was supposed to administer electric shocks as punishment. 

While the level of shock began as "slight", the teacher was supposed to increase the shock level with each new error, eventually reaching a very intense, very dangerous level labeled "XXX".  At this high level, the "learner" was instructed to complain, and then go silent.  If the "teacher" balked at increasing the shock level, the experimenter simply replied that "the experiment requires that you continue".  In his original experiment, Milgram reported that roughly 65% of subjects followed orders, administering the very highest level of shock.

22Milgram2.JPG (51134
            bytes)Subsequent 23Milgram3.JPG (60619
            bytes)experiments examined the conditions of obedience to authority.  For example, Milgram found that the percentage of obedient "teachers" varied depending on their proximity to the "learner", on the proximity of the authority (i.e., the experimenter), and on the institutional context in which the experiment took place.  Throughout the series of studies, Milgram's emphasis was on features of the situation controlling the subject's obedient behavior -- as he put it, "The sheer strength of obedient tendencies manifested in this situation".

The Milgram experiment has been given the move treatment not once but twice: first as the Tenth Level, a 1976 television movie, starring William Shatner (of Star Trek) as Milgram; then, in 2015, in Experimenter, starring Peter Sarsgaard as Milgram (and Lellan Lutz as Shatner)

A related adventure in psycho-Hollywood is The Stanford Prison Experiment, starring Billy Crudup as Phillip Zimbardo, a psychologist at Stanford whose experiment was inspired by Milgram's.

Milgram's experiment aroused considerable controversy, partly on ethical grounds -- it was the immediate stimulus to the institutionalization of "institutional review boards" for the protection of human subjects in research.  But it also aroused criticism on methodological grounds -- specifically, that Milgram's experimental situation contained cues that suggested that things were not really as they seemed on the surface.  If his deception was so transparent, then what looks like unquestioning obedience to authority might not be obedience after all.  

Chief among these critics was Martin T. Orne (Orne & Holland, 1968), who argued that the subject's perception of the experimental setting is a critical determinant of his behavior within that situation (full disclosure: Orne was my mentor in graduate school).  From Orne's point of view, Milgram's experiment contained three critical cues.  (1) With respect to the ostensible purpose of the experiment, to study the effect of punishment on learning: the "teacher" was not doing anything that the experimenter couldn't have done perfectly well himself.   Put another way, the "teacher" might well have asked himself, "What am I doing here?".  (2) Although the ostensible purpose of the experiment was to study the effect of punishment on learning, in fact the experimenter stayed in the room with the "teacher", and did not make any observations of the "learner".  This might well have communicated that the "teacher", not the "learner", was the real subject of the experiment.  (3) When the "learner" began to complain about the intensity of the shock, the impassive behavior of the experimenter is totally at odds with the situation apparently unfolding.  For example, the experimenter did not even bother to check on the "learner" when he stopped responding.  Taken together, this constellation of cues must clearly have communicated to the "teacher" that there was something "fishy" about the whole business.

27OrneAnalysis.JPG (103761 bytes)Orne's critique of Milgram was framed by his social-psychological analysis of psychological research in general (see Orne, 1962, 1970, 1973).  All too often, Orne argued, psychologists treat experimental subjects as if they were passive recipients of experimental manipulation -- beakers, if you will, filled with chemicals to see what reaction will occur.  Instead, Orne argued, experimental subjects are sentient beings, actively involved in the social interaction known as "taking part in an experiment".  For that reason, Orne argued, the experimental setting has to be seen "from the subject's point of view". 

Orne's emphasis on viewing the experiment "from the subject's point of view" is clearly related to Thomas's emphasis on "the definition of the situation", described below).  As such, Orne's analysis of the social psychology of the psychological experiment constitutes an early example of the revival of the cognitive perspective on social interaction.  to repeat:

The cognitive perspective was quickly translated into actual experimental research.  Consider, for example, the classic research on bystander intervention, a form of altruism, published by Darley and Latane (1968).  In these experiments, subjects were recruited for an experiment involving the completion of some personality questionnaires.  Some subjects were run alone in a research cubicle, others were run in groups.  During the experimental session, after the subject(s) were left alone to their work, the experimenters contrived an emergency -- smoke blowing into the room through ventilation ducts, or a research assistant falling in an adjacent room.  The principal finding of the experiment was that subjects were more likely to seek or render assistance when they were alone, than when they were in a group -- in other words, that the presence of other people deterred helping behavior.

A result like this could admit of a purely situationist interpretation, a la Milgram -- that behavior was influenced by some objective feature of the situation, such as whether others were present or not.  Instead, Darley and Latane offered a cognitive interpretation of the result, in terms of an analysis of the deterrents to helping:

The situation is, first and foremost, one that is ambiguous.  Two people fighting in the park may be intent on injuring each other; or it may be horseplay among friends; or they may be rehearsing for a school play.  Accordingly, we tend to look to other people for clarification.  But in this instance, the other people are doing the same thing -- a condition that Darley and Latane called pluralistic ignorance.  Everybody's looking around for clarification, but nobody's doing anything -- and that lack of action helps define the situation as a non-emergency.  

Even if the situation has been defined as an emergency, other cognitive factors may come into play to create the bystander intervention effect.  For example, the presence of other people may lead to inaction through diffusion of responsibility -- if each actor believes that someone else has already taken action, then nobody will think that any action is necessary.  Finally, the individual's self-efficacy beliefs may preclude action (Darley and Latane did not use this precise term, which was coined by Albert Bandura, but this is what they meant): we may easily believe that someone else present in the situation has more skills than we do to render proper assistance.  

Viewed objectively, the presence of other people deters helping behavior.  But from a psychological point of view, the actual determinants of behavior are not situational, but rather cognitive in nature, because they lie in people's individual beliefs and expectations concerning themselves and the situation in which they find themselves.  

Orne's analysis of the experimental situation, and the bystander intervention experiments of Darley and Latane, exemplify the cognitive perspective on social interaction as it re-emerged in the wake of the cognitive revolution.  The cognitive perspective, put bluntly, is that cognition mediates the person's response to environmental events.


Persons, Environments, and Behavior:

A Framework for Integrating 

Personality and Social Psychology

Link to background on personality and social interaction.

The classic framework for the analysis of social behavior was provided by Kurt Lewin (1890-1947).  Lewin took his PhD from the University of Berlin in 1914, trained in the tradition of Wundtian structuralism, but soon shifted his allegiance to the Gestalt school of psychology.  He emigrated to the United States in 1933, a refugee from Hitler's Europe, at which time he Americanized the pronunciation of his name to Loo-win (though, I suppose the adjectival form is still Levinian!).

Lewin initially taught at Iowa, then founded the Research Center for Group Dynamics at MIT (it subsequently moved to Michigan, where it remains as a component of the Institute for Social Research).  Through his American students, particularly Leon Festinger (1919-1989), and Festinger's students (who include Stanley Schachter and Philip Zimbardo), Lewin became widely influential in American social psychology.    His point of view is best represented in his early books, A Dynamic Theory of Personality (1935) and Principles of Topographical Psychology (1936).

Employing the conventions of mathematics, Lewin asserted that



B = f(P, E),



B = the individual's overt behavior:  behaviors that are publicly observable.  

For Lewin, every behavior is a social behavior, in that the individual's behavior is always in some way directed toward another person.

In other words... P = personal determinants: mental (cognitive, emotional, and motivational) states and dispositions residing within the individual's mind, such as beliefs, feelings, motives, traits, and attitudes.  

For Lewin, P represents all the causal factors that reside within the individual.


is a function of both 

personal and environmental


E = environmental determinants: factors impinging on the individual from outside, including aspects of the physical ecology (temperature, humidity, altitude, etc.) and aspects of the sociocultural ecology (the presence and behavior of other people,  constraints imposed by social structures,  social roles, situational demands and expectations, social incentives, etc.  

For Lewin, E represents all the causal factors that reside in the world outside the individual.  But for Lewin, because every behavior is social behavior, every situation is really a social situation, whose dominant features are the behavior of other people, as well as wider social and cultural forces.  


This bit of pseudo-mathematics represents the idea that personal and environmental determinants combine somehow to cause individuals to do what they do.  The comma (,) in the equation indicated that Lewin was open as to precisely how these factors combine -- which turns out to be a nontrivial detail! 

Perhaps the easiest way to think about how personal and environmental determinants combine to produce individual behavior is to think of them as independent of each other.  This is certainly the perspective adopted by traditional personality and social psychology.

Borrowing the phrase coined by C.P. Snow (1959, 1963) in his analysis of the relations between the sciences and the humanities, traditional personality and social psychology developed as "two cultures" -- each having little to do with the other (not surprisingly, given that each considered the other to be concerned with "noise").  


The Doctrine of Traits

            (40924 bytes)As subfields within psychology, personality and social psychology have historically emphasized different aspects of Lewin's formula.


Traditional personality psychology assumes that behavior is primarily determined by features of the person such as his or her beliefs, attitudes, values, emotions, motives, and traits, and that situational factors are largely irrelevant.  

B = f(P).

The canonical method of traditional personality psychology is to construct a "psychological test" to measure some personality trait, and then to use this information to predict individual behavior in some specific situation.  The test might take the form of a self-report questionnaire, a rating scale (completed by the subjects themselves or by others who know them well), or even a sample of actual behavior.  

tradperson.gif (7018
            bytes) So, for example individual differences in friendliness, assessed by means of a self-report questionnaire, would be used to predict whether a person would smile in some situation.  In this research, which often uses the technique of multiple regression analysis (a variant on the correlation coefficient), the trait measure (e.g., friendliness) serves as the predictor variable, and the behavioral measure (e.g., smiling) serves as the criterion variable.

A similar method is used to study the relationship between attitudes and behavior.  For historical reasons, having largely to do with an interest in attitude change in response to persuasive communications, attitudes have primarily been studied by social psychologists.  But the logic is the same: attitudes, which are internal dispositions to evaluate certain objects or ideas positively or negatively, are held to cause the individuals who hold these attitudes to behave in particular ways.  Thus, in the 2000 election, a registered Democrat was (probably) more likely to vote for Al Gore than for George Bush -- though not enough of them did, from Gore's point of view.

These dispositions are commonly studied in the form of traits and attitudes.  However, other dispositions are also relevant to behavior, such as moods, motives, values, and beliefs.   

The canonical method of traditional personality psychology exemplifies the doctrine of traits, derived to a great extent from the work of Gordon Allport (1937):

Social behavior varies as a function of internal dispositions that render it coherent, stable, consistent, and predictable.  

011Allport.jpg (66234
            bytes)G. Allport (1937) defined a personality trait as:



"a generalized and focalized neuropsychic system... with the capacity to render many stimuli functionally equivalent, and to initiate and guide consistent (equivalent) forms of adaptive and expressive behavior."

For Allport, there is an analogy between personality traits and physical traits.  Just as physical traits are stable dispositions to appear in a particular way, so personality traits are stable dispositions to behave in particular ways.  Traits are internal to the person.  Although not necessarily genetic in origin -- they could be acquired through a history of learning -- they are somehow represented in the nervous system.  These personal characteristics, once established, then mediate between the environment and behavior.  As Allport put it, traits "render situations functionally equivalent", in that they dispose the person to display similar sorts of behaviors in them.

Further, Allport contrasted two views of traits:

It's pretty clear where Allport got the term "biophysical": according to his view, traits have physical existence as biological entities.  The term "biosocial" is a little more mysterious: if traits are social labels, existing only in the individual mind, then one might think of them as "psychosocial".  At the same time, perception makes reference to the real world, and with "biosocial" Allport might have wanted to include both personality as it exists in the real world and the perceiver's mental categorization of it.

Allport himself preferred the biophysical view: for him, personality traits were real in precisely the same way that physical traits were real, and were subject to measurement in precisely the same way that physical attributes were.  This view is very popular, especially at Berkeley -- it's the premise of the Institute for Personality Assessment and Research, forerunner to the Institute for Personality and Social Research -- but it is also very controversial, for reasons that will become clearer later.

In this course, I take an agnostic position on the biophysical view: personally, I do not believe that traits are important determinants of behavior, and so I think it is a mistake to make them the center of personality research.  But even if traits exist, in the biophysical sense that Allport believed they existed, they are also social constructions.  We, all of us and every day, label and categorize people in terms of their traits.  And it is this categorization process that is a topic for research in social cognition. 

In fact, even from the biophysical view, personality assessment can be viewed as a process of social judgment -- in which the judge attributes traits to a person based on his or her scores on various personality tests.  A great deal of research within traditional personality psychology has been devoted to the question of how accurate these attributions are -- a line of research that implicitly assumes that traits have an existence independent of the judge.  But in this course, we will set aside the important and interesting question of accuracy, and focus on the cognitive processes by which trait attributions are made.


The Doctrine of Situationism

Traditional social psychology, by contrast, assumes that behavior is primarily determined by features of the environment, and especially features of the sociocultural ecology, such as interpersonal, organizational, and cultural factors, and that individual differences in personality are largely irrelevant.

B = f(E).

This viewpoint, which is congenial to the behaviorism espoused by John B. Watson and B.F. Skinner, is exemplified by traditional research on social influence, or the effects on behavior of the presence or behavior of other people. 

The canonical method of traditional social psychology is to manipulate some aspect of the social environment (such as whether behavior is private or public, or whether the subject receives information about other people's attitudes and opinions), and observe the effects of this manipulation on behavior in some specific situation.  All subjects might be exposed to all conditions (this is known as a within-subjects design), or different groups of subjects might be randomly assigned to each condition (this is known as a between-groups design).  

BfE010.jpg (60576 bytes) So, for example, we might arrange an encounter between a subject and an acquaintance or a stranger, and see if smiling occurs more often in one situation than the other.  In this research, which often uses the technique of analysis of variance (a variation on the t-test), the manipulated variable (e.g., presence of acquaintances or strangers) serves as the independent variable, and the observed variable (e.g., smiling) serves as the dependent variable.


Just to confuse things, sometimes in regression analyses, the predictors are labeled as independent variables, and the criteria are labeled as dependent variables.  This is because, mathematically, multiple regression is formally equivalent to the analysis of variance.

The canonical method of experimental social psychology exemplifies the doctrine of situationism:

Social behavior varies as a function of features of the external environment, particularly the social situation, that elicit behavior directly, or that communicate social expectations, demands, and incentives.

These features of the situation may be found in the external physical environment.  More likely, though, they are to be found in the external social environment, such as the presence and activities of other people, social demands, and social rewards.

The doctrine of situationism is sometimes attributed to Kurt Lewin himself -- see, for example, The Person and the Situation: Perspectives of Social Psychology (1991) by L. Ross & R.E. Nisbett.  It is true that, as Ross and Nisbett note, "the main point of Lewin's situationism was that the social context creates potent forces producing or constraining behavior" (p. 9).  But Lewin's field theory held that the person and the environment were part of a single "field", and so the idea of the environment acting on the person isn't really consistent with his views.  Lewin's views are actually more compatible with another doctrine, concerning interactionism, discussed later.

            (104376 bytes)A better source for situationism in social psychology is found in the radical behaviorism of B.F. Skinner -- a point made by Zimbardo ("Experimental social psychology: Behaviorism with minds and matters" in Reflections on 100 Years of Experimental Social Psychology, ed. by A. Rodriques & R.V Levine, 1999).  Consider, for example, the following quotation from Skinner's introductory psychology text, Science and Human Behavior (1953):

The free inner man who is held responsible for the behavior of the external biological organism is only a prescientific substitute for the kinds of causes which are discovered in the course of a scientific analysis.  All these alternative causes lie outside the individual (emphasis added).

Skinner mostly studied learning in nonhuman animals (chiefly rats and pigeons), but he had no difficulty generalizing from the nonhuman to the human case.  No matter the organism, behavior is under the control of eliciting and discriminative stimuli in the environment, and subject to selection by the organism's history of reinforcement.  Accordingly, there is no need to make reference to any mental states (such as belief, feeling, or desire), or for that matter any trait (such as neuroticism or extraversion) as either initiating behavior or as mediating between environmental stimulus and organismal response.  To the extent that Skinner considered such matters, he viewed personality traits as habits established through learning.

Despite the cognitive revolution in psychology, which displaced Skinnerian behaviorism from its hegemonic position in the field, situationism remains powerful in social psychology today.  In a tutorial on social psychology prepared for neuroscientists, Lieberman (2005) reasserted the power of the situation, as well as the related doctrine of situation blindness:

Anyway, beginning with the now-classic experimental demonstrations of social influence by Sherif (1937) and Asch (1951), and others, found others, situationism took hold in social psychology at the height of the hegemony of behaviorism in American psychology generally.  And even though the cognitive revolution overthrew behaviorism in the 1960s, the crypto-behaviorism of situationism retains its hold over social psychology, with social psychologists' frequent references to "the power of the situation" to influence behavior.  

This situationist viewpoint is exemplified by the classic topics in social psychology -- especially the literature on social impact, conformity, and compliance. But situationism also lies at the core of what we might think of as the "Four As" of social psychology:

The general picture in traditional social psychology, then, is of the individual responding to outside forces.  This view is generally compatible with the viewpoint of functional behaviorism, a variant of stimulus-response theory which holds simply that the organism's behavior varies as a function of environmental variables, including the presence of unconditioned, conditioned, generalized, and discriminative stimuli, and the organism's history of reinforcement.  This stimulus-response viewpoint, in turn, accounts for the common observation of cross-situational variability in behavior.  It is a characteristic feature of human social behavior that it is closely tuned to subtle variations in the situation, and situationism offers one account of why this is so.  


The Trait-Situation Controversy

For most of the 20th century, personality and social psychology proceeded largely independently of each other (in the 1930s, Gordon Allport wrote a seminal text on personality, while his brother Floyd did the same for social psychology).  In many psychology departments, personality and social psychology were represented by different groups of faculty, the same way that cognitive and clinical psychology are.  And, pretty much, each group treated the other with benign collegial neglect.  But in the 1960s, partly as a late result of the hegemony of behaviorism in psychology, as well  there arose a trait-situation controversy over which factors were more powerful predictors of behavior -- internal traits or external situations.  This debate, which focused on statistical comparisons of the percentage of behavioral variance that was accounted for by traits and by situational factors, came to a head in the late 1970s and 1980s (long after behaviorism had passed from the scene) devolved into a contest over whose "effect size" was bigger.

Mischel (1968), in a review of available research, concluded that the modal correlation between subjects' scores on a personality test and their actual behavior in some specific test situation was about r = .30 -- a figure indicating that traits account for about 10% of behavioral variance.  Mischel famously (and derisively) dubbed this figure the personality coefficient.  Mischel also suggested that the perceived situation would account for more behavioral variance than traits, but he did not actually test this proposal.

A counterattack by Funder and Ozer (1983) sampled from the classical social-psychological literature on situational influence, translated t values and F ratios into correlation coefficients, and determined that the effect of situational variance amounted to a correlation of about r = .45 -- a figure indicating that situations account for about 20% of behavioral variance.  So, most variance isn't accounted for by situations, either.  (Note, however, that Mischel was talking about the effect of the perceived situation, while F&O analyzed the effect of the objective situation, as experimentally manipulated).  Apparently, neither traits nor situations account for "most" behavioral variance.

So what began as a stereotypically masculine "Battle of the Correlation Coefficients", intended to determine whose was bigger, ended up looking more like a fight in an elementary schoolyard, with each side shouting "So's your mother" at the other one.  In retrospect, the Battle of the Effect Sizes was essentially a pointless exercise, and generated much more heat than light.  It's over now, and in most psychology departments personality and social psychologists work side by side, as indeed they do at Berkeley -- though they still keep their hands on their swords.


Putting the Person and the Environment Together

However, these traditional formulations are largely misleading.  Nobody believes that one factor is exclusively responsible for behavior, and the other is wholly irrelevant.  Dispositional and situational factors probably combine somehow to cause behavior to occur.

            (44408 bytes)As noted earlier, one possibility is that P and E are independent -- that is, that each set of factors exerts its own separate influence on behavior, without affecting the other in any way.  This notion lies at the heart of the traditional situation, in which personality and social psychology were situated as separate and independent subfields of psychology.   In such a situation, behavior is partly predicted by personality traits, and partly affected by situational manipulations.  In mathematical terms, personal and environmental factors are additive:

B = f(P + E).  

If P and E are independent:

In statistics, these two effects would be characterized as the main effect of personality and the situation, respectively.

IndepenIllus014.jpg (65264 bytes)Thus, friendly people may smile more than unfriendly people, and people may smile more at acquaintances than at strangers, but the difference between friendly and unfriendly subjects is constant across the two situations (subtract the means), and the difference between acquaintance and stranger targets is constant across levels of friendliness (again, subtract the means).  Statistically speaking, there is no interaction between these main effects.

But this was not Lewin's idea at all.   Lewin sought to apply the principles of Gestalt psychology to the study of social behavior.  The Gestalt school is known for its assertion that "the whole is greater than the sum of its parts".  Applied to perception, this means that perception encompasses the entire stimulus field.  Individual stimulus elements form a coherent, integrated whole, and cannot be isolated from each other.  Similarly, Lewin argued that social behavior is responsive to the entire field of social stimuli -- not just the other person immediately present, but also the wider social context in which the interaction occurs.  Lewin went even farther to assert that the social situation includes the person him- or herself: the person is part of the stimulus field to which he or she responds.

Lewin expressed this basic idea throughout his writings, in various ways:






006Lewin33.jpg (112735 bytes)

Lewin1939.jpg (77948 bytes)

Lewin1940.jpg (120758 bytes)

                    (124967 bytes)


                    (128198 bytes) Lewin1946Novel.jpg (105153 bytes) Lewin1946c.jpg (121416 bytes)

Statements like these show why claims that Lewin is the godfather of situationism in social psychology are, simply, wrong.  Lewin, influenced by Gestalt psychology, was a field theorist -- he believed that the person and the environment were interdependent elements constituting a unified psychological field.  The challenge is how to understand this interdependence.


The Doctrine of Interactionism

PxEInteraction.JPG (39302 bytes)The trait-situation controversy faded partly due to exhaustion of the participants, but also because psychologists began to consider a more interesting possibility -- that personal and environmental determinants interacted with each other in a variety of ways.  This comes closer to Lewin's own position.  Remember that he was heavily influenced by Gestalt psychology, and believed that the person and the environment constituted an organized and integrated field in which behavior takes place.  In this behavioral field, the person and the environment are inextricably intertwined.

            (107518 bytes) The Doctrine of Interactionism proposed by K.S. Bowers (1973), holds that people influence the situations that, in turn, influence their behavior.  As he put it:

"Both behavior and reinforcement are subject to selection by biocognitive structures.  These structures include the biological substrates of mental processes; and the cognitive system which organizes them.  Interactionists agree that a person's behavior is determined by the situation in which it occurs.  But they also assert that the situation itself is largely determined by the person....  An interactionist or biocognitive view denies the primacy of either traits or situations in the determination of behavior....  More specifically, interactionism argues that situations are as much a function of the person as the person's behavior is a function of the situation."

The doctrine of interactionism was originally intended to counter the doctrine of situationism: 

Personal and environmental factors are interdependent -- in particular, people create the environments to which they respond.    

Interactionism agrees that people's behavior is influenced by the situations in which they find themselves.  But because it views people as part of the environment, it holds that personal factors of the sort envisioned in the doctrine of traits can still play an important role in behavior.

From an interactionist perspective, different kinds of people show different patterns of response across different situations.  In mathematical terms, personal and situational factors are multiplicative:

B = f(P x E).

If P and E interact:

InteractIllus018.jpg (63450 bytes)Thus, for example, friendly people might smile more than unfriendly people, but this difference would be bigger when they encounter a stranger than when they encounter a friend.  Or, put another way, friendly people might discriminate less between the two situations than unfriendly people would.  Such a situation is known statistically as the person-by-situation interaction.

IneractOther020.jpg (51869 bytes)The person-by-situation interaction takes a number of forms.

Interactionism and the "S-R Inventory" Method

In some interpretations, the Person by Situation interaction is modeled on the statistical model of the analysis of variance, where independent variables influence dependent variables individually as main effects, or combined in interactions.  The attraction of this model is evident in early statistical analyses of the power of interactions, including some cited by Bowers himself, which are all based on the ANOVA model.

In the mid-1970s Norman Endler (1973, 1975) introduced "S-R inventories" of personality which assess the effects of situations and response modes, as well as of individual differences on expressions of traits such as anxiety or hostility.  These inventories asked subjects to report not only how likely a particular situation would elicit an anxious or hostile response (for example), but also how likely they would be to display anxiety or hostility in a particular manner in each situation.  When administered to a large group of subjects, the data generated by these inventories can be analyzed to yield estimates of the variance accounted for by various causal factors, including the main effect of persons, collapsed across situations (and response modes), the main effect of situations, collapsed across persons (and, again, response modes), and the interaction of the person and the situation (averaging across response modes), as well as individual differences in the pattern of behavior across situations.  

For example, Dworkin and Kihlstrom (1978) constructed an "S-R Inventory of Dominance" that included a number of different stimulus situations calling for dominant behavior:

For each situation, the questionnaire also posed a number of possible responses:

If we construe the entire set of ratings by a large number of subjects as the total variance of dominant behavior in the population, the questionnaire data can be analyzed to determine a number of different components of variance, including:
SRcomponents011.jpg (58368 bytes)When Dworkin and Kihlstrom analyzed the components of variance in dominance, they discovered that individual differences among persons accounted for about 10% of the total population variance in dominant behavior, while situational differences accounted for about 8%, and response modes accounted for about 7%.  



This is the general pattern of results from S-R Inventory studies that have been conducted in various domains, and collectively these studies have been taken as evidence that, indeed, the person-by-situation interaction is more powerful than either persons or situations taken in isolation -- or, for that matter, the sum of persons and situations taken independently.  

These results are interesting, and they helped to break through a seemingly endless person-vs.-situation debate in the 1970s.  But they also miss the entire point of interactionism -- which, in the Lewinian point of view, that persons are part of the situations to which they respond -- or, put another way, that persons and situations together constitute a unified field in which behavior takes place.  

The ANOVA model is simply blind to the dynamic interplay of persons and situations: it has no way of revealing how persons create the situations to which they respond.  Moreover, because the statistical model of ANOVA assumes that causality is unidirectional -- that is, that it proceeds from independent variable to dependent variable -- it misses the complexity of causal relations.  These deficiencies are corrected by the doctrine of reciprocal determinism, as well as by an analysis of the dialectic between the person and the situation.  

Interactionism as a Dynamic Process

But Bowers wasn't really talking about statistical interactions (just as Lewin wasn't talking about P and E as independent variables).  Bowers spent so much time discussing the results from S-R inventory method because, at the time, that was the only data he had available.  But, like Lewin, Bowers had something else in mind.  Bowers' interactions refer to the dynamic interplay between the person and the situation, in which people help create the situations to which they in turn respond.  

In Lewin's terms, P ==> E, and then E ==> B.

So how do people shape their environments?

David Buss (Journal of Personality & Social Psychology, 1987) has identified three ways in which people affect their own environments: evocation, selection, and manipulation.



The mere presence of a person in an environment alters that environment, independent of his or her traits, attitudes, or behaviors -- or even in the absence of any behavior a all.  In evocation, the individual unintentionally (and even unconsciously) evokes behavior from others which, changes the situation for the evoking person.  The environment is not changed by the person's deliberate, voluntary acts; the effects of evocation are probably related to the person's physical appearance.  Because the environment consists of other people, evocation effects are mediated by others' cognitive structures and processes, such as their beliefs and expectations.  As an example, the physical appearance of a newborn baby's external genitalia structures the environment around his or her parents' and culture's beliefs about gender roles.    

Link to an expanded discussion of evocation.



People deliberately choose to enter one environment as opposed to another, perhaps out of a desire to match their environments with their individual personalities.  The point of selection is that the match between the person and the environment is nonrandom.  Individuals choose environments that are congruent with their own personalities, supporting and promoting their own preferences and tendencies.  Each choice pre-empts alternatives (recall the character played by Gwyneth Paltrow in the film Sliding Doors).  Thus, the individual has little or no opportunity to engage in new behaviors.  In any event, through selection processes the environment is to some extent of the person's own making, because he or she actively chooses to be in one environment as opposed to others.  The principle lies at the core of clinical behavior therapy: a patient who wants to change his or her behavior must put him- or herself in an environment that will support the change, and avoid environments that will oppose it.  Note, too, that environmental selection is often more complex than a simple act of will.  Sometimes choices simply aren't available, and sometimes the selection is made by external forces.  Again, a familiar example are the constraints imposed by culture-specific gender roles.  The effects of prejudice and discrimination against racial minorities and other outgroups is partly a matter of evocation (because racial outgroups differ in appearance from racial ingroups), and partly a matter of selection by virtue of "choices" for the minority outgroup made by the majority ingroup.    

Link to an expanded discussion of selection.



People engage in overt behavioral activities that alter the objective environment -- that is, the environment as it is publicly experienced by everyone in it.  Here we have deliberate, overt behavior that is intended to alter the environment.  Manipulation goes beyond the choice among available environments, and has the effect of creating an environment that would not otherwise be available.  Finding themselves in a particular environment, and unable to select a different one, people engage in behaviors that will modify the character of their environment, as it would be objectively described by an independent observer.  Environmental manipulation underlies all acts of instrumental or operant behavior, where the organism's behavior operates on the environment, changing it in some way, so that it more closely conforms to the organism's goals and purposes.    

Link to an expanded discussion of manipulation.



Evocation, selection, and manipulation all change the environment through behavior: either the behavior of the person him- or herself or that of other people.  In each case, someone does something overtly that changes the objective character of the environment -- that is, changes the environment for everyone in it, not just for the person itself.  But these three modes do not exhaust the effects of the person on the environment.  There is a fourth mode, one that Buss himself does not explicitly recognize, perhaps because his evolutionary perspective effectively blinds him to it (Cantor & Kihlstrom, 1987).

People engage in covert mental activities that alter their mental representations of their subjective environment -- that is, the environment as they privately experience it.  As opposed to behavioral manipulation, cognitive transformation does not act on the objective environment -- the environment as it would be described in the third person by an objective observer.  Rather, transformation acts on the subjective environment.  Through cognitive transformations, people can change their internal, mental representations of the external physical and social environment -- perceiving it differently, categorizing it differently, giving it a different meaning, than before.  In cognitive transformation, the objective features of the environment remain intact -- they have not been altered through evocation, selection, and manipulation.  Rather, the person's covert mental activity has altered the environment for that person only; the environment is unchanged for everyone else -- unless and until the cognitive transformation leads the person to engage in (evocative, selective, and manipulative) behavior that will, in fact, change the environment for everyone.  

Link to an expanded discussion of transformation.


Illustration: Delay of Gratification

The difference between the objective and the subjective environment, and between behavioral manipulation and cognitive transformation, is illustrated by some classic research on delay of gratification in young children .  Delay of gratification has to do with people's ability to tolerate frustration and control their impulses.  It's a pretty important aspect of socialization.  Max Weber, a pioneering sociologist, thought that delay of gratification was the basis for the "Protestant ethic" of self-restraint and the negation of pleasure (obviously, he didn't know many Protestants!) which he thought lay at the basis of capitalism.  But every culture requires some ability for ego-control: to plan ahead and tolerate delays.  This ability is generally acquired early in life, as a result of socialization.  Of course, even within a culture individuals will differ in their ability to delay gratification: as traditional personality psychologists might put it, some people have it, some people don't.


Predicting Delay from Personality Traits

In a study by Funder, Block, & Block (1983), conducted at UCB, nursery-school teachers administered a test of intelligence, and also rated their pupils' personalities on an instrument (known as the California Q Set) that provided measures of two higher-order dimensions of personality:


 A "Big-Five Contrarian"

Why didn't Funder et al. assess the children's personalities in terms of The Big Five?  At the time the study was done, consensus had not yet developed around The Big Five as the structure of personality.  There were other competing systems, including Hans Eysenck's four-factor proposal (neuroticism, extraversion, psychoticism, and intelligence) and the two-factor system (ego-control and ego-resiliency) proposed by Jack and Jean Block, who after all were Funder's co-investigators.  These two competing systems still undergird much research, and in fact Jack Block has on numerous occasions (e.g., 1995, 2001) pronounced himself a "Big Five Contrarian", and offered trenchant critiques of the research supporting The Big Five as a universally applicable structure for personality description.

Around the same time as the assessments were made, Funder et al. engaged the children in an experimental assessment of their ability to delay gratification employing two different situations:


Predictor Variable r
IQ .21
Ego Control .25
Ego Resiliency .23
"Is unable to delay gratification" -.27

All four predictors -- IQ, ego control, ego resiliency, and the specific item "is unable to delay gratification" correlated in the range of .20 < r < 30 with actual delay behavior (the correlation between delay and the specific item was, of course, negative).

Note that the magnitude of these correlations is in line with Mischel's conclusion about the "personality coefficient": that a correlation of about r = .30 is the upper limit on the relationship between personality traits and behavior.  Personality in general predicts behavior in particular, but that prediction is relatively weak.  Put another way, there is a ceiling on the extent to which we an predict behavior in a particular situation, knowing the individual's personality traits.  Put more bluntly, there's more to behavior than personality traits.

The "Power" of the Situation

Research on delay of gratification in children also illustrates the power of the situation.  In a study by Mischel and Ebbesen (1970), children were asked which of two rewards, cookies or pretzels, they preferred.  Then the children were told that the experimenter would go away for  while (actually, about 15 minutes).  If they waited for the experimenter to return, they would receive their preferred reward.  But if they could not wait, they would receive heir nonpreferred reward.  Then the experimenter left.  In one condition of the experiment, he took the cookies and pretzels with him.  In other conditions, he left one ,or the other, or both behind.  How long did the children wait before signaling the experimenter to return?

In fact, children who waited in the absence of both rewards were able to wait a fairly long time -- some of them even outlasted the experimenter!  If either the preferred or the nonpreferred reward was left with the child, waiting time decreased.  And if the child was left to wait in the presence of both rewards, waiting time dropped almost to zero.  Young children cannot delay gratification long in the presence of a reward.  But this experiment shows clearly the influence of the situation -- whether rewards are present or not -- on the child's behavior. 

Another counterattack in the trait-situation debate, by Funder & Harris (1986), re-analyzed data collected by Mischel and his colleagues on the "situational" determinants of delay of gratification, yielding a weighted average effect size of about .45, again corresponding to about 20% of variance explained.  Funder and Harris conceded that Mischel's research yielded dispositional effects accounting for only about 8% of the variance (i.e., almost exactly what would be expected on the basis of a personality coefficient of .30), but suggested that stronger dispositional effects would be obtained in studies that improved the assessment of dispositional characteristics, and that assessed behavior closer in time to the assessment of the dispositions.  However, the delay-of-gratification data collected by Funder, Block, and Block (1983), which met those requirements, yielded correlations in the range of .11 to .27 -- very similar to those obtained by Mischel (Kihlstrom, 1986).

Selection of the Environment

In the experiment just described, the children themselves have no choice in their environment: they are forced to wait in the presence of one or the other reward, or both, or neither.  But consider a child who is given a choice  If he is asked whether he would like to wait in the presence or absence of the reward, or in the presence of the preferred or nonpreferred reward, his choice will affect his behavior by determining whether he will wait in a situation in which long delays are easy or hard to achieve.  If he chooses to wait in the presence of both rewards, he may be unable to wait for very long.  If he chooses to wait in the absence of both rewards, he may very well outlast the experimenter!

Behavioral Manipulation

But what if you find yourself in a situation you didn't choose, or for some other reason the environment is wrong?  It turns out that even young children have the capacity to manipulate the environment through their overt behavior, affecting changes in the actual, objective, character of the environment.  Rather than looking to children's personality traits to understand delay of gratification, Mischel and his colleagues (1989) have asked the question: "What do 'high-delay' children do?".   Consider, for example, children in an experiment in which they must wait in the presence of a promised reward.  We know from a previous study that, on average, children do not wait very long in such a situation.  But some children do: How do they do it?  Traditional personality psychologists would attribute these differences in delay of gratification to differences in personality traits, sch as ego-control and ego-resiliency.  But while traditional personality psychologists focus their analyses on the traits that children have, interactions would focus on what the children do.

Mischel and his colleagues (Mischel et al, 1989) observed high-delay children engaging in a number of interesting behaviors:

In other words, these children were behaving in such a way as to put the rewards out of sight -- to change the environment into one in which the reward, though physically present, is not really there.  

106SelfDistrac.jpg (50573 bytes)A formal experiment by Mischel, Ebbesen, and Zeiss (1972, Exp. 1) makes the same point.  Children were given the choice of marshmallows or pretzels, and then asked to wait for the experimenter to return before he could receive the preferred reward.  The child could also signal for the experimenter to return, in which case he would get the nonpreferred reward.  Children who were given no distraction could not wait very long.  But children who were told to play with a "Slinky" toy were able to wait much longer. 


109OverrtCovert.jpg (52999 bytes)The distraction doesn't even have to involve overt behavior, like playing with a toy.  In another condition of the "Slinky" experiment, children were simply instructed to spend their time thinking of "anything that's fun to think of".  These children were able to delay gratification even longer than those who played with the "Slinky" toy.



By putting the reward out of sight, or by directing their attention elsewhere (e.g., to a game, a song, or a Slinky toy, or just "fun" thoughts), they have effectively altered their environment from one in which they cannot wait very long to one in which they can.  


Cognitive Transformation

Even in the absence of overt behavior, it is possible to transform the subjective mental representation of the situation in such a way as to promote delay of gratification.  An experiment on the role of ideation in delay of gratification by Mischel and Baker (1975) illustrates the power of such  cognitive transformations.  In this experiment the children were given the usual choice of marshmallows vs. pretzels, and during the waiting period the preferred reward remained in sight.


Ideational Instructions

Consummatory Transformative
Look at the marshmallows. They are sweet and chewy and soft. When you look at marshmallows, think about how sweet they are when you eat them.... When you look at marshmallows, think about how soft and sticky they are in your mouth when you eat them..... When you look at marshmallows, think about how white and puffy they are. Clouds are white and puffy too -- when you look at marshmallows, think about clouds.... The moon is round and white. When you look at marshmallows, think about the moon....


Look at the pretzels; they are crunchy and salty. When you look at pretzels, think about how crunchy they are. When you look at pretzels, think about how salty they taste when you lick them or chew them.... When you look at pretzels you can think about how long and brown they are. A log is long and brown. When you look at pretzels, think about logs and tree trunks. Or you can think about how round and tall they are. A pole is round and tall....

            (51345 bytes)The results of the experiment were very striking.  Children who focused their thoughts on the consummatory aspects of their preferred reward were not able to delay gratification for very long.  Those who transformed their preferred reward into something that did not taste sticky-sweet, or crunchy-salty, were able to delay for a long time.  What caused their delay (or lack thereof) was not their personality traits, or the environment, or even their own behavior.  The important factor was how they thought about the rewards.



Thus the notion of delay of gratification as a trait that people have, proves to be an oversimplification.  Ego-control is not just a personality characteristic, it is also a product of strategic activity.  Delay of gratification is accomplished through a combination of selection, manipulation, and transformation, all oriented around the general principle of "out of sight, out of mind".

Through the process of social learning, children acquire both knowledge of effective behavioral and cognitive strategies for delay of gratification, and the ability to deploy these strategies effectively.  


The Doctrine of Reciprocal Determinism

            (58562 bytes)Traditionally, the person-by-situation interaction has been characterized as unidirectional.  That is to say, in Bowers' preferred formulation, people somehow affect the situations they are in, and this interaction, along with the main effects of the person and of the situation, causes behavior to occur.  But of course, it may also be that, in addition to the person influencing the situation to which he or she responds, the environment also shapes the internal states and dispositions of the person in it.  If people can influence their environments, why shouldn't their environments influence them?

BanduraDocRecipDet.JPG (96771 bytes)Albert Bandura has called this state of affairs reciprocal determinism.  In reciprocal determinism, causality is bidirectional:

TriadicReciprocality.JPG (64082 bytes)The doctrine of reciprocal determinism is essentially a more dynamic extension of the doctrine of interactionism:

Where interactionism asserts that people are a part of their own environment, reciprocal determinism asserts that people, their environments, and the behavior that takes place within those environments form a complex, dynamic, interlocking system characterized by nonlinear, bidirectional, causal relations.   

In a very real sense, reciprocal determinism is the state of affairs envisioned by complexity theory (also sometimes known as chaos or catastrophe theory).

But if the causal relations between P and E can be bidirectional, why can't the causal relations between P and B, and between E and B, be bidirectional as well?

04S021BPETriadic.jpg (31743 bytes)In other words, each element in Lewin's formula -- not just P and E but B as well, is both cause and effect.  Because the reciprocal causal relations involve three elements, Bandura also has labeled this expanded notion of reciprocal determinism triadic reciprocality



Reciprocal determinism in general, and triadic reciprocality in particular, entails a very interesting situation in which everything is simultaneously both the cause and the effect of everything else. 

            (71504 bytes)Still, when everything in a system affects everything else, things are going to get awfully complicated awfully quickly.  For this reason, it is rarely possible to study triadic reciprocality in all of its dynamic glory.  Fortunately, the lack of simultaneity works to our advantage, permitting an analytic decomposition in which we break triadic reciprocality down into its bidirectional segments, or dialectics (from the Greek word meaning "dialog").  This yields The Three Dialectics in Social Behavior:


P <---------------> B The dialectic between the person and his or her behavior.
E <---------------> B The dialectic between the environment and the behavior that occurs in it.
P <---------------> E The dialectic between the person and the environment in which his or her behavior takes place.

Link to a more complete analysis of the three dialectics.


The General Social Interaction Cycle

According to the doctrine of interactionism, people create their own situations through their thoughts and actions; and according to the doctrine of reciprocal determinism, a person's behavior feeds back to alter the person him- or herself.  These dynamics are played out even in what the social psychologist Henri Tajfel has called the minimal group situation -- a simple dyadic relationship consisting of only two people interacting with each other.

04S026GSIC.jpg (57822
            bytes)The general social interaction cycle (Cantor & Kihlstrom, 1987) is a conceptual scheme for representing any dyadic social interaction, whether as mundane as buying a toothbrush or as monumental as proposing marriage.  In the scheme, two participants are assigned the role of Actor and Target, respectively.  This assignment is of course somewhat arbitrary, because each individual is both an actor and the target of the other's actions.  For convenience, the Actor role is assigned to the individual who initiates the social interaction, as in the following illustration.


The General Social Interaction Cycle is derived from earlier descriptions of a General Social Interaction Sequence by Darley and Fazio (American Psychologist, 1980) and by Jones (Science, 1986).  However, for reasons that will be come clear presently, the "sequence" of transactions between Actor and Target is better conceptualized as a cycle -- even a set of embedded cycles.

04S026GSIC.jpg (57822
            bytes)First, the Actor enters the situation -- the immediate context in which he or she physically encounters the target (from this point on, for simplicity in exposition, we'll call the Actor "she" and the Target "he".

Some of these skills are cognitive in nature, such as her ability to "read people"; others are motoric, such as a particular way of walking, or using her hands.This fund of social knowledge may be characterized as social intelligence (Cantor & Kihlstrom, 1987, 1989; Kihlstrom & Cantor, 1989, 2000). 

Social intelligence is not to be confused with intelligence in the sense of IQ tests.  In referring to social intelligence, Cantor and Kihlstrom are not interested in whether people are smart or stupid in social situations.  Rather, they use the word intelligence in a manner closer to its military sense of information that is used to guide action. 

Social intelligence comes in two types:

The social intelligence view of personality begins with the proposition that personality is reflected in social behavior.  Individual differences in social behavior do not reflect individual differences in personality traits (whatever they are), but rather reflect individual differences in the individual's cognitive resources -- the fund of declarative social knowledge that the individual brings into the social situation, and the repertoire of procedural social knowledge that the individual brings to bear on social interaction.

In this respect, social cognition is the study of social intelligence.

As she begins the interaction, the Actor forms an impression of the situation -- of the target, and of the immediate environmental context: Does he still seem interested?  Is this a good time to ask?  This impression combines knowledge derived from two general sources:

04S026GSIC.jpg (57822
          bytes)Finally, the Actor acts on the basis of her impression.  She may approach the target or shy away, she may pop the question or not.

Assuming that the Actor has asked him for a date, attention now shifts to the Target, who now has to do something in response to the Actor's initial salvo.

The Target knows he's free Friday night, but that's not decisive.  Should he play hard to get?  Should he wait to see if he gets a better offer from someone else?

On the basis of the impression he's formed, the Target responds: He decides to keep his options open for Friday night, but doesn't want to spurn the Actor entirely, so he says he can't see her Friday, but proposes that they go out on Saturday instead.

04S026GSIC.jpg (57822
            bytes)Now attention shifts back to the Actor, who has to interpret his response, and revise her impression of the situation accordingly.

On the basis of her impression, the Actor responds to the Target:

04S026GSIC.jpg (57822
            bytes)Now the ball is back in the Target's court: he has to interpret her response, revise his impression, and figure out what to do next.



And so it goes, with the cycle of exchanges continuing:
But the General Social Interaction Cycle also transpires at another level, within each individual participant.  According to the Doctrine of Reciprocal Determinism, behavior does not simply affect the person toward whom it is directed; it also feeds back to affect the person who emitted the behavior.

In any event, each participant in this social interaction is behaving in accordance with his or her construal of him- or herself, and of the other, and of the situation in which they meet.  Each of these construals is modified by the other's behavior, and his or her own.  And it's these individual construals, in the end, that lead the participants to behave the way they do.

The way to understand an individual's behavior is to understand the individual's construal of the situation.


The Self-Fulfilling Prophecy

The foregoing analysis of the General Social Interaction Cycle shows how a person's beliefs, translated into behavior, can influence another person's behavior in such a way as to change the environment for both of them.  Cognitive transformations of the environment are not entirely private, because they lead to overt, public behavior which can alter the objective situation for other people as well.  This process is exemplified by what the American sociologist Robert K. Merton (1947) called the "self-fulfilling prophecy", and what social psychologists have come to call expectancy confirmation effects.


Origins of the Self-Fulfilling Prophecy

The self-fulfilling prophecy was originally described by a sociologist, Robert K. Merton, in 1948:

Definitions of a situation... become an integral part of the situation and thus affect subsequent developments....  The self-fulfilling prophecy is, in the beginning, a false definition of the situation evoking a new behavior which makes the originally false conception come true.  The specious validity of the self-fulfilling prophecy creates a reign of error. For the prophet will cite the actual course of events as proof that he was right from the very beginning. Such are the perversities of social logic.

(Merton was one of the few social scientists to win the National Medal of Science.)

Although the most dramatic instances of the SFP occur when a false expectation is made to come true, Gavriel Salomon (1981) has pointed out that the original prophecy doesn't actually have to be false.  The prophecy could be a true definition of the situation, which leads to behaviors that maintain that situation, even though it would otherwise change.  In which case, we would be talking about a self-sustaining prophecy

Whereas self-fulfilling prophecies lead to new behaviors, self-sustaining ones reinforce and maintain already existing behaviors... at the expense of other behaviors.  What characterizes self-sustaining prophecies and distinguishes them from self-fulling ones is that they introduce no dramatic changes in the target person's behavior, and hence are not easily detectable.

Either way, whether self-fulfilling or self-sustaining, the definition of the situation is what matters.

The self-fulfilling prophecy is crucial to the analysis of social interaction.  Each participant brings into the situation a set of conceptual baggage, variously defined as beliefs and expectations (or, in Merton's terms, "prophecies"), which guide his or her subsequent behavior -- behavior that, in various ways, tends to elicit behavior from others that is consonant with the actor's beliefs and expectations.  In Merton's analysis, the self-fulfilling prophecy is unconscious: while the actor may well be aware of his or her beliefs and expectations, and of the relation between these mental states and his or her own behavior, the actor may well be quite unaware of the effects of his or her behavior on the behavior of the other person.  


Experimenter Bias

Merton's analysis was very provocative, but his evidence for self-fulfilling prophecies was, admittedly, somewhat impressionistic.  Among the earliest (and also very provocative) experimental demonstrations of the self-fulfilling prophecy were conducted by Robert Rosenthal (1963).  Rosenthal had first raised the issue of experimenter bias in his doctoral dissertation on defense mechanisms, at UCLA (1956), and be began formal work on the problem in his first faculty job, at the University of North Dakota (1957-1962),  But he had difficulty publishing his research until he took a job at Harvard -- at which point people began taking him seriously!

Clever Hans

04S039ClvrHns.jpg (107719 bytes)Rosenthal was inspired, in part, by the example of Clever Hans, a "counting horse" who was studied in the early 20th century.  Amazingly, Hans could give correct answers to even very complicated questions, including arithmetic problems (involving either integers or fractions), reading and spelling, and identify musical tones and intervals: so long as the answer could be converted into a number, Hans could tap out the answer with his hoof.  A scholarly investigation in 1904 found no evidence of fraud: there was no evidence of intentional cuing of Hans by his owner, Mr. Von Osten; nor did the committee notice any unintentional cuing on Von Osten's part.  However, a further investigation by Otto Pfungst, aided by the psychologist Carl Stumpf (who had also been a member of the original investigating committee) revealed that unintentional cuing was, in fact, going on.  Pfungst showed that certain controls degraded Hans' performance: if he were blindfolded, or the questioner was positioned at some distance from the horse, or the questioner was ignorant of the correct answer, Hans' ability to answer questions correctly was markedly diminished.  

It turned out that Hans was actually responding to subtle cues provided by the questioner (who did not have to be his owner).

  • Pfungst noticed that after posing a question, Von Osten (and others) would lean forward to count Hans' hoof-taps.
  • At this cue, Hans would start tapping.
  • Pfungst also noted that Von Osten would lean further forward when he expected that Hans' answer would be long, which cued Hans to tap at a faster rate.
  • As Hans approached the correct answer, Von Osten would straighten up, cuing Hans to stop.     

All of this was completely unconscious on Von Osten's part, and other questioners tended to do the same thing (of course, Hans was probably conscious of what was going on, but there was no way to ask him!).

Pfungst confirmed his suspicions in a laboratory experiment, in which he himself tapped out answers to questions posed by others, and observed the same sort of unintentional cuing in 23 of 25 questioners.  

Again, it is important to understand that it is fairly clear that fraud was not involved in the Clever Hans episode.  Mr. Von Osten welcomed both investigations.  Unfortunately, he died soon after Pfungst published his analysis later in 1904 (Pfungst's book was republished in 1965, with an introduction by Rosenthal, and makes for wonderful reading).  We don't know what happened to Hans, but presumably he had a nice retirement on the farm.

Rosenthal's first demonstration of experimenter bias involved 12 undergraduate experimenters who were running rats in a maze-learning experiment in the context of an undergraduate course in research methods (Rosenthal & Fode, 1963).  All the rats were genetically identical (as identical as lab rats can be), and all were treated identically before the experiment.  Nevertheless, the undergraduate researchers were told that the rats were a specially bred "Berkeley strain" of animals; half were told that their rats had been bred to be "maze bright", and half were told that their rats were "maze dull".  As it turned out, rats labeled as "maze bright" learned the maze faster than those who had been labeled as "maze dull" -- despite the fact that the rats were identical in every respect.  Special controls insured that this outcome was not an artifact of mere error or deliberate cheating.  Rather, the experimenters seemed to treat rats differently, in line with their expectations, and this differential treatment enhanced or retarded performance in fact.

Similar results were obtained in a study of person perception (Rosenthal, 1963), in which subjects were asked to rate the people pictured in photographs in terms of the degree of success experienced by the person.  Again the data was collected by undergraduate research assistants, and again Rosenthal manipulated the experimenters' expectancies.  Half were led to expect that "most" subjects would see the targets as failures; half were led to expect that "most" would see them as successes.  The result of the experiment was very dramatic: nonoverlapping distributions of perceived success.  Subjects who were expected to perceive failure actually did, and subjects who were expected to perceive success actually did as well.

04S040Burnham1966.jpg (53199 bytes)In a very dramatic experiment on experimenter bias, Burnham (1966) had experimenters test rats whom they believed to have received surgical brain lesions that would impair their task performance (control experimenters tested rats whom they believed to be neurologically intact).  In fact, half the rats were lesioned, but the other half were intact.  Intact rats whom the experimenters believed to be lesioned performed worse than intact rats believed to be intact; and lesioned rats whom the experimenters believed to be intact performed better than lesioned rats believed to be intact.  In fact, the experimenters' beliefs about their rats' neurological status had a somewhat larger effect on the rats task performance than their actual neurological status did!

04S041RsnthlRubn78_1.jpg (59250 bytes)The experimenter bias experiments were highly controversial -- in fact, Rosenthal initially had a hard time getting them published -- and they stimulated a large number of conceptual and methodological critiques, as well as some empirical failures to replicate.  Nevertheless, there have been a large number of successful replications of the basic "E-bias" effect in many different domains.  Rosenthal and Rubin (1978) summarized the results of some 345 studies of experimenter bias.  Using an early version of a statistical technique known as meta-analysis, which combines the results of a number of different studies, Rosenthal and Rubin determined that the mean size of the experimenter bias effect was approximately 0.70 standard deviations.  This is a huge effect by the standards of social science.  Rosenthal and Rubin acknowledged the failures to replicate, but they calculated that the overall effect in 345 students was so strong that it would take 65,122 additional failures to reduce the magnitude of the E-bias effect to zero!  Some investigators may not like it, but there is no doubt that experimenter bias is real and nontrivial. 

How does experimenter bias affect the actual outcomes of experiments?  Rosenthal (1963) discussed a number of factors.

Pygmalion in the Classroom

Rosenthal's next shot in the E-bias debate was a study of "Pygmalion in the Classroom" (Rosenthal & Jacobson, 1968), an experiment involving children (grades 1-6) enrolled in a number of public elementary schools.  The schools involved a "tracking" system in which children of low, medium, and high ability were segregated into different classrooms.  Early in the fall semester of one school year, in addition to the usual standard IQ test, the children were administered what was presented to the teachers and (administrators) as a nonverbal IQ test that would predict educational progress and identify "intellectual late-bloomers" -- that is, children who might not be doing well now, but who could be expected to catch up and even exceed the achievement of their age-peers.  In fact, the test had nothing to do with IQ, or the prediction of late-blooming.  It was a fake.  Nevertheless, the children's teachers were given false feedback about the children's test scores, identifying a random 20% of pupils in each class (that is, regardless of ability level) as "late-bloomers".  

pygmalion2.gif (4520
            bytes)Later, in the spring of that same academic year, the children were retested with the standard IQ test.  On average, children in all classes and grade levels showed increases -- school really does make you smarter!.  But those children who had been falsely identified as intellectual late-bloomers showed greater gains than the other children.  This effect occurred in all grades, and in all tracks within each grade, but was most significant for grades 1 and 2.


The "Jacobson" of Rosenthal & Jacobson was Lenore Jacobson, who was a principal of an elementary school in South San Francisco, and actually initiated the study.  She had read about Rosenthal's work on experimenter bias, and had the idea that the same sort of thing might happen with teachers' expectations of students.  She wrote to Rosenthal about her idea, and the rest is history.

The Myth of Pygmalion and Galatea

pygmalion1.jpg (25680 bytes)Rosenthal and Jacobson's study took its name from the myth of Pygmalion and Galatea, as told by the Roman poet Ovid (47 BC - 7 AD).  Pygmalion, the sculptor-king of Cyprus, hated women and resolved never to marry.  But he also produced a statue depicting the most beautiful woman in the world.  Despite his misogynistic attitudes, he fell in love with his creation, and pretended that it was a real woman.  At one point, he asked Venus, the goddess of love, to give him a maiden just like his statue.  Venus, impressed by his change of heart, brought the statue to life.  Pygmalion named the maiden Galatea: they married and lived happily ever after.

There are many modern variants on the Pygmalion myth, including:

  • George Bernard Shaw's play Pygmalion (1913), in which Professor Henry Higgins transforms Eliza Doolittle, a Cockney flower girl, into a paragon of high society.
  • Pygmalion, in turn, was the basis for My Fair Lady, a famous Broadway musical by Lerner and Lowe.
  • Bedtime for Bonzo (1951), a movie starring Ronald Reagan (long before he became Governor of California and then President of the United States -- in fact, long before he became a Republican!).  In the film Reagan plays a psychology professor who raises a chimpanzee like a child to prove that environment is more important than heredity.  He does this not just out of academic interest, but also to overcome the fact that his father was a convicted criminal, so he can marry the Dean's daughter.

04S046RsnthlRbn78_2.jpg (61338 bytes)Rosenthal and Jacobson's study aroused immediate controversy in the teaching profession, and some criticism -- mainly statistical -- from psychologists and other social scientists -- particularly Elashoff and Snow (1970, 1971).  However, the essential point of the Pygmalion experiment -- that expectations can become self-fulfilling prophecies -- has been borne out many times since.  Rosenthal and Rubin (1978) counted 112 studies of "Pygmalion"-type expectancy effects in classrooms (students who are expected to perform better do), in clinics (patients who are expected to get better do), and workplaces (employees who are expected to perform better do).  These are the studies of "Everyday Situations" included in their review, which yielded a very substantial mean effect size averaging 0.88.  So, again, there is no doubt that Pygmalion effects occur, and that they are nontrivial.

Not all studies of experimenter bias and Pygmalion effects yield positive findings, raising questions about the reliability of the effects.  Rosenthal and Rubin essentially invented meta-analysis precisely to demonstrate that, aggregated across all the available studies expectancy confirmation effects were real and couldn't be ignored.  A little later, Glass et al. (1980) put meta-analysis to the same purpose in studies of psychotherapy outcome.  These days, meta-analysis is the chief means for constructing a "quantitative" (as opposed to a "qualitative", narrative) review some segment of the scientific literature.  And it all began with experimenter bias and Pygmalion effects.

Pygmalion in the Real Classroom

The original Pygmalion studies employed experimental designs, essentially turning classrooms into laboratory environments -- raising the question of whether, and how strongly, Pygmalion effects occurred under more natural classroom conditions.  

Madon, Jussim, and Eccles (1997) performed the largest such study to date, employing 1,539 6th-grade math students enrolled in the Michigan Study of Adolescent Life, a long-standing longitudinal study based at the University of Michigan.  For this purpose, the investigators extracted a number of variables, among which were:

  • Scores on a standardized test of math ability administered at the end of 5th grade, as a measure of baseline math ability.
  • Math teachers' perceptions of their students, recorded at the beginning of 6th grade. 
  • Student self-ratings of math ability, and motivation and effort in math.
  • Student performance on another standardized math test, administered in 7th grade.

The study was complex, but the general idea was to determine whether teachers' expectations of student performance influenced actual student performance, taking into account baseline differences in math ability.

  • The study confirmed the Pygmalion effect: overall teachers' perceptions of students' math ability predicted the students actual math performance, even when taking account of previous achievement in mathematics.
    • The magnitude of the effect was relatively modest, with a representative d < .30 -- an effect size considerably smaller than what Rosenthal and Rubin found in their review of "experimental" studies.
    • Interestingly, positive expectations had a stronger effect than negative expectations.  That is, the self-fulfilling prophecy was stronger when teachers expected a student to do well.
  • The overall Pygmalion effect was moderated by other factors.
    • Overestimates predicted better than underestimates. The self-fulfilling prophecy was stronger when teachers overestimate a student's ability.
    • Low-achievers were more susceptible to the Pygmaliion effect that high-achievers.  Teacher's expectations predict achievement more strongly for low-  than for high achievers.
  • The strongest finding in the study, however, was of accuracy: teachers' predictions.  That is, teachers' expectations predicted students' subsequent performance largely because their expectations were accurate reflections of students' ability.

Despite the evidence for accuracy, the evidence for the self-fulfilling prophecy should not be dismissed.  Madon et al. (1997) presented a conceptual model of how the Pygmalion effect occurs.

  • Teachers derive their perceptions of students' ability from various sources
    • Previous grades and standardized test scores
    • Assessments of student motivation
    • Assessment of student self-esteem
    • Student physical attractiveness (the halo effect, to be discussed later in the lectures on Social Perception).
    • Student demographics, such as socio-economic status.
    • Student personality characteristics, such as agreeableness and conscientiousness.
    • Student home life.
  • Teacher perceptions of students vary along several separate dimensions:
    • Actual performance
    • "Latent" talent
    • Effort on task
  • Students' performance outcomes also vary along several dimensions:
    • Future grades and standardized test performance
    • They didn't include this, but another important set of outcomes has to with student interest in a subject and motivation to do well.
  • The effect of teacher perceptions on student performance may be moderated by several factors
    • Whether the teacher over- or underestimates student ability
    • Whether the teacher has high or low expectations of the student (this is not the same thing)
    • Student self-assessment of ability in the subject at hand.
    • Student's level of previous achievement in the subject

The important point of the Pygmalion study is that the children weren't really "late-bloomers" -- but their teachers believed that they were, and treated them accordingly, and the students themselves responded in kind.  This is the three-stage model of the self-fulfilling prophecy, in which the effect is mediated by the teachers' behavior toward the students.

Thus, the effect of the cognitive transformation (identifying some students as late-bloomers) was mediated by behavioral manipulations (treating students identified as late bloomers differently than the other students).  These behavioral differences boiled down to four factors:

In other words, teachers who believed that a child was a late bloomer behaved as if that child were already doing well -- in short, they treated late-bloomers as if they were already smart.  In so doing, they created a different environment for the late bloomers than for the other children -- who after all were in the same physical classroom.  And, not surprisingly, the late-bloomers responded by -- well, blooming.

Little or none of this is done verbally.  As a result, Rosenthal became an early leader in the study of nonverbal communication through facial expressions, posture, and movement.

Subsequent analyses by Jussim (1986) and others suggest that the process is actually more complicated than this.

In the Pygmalion experiment, the teacher expectations were derived simply from information provided by the researchers, that some students in each class had been identified by a test as "intellectual late-bloomers".  But in real life, these expectations have a variety of sources:

And once the teachers begin to actually interact with students, these expectations are themselves subject to processes that may maintain or change them:

Regardless of whether they change, teachers' differential expectations concerning students lead to differential treatment of these students.  These are, essentially, the same sorts of factors identified by Rosenthal and Jacobson:

The pathway between expectations and differential treatment is itself mediated by a number of factors:

And, of course, the students will react to the way they are treated:

And the pathway between treatment and reaction is also mediated by a number of factors:

No matter how complicated, the self-fulfilling prophecy begins with a cognitive transformation -- in this case, by putting children in one category rather than another.  The teachers then behaved in accordance with this cognitive categorization, and the children responded by behaving in a manner consistent with how they were treated.  In other words:

Private belief created public reality.

Stereotype Threat: A Self-Fulling Self-Prophecy?

Perceivers can engage in expectancy-confirmation processes with respect to targets: that's what the classic self-fulfilling prophecy is all about.  And targets can react to perceivers by engaging in self-verification activity.  But it also seems that people can be both actors and targets of their own actions, and bring self-fulfilling prophecies down on themselves.  This is illustrated by the phenomenon of stereotype threat -- a concept introduced by Claude Steele, a social psychologist who, in 2014, was appointed Vice Chancellor and Provost here at UCB.

According to Steele and Aronson (1995), stereotype threat begins with an individual's awareness of some group stereotype applied to a group of which he or she is a member -- that, for example, or that African-Americans generally have low intellectual ability, or that girls and women carry a biologically indisposed toward math, science, and engineering -- or, for that matter, that white men can't jump (interestingly, it's hard to find negative stereotypes about white males in general).  When members of stereotyped groups take tests where their performance is diagnostic of their individual ability, the possibility of failure poses a double threat: (1) the shame of confirming the group stereotype; and (2) the humiliation attached to personal failure.  This "double whammy" increases anxiety, which in turn impairs performance -- especially when the performance category is highly salient, when the domain is highly self-relevant, and the test is believed to be diagnostic of one's own personal abilities.

The first experimental demonstration of stereotype threat relied on the stereotype of African-Americans as relatively low in intellectual ability -- as reflected, for example, in the infamous black-white differential in SAT and GRE scores (Steele & Aronson, 1995).  In the study, black and white college students were recruited for an experiment in which they were administered a difficult test of verbal ability. 

The results showed clear evidence of the predicted stereotype-threat effect:

Another experiment focused on gender rather than racial stereotypes -- one component being the idea that girls and women simply lack the talent for math and science that boys and men have.  As in the earlier experiment, male and female college students were recruited for a study involving a difficult test of mathematical ability. 

You can think of stereotype threat as a sort of self-fulfilling self-prophecy: a person subject to a negative group stereotype will behave in such a manner as to confirm that the stereotype is true.

Social Construction and the

Faith-Based Presidency of George W. Bush

Social cognition explores how people acquire, represent, and use social knowledge -- how they cognize the social environment in which they live.  And it is an assumption of cognitive psychology in general that reality is independent of the perceiver -- that the goal of the perceiver is to construct a valid mental representation of reality.  But one of the important insights of personality and social psychology is that, to some extent at least, people create the environment to which they respond.  In other words, reality isn't so independent of the perceiver after all.

In the run-up to the 2004 presidential election, Ron Suskind, a reporter for the Wall Street Journal, wrote an article about the 43rd president's "preternatural, faith-infused certainty in uncertain times".  In addition to his political insights, Suskind also captured succinctly the difference between social cognition as an empirical process of acquiring knowledge about the world and social construction as a process of creating that world through action and belief.  Suskind writes ("Without a Doubt", New York Times Magazine, 10/17/04).  

In the summer of 2002... I had a meeting with a senior advisor to Bush... [who] told me something that at the time I didn't fully comprehend -- but which I now believe gets to the very heart of the Bush presidency.

The aide said that guys like me were "in what we call the reality-based community," which he defined as people who "believe that solutions emerge from your judicious study of discernible reality."  I nodded and murmured something about enlightenment principle and empiricism.  He cut me off.  "That's not the way the world really works anymore," he continued.  "We're an empire now, and when we act, we create our own reality.  And while you're studying that reality -- judiciously, as you will -- we'll act again, creating other new realities, which you can study too, and that's how things will sort out.  We're history's actors... and you, all of you, will be left to just study what we do"


Expectancy Confirmation Effects 

There are two fundamental aspects of social cognition:

In large part, social reality is created through expectancy confirmation processes related to Merton's notion of the self-fulfilling prophecy:

Definitions of a situation... become an integral part of the situation and thus affect subsequent developments....  The self-fulfilling prophecy is, in the beginning, a false definition of the situation evoking a new behavior which makes the originally false conception come true.  The specious validity of the self-fulfilling prophecy creates a reign of error. For the prophet will cite the actual course of events as proof that he was right from the very beginning. Such are the perversities of social logic.  

Following Darley and Fazio (1980) and Jones (1986), we can distinguish between two different expectancy confirmation processes operating at the individual level:

            (54149 bytes)These two modes can be related to the person-situation interaction:

Through behavior, private beliefs can create a public reality.   


Beliefs Create Reality

Experimenter bias and Pygmalion effects are part of a larger class of phenomena called expectancy confirmation effects.  Expectancy confirmation effects come in two broad forms:

Perceptual confirmation effects are one way that beliefs can create private reality.  But if we are interested in the creation of public reality, we should be especially interested in behavioral confirmation effects.

Perhaps the most analytic studies of behavioral confirmation effects have been performed by Mark Snyder, Bill Swann, and their colleagues (Snyder was Swann's mentor in graduate school). 

Many of these studies have employed the "getting acquainted" paradigm, in which subjects are assigned to interact with a previously unknown partner.  The getting acquainted paradigm permits social psychologists to study a basic social process, by which one person gets acquainted with another, under controlled laboratory conditions.  


Snyder & Swann (JPSP, 1978)

Note: Just to keep things straight, Snyder & Swann published two different studies in 1978:

  1. one in the Journal of Personality & Social Psychology (JPSP), and
  2. the other in the Journal of Experimental Social Psychology (JESP).

In one "getting acquainted" study, the subjects were given a personality profile that described their female partner (who, in this experiment, didn't actually exist) as somewhat extraverted.  Then the subjects were asked to select some questions, from a larger set provided by the experimenter, that they could use to find out what the person is really like -- to fill in the details of the original personality profile.

021Bias1.jpg (44481 bytes)The study revealed a distinct bias in question-selection, whereby subjects would ask questions that were consistent with the way their partners were portrayed:

The point of the study is that almost everyone is capable of giving a positive response to any of these questions.  The most introverted person knows how to liven up a dull party (even if he wouldn't actually do it), and the most extraverted person sometimes has trouble opening up to others (if not very often).  Because everyone can give a positive response to any of these questions, they do not really reveal anything in particular about their personalities.  Introverts can look extraverted, if they're asked "extraverted" questions, and extraverts can look introverted, if they're asked introverted questions.  But another way, anyone can appear either extraverted or introverted, depending on what they're asked.  And what they're asked depends on the questioner's expectations.  


023Bias2.jpg (45503
            bytes)In an extension of the getting-acquainted paradigm, Snyder and Swann again presented subjects with profiles depicting their partners as extraverted or introverted, and again allowed them to select questions in order to fill in the details and find out what their partner was really like.  And again they observed a bias in question-selection, in which extraverts were asked questions that were congruent with extraversion, and introverts asked questions congruent with introversion.  But this time, there really was an interaction partner (randomly assigned to conditions, of course), and she really got to respond to the questions.  68S&S782Ratings.JPG (51964 bytes)The conversation took place over a telephone (again, simulating a real-life "getting acquainted" scenario), permitting the partner's responses to be recorded and evaluated by judges who were kept blind to the expectations provided to the subjects or the subjects' questions (i.e., all they heard were the partners' answers).  Again, partners whom the subjects believed to be extraverted received higher ratings on extraversion than partners whom the subjects believed to be introverted -- and vice-versa for introversion.  

Here is the self-fulfilling prophecy in operation: actors who believed that targets were extraverted (or introverted) behaved in such a manner as to elicit extraverted (or introverted) behavior from the targets.


Snyder, Tanke, & Berscheid (JPSP, 1977)

Similar findings were obtained in another study on social stereotypes associated with attractiveness (Berscheid is an expert on the social psychology of physical attraction and romantic love). In this study, male subjects were presented with a photograph of their interaction partner, instead of a narrative profile.  The photograph depicted their female partners as either relatively attractive or relatively plain.  Of course, the partners were randomly assigned to conditions, independent of their actual physical attractiveness, and because the entire interaction again took place over the telephone, the subjects were none the wiser.  

70Anticipatory.JPG (41479 bytes)Even before the conversation took place, the subjects were asked to describe their "anticipatory images" of their partners.  Following the "halo effect", in which there is a tendency to see socially desirable qualities as correlated, the subjects expected attractive partners to be sociable, poised, humorous, and socially adept than their plain partners.



71Ratings.JPG (40749
            bytes)Then they actually interacted with their partners over the telephone: as before, naive judges listened to the partners' side of the conversation, and then rated their personalities.  Partners who had been labeled as physically attractive received higher ratings of extraversion (e.g., more sociable, poised, sexually warm, and outgoing)  than those who had been labeled as plain.  



Again, we can see the self-fulfilling prophecy at work: Actors who believed that targets were extraverted (simply because they were physically attractive) behaved in such a way as to elicit extraverted behaviors from the targets.

These two experiments, taken together, clearly demonstrate the mechanisms of the self-fulfilling prophecy:


Snyder & Swann (JESP 1978)

Another experiment (one of my personal favorites in the entire social-psychological literature) provides a more detailed sequential analysis of the unfolding of a self-fulfilling prophecy.  This experiment employed a noise-gun paradigm, adapted from the experimental study of aggression, and involved three subjects, run two at a time.  The subjects, who were initially unknown to each other, were recruited for an experiment on reaction times.  Such experiments are often boring, and so in order to "keep it interesting" the task was presented as a game with some special features:

The experiment was run in two phases, each with two subjects at a time.

Phase 1 of the experiment involved two subjects selected at random (the third waited to participate in Phase 2).  Both subjects filled out a hostility questionnaire, and then each was randomly assigned a role. 

On the first trial, the Target (T) was seated at the reaction time apparatus, and the Labeling Perceiver (LP) was seated at the noise gun.  More LPs who anticipated interacting with a hostile opponent selected the highest intensity of the noise gun for this very first trial, compared to LPs who anticipated interacting with a cooperative opponent.  Thus, LPs treated Ts in a manner that was consistent with their expectations.


On the next trial, the subjects exchanged places, with the Target seated at the RT apparatus.  Targets who had been labeled as hostile (and treated as such) selected a higher noise gun intensity than those who had been labeled cooperative (and also treated as such).  



During the remainder of Phase 1, the Target and the Labeling Perceiver took three more turns each, over which Snyder and Swann generally observed an escalation of the labeling effect.  

At the conclusion of Phase 1, Labeling Perceivers rated targets who had been labeled as hostile as more hostile, in fact, than those who had been labeled as cooperative.  Thus, individuals who had been labeled as hostile (or cooperative), and treated as such, actually came to behave in a hostile (or cooperative) manner.

In Phase 2 of the experiment, the Labeling Perceiver was dismissed, and the Target continued the experiment with the third subject, who was assigned the role of Naive Perceiver, because he received no information about the Target at the outset of the experiment.  

For the first trial of Phase 2, the Target was seated at the noise gun, and the Naive Perceiver at the RT apparatus.  As in Phase 2, Ts and NPs exchanged places for a total of 4 trials each.  

Across the four trials of Phase 2, Targets who had been labeled as hostile at the outset of Phase 1 chose higher noise-gun intensities than those who had been labeled as cooperative, but the Target's label interacted with the Target's attribution for his own behavior.  Targets who had been labeled as hostile in Phase 1 continued to behave in a hostile manner in Phase 2, but only those who had been encouraged to view their own behavior as a product of their personality dispositions.  There was no effect of label among subjects who had been encouraged to view their own behavior as a product of the situation.


At the conclusion of Phase 2, the Naive Perceivers rated Targets who had been labeled hostile as more hostile compared to those who had been labeled as cooperative -- but again, this effect occurred only for Targets who had been encouraged to view their own behavior in dispositional terms.

Putting the two phases together, we can see the full spectrum of self-fulfilling prophecy effects:

So, in this experiment, the self-fulfilling prophecy arises depending on the Labeling Perceiver's first impression of the Target, but the self-fulfilling prophecy persists depending on the Target's interpretation of his own behavior.


Self-Verification Effects

There are many other experimental analyses of the self-fulfilling prophecy, and of the General Social Interaction Cycle, including:

Many of these studies illustrate a general characteristic of expectancy confirmation, which is an amplification effect -- a vicious (or, in some cases), a virtuous cycle, in which initial expectations lead to a little change in reality, which reinforces the expectation, which leads to a bigger change in reality, and so on.

Whether amplification occurs over time or not, the important point is that people who have expectations concerning other people tend to treat them in such a way as to elicit from those people behavior that tends to confirm their initial expectations.  The self-fulfilling prophecy can be quite powerful: perceivers themselves are generally not aware of the effect, and so attribute the target's behavior to the target rather than to their own actions.  And targets, for their part, are generally unaware of the perceivers' expectancies, so they have no opportunity to correct the perceivers' misperceptions.  

This last point raises the question of the target's role in the self-fulfilling prophecy.  The prophet's beliefs and expectations are important, but it's not the case that the target is passive in all of this.  The Snyder & Swann experiment shows that the target's attributions for his or her own behavior can make a difference to whether the self-fulfilling prophecy actually gets fulfilled.  Additional experiments on self-verification, mostly performed by Swann and his colleagues, show what can happen when the target actively tries to counteract the self-fulfilling prophecy.  

In Swann's view, there are two aspects of the process by which beliefs create reality:

Through research on self-verification, we gain further insight into the target's role as actor.

Hilton & Darley (1985)

Here's a simple example of the type: a study by James Hilton and John Darley (1985) employing the standard "getting-acquainted" paradigm, involving two genuine subjects (no confederates). 

After an uncontrolled interaction, the perceivers were asked to rate the targets.  Naive targets generally confirmed the perceivers' expectations.  Those who were expected to be cold were rated as colder.  But the informed targets did not show this effect -- if anything, they reversed it.  Apparently, during the uncontrolled interaction, they strategically behaved in such a manner as to counter the perceivers' initial expectations.

To learn more about exactly what targets can do to counter perceivers' expectations, we turn to a series of studies by Bill Swann and his colleagues -- one of which is also on my "Faustian" list.

Swann & Read (JPSP, 1981; JESP, 1981)

In this experiment, subjects were classified as "likeable" or "dislikable" based on a set of adjective self-ratings.  After being placed in the getting-acquainted situation, they were led to suspect that their interaction partner liked or disliked them (there was also a no-expectation control).  In fact, subjects classified as likeable elicited more positive reactions from their interaction partners -- especially when the partner's appraisal was incongruent with their self-conception.


Swann & Hill (JPSP, 1982)

This experiment made use of Mastermind, a parlor game popular in the 1970s in which one player (the "codemaker") sets a pattern of colored pegs in a pegboard, and the other player (the "codebreaker") has to determine what the pattern is.  Players indicate their guesses by placing colored pegs in holes, and receive feedback from their opponents (it's a great game, and you can play it on the Internet as well as at home).  In this experiment, female subjects first completed self-ratings of dominance and submissiveness, and then were paired up as codebreakers to play against the experimenter, serving as codemaker.  Unbeknownst to the subjects, their partners were really confederates of the experimenter.  During the first phase of the experiment, the subject and confederate played against the experimenter, alternating roles of leader (who decided what guess to make) and assistant (who made suggestions).  

SwannHillDesign.JPG (64761 bytes)After a break in the game, they were asked to decide for themselves who would be the leader, and who the assistant, for the next set of trials (Phase II).  The confederate then suggested either that the subject serve as leader ("You seem to be the forceful, dominant type") or assistant ("You don't seem to be the forceful, dominant type").  Of course, this feedback was completely independent of the subjects' self-concepts.  Some subjects who identified themselves as dominant received feedback that they were submissive, and vice-versa.  Naturally, this "misperception" created some consternation on the part of the subjects. 


At this point, half the subjects were given the opportunity to respond to the feedback -- through protestations of dominance (or, for that matter, submissiveness), statements, and queries to the confederate.  Later, these interactions were rated by judges who were blind to the experimental condition in which the subjects were run.

            (43863 bytes)Subjects who received discrepant feedback were more resistant to the feedback than those who received feedback that was congruent with their self-images -- regardless of whether their self-image was dominant or submissive.  Submissive subjects were quite aggressive in defending their submissiveness!


            (40711 bytes)Similarly, dominant subjects who received discrepant feedback were more dominant, compared to submissive subjects in the same condition.


043Change.jpg (43356
            bytes)At the very end of the experiment, subjects were asked to rate their personalities again.  Subjects who received feedback that was consistent with their self-concepts showed little change from the pretest to the post-test.  Subjects who received discrepant feedback showed more change, but those who had been given the "interaction opportunity" to correct the discrepant feedback showed less change than those who had not received such an opportunity.


Thus, when given an opportunity to do so, subjects will behave in such a manner as to correct another's erroneous perceptions of them, and to conserve their own self-concepts.  What emerges is what Swann has characterized as a battle of wills between perceptual and behavioral confirmation effects on the part of the perceiver, and self-verification effects on the part of the target.


Swann and Ely (1984)  

This line of research came to a climax with a study that directly pitted expectancy confirmation against self-verification, and showed that targets' self-verification efforts can actually alter perceivers' impressions of targets.  In this study, undergraduate women were recruited for a study of the interview process.  The procedure is complicated, so follow carefully!

In Phase 1 of the experiment, the experimenters manipulated the perceivers' expectancies: ostensibly, based on prior ratings by others (parents, family, friends, psychologists), the targets were described as extraverted or introverted.
Slide97.JPG (55627
          bytes)Then the perceiver was provided 12 possible questions, and asked to select 5 of these to ask the target: 6 of these were related to extraversion and 6 of these were related to introversion, though the questions were not quite as blatant as in the earlier experiment by Snyder & Swann (JPSP 1978).  In line with the earlier study, the perceivers showed a "confirmatory" bias, choosing to ask questions about extraversion to targets who had been described as extraverted.  This was especially the case when they had high certainty about the target's personality.

Slide98.JPG (58255
            bytes)Swann and Ely also had blind judges rate the targets on the basis of their end of the conversation -- that is, they had no idea what questions the perceiver had asked -- just what they had said in reply.  This time, in contrast with the earlier experiment, the targets generally behaved consistently with their self-concepts -- especially when their "self-certainty" was high.  When the targets were low in self-certainty, their behavior was less determined by their self-concepts -- and especially when the perceiver's level of certainty was high.  So, already we see a conflict between expectancy confirmation processes and self-verification processes.  Targets can resist the perceiver's expectancies, and behave in conformity with their own self-concepts -- especially when they are more certain about their self-concepts.

Phase 2 of the experiment essentially repeated the procedure, with the perceiver selecting from a new batch of questions, and the judges blindly rating the target's responses.  

            (68136 bytes)On the perceivers' side, the results were complex.  Uncertain perceivers generally shifted to a "disconfirmatory" strategy: even when they expected to be interacting with an extravert, they tended to ask fewer "extraverted" questions; this was especially the case when target self-certainty was high.   Highly certain perceivers tended to continue with a "confirmatory" strategy, asking "extraverted" questions of targets presumed to be extraverts -- but only when the target's self-certainty was low.  It's as if he gave up on targets who didn't confirm their expectancies on the first phase, and persisted only with targets who actually showed a tendency toward expectancy confirmation.

            (57751 bytes)On the target's side, the judges' ratings continued as in Phase 1.  Despite whatever the perceivers were doing with their questions, targets who were more certain about their self-concepts continued to behave in line with those self-concepts.  Uncertain targets, interacting with a highly certain perceiver, showed some reversal -- with extraverts actually behaving in a somewhat introverted manner.


The procedure was repeated once more in Phase 3, with perceivers selecting from yet a third set of questions, and targets' third set of responses rated by blind judges.

            (63009 bytes)This time, the perceivers' behavior was essentially independent of their expectations -- especially with the less-certain perceivers.  When perceiver certainty was high, it's as if they made one last, halfhearted attempt to elicit expectancy-confirming behavior from the targets.



            (57840 bytes)And the judges' ratings of the targets continued as before.  When the target self-certainty was high, her behavior was congruent with her self-concept, regardless of the perceiver's initial level of certainty.  When target self-certainty was low, there was no effect of her self-concept -- but then again, it's doubtful that the subjects had a concept of themselves in this domain in the first place!  The important point is that there is no effect of expectancies, either.


            (52105 bytes)Finally, the perceivers were asked to rate their final impressions of their targets.  Remember, at the outset of the experiment the perceivers' expectancies were incongruent with the targets' self-concepts.  Initially certain perceivers ended up more even-handed in their judgments, giving rating targets who considered themselves to be extraverts as no more extraverted than those who considered themselves to be introverts.  And less-certain perceivers actually reversed their expectancies, so that their ratings fell more clearly in line with the targets' own self-concepts.


In the final analysis, then, in the "battle of wills", the target will eventually win out.  Given an opportunity, targets will tend to correct the perceiver's erroneous expectations.  In extreme cases, targets will revise these expectations entirely.  At least, they will dampen the usual expectancy-confirmation processes.  But this countercontrol by the target requires that two conditions be in place:

  1. The perceiver has to be somewhat uncertain about his or her expectations.
  2. The target has to be quite certain about him- or herself.
A footnote: It's fairly clear that the experiment didn't turn out quite as intended.  It's essentially a standard 2x2 design, with high or low levels of perceiver certainty crossed with high or low levels of target self-certainty.  The experimenters almost certainly wanted to see expectancy confirmation in at least one cell -- where the perceiver is highly certain about the target, but the target is relatively uncertain about herself.  But they didn't really get this.  Instead, the perceivers' expectancies got revised under all conditions of the experiment.  And, I suspect, they also wanted the effects to be symmetrical: that the performance of introverts would mirror that of extraverts.  They didn't really get this effect either -- perhaps because their sample of targets, being drawn from the population of college students, was somewhat biased toward extraversion in the first place.  No matter: it's still a beautiful experiment.

The Fate of the Self-Fulling Prophecy

Expectancy confirmation effects -- whether in the form of experimenter bias or the Pygmalion effect have been very controversial.  I've already reference the critique of Elashoff and Snow (1970, 1971), who were very vigorous in their criticisms of the original Pygmalion experiment.  As we might expect, most critical analysis has focused on the classroom, and the influence of teacher expectations on student performance. 

The most sustained analysis of classroom self-fulfilling prophecies has come from Lee Jussim and his colleagues.  Here are some examples.
Perhaps the most sustained advocate of SFPs is Bill Swann, but even he emphasizes that expectancy confirmation effects are counteracted by self-verification effects. 

As Madon et al. (2011) note, social constructivism, including the self-fulfilling prophecy and various kinds of expectancy-confirmation effects, is almost axiomatic in social psychology.  We don't simply perceive the social situation, including the others we encounter in it: we construct a mental representation of the situation by virtue of our perceptual-cognitive processes; and this perceptual-cognitive activity has consequences.  Because we behave in accordance with our mental representation of the situation, our behavior can alter the situation itself, shaping the situation along the lines of our perception of it. 

Social constructivism has been controversial, however, for a couple of reasons.

Of course, to say that expectancy confirmation plays some role in social interaction is not to say that there's nothing else going on.  Experiments like Rosenthal's, Pygmalion in the Classroom, and the studies of Snyder and Swann are important because they illuminate an aspect of social interaction that would otherwise go unnoticed. 

Most of the critical fire directed at the SFP has been aimed at Pygmalion in the classroom: Elashoff and Snow's early critique, and the more recent studies of Jussim and his associates.  The critical position is that SFPs are a lot less powerful than we might have thought they were.  That may be true, but it may be true for other reasons.  The Pygmalion study had an enormous influence on education schools, one result was that teachers were taught to be more aware of, and to try to overcome, any stereotypes and prejudices that they might harbor toward their students.  "Every child can learn, and no child should be left behind".  Sound familiar.  In this new educational environment, with its emphasis on mainstreaming and avoidance of "tracking", it would be surprising to find that SFPs weren't relatively weak in educational settings.  But that doesn't mean that they don't still affect what goes on in the classroom, and any weakness of SFPs in the classroom says nothing about their strength in other social situations.

So the position taken in this course is emphatically not that there is no objective world out there, and it is emphatically not that we can perceive anything that we want to, unconstrained by any sort of objective reality; it is emphatically not that expectations are decisive for social interaction. 

We'll come up against this issue again, when the question concerns the accuracy of Social Perception.

Social Cognition and Social Construction

Jones (1990), summarizing these sorts of studies in an expansion of his 1986 list, described three different expectancy confirmation effects:

  1. Behavioral Confirmation: where the perceiver's beliefs and expectancies cause him to behave in such a way as to elicit behavior from the target, in turn, which objectively confirms the perceiver's expectations.
  2. Perceptual Confirmation: where the perceiver's behavior is clear, but the target's behavioral response is ambiguous, and the perceiver's beliefs and expectancies influence the perceiver's interpretation of the target's behavior.
  3. Self-Verification: where the target challenges expectancy-confirmation effects, and displays behavior that will reinforce his or her own self-concept, and correct the perceiver's erroneous expectations and beliefs.

At a somewhat higher level of analysis, these studies reinforce the distinction between the physical environment -- temperature, humidity, elevation, pollution, noise level, and the like -- which is the province of ecological (or environmental) psychology, and the social environment of other people -- their presence and their activity, their expectations, demands, and rewards -- which is the province of social psychology.  But while much of classic experimental social psychology focused on the objective social environment, as it would be described by a third person, the cognitive perspective on social interaction focuses on the subjective social environment, as it is cognitively constructed by the individual actors in a social situation.  

From a cognitive point of view, it is the subjective environment -- whether physical or social -- that really determines the individual's behavior.

But at an even higher level of analysis, these studies, and many more like them, underscore the bidirectional relationship between the person and the environment. 

Beliefs shape, and sometimes create, reality.


This page last revised 02/05/2016.