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


Midterm Examination

Narrative Review


In the past, my practice has been to conduct a pre-exam review in class, accompanied by some illustrations, followed by a question-and-answer session.  That takes a lot of time, however -- time that might be better devoted to students' questions.  Accordingly, I now prepare written "narrative" review, prior to each exam, which will leave extra time for questions in the review session itself.  

Click here for general information about exams in this course.

Link to the slides used in the in-class "Q&A" session.

The Exam

The exam is scheduled for a Wednesday.  There will be a "Q&A" review, conducted in class, on the Monday before the exam.  Students with disability accommodations should get these to me as soon as possible, so that we can make the appropriate arrangements.



The exam will cover all lectures and required readings to date:

as well as the corresponding chapters of the required texts:

There are lots of resources available for the examination, in addition to the in-class review session:

In addition, students are encouraged to post questions to the Forum area of the course website on bCourses.  I will do my best to respond to them, provided that they are posted no later than noon on Tuesday, March 16.  Do not send questions by private Email to either me or the GSIs: we want to make sure that everybody in class has equal access to the exchanges.


Exam Construction and Scoring

The exam will consist of roughly 15 short-answer questions Roughly half will be drawn directly from the lectures, roughly half will be drawn directly from the text.  Of course, there is some overlap between lectures and readings. As a rule, I try to have at least one question from each lecture, and at least one question from each chapter of the required reading.  

The exam requires only short answers.  In no case are more than five (5) sentences required to answer a question; often, the job can be done in fewer than five sentences. 

Answers should be written in ink on the exam itself.  Exams written in pencil are not eligible for regrading.  No "blue books" are required for the exam.

The focus of my exam is on basic concepts and principles.  There are no questions about names or dates (though names and dates may appear in questions).  There are no questions about picky details.  There are no questions about specific experiments, though you should be able to recognize the implications of the phenomena revealed by some classic experiments.  There are no intentionally tricky questions: I want you to understand basic concepts and principles, not the exceptions to the rules.

The exam will be scored following a guide which I have prepared, and which will be posted to the course website following the exam.  The guide is just that: alternative approaches to answering a question will also be considered.  We will do our best to have grades posted within a week of the exam.  

When grades are posted, there will also be an announcement to this effect, and then students will have an opportunity to correct any errors that may have occurred in the grading.

Exams will be returned in discussion sections.


The introductory lecture focused on the parallels, and the differences, between cognitive psychology and social cognition.  

You should understand the parallels between social cognition and a typical cognitive psychology course.  Both include topics such as perception, memory, language, reasoning, problem-solving, judgment, and decision-making; the development of social cognition, individual differences in social intelligence, and social-cognitive neuroscience.  One way to think about social cognition is that it is cognitive psychology applied to the world of social objects -- meaning people (including ourselves), their behaviors, and the situations in which we encounter them.

But social cognition differs from "nonsocial" cognition in a number of ways.  

Some of these are quantitative, meaning that social and nonsocial cognition share the same features, but differ in degree:

Other differences are qualitative, meaning that social cognition has features that simply aren't present in the nonsocial case:


The Cognitive Perspective on Social Interaction

The cognitive perspective on social interaction begins with the simple proposition that an individual's behavior in some situation depends on the meaning that situation has for him or her.  Therefore, in order to understand an individual's social behavior, we must understand the social situation from that individual's point of view (a point which is expressed nicely in a famous quote from Harper Lee's To Kill a Mockingbird)..

Personality and social psychology initially developed as "two cultures", each representing one aspect of "Lewin's Grand Truism" that B = f(P, E).  For personality psychologists, behavior was determined by personal features such as traits and attitudes: this is known as the doctrine of traits..  For social psychologists, behavior was determined by features of the physical and, especially, social situation: this is known as the doctrine of situationism..  Put another way, traditional personality and social psychology viewed between P and E as acting independently of each other in the determination of behavior.  In particular, experimental social psychology tended to analyze such behavioral phenomena as conformity, obedience, aggression, and interpersonal attraction as a function of the situation.

At least in America, experimental social psychology soon came under the sway of behaviorism, with its emphasis on social behavior as responses controlled by environmental stimuli.  This led to a redefinition of social psychology as the psychology of social influence, as expressed in the Doctrine of Situationism and various assertions of "the power of the situation" and "situation blindness". 

The classic expression of situationism in social psychology is the Milgram experiment on obedience to authority.  But understanding the experiment "from the subject's point of view", as described by Martin Orne, shows how situational influence is mediated by cognitive processes. 

But the assertion of person-situation independence was distinctly un-Lewinian, because Lewin himself insisted that the person and the environment were interdependent variables constituting a single psychological field -- one in which, following Gestalt psychology, "the whole is greater than the sum of its parts".  The Lewinian view was revived by Bowers' Doctrine of Interactionism: "Situations are as much a function of the situation as the person's behavior is a function of the situation".

You should understand the four modes by which persons affect their environments: Evocation, Selection, Behavioral Manipulation, and Cognitive Transformation.  Of course, in a course on social cognition, we're particularly interested in cognitive transformations. 

Bandura's Doctrine of Reciprocal Determinism expands Bowers' Doctrine of Interactionism.  Lewin construed causality as unidirectional, flowing from P and E to B; Bowers did as well, with causality flowing from P to E as well.  But the environment can influence the person as well.  More important, Bandura pointed out that behavior can alter both the environment in which it occurs and the person who engages in it.  In reciprocal determinism, all three elements in Lewin's Grand Truism serve as both causes and effects of the others -- a situation he calls "triadic reciprocality".

You should understand the distinction between objective and subjective environments, as expressed in the Thomas Theorem and symbolic interactionism, and its implications for the situational control of behavior.That is to say, it's not the situation that controls behavior -- it's the perceived situation.  And the perceived situation is a cognitive construction.

The cognitive perspective on social interaction, including reciprocal causality, can be illustrated in the General Social Interaction Cycle, in which two people are arbitrarily assigned the roles of Actor and Target, respectively; but it soon becomes apparent that the Target is also an actor, and that the Actor is the target of the Target's actions.  It's also clear that cognition, in the form of expectations, beliefs, and impressions, permeates every aspect of the cycle.

You should understand how the concept of the self-fulfilling prophecy (which begins with the prophet's definition of the situation) leads to an experimental model for studying cognitive transformations under the guise of expectancy confirmation effects, as in Rosenthal's work on "Pygmaliion in the Classroom".  In the Pygmalion experiment, teachers' beliefs about their students leads them to behave in such a way towards their students as to elicit behavior which confirms those very same, initially false, beliefs.

And finally there is Steele's concept of stereotype threat, which can be thought of as a self-fulfilling prophecy about oneself.

The self-fulfilling prophecy is also illustrated by Snyder and Swann (the "Noise Gun" experiment), among others, on interpersonal expectancy effects, which come in two forms:

But the relation between Perceiver and Target is not solely a matter of confirming the perceiver's expectations -- not least because the Target, being a sentient being, is trying to control the perceiver's perception.  In research on self-verification (e.g., the "Mastermind" experiment), Swann and his colleagues have shown how a target, when given an opportunity to correct the perceiver's erroneous misconceptions, can do so.  In what Swann calls a "battle of wills", self-verification will often win over expectancy confirmation.

All of this research illustrates the basic principle that social reality is not independent of the person experiencing it.  Rather, social reality is cognitively constructed, in the mind of the perceiver; and it is this cognitive construction that determines the perceiver's behavior. 



Fiske and Taylor present an alternative view of some of this historical material in their Chapter 1.  For example, their treatment of Lewin is very good, and complements material presented in lecture.

"Brains matter", as F&T say, but not for this midterm.  However, keep this section in mind for when we get to the lectures on Social-Cognitive Neuroscience.  Cultures do matter, though, and you should understand the distinction between independence and interdependence.

Dual-process theories abound in social psychology, and F&T's Chapter 2 is devoted to them, especially as they are relevant to person perception, causal attribution, and attitudes..  You should understand the distinction between automatic and controlled processes, , the relevance of priming (typically construed as automatic), and the implications of automaticity for conscious control (some theorists think that automaticity dominates control, and consciousness counts for little or nothing in behavior).  But setting that issue aside, there are certain features of motivation that help determine whether processing is automatic or controlled.

Zerubavel's Chapter 1 lays out the agenda for cognitive sociology.  Social cognition is cognitive psychology with people as objects, but both social cognition and cognitive psychology focus on the individual's mind and behavior.  Cognitive sociology (also known as the sociology of knowledge) looks at collective, rather than individual, cognition -- that is, how members of groups (what Zerubavel calls "thought communities") come to perceive, remember, and think about the world in the same way through what he calls "cognitive socialization".  Figure 1.1 affords a good summary of the chapter.

Social Perception

The lectures on social perception began by repeating themes encountered in any course on perception:

The question of where knowledge comes from, as represented by the philosophical debate between nativists and empiricists.  Psychology has its roots in empiricism: hence the emphasis in early experimental psychology on sensation and perception: knowledge is acquired through experience, and experience is mediated by sensory-perceptual mechanisms.  But, as we'll see, there's room for nativism as well.

Cognitive psychology offers two views of perception:

The distinction between sensation and perception:

So often, our impressions of other people are represented linguistically in a set of trait adjectives and other words, or short phrases, which describe what a person is like -- physical attributes, behavioral information, social relations, characteristic situations, origins, and functional properties. 

The term "person perception" was coined by Bruner and Tagiuri, but the concept was studied earlier by Asch, who called it "impression formation".  You should understand the basic features of Asch's impression formation paradigm, and its principal results -- in particular his distinction between central and peripheral traits, and his discovery of order effects.

The next question is why some traits are central to impression formation and not others?  Wishner concluded that central traits were more highly correlated with other traits, so that they provided more information about the person than peripheral traits.  Knowing that a person is warm or cold, intelligent or unintelligent, simply tells you more about that person than knowing that he or she is polite or blunt.  Later, Rosenberg and Sedlak refined this conclusion: warm-cold and intelligent-unintelligent are central because they load highly on the two major dimensions of trait space, social and intellectual good-bad.  This two-dimensional solution was reiterated in Fiske's argument that the dimensions of warm-cold and competent-incompetent organize our impressions of other people, including group stereotypes.

A more recent proposal, which has not been tested (hint, hint) is that traits will be central to impression formation that load highly on any of the "Big Five" personality traits: neuroticism, extraversion, agreeableness, conscientiousness, and openness to experience.  the Big Five dimensions are ubiquitous in personality ratings; but interestingly, they also emerge in ratings of complete strangers, which suggests that the Big Five resides in the head of the perceiver as well as in the personality of the perceived.

Note: The notion of central traits implies that we carry around in our heads an "implicit personality theory" (a term also coined by Bruner and Tagiuri) -- a description of "the Generalized Other" (a phrase Cronbach borrowed from G.H. Mead).  You should understand what it means to have an implicit personality theory.  Osgood's tridimensional model of implicit personality theory, involving dimensions of evaluation, potency, and activity was rejected by Rosenberg in favor of a structure involving two correlated dimensions of evaluation: social (warm-cold) and intellectual (intelligent-unintelligent).  An alternative structure of implicit personality theory, based on The Big Five, is suggested by a study in which those five dimensions, which emerge strongly when judges make ratings of people they know well, also emerge when judges make ratings of perfect strangers.  I didn't get to talk about IPT in lectures, so you're not responsible for it, but it's relevant to central traits.  For those who are interested, there's a big discussion in the Lecture Supplements.

Note: Asch, as a Gestalt psychologist, viewed impressions of personality as a whole that was "more than the sum of its parts".  But another approach, identified with Anderson's "Cognitive Algebra", is that the whole is exactly the sum of its parts -- or, more precisely, that global judgments of likeability are given by the weighted average of the likeability values associated with the individual stimuli in the trait ensemble.  Anderson then systematically tested the adding, averaging, and weighted averaging models of impression formation.  I didn't have time to talk about cognitive algebra, either, but it's a good example of mathematical modeling of social cognition, and interested students will find a discussion in the Lecture Supplements. 

The impression-formation paradigm, with its emphasis on implicit personality theory as a cognitive background, constitutes a Helmholtzian, constructivist approach to person perception.  But the Gibsonian, ecological view also is represented in person perception, particularly by the work of McArthur and Baron.

Ekman's work on facial expressions of emotion suggests that we have evolved a perceptual apparatus that enables us to perceive basic emotions like joy, sadness, fear, anger, surprise, and disgust from the expressions on people's faces -- or, more specifically, by the stimulus information provided by the movements of the facial musculature.  You should know, for example, the difference between the "Duchenne smile" and the "Pan-American smile". 

Emotion recognition appears to be highly accurate, even across cultures, giving rise to the Universality Thesis that facial expressions are an innate product of our evolutionary heritage, and our ability to recognize them is, likewise, a product of an evolved perceptual module.  The Universality Thesis is widely accepted, but it has some problems.  For example, accuracy is not constant across emotions.  Happiness appears to be universally detected, but the negative emotions show more cross-cultural variability.  Moreover, the strongest evidence for universality is based on studies which employ posed expressions and forced-choice response formats.  Spontaneous expressions are less accurately perceived, especially when perceivers must generate their own responses. 

The detection of deception (otherwise known as lying) also illustrates the Gibsonian, ecological view.  Again, the idea is that we have evolved the ability to extract information from the face, voice, posture, and gesture that indicate whether the person is telling the truth.  It's a good story, but it confronts one empirical problem, which is that we're not actually very good at detecting deception; and another, which is that we're particularly poor at detecting deception from facial and other cues.  Signal-detection analyses suggest that accuracy in lie-detection, as bad as it is, may actually be inflated by the occurrence of many false alarms among successful lie-detectors.  You don't have to know how signal-detection parameters, such as sensitivity (e.g., d') and bias (e.g., C) are calculated, but you should understand the basic concepts involved.

 An even bigger problem is that accuracy is often examined in targets who "leak" cues to de
ception -- and, in fact, liars "leak" relatively infrequently.  So one reason why perceivers are relatively poor lie-detectors is that targets are pretty good liars!  But lie-detection is possible: an objective statistical algorithm improves detection accuracy above what can be accomplished by perceivers relying on their subjective impressions.

All of this can be understood in terms of Brunswik's "lens model" for perception and judgment.  You should know the basic elements of the lens model, including the concepts of ecological validity and cue utilization.  The problems with lie-detection, or any other aspect of person perception, are manifold.  First, the distal stimulus has to provide ecologically valid cues; then the perceiver has to actually utilize these ecologically valid cues in order to perceive the stimulus accurately.  If the stimulus does not provide ecologically valid cues, or if the perceiver does not utilize them - -even worse, if the perceiver utilizes cues that are not ecologically valid -- then perception will not be accurate.  A study of lie-detection based on the lens model indicates that there are ecologically valid cues to lying, but that they are relatively weak; unfortunately, perceivers tend to think these cues are more valid than they actually are.

Similar problems crop up in studies of "gaydar".  Perceivers are better than chance at distinguishing between heterosexuals and homosexuals, presumably based on the configuration of targets' facial features.  But we know very little about the actual patterns of ecological validity and cue utilization involved.  One hypothesis is that gaydar seems to be mediated primarily by whether the person's face seems stereotypically "masculine" or "feminine".

The accuracy of gaydar can also be inflated by methodological issues.  The base rate of homosexuality in the population is relatively low, and using Bayes' Theorem to take account of base rates suggests that the accuracy of gaydar is actually pretty low.  You don't need to know the formula for Bayes' Theorem, any more than you need to know the formulas for Signal Detection Theory, but you should be clear on the basic concepts -- not least because, once you begin to take account of base rates, the accuracy of lie-detection seems even worse than the most optimistic view.  



F&T discuss person perception in their Chapter 3, as "Attention and Encoding".  The idea is that you perceive something by paying attention to it, and that perceiving something encodes a representation of it in memory.  

F&T's Chapters 9 and 10 are about attitudes, and frankly I wouldn't have included this material if I had been writing the book -- because attitudes are fundamentally emotional, not cognitive, in nature.  The link is that we have cognitions about attitudes, and that attitudes can change depending on our cognitive processing of persuasive communication.  In any event, there are some things you should pay particular attention to:

Zerubavel's Chapters 2-3 begin to instantiate the parallels he seeks to draw between cognitive psychology and cognitive sociology.  Chapter 2 represents (visual) perception at the collective level (i.e., the way groups "look" at the world", rather than the individual level, and Chapter 3 does the same thing for attention (i.e., the way group members "open" their minds to some things, and "close" them to others).  

But most important, you should understand how the concepts discussed in Chapters 2-3 illustrate Zerubavel's basic points: that there are some thoughts that people have by virtue of their membership in certain groups; and that groups are defined, cognitively, by the fact that their members share certain thoughts.


Social Memory

Perceptual activity depends on memory, and it also leaves a trace in memory.  

How are percepts represented in memory?  Marr and Nishihara proposed that most objects, such as the human body, could be represented as constructions of cylinders.  A related proposal, offered by Biederman, is that percepts can be represented as combinations of a small number of geons, or basic shapes.  But mostly, when we talk about memory as a byproduct of perception, we are talking about verbal descriptions of the distal stimulus.

The connections between perception and memory may be illustrated by the process of face recognition, as exemplified by the model of Bruce and Young.

B&Y argue that there are close parallels between face perception and the perception of both objects and words.  In each case, perception goes through the same set of basic processes, ending with some sort of naming response.

Cognitive psychology offers a taxonomy of the kinds of knowledge represented in (long-term) memory:

In social cognition, social memory boils down to person memory -- that is, memory for factual knowledge concerning the characteristics and behaviors of other people. So that makes it declarative memory.  Memory for a person's general characteristics is semantic in nature.  Memory for a person's specific behaviors and experiences is episodic in nature.  

I didn't lecture on the effects of mental set -- memorization vs. impression-formation -- on person memory, so you're not responsible for it.  But the general finding is that asking subjects to form an impression of the person described in the stimulus materials leads to better memory than asking the subject to memorize the description.  This is probably because the impression set leads to greater elaborative and organizational activity at the time of encoding.

Most of the lectures concerned representational issues -- that is, how is declarative knowledge concerning persons represented in memory.  The general framework for this discussion is provided by generic associative-network models of memory, in which individual people are represented as nodes in the network; these nodes are linked to other nodes describing their traits and behaviors by associative links.  The general structure is illustrated by the fan effect, in which response latency is a function of the number of links fanning out from the person node.  The general idea is that associative links are searched serially, and it takes time to trace down each associative link; therefore, the more you know about a person, the longer it will take to verify any particular fact about that person.  More broadly, response latencies can be employed to make inferences about the structure of knowledge stored in memory.

An important methodology in this respect is priming, where performance of one task facilitates (or, in the case of negative priming, impairs) performance on a subsequent task.  You should be familiar with the logic of priming and its application to studies of the structure of knowledge representation.

In a generic associative network model of memory, which is all we really care about in this course, each "person" is represented by a different node.  When two nodes refer to the same person, as in the statement that James Bartlett is the lawyer, facts about the lawyer aren't simply imported to the James Bartlett node.  Rather, a new associative link is established between the James Bartlett node and the node representing the lawyer.  That is why, if you know that James Bartlett rescued the kitten and that The lawyer cursed the salesgirl, it takes extra time to verify that James Bartlett cursed the salesgirl.

Another important line of research concerns schematic effects on person memory -- where the term schema (plural schemata) refers to the background knowledge, expectations, and beliefs that serve as the cognitive framework for perception, memory, and thought.  That is, how does our background knowledge of what a person is like in general -- semantic person memory -- affect our memory for what a person has done and  experienced -- episodic person memory?  The general findings of this line of research are:

One explanation of these findings is as follows:

Another explanation is in terms of an associative network model of memory:

In a test of the associative-network model, it was found that:

These experiments involved episodic person memory.  The next question is: how is semantic (trait) and episodic (behavioral) information represented in memory?  Here there are two general possibilities:

But when subjects recall behavioral episodes, they do not cluster their recall according to the traits the behaviors represent.  This lack of clustering is inconsistent with the organizational view, but consistent with the independence view.

And retrieval of trait knowledge about a person does not prime retrieval of behavioral knowledge about that same person, which is also inconsistent with the organizational view, and consistent with the independence view. 

Finally, amnesic patients, who lack episodic knowledge about their past behaviors and experiences, nonetheless retain semantic knowledge about their personality characteristics.  This dissociation (which is what it's called in neuropsychology) is especially dramatic in Tulving's case of K.C., and also (if less dramatically) in Klein's case of W.J.

So, we conclude that, in person memory, items of trait (semantic) and behavioral (episodic) knowledge are represented independently of each other.  In this view, each node in the network is a symbol that stands for some fact about the person.

Note. There are other views of the representation of person knowledge in memory.  I didn't talk about these in class, but they are discussed by Fiske and Taylor.  Here's a sketch for those who are interested:

The proceduralist and connectionist views have their advocates, but to date they haven't guided much research on person memory.  Instead, the symbolic and connectionist views form the theoretical background for asking about the neural representation of person memory in the brain. 

Ever since Lashley announced his Law of Mass Action, accounts of the neural substrates of memory have favored the connectionist view -- which is why connectionist models of memory are often touted as "neurally plausible".  

However, a recent neuroscientific study involving single-unit recording of brain activity in conscious patients yielded evidence supporting the locationist view.

Results such as these suggest that, perhaps, there is a "Grandmother Neuron" after all -- a small neural unit that represents, in neural form, all your knowledge about our grandmother.



The first part of Fiske & Taylor's Chapter 4 presents another overview of the associative-network model of memory that organized the lectures.  

F&T also present some information on proceduralist and connectionist (parallel processing) alternatives, but de-emphasize these.  Their comparison of the advantages and disadvantages of declarativist, proceduralist, and connectionist models is quite good.  But you don't have to know any details about how the models work.  

Repeat: Don't get hung up on proceduralist and connectionist models.  They're interesting, but almost all the work on person memory has been based on declarativist models.

More important is the material on the embodiment of person memory.  You should have some idea of what it means to say that person memory (or any aspect of social cognition) is "embodied".  Then you can explain it to me, because I really don't get what all the fuss is about.  

There's also a lot of material on social categories in this chapter, as an example of semantic social memory, but it is more relevant to the lectures on social categorization.  Accordingly, this aspect of the chapter is more relevant to that section of the course.  Perhaps the most relevant material is the comparison of the three models of category structure: the category view (characterized in lecture as the classical proper-set view), the prototype view, and the exemplar view.  

Zerubavel's Chapter 6 discusses the most provocative aspect of cognitive sociology: the concept of collective memory.  Again, the idea is that the past is not merely an objective fact, but is rather a socially constructed reality -- but one constructed by groups, not by individuals.  Members of different social groups will remember different events, and will remember the same events differently, than those of other social groups.  And social groups are defined, in part, by what they choose to remember and forget, and how they choose to remember it.  While individual memories are represented as associative networks (or connectionist networks), or as neural networks in the brain, collective memories are represented in such "impersonal sites" as history textbooks, memorials, and monuments.  Of particular interest, I think, is Zerubavel's concept of the "sociobiographical" memory, in which individuals remember events that happened to their groups as if they had happened to themselves.  Think of the Serbs in Kosovo: "We'll never forget 1389!".  Or the motto of Holocaust survivors: 'Never forget!".  

Such social phenomena make clear Zerubavel's essential points: 

Social Categorization

After forming an impression of the person (and the situation), and encoding this information in memory, the fundamental cognitive task for the social perceiver is categorization. After all, as Bruner pointed out, every act of perception is an act of categorization: in the course of giving meaning to the stimulus, the perceiver categorizes the stimulus as similar, in at least some respects, to other stimuli encountered in the past (and different from others). Moreover, categorization is critical for social interaction: persons and situations that fall into the same category will elicit similar sorts of behaviors.

You should understand the technical difference between a category, which partitions the real world into equivalence classes, and a concept, which is the mental representation of a category. But psychologists tend to use the terms concept and category interchangeably -- which is why these lectures are entitled "Social Categorization" instead of "Social Conceptualization".

You should also understand (from your earlier courses in psychology or cognitive science) the different models of conceptual structure: in particular, the classical view of categories as proper sets; the revisionist "prototype" view of categories as fuzzy sets; the exemplar view. There is a good comparison of the prototype and exemplar views of concept representation in Fiske and Taylor's Chapter 4.

A basic feature of social cognition is the distinction between ingroup and outgroup, or "us" versus "them" -- or, in technical terms, between members of one's ingroup and members of any outgroup.

You should be familiar with the Robbers Cave Experiment and studies of the minimal group paradigm.

You should have some feeling for what the natural categories are in the social domain -- categories like:
Some categories, like sex (gender) look straightforward, but have hidden complexities: there may be more than two sexes, and gender identity, gender role, and sexual orientation may be orthogonal to biological sex. But the important point of the bulk of these lectures was to illustrate how widely various "natural" categories differ across cultures and time.
We usually think of "natural" categories as residing in the world, existing independently of the mind, and represented by mental categories. But social categories look a lot like cognitive and social constructions, imposed on the world, not perceived in the world.
Note: Setting the issue of the ontological status of social categories aside, there is the matter of the structure of social categories.  The literature on social categorization recapitulates the literature on nonsocial categorization. For example, personality types like melancholic and phlegmatic don't seem to be structured as proper sets, but rather seem to more closely fit a prototype view in which two continuous dimensions, strength and speed of emotional response, underlie what looks like a discrete typology.  There's a lot on this issue in Fiske and Taylor, and also in the Lecture Supplements.
  • Psychiatric diagnoses, which can be viewed as fuzzy sets, also look more like fuzzy sets, organized by a principle of family resemblance. While psychiatric novices may rely on prototypes when making diagnoses, psychiatric experts may rely on exemplars.
  • Personality traits like agreeable and aloof can also thought to be categories -- but categories of actions rather than of personsBuss and Craik, in their "act frequency" model,  argued that these, too, seem to be structured as fuzzy sets, represented by prototypes.
I didn't get to talk about exemplars much, which is OK, because most social-cognitive work on social categories is based on one version or another of prototype theory (e.g., the dimensional view or the featural view). But Fiske and Taylor do comparison the prototype and exemplar views.  Although these are alternate models of conceptual structure, research suggests that while novices in a domain rely on prototypes for categorization, experts tend to rely on exemplars (or, at least, subordinate-level categories).

The most prominent social categories seem to be stereotypes of various sorts.  You should know something about the findings of the "Princeton Trilogy" of studies, and the re-assessment of these studies by Devine and Elliot -- particularly their distinction between one's knowledge of a cultural stereotype and one's personal belief.

Stereotypes also seem to be organized as fuzzy sets, represented by prototypes: nobody thinks that every German is industrious, but people do think that Germans are more industrious than people general.

Of course, when we talk about stereotypes we're talking about people's beliefs, rather than objective reality -- beliefs which may, in fact, diverge radically from what is objectively the case (Germans might not be more industrious than the average bear).  So where do stereotypes come from?  Stereotypes can be explained in economic terms (like ethnocentrism) and motivational terms (like social identity theory), but this course focuses on stereotyping as an inevitable outcome of categorization.  Stereotyped beliefs (let's not call them knowledge, exactly) sometimes reflect social learning, socialization, and acculturation processes.  They might also be based on direct experience personal encounters with the objects of stereotypes -- so that sometimes stereotypes have a "kernel of truth" to them. 

The accuracy of stereotypes can be measured in a number of different ways, including accuracy.  A stereotype may be inaccurate, in that it is an over- or underestimation of a group's standing on some dimension; but even so, stereotype beliefs may still be highly correlated with a group's actual standing.  Judd and Park's "full accuracy design" is probably the best way to assess accuracy.  In one application of the full accuracy design, J&P found that both Democrats and Republicans overestimated the liberalism of Democrats and underestimated the conservatism of Republicans; each group stereotyped the other more than they did themselves; and the degree of stereotyping was greater for those with stronger party affiliation.

In another analysis, Jussim argued that racial, ethnic, and gender stereotypes were actually pretty accurate, compared to self-reports from the stereotyped groups.  But there are not very many such studies, perhaps because of the pervasive assumption that stereotypes are fundamentally inaccurate, and they've got a variety of methodological problems, so it's probably too early to draw any conclusions about stereotype accuracy.  But that's where the evidence seems to stand at present.

But at the very least, stereotypes are grossly exaggerated beliefs, so where do they come from?  One prominent theory is that they reflect the formation of an illusory correlation, based on the feature-positive effect.  That is, people pay more attention to the presence of a feature than to its absence, and they pay more attention to the conjunction of unusual features -- thus inflating the perceived correlation between them.  I didn't lecture on this material, but if you want to know more, take a look at the relevant portion of the Lecture Supplement.

While we talk mostly about the stereotypes that perceivers carry around in their heads, it's also important to understand that stereotypes have real consequences for the objects of perception -- and not just in terms of outright prejudice and discrimination.  There are expectancy confirmation effects, for example, which might elicit stereotype-consistent behavior from stereotyped individuals.  There's the problem of attributional ambiguity, stereotype avoidance, and stereotype vulnerability.  And then there's the problem of stereotype threat, which I discussed in the lectures on "The Cognitive View", and also discussed at length in Fiske & Taylor).  Interestingly, stereotypes can also have an effect on the perceiver, in the form of stereotype lift.

Assuming that stereotypes are initially acquired through some sort of social learning process, it makes sense to think that they can be changed through social learning as well.  As ingroup members gradually learn more about outgroup members, their beliefs may become more accurate. 

However they are formed, it's commonly held that stereotypes are automatically elicited by members of the stereotyped group, and they automatically bias our impressions and behaviors toward those groups.  However, Devine argued that even after they are activated, we can exert conscious control over the effects of these stereotypes on our thoughts and actions regarding outgroup members.  Again, I didn't get to lecture on this material, but if you want to know more, take a look at the Lecture Supplement.

Banaji and Greenwald have argued that stereotypes and other attitudes an also operate unconsciously, so that we don't even know that we harbor stereotypes and prejudices.  They argue that unconscious stereotypes can be assessed by procedures such as the Implicit Association Test.  But it's not at all clear that the IAT measures personal beliefs as opposed to knowledge of social stereotypes (Devine's distinction again); and it's not at all clear that any "unconscious" stereotypes revealed by the IAT differ from our conscious attitudes.



Remember to review the material in Fiske & Taylor's Chapter 4, on theories of category structure.  You should know what the problems are with the classical view of categories, what the "prototype" and "exemplar" revisionist theories are all about, and how they solve the problems with the classical view.

F&T's Chapter 11 focuses on social stereotyping, with a very nice presentation of social identity theory, ingroup/outgroup effects, ingroup favoritism and outgroup homogeneity. Of particular interest is their discussion of "entitativity", and group essentialism -- the idea that there are essential differences between groups that, fundamentally, exist only as social constructions.

Also pay attention to the forms of "subtle stereotyping" -- automatic, ambiguous, and ambivalent. Finally, the literature on stereotype threat, which is a mechanism by which outgroup members actually seem to incorporate the ingroup's stereotype of them.  The material on automatic stereotyping again refers to the IAT.

There is also a nice discussion of stereotype threat, which I discussed in the lectures on Social Categorization.

Fiske's stereotype Content Model, covered in both Chs. 11 and 12, is derived from Rosenberg's work on central traits (intellectual and social good-bad), as discussed in the lectures on Social Perception.

Stereotyping leads almost inevitably to prejudice, the emotional counterpart of cognitive stereotyping, and which is the subject of F&T's Chapter 12. You should understand something of how cognitive and emotional factors interact in the cases of racial, gender, age, and sexual prejudice

An interesting point made throughout this chapter has to do with the effects of prejudice on the prejudiced person, as well as the target of the prejudice -- phenomena like attributional ambiguity and stereotype threat

Zerubavel's Chapters 4 and 5, on social categorization, continue the theme opened up in his earlier chapters on perception and memory. Zerubavel shows how categorization can follow social norms, as in kosher laws about edibility, or state laws about eligibility for marriage. His comparison of rigid, fuzzy, and flexible divisions between categories recapitulates (without actually saying so) the discussion of proper sets versus prototypes. Don't worry about the semiotic square in Chapter 5.  If you get it, perhaps you'll explain it to me; but the distinctions between.indicators, icons, and symbols is useful.


Social Judgment and Inference

Categorization is just one of many judgments made in the course of social interaction. After all, impression -- formation  deciding whether a person is intelligent or not, extraverted or introverted, is as much a matter of judgment as it is a matter of perception. Another fundamental social-cognitive task is causal attribution -- that is, explaining why people do the things that they do. In this course, causal attribution serves as a particularly well-studied example of social judgment in general.  When attribution is made to a person, another task comes into play -- moral judgment.  I don't have time to talk about moral judgment, but there is some material on the topic in the Lecture Supplements for those who are interested.

Remember that we're talking about what Heider called phenomenal causality -- that is, what ordinary people believe caused something to happen. Determining actual causality is a matter for formal science. Based on "Lewin's grand truism", B = f(P, E), Heider asserted that people tend to attribute behavior to either personal or situational factors, and this person-situation dichotomy governed research and theory in the area of causal attribution ever since (at least, until recently).

You should understand how Kelly's "covariation calculus" for causal attribution works: how information about consensus, consistency, and distinctiveness, derived from multiple observations of behavior, leads people to attribute behavior to the actor, target, context, or some combination of these. You should also know how McArthur tested Kelly's model, and the general results of her research.

Note: Kelly's model cannot apply to single observations of behavior, leading to the use of causal schemata such as multiple sufficient causes and multiple necessary causes to make inferences about causality. The Lecture supplements contain information on how the common judgment heuristics -- representativeness, availability, simulation, and anchoring and adjustment -- can play a role in causal attribution.  But you're not responsible for it.

Note:  Nor are you responsible for knowing about the causality implicit in language -- in language semantics, rather than language syntax. Behavioral action verbs tend to invoke the agent-patient schema, by which causality tends to be attributed to the agent (who is usually the grammatical subject); mental state verbs tend to invoke the stimulus-experiencer schema, by which causality tends to be attributed to the stimulus (who can be either the grammatical subject or the grammatical object). You should also know why this is not a "Whorffian" result, but actually seems more like a "contra-Whorffian" result.

People tend to depart from the covariation calculus, even when there is enough information to apply it, resulting in certain "errors", or "biases", in causal attribution. You should understand the three most famous of these:

The covariation calculus exemplifies normative rationality, so you should know what that entails.  Documenting departures from prescriptions for normative rationality, like the fundamental attribution error, constitute a long-standing theme in social cognition research.  

Note: You should have an appreciations for this "errors and biases" literature, but you don't have to agree with me that there exists a substantial "People Are Stupid" school in social psychology.  That's my thing, and I do want to plant the idea in your mind, so that you'll recognize stupidism when you see it, but I'd never test you on it.

At the same time, the evidence for these errors and biases should not be overstated -- at least, if Malle's (2006) meta-analysis is right. Malle found that evidence for the Actor-Observer Difference in causal attribution appears to be entirely lacking: people appear to employ the same balance of internal and external attributions about their own behavior as they do for other people. There is some evidence for a (weak) self-serving bias -- but it may have more to do with protecting self-esteem, and isn't properly characterized as an error -- that is, as a mistake.

In the final analysis, though, the whole business of discovering errors in causal attribution may be based on a misinterpretation of "Lewin's grand truism", because Lewin didn't consider P and E to be independent causal factors, but rather interdependent -- just as the Doctrine of Interactionism says they are. If Ps affect their environments through evocation, selection, behavioral manipulation, and cognitive transformation, then the distinction between P and E is a false one. To the extent that E is caused by P, behavior that might seem to be caused by E may in fact be ultimately caused by P. Moreover, from the cognitive point of view it's the perceived situation, not the objective situation, that causes behavior -- and perception is internal to P, not a part of the external E. Finally, voluntary behaviors are intentional, and goal-directed, and intentions and goals are always features of the person. The point is that, from a psychological standpoint, correct attributions are always to P, not E.

You should understand something about the "folk-conceptual" theory of causal attribution offered by Malle as an alternative account of how people actually reason about behavior. First, people appear to distinguish between unintentional behavior, which is explained by more-or-less mechanical causes, and intentional behavior, which is explained by reasons -- that is, by mental states of belief, desire, and value (however, reasons are not always linguistically marked, which may lead researchers to misinterpret them as external, situational causes). So far as causes are concerned, they can be further classified along such dimensions as internal-external, stable-unstable, and global-local -- dimensions that apply to physical causality as well as to behavior. Enabling factors stand between intentions and actions, and between causes and behaviors. And the reasons underlying intentional actions may themselves have causal histories, which give a more-or-less mechanical explanation of how the person came to hold whatever beliefs, desires, and values that led him to do what he did.

Note: Malle's folk-conceptual theory of causal attribution reveals a number of Actor-Observer asymmetries that are not encompassed by theories of the traditional P-E type. But the most important feature of Malle's approach is that it attempts to understand how ordinary people reason about causes. From this perspective, the "Fundamental Attribution Error" -- attributing behavior to persons rather than to situations -- is not an error; but it is fundamental to understanding how real people explain real behavior.  But I didn't lecture on this material, in order to make room for new material on automaticity, so you're not responsible for it.

Kelley's and Malle's theories give the impression that causal attribution occurs by reason of careful, conscious reasoning.  But there's another point of view, which is that social cognition mostly occurs automatically and unconsciously. 

You should know the four "canonical features" that distinguish automatic from controlled processes:

Automatic processes are triggered by the environment, and play out without drawing attention to themselves.  Cognition still mediates between stimulus and response, but it does so unconsciously, automatically, without conscious, deliberate thought.  The chief proponent of this view is John Bargh, though there are many other adherents as well. 

The concept of automaticity is very appealing, especially to those social psychologists who believe that scientifically correct causal attributions are to the situation.  But this high level enthusiasm may be unwarranted.

Most important, though, investigators rarely make any attempt to measure the comparative strength of automatic and controlled processes.  If automaticity pervades social cognition and behavior, we would expect that comparative experiments would show this.  But they don't, not really, as illustrated by experiments involving Jacoby's process-dissociation paradigm. 

It should surprise nobody that automatic processing dominates when attention is divided, or retrieval is difficult, or responses have to be made within half a second.  But are these conditions representative of social cognition and behavior in the ordinary course of everyday living?

Note: These experiments make use of the process-dissociation paradigm, which is more complicated than I can get into in lecture.  There's a fuller discussion in the Lecture Supplements (look in the separate supplement on Automaticity as well), but for present purposes, all you need to know is that the PDP pits automatic and controlled processes against each other, permitting a measurement of their comparative strength.



F&T's Chapter 6 contains a very good summary of the traditional literature on causal attribution, including Kelley's covariation calculus, and his ideas about causal schemata (Taylor was Kelley's graduate student, and Fisk was Taylor's, so that make Kelley Fiske's intellectual grandfather).  You should also know something about Jones and Davis's "correspondent inference theory", and Weiner's work on causal attributions for success and failure (i.e., achievement). Don't worry about the stage models of the attribution process, such as Trope's and Gilbert's. 

But do pay attention to the material on the theory of mind and mind perception, as that will come up again later in the semester, in the lectures on Social-Cognitive Development.

F&T's Chapter 6 also contains an extended discussion of the FAE, the Actor-Observer Effect, and the Self-Serving Bias.   In fact, the existence of errors and biases in social judgment is a major theme running through the Fiske & Taylor text -- as in their characterization of social perceivers as "cognitive misers".

The existence of these errors and biases have led some theorists to doubt that people are rational decision-makers -- and this theme, too, crops up often in the F&T text. But others have countered that most social judgments take place under conditions of uncertainty, where a reliance on "fast and frugal" judgment heuristics is actually rational -- bounded rationality, but still rationality.

F&T's Chapter 7 presents an excellent account of the judgment heuristics approach to social judgment, inference, and decision-making. Don't worry about prospect theory: it's incredibly important, and got Kahneman the Nobel Prize, but it's not all that relevant to this course. But do pay attention to the material on when heuristics are used, and when they fail us. Also, the effects associated with prospect theory, such as the conjunction fallacy, and and the assessment of covariation (including illusory correlation).

F&T's Chapter 8 continues the discussion of error and bias, and has a very nice discussion of the debates over the role of normative rationality in social reasoning, including the comparison of linear and nonlinear algorithms with human judgment.  One response is to be alarmed over the pervasiveness of error and bias in social judgment.  Another is to downplay the evidence on the grounds that it's based on experimental paradigms that might not be ecologically valid.  A third is to point to the accuracy of quick, intuitive judgments -- a point of view stressed in Malcolm Gladwell's best-selling book, Blink

In addition, while traditional discussions of social cognition appear to assume that causal attributions and other social judgments are (or should be) a product of conscious, deliberate cognitive processes, more recent theorists have focused on the role of automatic processes in social judgment. The automaticity of social cognition, which I discussed in lectures on Social Judgment, is also a prominent theme in the F&T text.


Questions and Comments

As you prepare for the exam, feel free to post questions to the "Questions and Comments" Forum in the course website.  Follow the same procedure as for your required Forum Postings.  Questions posted before noon on the Tuesday immediately prior to the exam will get a reply before 5:00 PM that day.  After noon, though, there are no guarantees.  

Most of the time, I, or one of the GSIs, will reply to the question or comment.  But if you also want to get in on the act, feel free to post a response to another student's posting.  If you want to reply to or posting, or comment on a reply, click on "Reply to Thread", and proceed as above.  Just keep it informative and constructive.  Internet bullying is not very nice.  


This page last revised 10/14/2015.