martin seta crelia effects of motivation.pdf

Journal of Personality and Social Psychology1990, Vol. 59, No. 1,27-37

Copyright 1990 by the American Psychological Association, Inc.O022-3514/90/SO0.75

Assimilation and Contrast as a Function of People's Willingnessand Ability to Expend Effort in Forming an Impression

Leonard L. MartinUniversity of Georgia

John J. SetaUniversity of North Carolina at Greensboro

Rick A. CreliaUniversity of Georgia

Set/reset (Martin, 1986) hypothesis that contrast demands more cognitive effort than does assimila-tion was examined. In Exp. 1, the impressions of distracted Ss showed assimilation toward blatantlyprimed concepts, whereas the impressions of nondistracted Ss showed contrast. In Exp. 2, Ss toldthat their ratings would be lumped into a group average showed assimilation, whereas Ss told thattheir ratings would be examined individually showed contrast. In Exp. 3, the impressions of Ss lowin need for cognition showed assimilation, whereas the impressions of Ss high in need for cognitionshowed contrast. Exp. 1 also showed that the results were not due to differences in recall of thetarget information, and Exp. 3 showed that the results were not due to differences in recall of thepriming stimuli. Together, the results suggest that the processes involved in contrast demand morecognitive effort than do the processes involved in assimilation.

The context in which a stimulus is embedded can have signifi-cant effects on people's judgment of that stimulus. The twoeffects that have been most reliably demonstrated are assimila-tion and contrast. Assimilation refers to a positive relation be-tween the value people place on a target stimulus and the valuethey place on the contextual stimuli that accompany the target.Contrast refers to a negative relation between these two values(cf. Sherman, Ahlm, Berman, & Lynn, 1978). The terms assimi-lation and contrast are generic in the sense that they refer simplyto the direction of context effects, and not to the processesunderlying them. It is not unreasonable to assume, therefore,that there are different kinds of assimilation and contrast thatinvolve different kinds of processes (cf. Wedell, in press).

The factor most frequently considered in the explanation ofassimilation and contrast is the distribution of the contextualstimuli (Helson, 1964; Manis, Nelson, & Shedler, 1988; Par-ducci, 1965; Upshaw & Ostrom, 1984). The typical finding isthat individuals rate stimuli of moderate value lower on thedimension of judgment (e.g, height, weight, or degree of persis-tence) if they have previously rated stimuli high on that dimen-sion than if they have previously rated stimuli low on that di-mension. As an example, a moderately tall person might berated as short by subjects who have just rated a very tall person,but as tall by subjects who have just rated a very short person.This would be a contrast effect.

The effects of the distribution of contextual stimuli can bemoderated, however, by several factors. For example, rating of atarget of ambiguous value tends to show contrast only when

We thank Dan Gilbert, Doug Wedell, and an anonymous reviewerfor their valuable comments on an earlier version of this article.

Correspondence concerning this article should be addressed toLeonard L. Martin, Department of Psychology, University of Georgia,Athens, Georgia 30602.

that rating is made in the context of stimuli that are very ex-treme on the dimension of judgment. When ambiguous stimuliare rated in the context of moderately extreme stimuli, assimila-tion may occur. As an example, an animal of ambiguous size(e.g., a fictional animal) would be rated as relatively large bysubjects who have previously rated the size of a horse but wouldbe rated as relatively small by subjects who have previouslyrated the size of a whale (Herr, Sherman, & Fazio, 1983).

The effects of contextual stimuli can also be moderated bythe number of categories with which subjects are presented torecord their judgments. In general, the effects of the contextualstimuli decrease as the number of available categories increases(e.g, Wedell & Parducci, 1988). Finally, the magnitude of contex-tual effects may depend on how the stimuli are presented. As-similation may be more likely when the target and contextualstimuli are presented simultaneously, whereas contrast may bemore likely when they are presented successively (Martin &Seta, 1983; Wedell, Parducci, & Geiselman, 1987).

Decisions in the Use of Contextual Information

In recent years, a number of researchers (e.g, Lombardi, Hig-gins, & Bargh, 1987; Martin, 1986; Newman & Uleman, 1990;Strack, Schwarz, Bless, Kubler, & Wanke, 1989) have been find-ing evidence of both assimilation and contrast under condi-tions in which none of the factors just described are manipu-lated. Martin (1986, Experiment 3), for example, presented sub-jects with a series of either positive or negative self-referentstatements and asked them to write another sentence in theirown words that communicated the same meaning as theprinted statement. Some subjects were given eight statements toreflect, whereas others were given four. All subjects were al-lowed to reflect four statements and then were asked to put thattask aside and begin a different task. This means that subjectswho had been asked to reflect four statements had finished the

27

28 L. MARTIN, J. SETA, AND R. CRELIA

A. Initial Representation

of Ambiguous

Target Person

B. Representation After

Priming with a

Positive Concept

C. Representation

After Resetting

Figure 1. Metaphorical representation of a set-reset contrast effect.

reflection task, whereas subjects who had been asked to reflecteight statements were only halfway through that task.

Following the interruption or completion of the reflectiontask, subjects were asked to form an impression of a targetperson described in terms that were ambiguous with respect totheir favorableness. These impressions were assimilated towardthe implications of the primed concepts when subjects hadbeen interrupted but were contrasted with those implicationswhen subjects had been allowed to complete the priming task.

Martin (1986) explained these results in terms of the set/resetmodel. This model is based on the assumption that when indi-viduals take seriously the goal of forming an impression, theyattempt to assess their own genuine reaction to the target per-son. If they find themselves thinking negative thoughts becauseof their exposure to negative contextual stimuli, for example,then they avoid these thoughts in forming their impression.Such thoughts are not their genuine reaction to the target. Whatthey do instead is partial out this contextual influence andattempt to assess their own context-independent reaction to thetarget person. The partialingout of reactions deemed inappro-priate to a judgment is called resetting.

According to Martin (1986), resetting may lead to contrast,as depicted schematically in Figure 1. The circle in Panel Adepicts a hypothetical reaction to a target person described interms ambiguous with respect to their favorableness. The reac-tion has both positive and negative features. In Panel B, the

effects of priming subjects with a positive concept are shown asan overlapping circle on the positive component of the ambigu-ous reaction (e.g, priming the concept "adventurous" for sub-jects forming an impression of an adventurous-reckless per-son). Because of the close overlap, subjects may not make aperfect discrimination between their reaction to the primingtask and their reaction to the target. Thus, when they attempt topartial out their contextually induced reaction, they may inad-vertently partial out aspects of what otherwise would have beentheir genuine reaction to the target. That is, they may respondto the target predominantly in terms of the target's context-dis-tinct features (see Panel C). According to Martin (1986), thiscontext-distinct residual is then used as a set of probe cues toaccess a new concept with which the individuals interpret thetarget information. Thus, there is a contrast effect on the sub-jects' interpretations of the target information (reset contrast).

Because reset contrast involves the suppressed use of theprimed concepts, subjects would be more likely to successfullyreset when they are not motivated to continue thinking aboutthe primed concepts. The purpose of the interruption/comple-tion manipulation in the Martin (1986) study was to systemati-cally vary the subjects' tendencies to perseverate on the primedconcepts. Research has shown that individuals are more likelyto continue thinking about tasks they have not completed thantasks they have completed (Marrow, 1938; Martin & Tesser,1989; Zeigarnik, 1927/1938). Thus, subjects in the interrupted-task conditions should be more likely than those in the com-pleted-task conditions to perseverate about the priming taskduring the subsequent formation of their target impression. Asa result, they would find it difficult to reset. The end result isthat subjects in the interrupted-task condition are more likelythan those in the completed-task condition to use the primedconcepts in forming their impressions.

These findings are difficult to explain in terms of the modelsof context effects described earlier (e.g, Helson, 1964; Parducci,1965; Upshaw & Ostrom, 1984). None of the factors consideredcritical in those models was manipulated. Subjects were ex-posed to the same priming stimuli (within the negative andpositive conditions) and the same target information in thesame sequence, and they indicated their impressions on thesame kinds of response scales. The only difference between theconditions that produced assimilation and those that producedcontrast was whether the priming task was interrupted or com-pleted. According to Martin (1986), the blatant nature of thepriming task caused subjects to attempt to partial out theprimed concept from their impression of the target. Subjectswho had completed the priming task were successful in theirattempts, whereas subjects who were interrupted during thepriming task were not.

It should be noted that not all instances in which subjectspartial out a contextually induced reaction are a function oftheir desire to assess their context-independent reaction to thetarget. As Strack, Martin, and Schwarz (1988) showed, a sub-ject's adherence to the maxims of communication (e.g, H. H.Clark, 1985; Grice, 1975) may produce conceptually similarresults. Having just answered a question about their happinesswith their dating life, for example, subjects can assume that thequestioner has the information he or she asked for with regardto that topic. Consequently, when they immediately encounter

ASSIMILATION AND CONTRAST 29

a question about their happiness with life as a whole, subjectsmay interpret this question as a request for information aboutaspects of their lives other than dating (e.g, school, health, andfamily). Consequently, they may partial out their happinesswith their dating life from their happiness with life in general.

Taken together, the results of Martin (1986) and Strack et al.(1988) suggest that individuals will partial out contextuallyprimed information (e.g, concepts or previous answers) if use ofthat information appears inappropriate to their current process-ing objectives. One cannot give his or her own spontaneousreaction to the target by knowingly responding in terms of acontextually primed concept. Nor can one give new informa-tion by responding with previously given information.

Contrast and Cognitive Effort

One implication of the set/reset view of contrast is that con-trast involves more cognitive steps than does assimilation. Withreset contrast, subjects avoid the use of the primed concept andsearch for another concept that matches the implications of thetarget information. With assimilation, on the other hand, sub-jects merely use the primed concept to interpret the target in-formation. This difference implies that reset contrast demandsthe expenditure of more cognitive effort than does assimilation.If this is true, then limiting the amount of cognitive effort sub-jects are willing or able to put forth in forming an impressionshould undermine their ability to achieve reset contrast. Withlimited cognitive resources, subjects should default to the lesseffortful option of using the most accessible, applicable con-cept (cf. Gilbert, Pelham, & Krull, 1988; Martin & Harlow, inpress). A schematic representation of the set/reset model is con-tained in Figure 2.

We report on three experiments that explored the role ofcognitive effort in the occurrence of reset contrast. If reset con-trast really does demand more cognitive effort than does assimi-lation, then under conditions that ordinarily result in contrastwhen subjects are not distracted, there should be assimilationwhen they are distracted. We tested this hypothesis in Experi-ment 1.

Experiment 1

In Experiment 1, subjects engaged in a pencil-and-paper taskthat blatantly primed either positive or negative concepts (seeMartin, 1986, Experiment 3). Then, they formed an impressionof a person whose actions could be construed in either positiveor negative terms. Because of the blatant nature of the primingtask, we expected subjects to be predisposed toward formingimpressions that were inconsistent with the implications of theprimed concepts. That is, they would be predisposed towardcontrast (see Lombardi et al, 1987; Martin, 1986; Newman &Uleman, 1990; Strack et al, 1989). If the cognitive effort hy-pothesis is correct, however, then this contrast effect may bemodified by the amount of cognitive effort subjects exert whileforming their impression.

To test for this, we asked half of the subjects in each primingcondition (positive vs. negative) simply to form an impression,but we asked the remaining subjects to listen to a tape record-ing while forming their impressions. On the recording, a male

Observetarget

•Have

reaction

/ Reaction \V inappropriate? /

No UseReaction

Formatresponse

Yes

Exert \ Nocognitive

effort? .

"Restt"

Use residualas probe

cue

Figure 2. Schematic representation of the set-reset model.

voice was reciting a random string of letters and digits. Thesubjects' task was to keep track mentally of the number of digitsrecited. Following their reading of the target description (withor without distraction), all subjects were asked to indicate theirimpression of the target in an open-ended format. That is, theywrote down in their own words what they thought of the target.

If contrast demands more cognitive effort than does assimila-tion, then subjects' impressions should be contrasted with theimplications of the primed concepts when subjects are not dis-tracted but should be assimilated toward those implicationswhen subjects are distracted.

Method

Subjects and design. Forty-six men and women were recruited fromintroductory psychology classes at the University of Georgia. Thesesubjects were given partial course credit in return for their participa-tion. The subjects were randomly assigned to one of four conditions ina 2 X 2 design. The four conditions were as follows: (a) primed with apositive concept, no distractor; (b) primed with a negative concept, nodistractor; (c) primed with a positive concept, distractor; and (d)primed with a negative concept, distractor. There were approximatelyequal numbers of subjects in each cell and the ratio of men to women ineach cell was about equal.

Procedure. The procedure was similar to that used by Martin(1986). Subjects were brought to the laboratory in groups of 4 to 6 andwere seated individually at desks. They were assigned to either thepositive or the negative priming condition through the counterba-


Table 1Impressions as a Function of Valence of the PrimedConcepts and Level of Distraction

Primed concepts

Level of distraction Negative Positive

Dis tractorNo distractor

2.7.3.3. b

2-5c

Note. Subjects' open-ended descriptions were coded for favorablenesson a 7-point scale. The higher the value, the more favorable the impres-sion. Logical comparisons that do not share a common subscript differat p < .05.

lanced distribution of stimulus booklets. The level of distraction wasvaried between experimental sessions.

The booklet for the first task was composed of four pages on whichself-referent positive or negative statements were listed (see Martin,1985). Subjects were told that these statements were part of a "socialempathy task" and that the ostensible purpose of the task was to seehow well the subjects could match their feelings with those of anotherperson. For each of the four statements in the booklet, subjects were towrite, in their own words, a sentence that captured the same feeling asthe printed statement. This task has been shown to be effective inproducing contrast effects (see Martin, 1986, Experiment 3). The sub-jects were paced through this task one statement at a time. When theyhad completed the last statement, they were told to place their book-lets face down on their desks.

The experimenter then gave subjects the instructions for the impres-sion-formation task. In the no-distractor conditions, subjects were in-structed simply to read the description and to form an impression ofthe person described. Subjects in the distractor conditions, on theother hand, were told that the experimenter was examining the effectsof distraction on people's ability to form impressions—"the kind ofthing that might happen if you met a person for the first time at a loud,crowded party." The experimenter then explained that to test for this, atape recording would be played while the subjects were trying to formtheir impressions. In the recording, a male voice was reciting a randomstring of digits and letters. The subjects were told to keep track men-tally of the number of digits spoken. They were instructed to pay equalattention to the impression-formation task and the digit-counting task.In both the distractor and the no-distractor conditions, subjects weregiven Vh min to read the description of the stimulus person. When thistime had elapsed, the recorder was turned off in the distraction condi-tions, and the experimenter asked the subjects in all conditions to turnto the next page of the stimulus booklets.

On the next page of the booklet, subjects were asked to describe thetarget person in their own words. Subjects were given 3 min for thistask. When all subjects had finished this task, they turned to the nextpage of the booklet, where they were asked to recall everything theycould about the target person. They were asked to be as complete andaccurate as possible and to recall the description in its original word-ing. Subjects were given 3 min for this task. When the recall task wascompleted, subjects were debriefed and excused from the experiment.

Results and Discussion

Subjects' open-ended descriptions were presented to two re-search assistants blind to the conditions under which the de-scriptions were written. These assistants rated each descriptionon a 7-point scale ranging from very negative (1) to very positive

(7). These ratings (interrater reliability = .79) were averaged andsubmitted to a 2 (positive prime vs. negative prime X 2 (distrac-tor vs. no distractor) between-subjects analysis of variance(ANOVA).

The ANOVA revealed only one significant effect, an interac-tion between the valence of the priming stimuli and the pres-ence-absence of distraction, F{\, 42) = 11.76, p < .005. Thisinteraction was decomposed through the use of planned com-parisons (see Table 1). These comparisons indicated that sub-jects who had been distracted while forming their impressionsformed more favorable impressions following priming with apositive concept (M = 4.7) than following priming with a nega-tive concept (M = 2.7). Subjects who had not been distractedformed less favorable impressions following priming with a pos-itive concept (M = 2.5) compared to a negative concept (M =3.3). In short, there was assimilation when subjects were dis-tracted and contrast when they were not distracted. These re-sults support the set/reset hypothesis prediction that the pro-cesses involved in contrast demand more cognitive effort thando the processes involved in assimilation.

The subjects' exact recall of the target paragraph was alsoanalyzed (see Table 2). This was done to assess an alternateinterpretation of our data. It is possible that subjects distractedwhile reading the target description retained less of the targetinformation than did nondistracted subjects. If so, then thesesubjects may have been more influenced by the primed con-cepts than the nondistracted subjects simply because they hadless target information on which to base their judgments (cf.Herr et al, 1983). If this interpretation is correct, then our re-sults tell us little about the role of cognitive effort in assimila-tion and contrast.

To assess this alternate hypothesis, two assistants counted thenumber of events the subjects correctly recalled. These assis-tants were blind to the conditions from which the subjects'recall protocols came. A response was counted as correct only ifit referred to a specific event or personality characteristic in thedescription. For example, "climbed a mountain" would becounted as a hit, but "does wild things" would not. The codings(interrater reliability = .95) were averaged to form a single com-posite of each subjects' recall and submitted to an ANOVA.

This ANOVA revealed that subjects who were distracted re-called slightly less information about the target (M= 3.6) thandid subjects who were not distracted (M= 4.5), but this differ-ence was not significant, F(l, 42) = 1.38. There was also nosignificant effect of prime on recall, F(l, 43) = 1.06, nor wasthere a significant interaction between prime and distractor,

Table 2Recall of the Target Information as a Function of Valence of thePrimed Concepts and Level of Distraction

Level of distraction

DistractorNo distractor

Primed concepts

Negative

4.04.8

Positive

3.34.1


Moreover, when the recall scores were entered as a covariateinto the analysis of subjects' impressions, they did not signifi-cantly alter the results compared to when no covariate was used(covariate: F < 1). Thus, the results of Experiment 1 appear to bedue to differences in the way subjects processed the informa-tion they had rather than to differences in the information theyhad to process.

Experiment 2

The results of Experiment 1 are consistent with the hypothe-sis that reset contrast demands more cognitive effort than doesassimilation. If a mental process is truly effortful, however,then to perform that process, people must not only have a suffi-cient amount of cognitive capacity available but also be willingto use it. In Experiment 2, we held constant the amount ofcapacity subjects had available, but we manipulated their will-ingness to use that capacity.

As in Experiment 1, subjects were primed with either positiveor negative concepts and were asked to form an impression of aperson described in terms that were ambiguous as to their fa-vorableness. Subjects were run in small groups (e.g, 5 to 7), andmost of the subjects in each group were told to put no name orany other identifying mark on their papers. They were told thattheir ratings would be averaged with those of the othermembers of their group and that the experimenter was inter-ested only in the overall group response. The 1 or 2 remainingsubjects (depending on the size of the group) were asked to puttheir names on their stimulus booklets and were told that theexperimenters were concerned with their personal responses.

Research on social loafing has shown that individuals whoare engaged in a group task in which their individual contribu-tions cannot be identified exert less cognitive effort than doindividuals whose outputs can be identified (Petty, Harkins, &Williams, 1980; Petty, Harkins, Williams, & Latane, 1977). Onthe basis of this research, we hypothesized that subjects whowere told that their responses would be examined individuallywould exert more effort in forming their impression thanwould subjects told that their responses would be lumped into agroup average. So, if contrast really does demand more process-ing capacity than does assimilation, then impressions formedby individually identified subjects should show contrast,whereas those formed by the unidentified, grouped subjectsshould show assimilation.

Method

Subjects. Sixty-two men and women were recruited from introduc-tory psychology classes at the University of Georgia. These subjectswere given partial course credit in return for their participation. Sub-jects were randomly assigned to one of four conditions in a 2 X 2design. The conditions were as follows: (a) identified individualsprimed with a positive concept, (b) identified individuals primed witha negative concept, (c) unidentified group members primed with apositive concept, and (d) unidentified group members primed with anegative concept. There were approximately equal numbers of subjectsin each cell, and the ratio of men to women in each cell was aboutequal.

Procedure. Subjects were brought to the laboratory in groups rang-ing from 5 to 7. The subjects were seated individually at desks and

handed two stimulus packets. The first packet contained the primingmaterials, and the second contained the target paragraph and thescales on which subjects were asked to indicate their impressions. Thepriming materials and the target paragraph were the same as thoseused in Experiment 1.

Subjects were run in groups varying in size from S to 7. The experi-menter randomly selected 4 or 5 in each group and stressed to thesepeople that they were to put no name or any other identifying mark ontheir papers. It was emphasized that their ratings would be averagedwith those of the other members of their group and that the experi-menter was interested only in the overall group response. The 1 or 2remaining subjects were asked to put their names on their stimulusbooklets and were told that the experimenter wished to compare theirindividual responses with the group average.

Subjects then read the description of the target person and indicatedtheir impressions of this person by responding to four 9-point scales.These scales were bounded by bipolar trait adjectives that were applica-ble for construing the actions in the target description. These adjec-tives were adventurous-reckless, conceited-stubborn, aloof-indepen-dent, and stubborn-persistent. After subjects had indicated their im-pression on these scales, they were debriefed and excused from theexperiment.


It is typical in social loafing research to use the scores ofsingle individuals led to believe either that they are working in agroup or that they are working alone. We felt, however, that thesocial loafing manipulation would be more effective if subjectsactually performed in the group. We made sure that no experi-mental session yielded more than one score for any of the fourconditions by averaging the scores of the grouped subjects (2-3)in each session. In other words, the scores of subjects identifiedas individuals were entered into the analysis as separate datapoints, whereas the scores of subjects run as a group were aver-aged to yield a single score for each priming condition (positiveand negative) for each experimental session. Although the aver-aging of scores may make the scores in the grouped conditionmore stable than those in the individual condition, it does notaffect the direction of the effects. Consequently, this averagingdoes not represent a biased test of our hypothesis.

The data were submitted to a 2 (valence of prime) X 2 (individ-ual vs. group) X 4 (rating dimension) repeated measures ANOVA.The analysis revealed a significant main effect for rating di-mension, F(3,117)= 5.53, p < .001, which indicated that, as agroup, subjects rated the target person less favorably on thestubborn-persistent dimension than on any of the other di-mensions. It also revealed a significant Rating X Individual-Group interaction, F(3,117)= 3.08, p < .03. This interactionoccurred because subjects who were isolated as individualsrated the target more favorably on the conceited-self-confidentdimension than did subjects who were told they were part of agroup.

Theoretically more interesting, however, was the significantinteraction between the valence of the priming stimuli and theindividual-group manipulation, F(l, 39) = 7.36, p < .01. Thisinteraction was decomposed through the use of planned com-parisons (see Table 3). These comparisons revealed that theunidentified group subjects rated the target person as signifi-cantly more favorable following priming with a positive concept(M= 6.7) than following priming with a negative concept (M=


Table 3Impressions as a Function of Valence of the PrimedConcepts and Level of Identification

Primed concepts

Level of identification Negative Positive

Unidentified, groupedIdentified, individual

5.4.7.1h

6.7C

6.2C

Note. Subjects' ratings were made on 9-point scales. The higher thevalue, the more favorable the impression. Logical comparisons that donot share a common subscript differ at p < .05.

5.4). The individually identified subjects, on the other hand,rated the target person as significantly less favorable followingpriming with a positive concept (M = 6.2) compared to a nega-tive concept (M = 1.1).

In short, there was assimilation in conditions in which sub-jects have been shown in past research (Petty et al, 1980; Pettyet al, 1977) to exert little cognitive effort (unidentified grouped)and contrast in conditions in which subjects have been shownto exert more cognitive effort (individually identified). Theseresults conceptually replicate those of Experiment 1 and pro-vide additional support to the hypothesis that the processesassociated with contrast demand more cognitive effort than dothe processes associated with assimilation.

Experiment 3

The results of Experiments 1 and 2 are consistent with thenotion that contrast demands more cognitive effort than doesassimilation, but they do not provide unequivocal support forthat hypothesis. The problem is that the manipulations of ca-pacity used in those experiments may have inadvertently in-fluenced another factor known to determine the occurrence ofassimilation and contrast. Several studies (e.g, Lombardi et al,1987; Newman & Uleman, 1990; Strack et al, 1989) have shownthat subjects' impressions are likely to reflect contrast whensubjects can recall any of the priming stimuli (even one) but arelikely to reflect assimilation when subjects cannot recall any ofthese stimuli.

It is possible that the nondistracted subjects in Experiment 1and the individually identified subjects of Experiment 2 re-membered the priming stimuli, whereas the distracted andgrouped subjects did not. If so, then we might expect that thedistracted and grouped subjects would show assimilation,whereas the nondistracted and identified subjects would showcontrast. This, in fact, was the obtained pattern of results.

If the results are due to differences in memory for the prim-ing stimuli, then one could argue that we have little evidencethat contrast demands more cognitive effort than does assimila-tion. It may be that memory of the priming stimuli changes theperceived similarity of the target to the priming stimuli (orsome similar factor), and this, in turn, determines whether as-similation or contrast occurs (cf. Herr et al, 1983).

We attempted to clarify this issue in Experiment 3 by using amanipulation of processing capacity that was unlikely to inter-fere with the subjects' recall of the priming stimuli. In addition,

we assessed the subjects' recall of the priming stimuli to seewhether it correlated with the occurrence of assimilation andcontrast. We manipulated the subjects' tendencies to exert cog-nitive effort by blocking them on their need for cognition.

Need for cognition is an individual difference that reflectsone's tendency to engage in and enjoy effortful cognitive pro-cessing (Cacioppo & Petty, 1982). Considerable evidence sug-gests that subjects high in need for cognition exert more efforton cognitive tasks than do subjects low in need for cognition(see Heesacker, 1985, for a review). So, if contrast demandsmore cognitive effort than does assimilation, then the impres-sions of subjects high in need for cognition should show con-trast, whereas the impressions of subjects low in need for cogni-tion should show assimilation, and this should be true even ifthe level of prime recall is the same for both groups of subjects.

Method

Subjects. Fifty-two men and women were recruited from introduc-tory psychology classes at the University of Georgia. These subjectswere given partial course credit in return for their participation. Sub-jects either high or low in their need for cognition were randomlyassigned to one of two priming conditions (positive vs. negative). Therewere an equal number of subjects in each cell, and the ratio of men towomen in each cell was about equal.

Procedure. Subjects were run in groups ranging from 5 to 8. Theywere told that the experimenter was interested in examining how peo-ple form impressions of one another. The experimenter then distrib-uted booklets containing the priming task and the impression forma-tion task. The priming task was similar to that used in the earlierexperiments except that in Experiment 3 subjects were primed with theconcepts persistent and stubborn.

After all subjects had completed the priming task, the experimenterasked them to turn to the next page of their stimulus booklets and toform an impression of the person described on that page. The descrip-tion was ambiguous in terms of how stubborn or persistent the targetwas. When subjects felt they had read the description to their satisfac-tion, they turned to the next page of the booklet and indicated theirimpressions on four rating scales. These were 7-point scales bound bythe following adjectives: inflexible-flexible, nosy-curious, stingy-thrifty, and conceited-self-confident.

When all subjects had completed their ratings, they were asked toturn to the last page of the stimulus booklet and write down as many ofthe traits and behaviors as they could remember from the priming task.They were given 5 min for this task. Following the recall task, thestimulus packets were taken up, and the need-for-cognition inventories

Table 4Impressions as a Function of Valence of the PrimedConcepts and Need for Cognition

Need for cognition

LowHigh

Primed concepts

Negative

5.2.6.1.

Positive

6.4b

4.8C

Note. Subjects' ratings were made on 7-point scales. The higher thevalue, the more favorable the impression. Logical comparisons that donot share a common subscript differ at p < .05.


Table 5Recall of the Priming Stimuli as a Function of Valence of thePrimed Concepts and Need for Cognition

Need for cognition

LowHigh

Primed concepts

Negative

5.66.1

Positive

6.64.5

were distributed (Cacioppo, Petty, & Kao, 1984). After filling out theinventories, subjects were debriefed and excused.


There is no absolute criterion by which individuals can beclassified as high or low in need for cognition (Heesacker, 1985).Therefore, we blocked subjects on need for cognition using thehighest and lowest one third of our sample. The ratings fromthese subjects were analyzed using a 2 (high vs. low need forcognition) X 2 (positive vs. negative prime) X 4 (rating dimen-sion) repeated measures ANOVA. This analysis revealed two sig-nificant effects.

There was a main effect for rating dimension, FQ, 108) =6.18, p < .009, indicating that subjects rated the stimulus per-son more favorably on the inflexible-flexible (M = 6.4) and theconceited-self-confident (M - 6.4) dimensions than on thenosy-curious (M = 4.9) and the stingy-thrifty (M = 5.2) dimen-sions. Theoretically more interesting, however, was the signifi-cant interaction between valence of prime and level of need forcognition, F(l, 36) = 5.75, p < .02. This interaction was decom-posed through the use of planned comparisons (see Table 4).These comparisons indicated that subjects low in need for cog-nition showed assimilation, whereas subjects high in need forcognition showed contrast. That is, subjects low in need forcognition formed more favorable impressions after beingprimed with positive concepts (M = 6.4) than after beingprimed with negative concepts (M = 5.2). Subjects high in needfor cognition, on the other hand, formed more favorable im-pressions after being primed with negative concepts (M = 6.1)compared to positive concepts (M = 4.8).

Recall of the priming stimuli. To see if recall of the primingstimuli was accounting for the differences in our results, weattempted to correlate subjects' recall of the priming stimuliwith their ratings (e.g, Lombardi et al, 1987). In the process ofcoding the data, however, we found that all but 1 subject re-membered the priming stimuli. More specifically, all but 1 sub-ject recalled at least one of the primed-trait concepts and, onthe average, 1.9 of the 8 behaviors they had to categorize (seeTable 5). Thus, our results are not a function of whether subjectsdid or did not recall the priming stimuli. We then checked tosee if amount of prime recall predicted subjects' tendencies toshow assimilation as opposed to contrast. It did not. When therecall scores were submitted to an ANOV\, there were no signifi-cant differences between conditions, and when recall was usedas a covariate in an analysis of subjects' impressions, it did notsignificantly alter the results obtained when no covariate wasused (covariate: F(l, 35) = 1.4). In short, the tendency of our

subjects to show assimilation or contrast in their impressionswas not a function of their ability to recall the priming stimuli.

General Discussion

We began with the general question of whether the executionof a contrasted judgment demands more processing capacitythan does the execution of an assimilated judgment. Our re-sults suggest that it does (at least with reset contrast). In threeconceptual replications, the impressions of subjects who ex-erted cognitive effort showed contrast, whereas the impressionsof subjects who did not exert cognitive effort showed assimila-tion. Experiment 1 showed the effects of ability to exert cogni-tive effort, Experiment 2 showed the effects of willingness toexert cognitive effort, and Experiment 3 showed the effects ofdispositional tendencies to exert cognitive effort.

In addition, Experiment 1 ruled out the possibility that theresults were due to differences in the subjects' ability to recallthe target information, and Experiment 3 ruled out the possibil-ity that the results were due to differences in the subjects' abilityto recall the priming stimuli. Taken together, the results of allthree experiments support the hypothesis that the processesinvolved in contrast demand more cognitive effort than do theprocesses involved in assimilation.

Mechanism for Assimilation and Contrast

It is important to note that in all three experiments, assimila-tion and contrast were obtained when subjects were exposed tothe same contextual stimuli (within the positive and negativeconditions) and the same target person in the same sequence,and they indicated their impressions in the same way (i.e, thesame types of response scales or open-ended format). The onlydifference between the conditions that produced assimilationand those that produced contrast was the amount of cognitiveeffort the subjects exerted. Such a result is not readily explainedin terms of theories that assume that assimilation and contrastare a function of factors such as the distribution of the contex-tual stimuli (Helson, 1964; Manis et al, 1988; Parducci, 1965;Upshaw & Ostrom, 1984), the format in which the target andcontext stimuli are presented (Wedell et al, 1987), the ambigu-ity of the target stimulus (Herr et al, 1983), or the number ofresponse categories available to the subjects (Wedell & Par-ducci, 1988).

The results are also not readily explained by models thatassume that assimilation and contrast arise from a commonpsychological mechanism. Sherif and Hovland (1961), for exam-ple, proposed that both assimilation and contrast result fromplacement of the target stimulus in a category along the judg-mental continuum. Assimilation occurs when the target stimu-lus falls within the same judgmental category as does the contex-tual stimuli, whereas contrast occurs when the target stimulusand the contextual stimuli fall into different categories. Thereis no reason from such a model why contrast should demandmore cognitive effort than assimilation. Thus, their model (also,see Helson, 1964) cannot account for the present findings.

The present results are also difficult to explain in terms ofthe early models of concept priming effects (e.g, Higgins &King, 1981; Wyer & Srull, 1981). These models had suggested


that one's tendency to use a primed concept to interpret infor-mation isa function of the concept's accessibility and applicabil-ity. Accessibility refers to the concept's level of activation and isassumed to be a function of how recently or frequently a con-cept has been activated. Applicability refers to the descriptivematch between the concept and the target information. Inshort, the early models assumed that as long as a concept hadbeen recently or frequently activated and was applicable to thetarget information, individuals would use that concept to inter-pret the information.

This assumption was not supported in our experiments. Inall three experiments, the primed concepts were both accessi-ble and applicable. Yet, they were used to interpret the targetinformation only when subjects did not exert cognitive effort.Clearly, factors in addition to accessibility and applicability candetermine peopled use of concepts in forming impressions.

Appropriateness and Information Use

Martin (1986) argued that, in addition to accessibility andapplicability, factors in the general judgmental setting canprompt individuals into using or not using concepts. These fac-tors include (a) the relation between the target and the context(Martin & Seta, 1983; Seta, Martin, & Capehart, 1979; Strack,Schwarz, & Gschneidinger, 1985), (b) the desire to make a con-text-independent judgment (Kubovy, 1978; Martin, 1986), (c)instructional sets (Leach, 1974), and (d) communication rules(Strack et al, 1988). These factors can be thought of as influenc-ing one's estimate of the appropriateness of using contextuallyprimed information (Higgins, 1989; Strack & Martin, 1987;Strack etal, 1988).

In the present experiments, for example, one could say thatindividuals suppressed the use of the primed concepts andsearched for another because use of the primed concepts wouldhave been inappropriate. One cannot form one's own impres-sion by knowingly using a concept primed by exposure to thecontextual stimuli. One could do so unknowingly, however, aswhen the concept is primed very subtly (Bargh & Pietromo-naco, 1982; Lombardi et al, 1987).

A priming model incorporating assumptions similar to thoseof the set/reset model (Martin, 1986) was developed by Kubovy(1978) to explain how people generate a random number. Byrandom, Kubovy (1978) meant a number that seems to the per-son to have arisen "entirely from natural impulse, without anyexternal stimulus or constraint" (italics in original, Kubovy,1978, p. 359). He hypothesized that people first retrieve thenumber that comes most easily to mind (the accessibility step).Then, they check to see if the number appears random (therepresentativeness step). That is, they check to see whether theease with which the number came to mind was caused by someexternal event. If the first number they retrieve appears ran-dom, then they give it as their response. If it does not appearrandom, then they search for one that does. When they find it,they give it as their response.

To test this hypothesis, Kubovy (1978) asked some subjects to"name the first digit that comes to mind." He noted that 2.2% ofthe subjects in this condition chose the digit 1. In a secondcondition, Kubovy increased the accessibility of the digit 1 byasking subjects to "name the first one digit number that comes

to mind." Consistent with his accessibility assumption, Kubovy(1978) found that the percentage of subjects choosing 1 as theirresponse increased to 18%.

In a third condition, Kubovy (1978) asked subjects to "namethe first one digit number that comes to mind, like one." In thiscondition, the digit 1 is presumably highly accessible to sub-jects. It has just been mentioned twice. However, its accessibil-ity is also clearly related to an external event, the experimenter'sprompt. Thus, it is not representative of a random number. So,subjects should not give it as their response but instead shouldsearch for a number that seems more random (i.e, one not madeaccessible by the experimenter's prompt). Consistent with thistwo-step model, the number of subjects giving 1 as their re-sponse in this blatant priming condition decreased to 5.4%.

The commonality between Kubovy's (1978) model and theset/reset model (Martin, 1986) is that in both models individ-uals are assumed to engage in a judgment prior to using primedinformation. Both models imply that increasing the accessibil-ity of information (e.g., a concept, a number, or a prior answer) isnot a sufficient condition for increasing the use of that informa-tion. The present work extends both models by suggesting thatthe disuse of primed information demands more cognitive ef-fort than does the use of that information.

Controlled Versus Conscious Processing

The present results (also, see Lombardi et al, 1987; Martin,1986; Newman & Uleman, 1990; Strack et al, 1988; Strack et al,1989) suggest that the passive nature of concept priming inimpression formation may have been overemphasized by theearly priming studies. Our results, however, should not be takenas evidence that the use and disuse of concepts in impressionformation is a controlled process in the sense that it is open tointrospective awareness. Research on automatic and controlledprocesses (for summaries see Bargh, 1989; Shiflrin, 1988; Ule-man, 1989) has indicated that "virtually all complex and inter-esting processes of significant duration involve a mixture ofautomatic and control processes' (Shiffrin & Dumais, 1981, p.112). Thus, a process may be initiated without conscious inten-tion yet still demand cognitive effort to continue.

It may be that stimulus conditions like those previously men-tioned (i.e., blatant priming, communication rules) spontane-ously activate a reset procedure that then runs to completionbeyond the subjects' awareness (provided that subjects exertsufficient cognitive effort). Some indirect evidence for this as-sumption was obtained in our studies. During debriefing, oursubjects tended to deny that the priming stimuli had any influ-ence whatsoever on their ratings. They suggested instead thatpriming may have influenced the impressions of other subjects,but that they themselves had rated the target person as they didbecause they "really felt that he was that way" This leads us tobelieve that our subjects were aware of the priming stimuli andaware of their subsequent impression but were unaware of arelation between the two (cf. Nisbett & Wilson, 1977).

It is also important to note that the set/reset model (Martin,1986) assumes that individuals make a judgment of inappro-priateness, not a judgment of appropriateness (see Figure 2). Inother words, the model does not assume that people go aroundsecond guessing their use of each and every concept. Rather,


they assess appropriateness only when some factor (eg., aware-ness of the priming episode) brings that issue to their attention(cf. Schwarz & Clore, 1983; Strack et al, 1989). This may explainwhy assimilation, rather than contrast, tends to occur whenindividuals are not aware of being primed (Bargh & Pietromo-naco, 1982; Higgins, Rholes, & Jones, 1977; Lombardi et al,1987; Newman & Uleman, 1990; Srull & Wyer, 1979).

So, although awareness of the priming episode may generallybias people in the direction of a contrast (Lombardi et al, 1987;Newman & Uleman, 1990; Strack et al, 1989), the present for-mulation suggests that it will lead individuals to a contrastedjudgment only when (a) it prompts them to a decision of inap-propriateness, and (b) they exert sufficient cognitive effort topartial out the inappropriate reaction from their reaction to thetarget (i.e, they reset).

Capacity and Correction

The present results bear at least a superficial resemblance tothose obtained by Gilbert and his colleagues (see Gilbert, 1989,for a review), who examined the role of cognitive capacity inpeople's use of situational information when making disposi-tional inferences. In one experiment, Gilbert et al. (1988, Ex-periment 1) showed subjects a videotape of a target person act-ing in a rather nervous fashion. Half of the subjects were toldthat the target was discussing anxiety-provoking topics. Halfwere told that she was discussing relaxing topics. All subjectsrated the target in terms of how dispositionally anxious theyconsidered her to be.

It was assumed that subjects would attribute the target'sanxious behavior to the situation when she was portrayed asdiscussing anxiety-provoking topics but would attribute her be-havior to dispositional nervousness when she was portrayed asdiscussing relaxing topics. The results were consistent withthese assumptions—at least when the subjects were not dis-tracted. When subjects were distracted while forming their im-pressions, there was no difference in their ratings of the target'sdispositional anxiety. Gilbert etal. (1988)concluded that dispo-sitional inferences occur relatively automatically, whereas cor-rections for situational constraints on behavior involve morecontrolled processing. Thus, when people are distracted andcannot exert sufficient cognitive effort, they do not correct forsituational constraints on the behaviors of another person.

In our experiments, subjects were correcting for a situationalinfluence on their own behaviors (i£, influence of a primedconcept on their impressions of the target). These results, takenwith those of Gilbert et al. (1988), suggest that correcting for aninitial evaluation demands more cognitive effort than does go-ing with that initial evaluation. Of course, additional research isneeded to see whether the mechanisms involved in these differ-ent phenomena are really similar to one another.

Interruption, Rumination, and Cognitive Effort

The present results may also have some implications for Mar-tin's (1986) original interpretation of the set/reset findings. Asnoted earlier, Martin (1986) blatantly primed subjects with ei-ther positive or negative concepts and either interrupted themwhile they were performing the priming task or allowed them

to complete that task. When subjects had been interrupted,their subsequent impressions were assimilated toward the im-plications of the primed concepts. When they had not beeninterrupted, their impressions were contrasted with the impli-cations of the primed concepts.

Martin (1986) proposed that with interruption subjectstended to ruminate about the primed concepts (Martin &Tesser, 1989; Zeigarnik, 1927/1938). Because of this rumina-tion, it was difficult for them to suppress the primed conceptand search for a context-distinct concept. With completion ofthe priming task, on the other hand, there was no rumination.Hence, resetting was easier.

The present results raise the possibility that the effect of in-terruption may not be as concept-specific as originally thought.It may be that the rumination produced by the interruption hadits effect through the usurpation of general cognitive capacity.With lowered capacity, Martin's (1986) subjects, like those inthe present experiments, could not reset. If this revised inter-pretation is correct, then it suggests that almost any sort ofrumination (whether it be prime-related or not) following con-cept priming may undermine subjects' ability to achieve resetcontrast.

Caveats

It is important to note that the manipulations of cognitiveeffort in our experiments occurred during the subjects' at-tempts to form an impression, not during their performance ofthe priming task. There is reason to believe that differences inthe amount of cognitive effort expended during the primingtask may have effects different from those reported in this arti-cle. In fact, increased effort during the priming task may lead toassimilation rather than to contrast (cf. L. F Clark & Collins,1989; Strack etal, 1985).

It should also be noted that our argument that contrast de-mands more cognitive effort than does assimilation may belimited to certain types of contrast effects—namely, those thatresult from differences in people's use of information (i.e, resetcontrast). The same may not be true for other forms of contrast(Parducci & Wedell, 1986).

Summary and Implications

The present set of results suggests that the processes involvedin reset contrast demand more cognitive effort than do theprocesses involved in assimilation. One reason for this may bethat these forms of assimilation and contrast (i£, set and reset)arise from qualitatively distinct mechanisms. Assimilation mayinvolve interpretation of the target information in terms of thecontextually primed concept, whereas contrast may involve thesuppressed use of the primed concept and the interpretation ofthe target information in terms of an alternate concept (Martin,1986). Future work could concentrate on other factors that leadto resetting (e.g, communication rules and relation of the targetto the context) and on the processes that underlie it.

References

Bargh, J. A. (1989). Conditional automaticity: Varieties of automaticinfluence in social perception and cognition. In J. S. Uleman & J. A.Bargh (Eds.), Unintended thought (pp. 3-51). New York: GuilfordPress.


Bargh, J. A., & Pietromonaco, P. (1982). Automatic information pro-cessing and social perception: The influence of trait informationpresented outside of conscious awareness on impression formation.Journal of Personality and Social Psychology, 43, 437-449.

Cacioppo, J. T, & Petty, R. E. (1982). The need for cognition. Journal ofPersonality and Social Psychology, 42,116-131.

Cacioppo, J. T, Petty, R. E., & Kao, C (1984). The efficient assessmentof need for cognition. Journal of Personality Assessment, 48, 306-307.

Clark, H. H. (1985). Language use and language users. In G. Lindzey &E. Aronson (Eds.), The handbook of social psychology: Vol. 2. Specialfields and applications (3rd ed., pp. 179-231). New York: RandomHouse.

Clark, L. E, & Collins, J. E. (1989). Autobiographical memory as context:The role of understanding and mode of thinking on judgments of satis-faction. Paper presented at the annual convention of the MidwesternPsychological Association, Chicago.

Gilbert, D. T. (1989). Thinking lightly about others: Automatic compo-nents of the social inference process. In J. S. Uleman & J. A. Bargh(Eds.), Unintended thought (pp. 189-211). New York: Guilford Press.

Gilbert, D. T, Pelham, B. W, & Krull, D. S. (1988). On cognitive busy-ness: When person perceivers meet persons perceived. Journal ofPersonality and Social Psychology, 54, 733-740.

Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan(Eds.), Syntax and semantics 3: Speech acts (pp. 343-392). Orlando,FL: Academic Press.

Heesacker, M. (1985). Need for cognition scale. In D. J. Keyster & R. C.Sweetlend (Eds.), Test critiques (Vol. 3, pp. 466-474). Kansas City,MO: Test Corporation of America.

Helson, H. (1964). Adaptation-level theory: An experimental and system-atic approach to behavior. New \fork: Harper & Row.

Herr, P. M., Sherman, S. J., & Fazio, R. H. (1983). On the consequencesof priming: Assimilation and contrast effects. Journal of Experimen-tal Social Psychology, 19, 323-340.

Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectiv-ity and suffering from unconscious sources. In J. S. Uleman & J. A.Bargh (Eds.), Unintended thought (pp. 75-123). New York: GuilfordPress.

Higgins, E. T, & King, G. (1981). Accessibility of social constructs:Information processing consequences of individual and contextualvariability. In N. Cantor & J. F Kihlstrom (Eds.), Personality, cogni-tion, and social interaction (pp. 69-121). Hillsdale, NJ: Erlbaum.

Higgins, E. T, Rholes, W S., & Jones, C. R. (1977). Category accessibil-ity and impression formation. Journal of Experimental Social Psy-chology, 13,141-154.

Kubovy, M.(1978). Response availability and the apparent spontanietyof numerical choices. Journal of Experimental Psychology: HumanPerception and Performance, 3, 359-364.

Leach, C. (1974). The importance of instructions in assessing sequen-tial effects in impression formation. British Journal of Social andClinical Psychology, 13,151-156.

Lombardi, W J, Higgins, E. T, & Bargh, J. A. (1987). The role ofconsciousness in priming effects on categorization: Assimilationversus contrast as a function of awareness of the priming task. Per-sonality and Social Psychology Bulletin, 13, 411-429.

Manis, M., Nelson, T. E, & Shedler, J. (1988). Stereotypes and socialjudgment: Extremity, assimilation, and contrast. Journal of Personal-ity and Social Psychology, 55, 28-36.

Marrow, A. J. (1938). Goal tension and recall. Journal of General Psy-chology, 19, 3-64.

Martin, L. L. (1985). Categorization and differentiation: A set, re-set,comparison analysis of the effects of context on person perception.New 'York: Springer-Verlag.

Martin, L. L. (1986). Set/reset: Use and disuse of concepts in impres-sion formation. Journal of Personality and Social Psychology, 51,493-504.

Martin, L. L., & Harlow, T. E (in press). Basking and brooding: Themotivational effects of filter questions in opinion surveys. In N.Schwarz & S. Sudman (Eds.), Order effects in social and psychologicalresearch. New York: Springer-Verlag.

Martin, L. L., & Seta, J. J. (1983). Perceptions of unity and distinctive-ness as determinants of attraction. Journal of Personality and SocialPsychology, 44, 755-764.

Martin, L. L., & Tesser, A. (1989). Toward a motivational and struc-tural theory of ruminative thought. In J. S. Uleman & J. A. Bargh(Eds.), Unintended thought (pp. 306-326). New York: Guilford Press.

Newman, L., & Uleman, J. S. (1990). Assimilation and contrast effectsin spontaneous trait inferences. Personality and Social PsychologyBulletin. 16, 224-240.

Nisbett, R.E, & Wilson, T. D. (1977). Telling more than we can know:Verbal reports on mental processes. Psychological Review, 84,231-259.

Parducci, A. (1965). Category judgments: A range-frequency model.Psychological Review, 72, 407-418.

Parducci, A., & Wedell, D. H. (1986). The category effect with ratingscales: Number of categories, number of stimuli, and method ofpresentation. Journal of Experimental Psychology: Human Percep-tion and Performance, 12, 496-516.

Petty, R. E., Harkins, S. G, & Williams, K. D. (1980). The effects ofgroup diffusion of cognitive effort on attitudes: An information-pro-cessing view. Journal of Personality and Social Psychology, i5,81-92.

Petty, R. E, Harkins, S. G, Williams, K. D, & Latane, B. (1977). Theeffects of groups size on cognitive effort and evaluation. Personalityand Social Psychology Bulletin, 3, 579-582.

Schwarz, R, & Clore, G. L. (1983). Mood, misattribution, and judg-ments of well-being: Informative and directive functions of affectivestates. Journal of Personality and Social Psychology, 45, 513-523.

Seta, J. J., Martin, L. L., & Capehart, G. (1979). The effects of contrastand generalization on the attitude similarity-attraction relationship.Journal of Personality and Social Psychology, 37, 462-467.

Sherif, M, & Hovland, C. W (1961). Social judgment: Assimilation andcontrast effects in communication and attitude change. New Haven,CT: Yale University Press.

Sherman, S. 1, Ahlm, K, Berman, L, & Lynn, S. (1978). Contrasteffects and their relationship to subsequent behavior. Journal of Ex-perimental Social Psychology, 14, 340-350.

Shiffrin, R. M. (1988). Attention. In R. C. Atkinson, R. J. Herrnstein,G. Lindzey, & R. D. Luce (Eds.), Stevens' handbook of experimentalpsychology (2nd ed, pp. 739-811). New York: Wiley.

Shiffrin, R. M., & Dumais, S. T. (1981). The development of automa-tism. In J. R. Anderson (Ed.), Cognitive skills and their acquisition(pp. 111-140). Hillsdale, NJ: Erlbaum.

Srull, T. K., & Wyer, R. S. (1979). The role of category accessibility inthe interpretation of information about persons: Some determinantsand consequences. Journal of Personality and Social Psychology, 37,1660-1672.

Srull, T. K, & Wyer, R. S. (1980). Category accessibility and socialperception: Some implications for the study of person memory andinterpersonal judgment. Journal of Personality and Social Psychol-ogy, 38, 841-856.

Strack, E, & Martin, L. L. (1987). Thinking, judging, and communicat-ing: A process account of context effects in attitude surveys. In H. J.Hippler, N. Schwarz, & S. Sudman (Eds.), Social information process-ing and survey methodology (pp. 123-148). New'York: Springer-Ver-lag.

Strack, E, Martin, L. L.,& Schwarz, N. (1988). Priming and communi-


cation: Social determinants of information use in judgments of lifesatisfaction. European Journal of Social Psychology, 18, 429-442.

Strack, E; Schwarz, N, Bless, H, Kubler, A, & Wanke, M. (1989). Re-member the priming events! Episodic cues may determine assimila-tion versus contrast effects. Unpublished manuscript, University ofMannheim, Mannheim, West Germany.

Strack, E, Schwarz, N, & Gschneidinger, E. (1985). Happiness andreminiscing: The role of time perspective, affect, and mode of think-ing. Journal of Personality and Social Psychology, 49,1460-1469.

Uleman, J. S. (1989). A framework for thinking intentionally aboutunintended thoughts. In J. S. Uleman & J. A. Bargh (Eds.), Unin-tended thought (pp. 425-449). New York: Guilford Press.

Upshaw, H. S, & Ostrom, T. M. (1984). Psychological perspective inattitude research. In J. R. Eiser (Ed.), Attitudinal judgment (pp. 23-41). New York: Springer-Verlag.

Wedell, D. H. (in press). Methods for determining the locus of contexteffects in judgment. In J. P. Caverni, J. M. Faber, & M. Gonzales(Eds.), Cognitive biases. Amsterdam: North-Holland.

Wedell, D. H., & Parducci, A. (1988). The category effect in social

judgment: Experimental ratings of happiness. Journal of Personalityand Social Psychology, 55, 341-356.

Wedell, D. H., Parducci, A, AGeiselman, R. E. (1987). A formal analy-sis of ratings of physical attractiveness: Successive contrast and si-multaneous assimilation. Journal of Experimental Social Psychol-ogy, 23, 230-249.

Wyer, R. S., & Srull, T. K. (1981). Category accessibility: Some theoreti-cal and empirical issues concerning the processing of social stimu-lus information. In E. T. Higgins & M. P. Zanna (Eds.), Social cogni-tion; The Ontario Symposium (Vol. 1, pp. 161-197). Hillsdale, NJ:Erlbaum.

Zeigarnik, B. (1938). On finished and unfinished tasks. In W D. Ellis(Ed.), A source book ofgestalt psychology (pp. 300-314). New York:Harcourt, Brace & World. (Reprinted and condensed from Psycholo-gishe Fbrschung, 1927, 9,1-85.)

Received August 24,1989Revision received December 26,1989

Accepted January 19,1990 •

Documents

martin seta crelia effects of motivation.pdf