Journal of Personality and Social Psychology1978, Vol. 36, No. 6, 635-642

The Mindlessness of Ostensibly Thoughtful Action:The Role of "Placebic" Information in

Interpersonal Interaction

Ellen LangerHarvard University

Arthur Blank and Benzion ChanowitzThe Graduate Center

City University of New York

Three field experiments were conducted to test the hypothesis that complexsocial behavior that appears to be enacted mindfully instead may be performedwithout conscious attention to relevant semantics. Subjects in compliance par-adigms received communications that either were or were not semantically sensi-ble, were or were not structurally consistent with their previous experience, anddid or did not request an effortful response. It was hypothesized that unless thecommunication occasioned an effortful response or was structurally (rather thansemantically) novel, responding that suggests ignorance of relevant informationwould occur. The predictions were confirmed for both oral and written commu-nications. Social psychological theories that rely on humans actively processingincoming information are questioned in light of these results.

Consider the image of man or woman as acreature who, for the most part, attends tothe world about him or her and behaves onthe basis of reasonable inference drawn fromsuch attention. The view is flattering, perhaps,but is it an accurate accounting of coverthuman behavior?

Social psychology is replete with theoriesthat take for granted the "fact" that peoplethink. Consistency theories (cf. Abelson et al.,1968), social comparison theory (Festinger,1954; Schachter, 1959), and attributiontheory (Heicler, 1958; Jones et al., 1972;Kelley, 1967), for example, as well as gen-erally accepted explanations for phenomenalike bystander (non)intervention (Darley &Latane, 1968), all start out with the underly-ing assumption that people attend to their

The authors are grateful to Robert Abelson for hiscomments on an earlier draft of this manuscript andto Cynthia Weinman for conducting Experiment 1.

Requests for reprints should be sent to EllenLanger, Department of Psychology and Social Rela-tions, 1318 William James Hall, Harvard University,Cambridge, Massachusetts 02138.

world and derive behavioral strategies basedon current incoming information. The ques-tion raised here is not whether these formula-tions are correct, nor is it whether people arecapable of thoughtful action. Instead, wequestion how often people outside of the labo-ratory are actually mindful of the variablesthat are relevant for the subject and for theexperimenter in the laboratory, and by im-plication, then, how adequate our theories ofsocial psychology really are.

This article questions whether, in fact, be-havior is actually accomplished much of thetime without paying attention to the substan-tive details of the "informative" environment.This idea is obviously not new. Discussions ofmind/body dualism by philosophers and theconsequences that different versions of thisrelation have on its status as an isomorphic,deterministic, or necessary relationship be-tween the two are part of psychology's her-itage. However, the extent of the implicationsof this idea has not been fully appreciated norresearched. How much behavior can go onwithout full awareness? Clearly, simple motoracts may be overlearned and performed auto-

Copyright 1978 by the American Psychological Association, Inc. 0022-3514/78/3606-0635$00.7S

635

636 E. LANGER, A. BLANK, AND B. CHANOWITZ

matically, but what about complex social in-teractions?

The class of behavior of greatest interesthere is not that which is commonly under-stood to be automatic, such as walking ortypewriting, but rather that which is com-monly assumed to be mindful but may be, infact, rather automatic. We shall refer to ithere as mindless behavior—mindless in thesense that attention is not paid precisely tothose substantive elements that are relevantfor the successful resolution of the situation.It has all the external earmarks of mindfulaction, but new information actually is notbeing processed. Instead, prior scripts, writtenwhen similar information really was once new,are stcreotypically reenacted. Berne (1964)discussed the idea of scripts in a popularizedway, and Abelson (1976) rigorously elab-orated the concept in generating a computersimulation of belief systems. To Abelson,a script is a "highly stylized sequence oftypical events in a well-understood situation,. . . a coherent sequence of events expectedby the individual, involving him either as aparticipant or as an observer." (p. 33) (SeeAuthor's note, p. 642.)

The notion of a script was used -to describea study by Langer and Abelson (1972),where it was argued that asking a favor hadcertain script dimensions and that the successof getting compliance depended on the specificsyntax of the request rather than on thespecific content of the statement, fn thatstudy, the words making up the request wereheld constant, while the order of the wordsspoken was varied. The opening words deter-mined which script was followed, and compli-ance varied accordingly. Similar to the notionof script is Goffman's (1974) concept offrames, Harre and Secord's (1973) idea ofepisode, Thorngate's (1976) idea of carica-ture, Miller, Galanter, and Pribram's (1960)notion of plans, and Neisser's (1967) conceptof preattentive processing. Each of these for-mulations speaks 1,0 the individual's abilityto abide by the particulars of the situationwithout mindful reference to those particulars.

However, while Abelson has come closestto delineating the structure of scripts, no onehas yet experimentally determined the min-imum requirements necessary to invoke a par-

ticular script, nor has scripted behavior reallybeen demonstrated to be mindless. While theformer issue is not addressed in the presentarticle, the latter is the article's main concern,and we may shed some light on the require-ments for script learning and enactment oncethe mindlcssness of ostensibly thoughtfulactions has been demonstrated. This suggeststhat the essence of a script may not lie inrecurring semantics but rather in more generalparalinguistic features of the message. Whenwe speak of people organizing incoming in-formation, it is as important to take intoaccount what they systematically ignore as itis to take into account what they systemat-ically process. And when we speak of peopleignoring information, it is important to dis-tinguish between information that is ignoredbecause it is irrelevant and information thatis ignored because it is already known. It isknown because it has been seen many timesin the past, and aspects of its structure thatregularly appear indicate that this time is justlike the last. Thus, what is meant by mind-lessness here is this specific ignorance ofrelevant substance.

This article reports three field experimentsundertaken to test the mindlessness of osten-sibly thoughtful action in the domains ofspoken and written communication. It washypothesized that when habit is inadequate,thoughtful behavior will result and that thiswill be the case when cither of two conditionsis met: (a) when the message transmitted isstructurally (rather than scmantically) novelor (b) when the interaction requires an ef-fortful response.

Experiment 1

Method

The first experiment was conducted in the contextof a compliance paradigm, where people about touse a copying machine were asked to let another per-son use it first. The study utilized a 3 X 2 factorialdesign in which the variables of interest were thetype of information presented (request; request plus"placcbic" information; request plus real informa-tion) and the amount of effort compliance entailed(small or large).

Subjects. The subjects were 120 adults (68 malesand 52 females) who used the copying machine atthe Graduate Center of the City University of New

MINDLESSNESS OF OSTENSIBLY THOUGHTFUL ACTION 637

York. Each person who approached the machine onthe days of the experiment was used as a subjectunless there was a line at the machine when theexperimenter arrived or a person came to use themachine immediately after a subject had been ap-proached. (There was a minimum wait of S minutesbetween subjects). Half of the experimental sessionswere conducted by a female who was blind to theexperimental hypotheses, and the remaining sessionswere run by a male experimenter who knew thehypotheses.

Procedure. Subjects were randomly assigned intoone of the groups described below. The experimenterwas seated at a table in the library that permitted aview of the copier. When a subject approached thecopier and placed the material to be copied on themachine, the subject was approached by the experi-menter just before he or she deposited the moneynecessary to begin copying. The subject was thenasked to let the experimenter use the machine firstto copy either 5 or 20 pages. (The number of pagesthe experimenter had, in combination with the num-ber of pages the subject had, determined whether therequest was small or large. If the subject had morepages to copy than the experimenter, the favor wasconsidered small, and if the subject had fewer pagesto copy, the favor was taken to be large). The ex-perimenter's request to use the machine was madein one of the following ways:

1. Request only. "Excuse me, I have S (20) pages.May I use the xerox machine?"

2. Placebic information. "Excuse me, I have 5 (20)pages. May I use the xerox machine, because I haveto make copies?"

3. Real information. "Excuse me, I have S (20)pages. May I use the xerox machine, because I'm in arush?"

Once the request was made and either complied ornot complied with, the experimenter returned to thetable and counted the number of copies the subjectmade. The dependent measure was whether subjectscomplied with the experimenter's request.

If subjects were processing the informationcommunicated by the experimenter, then therate of compliance should be equivalent forGroups 1 and 2, since the amount of infor-mation conveyed is the same for both ofthese groups, but it might be different forGroup 3, since this group received additionalinformation. If, however, subjects are re-sponding to the situation on the basis of aprior script that reads something like "FavorX + Reason Y —> Comply," then the rate ofcompliance should be the same for Groups 2and 3 (placebic and real information) andgreater than for Group 1 (request only). Itwas predicted that the latter result would ob-tain. Thus, while the information given toGroup 2 was redundant in an information

Table 1Proportion of Subjects Who Agreed to Let theExperimenter Use the Copying Machine

Reason

Favor

Smalln

Bign

Noinfo.

.60IS

.2425

Placebicinfo.

.9315

.2425

Sufficientinfo.

.9416

.4224

theory sense (Shannon & Weaver, 1949), itwas predicted to be necessary, and thus notredundant, in a script sense.

As stated earlier, it was assumed that peo-ple would not behave in this pseudothinkingway when responding was potentially effort-ful. Then, there is sufficient motivation forattention to shift from simple physical char-acteristics of the message to -the semanticfactors, resulting in processing of current in-formation. Thus, it was predicted that as thefavor became more demanding, the placebic-information group would behave more likethe request-only group and differently (yield-ing a lower rate of compliance) from thereal-information group.

Results and Discussion

The proportion of subjects who complied ineach group was computed, and a 3 X 2 X 2(Request X Effort X Experimenter) analysisof variance was performed using 0 and 1 asscores (complied vs. did not comply). Thisanalysis yielded three main effects: communi-cation, F(2, 108) = 3.02, p< .05; effort,F(l, 108) = 43.40, p<.001; and experi-menter, F(l, 108) = 6.67, p < .01. The pro-portions of subjects who complied with thedifferent requests are presented in Table 1.Not surprisingly, the female experimenter hada higher rate of compliance than the maleexperimenter, but since there were no inter-actions between this variable and the others',the data are combined in the table for ease ofreading. A contrast analysis using planned,orthogonal comparisons was performed. Thecontrast analyses that were performed set thesmall effort/placebic-information group andthe small effort/sufficient-information group

638 E. LANGER, A. BLANK, AND B. CHANOWITZ

as equal to each other but distinct from thesmall effort/no-information group; the largeeffort/sufficient-information group was con-trasted with the large efforl/placebic-informa-tion group and the large effort/no-informa-tion group. These contrasts reflect the hy-pothesis that when there was small effortinvolved, the placebic-information groupwould be similar to the sufficient-informationgroup but that when effort was large, theplacebic-information group would be similarto the no-information condition. It was foundthat for the small-effort contrast, the meansof the placebic- and sufficient-informationconditions were virtually identical and signif-icantly different from the no-information con-dition, F(l, 114) =6.35, p < .05. For thecontrast comparing the more effortful favor,the no-information and placebic-informationgroups were identical and tended to be differ-ent from the sufficient-information group,^(1, 114) = 2.83, .10 <p> .05.

Also, and not surprisingly, for requests ofthe same type, small requests result in greatercompliance than larger requests.

The results support the hypothesis that aninteraction that appears to be mindful, be-tween two people who are strangers to eachother and thus have no history that wouldenable precise prediction of each other's be-havior, and in which there are no formal rolesto fall back on to replace that history, can,nevertheless, proceed rather automatically.1 f a reason was presented to the subject, he orshe was more likely to comply than if noreason was presented, even if the reason con-veyed no information. Once compliance withthe request required a modicum of effort onthe subject's part, thoughtful respondingseemed to take the place of mindlessness, andthe reason now seemed to matter. Under thesecircumstances, subjects were more likely tocomply with the request based on the ade-quacy of the reason presented.

Experiment 2

The next two experiments attempted to ex-tend the results of Experiment 1 to the do-main of written communications, since it isour contention that pseudothinking behavioris more the rule than the exception for prac-

tically all verbal behavior as well as nonverbalbehavior. The more one participates in anyactivity, the more likely it is that scriptlikequalities will emerge. Through repeated ex-posure to a situation and its variations, theindividual learns to ignore and remain igno-rant of the peculiar semantics of the situation.Rather, one pays attention to the scripted cuepoints that invite participation by {he indi-vidual in regular ways.

In Experiments 2 and 3, we sought to en-gage subjects in an activity that would havefor them scripted qualities. Specifically, theactivity we chose involved receiving and re-sponding to letters and memoranda that weresent through either the U.S. Mail or inter-office mail, depending on the study. As inExperiment 1, it was assumed that ostensiblythoughtful action would proceed mindlessly aslong as the structure of the activity involvedremained consistent with its scripted char-acter.

Following this assumption, we expected thatindividuals who received mail that asked fora response would return what was requestedif the communication was structurally phrasedso as to follow the commonly expected scriptfor mail. The return of the response wouldserve as evidence of the fact that the personhad read the material and engaged in theactivity of correspondence through the mail.If the communications to the subject weresemantically senseless and yet fulfilled thescript requirements for written communica-tion, we could safely assume that the returnof the mail signified that we had engaged thesubject in mindless behavior—that he or shehad not "thought about" the material but hadreturned it merely because it satisfied thestructural requisites for a habitual behavior.To make the case more strongly, we sent tothe subjects communications that were equally'senseless semantically but which varied intheir adherence to the structural requirementsof communications. If the responses varieddirectly with the adherence to structural con-sistency expected in communications, we couldinfer that the behavior that led to the sub-jects' returns was of a scripted character—entirely habitual, despite the fact that, on theface of it, if we observed the behavior we

MINDLESSNESS OF OSTENSIBLY THOUGHTFUL ACTION 639

would assume it was thoughtfully processedin character.

In Experiment 2, subjects were mailed ameaningless, five-item questionnaire. Thecover letter either demanded or requested thereturn of the questionnaire and was eithersigned (e.g., "Thank you for your help,George L. Lewis") or unsigned. It was as-sumed that signed requests and unsigneddemands were more congruent with the struc-ture of most written communications than un-signed requests and signed demands andtherefore would be more conducive to sustain-ing mindless behavior. The cover letter hadno letterhead and could not possibly, withthought, be construed as representing a legit-imate authority. Therefore, "thoughtful"processing of the cover letter would not un-cover any rational reasons for returning thequestionnaire.

In order to test whether habitual respond-ing was taking place, rather than merelypolite compliance, two groups of subjectswere selected who were assumed to vary intheir experience with written communications.It was predicted that the more experiencedsubjects (who were also the more educatedsubjects) would be more likely to return thequestionnaire when the structure of the re-quest/demand was consistent with their pastthan the less experienced subjects, for whomcongruency was not expected to matter.

Method

Subjects. Forty subjects were selected randomlyfrom the Manhattan telephone directory and consti-tuted the random-status group. Another 40 subjectswere chosen randomly from the "Physicians" sectionof the Manhattan Yellow Pages and constituted thehigh-status group.

Procedure. Each subject received a questionnairein the U.S. Mail consisting of the five following ques-tions:

1. The subway or bus is the more enjoyable modeof public transportation ?

2. Movies or plays are the more enjoyable form ofpublic entertainment ?

,3. Libraries or parks are the more enjoyable formof free public entertainment?

4. Forests or playgrounds are the more enjoyablepublic places to spend time ?

5. Cash or credit cards is the more efficient form ofpublic exchange of goods?

All subjects received the questionnaire at their resi-dence. Along with the questionnaire, the subjects

Table 2Proportion of Subjects Who Returned theQuestionnaire

Status

Condition

Congruentn

Iricongrueiifn

High

.5520

.3219

.2020

.3719

received a stamped envelope addressed to a postoffice box, as well as a cover sheet that varied in oneof the following four ways:

1. Congruent conditions, (a) Request/personal—"Iwould appreciate it if you would fill out the attachedquestionnaire and return it in the enclosed envelopeto me by September 10. Thank you for your help,George L. Lewis," (h) Demand/impersonal—"Theattached questionnaire is to be filled out and re-turned by September 10."

2. Incongnient conditions, (a) Request/impersonal—"I would appreciate it if you would fill out theattached questionnaire and return it in the enclosedenvelope to me by September 10." (b) Demand/per-sonal—"The attached questionnaire is to be filled outand returned in the enclosed envelope by September10. Thank you for your help, George L. Lewis."

Thus, the study was a 2 (random vs. high status)X 2 (request vs. demand) X 2 (personal vs. imper-sonal) factorial design. Again, it was predicted thathigh-status subjects who received congruent com-munications would be more likely to comply thanthe other groups.

Results and Discussion

Table 2 presents the proportion of subjectswho returned the questionnaire, by congru-ence and status.1 An analysis of variance wasperformed using 0 and 1 scores. Althoughthere were no main effects, a contrast that setthe high-status congruent group as differentfrom the remaining groups, which in turn wereequal to each other, was significant at p <.OS, F(l, 74) = 5.91. The congruent and in-congruent cells of Table 2 are broken downfor examination in Table 3. The analyses ofvariance of these data were not significant.However, there was a trend for a three-wayinteraction, F(l, 70) - 3.48, p < .08, whichindicates again that the congruency effect

1 Two of the original letters were returned withthe notice that the addressee no longer lived at theaddress. Hence, there were 78 subjects in the study.

640 E. LANGER, A. BLANK, AND B, CHANOWITZ

Table 3Proportion of Subjects Who Returned theQuestionnaire

High status Random status

Vet- 1m- Per- 1m-CoudiUon sonal personal sonal personal

Demandn

Requestn

.339

.7010

.4010

.3010

.449

.2010

.2010

.3010

tends to be modified by status. It appears thatour notion of what is congruent was correctonly for people like ourselves, who have hadan abundance of certain kinds of written com-munications and not others. That is, instead ofthere being a general script for written com-munications, there are probably several scriptspeculiar to individuals in their relation to so-cial institutions. In fact, on second thought,it seems that communiques sent from em-ployer to employee, or from manager to officeworker (the latter two probably comprisedmuch of the random-status group), wouldmore than likely be either of the demand/personal or request/impersonal sort, sincethese forms allow the sender to maintain hisor her status while still observing a modicumof civility.

Experiment 3 was undertaken to testagain, more rigorously, the mindlessness ofostensibly thoughtful actions in regard towritten communications. However, for thisstudy, the script was first determined em-pirically and then tested.

Experiment 3

Method

Eighty-three memoranda were collected from thewastepaper baskets of 20 secretaries of various de-partments at the Graduate Center of the City Uni-versity of New York. Sixty-eight percent of thesehad the request/impersonal form described earlier.While varying in content, each of these communica-tions requested rather than demanded that the secre-tary do something (e.g., "Please make 20 copies ofthis"), and none were signed at the bottom of therequest. Thus, for this group of people, the com-munication most congruent with their experiencewould be request/impersonal. Even though in theseinstances the receiver in all likelihood knew who thesender was, this kind of communication is still con-

sidered impersonal, since it stands in contrast tothose communications where the sender also is knownbut where the memo is signed just the same. Thedistinction between signed and unsigned memos isbeing drawn, in spile of the fact that in both casesthe sender is known, because small structural differ-ences of this kind arc predicted to either cue in ascript or not, depending upon one's past experience.The remaining 32% of the memos were virtuallyequally distributed among the other categories. Withthis in mind, 40 secretaries at the Graduate Centerwere sent, through interoffice mail, a senseless mem-orandum that was either congruent with their ex-perience or incongruent. In order to allow for com-parisons with Experiment 2, the same four forms ofwritten communication that were used previouslywere randomly sent to these subjects. However, nowthere were one congruent form (request/impersonal)and three incongruent forms (request/personal, de-mand/personal, domand/impcrsonal) :

Request. "1 would appreciate it if you would re-turn this paper immediately to Room 238 throughinteroffice mail."

Demand. "This paper is to be returned immediatelyto Room 238 through interoffice mail."

Half of each of these messages were signed ("Sin-cerely, John Lewis"), and half were unsigned andmerely had a number (R374-021-A) at the bottom ofthe message.

Nothing more was written on the memo. Subjectswere simply asked to return a piece of paper thatasked them only to return that paper to Room 238.The designated room did not exist in the building.The mailroom attendants put the returned lettersaside for us.

Thus, the study utilized a 2 (request vs. demand)X 2 (personal vs. impersonal) factorial design, with10 subjects in each cell.

Results and Discussion

Table 4 presents the proportion of subjectswho returned the letters as a function of thevarious conditions. To test the hypothesis thatmindless behavior will result when script re-quirements arc met, the proportions of sub-jects who returned the memo in the congruentcondition (.90) and the incongruent condi-tions (.60) were compared. Using 0 and 1scores, the analysis showed them to be signif-icantly different from each other, i(38) —1.78, p < .05. It should be noted that whatwe are calling congruent was determined bysampling a fraction of the secretaries' pastexperience with written communications.Sixty-eight percent of the memos fell into therequest/impersonal condition. Quite possibly,if we had mapped out first what was congru-ent for each secretary and then sent the ap-propriately structured-for-congruence memo

MINDLESSNESS OF OSTENSIBLY THOUGHTFUL ACTION 641

to her or him, the compliance might havereached 100%.

Experiments 2 and 3 provide support forthe mindlessness hypothesis in regard to writ-ten communications. It would seem thatthoughtful processing of the information com-municated to these subjects would have re-sulted in a nonresponse from them. Neverthe-less, when the script was congruent with sub-jects' experience, 55% of the physicians and90c/(, of the secretaries complied with 'themeaningless communication.

Conchisions

These studies taken together support thecontention that when the structure of acommunication, be it oral or written, seman-tically sound or senseless, is congruent withone's past experience, it may occasion be-havior mindless of relevant details. Clearly,some information from the situation must beprocessed in order for a script to be cued.However, what is being suggested here is thatonly a minimal amount of structural informa-tion may be attended to and that this in-formation may not be the most useful part ofthe information available. While the authorsdo in fact believe that people very oftennegotiate their interpersonal environmentsmindlessly, studies like these may simplydemonstrate that subjects are not thinkingabout what one thinks -they are thinking about(i.e., what is relevant), rather than demon-strating that their minds are relatively blank.If we knew all of the things subjects could bethinking about, we could use the present ex-perimental paradigm to at least test thisalternative. However, since there are an in-finite number of thoughts subjects may bethinking, this strong hypothesis will have toremain at the level of conjecture until otherexperimental methods are devised. The dif-ficulty of inventing such a methodologyshould not preclude efforts in that direction,since if mindlessness is the rule rather thanthe exception, man}' of the findings in socialpsychology would have to be reformulated(see Langer, 1978, for a more detailed discus-sion of this point).

While these studies ma}' be open to alterna-tive interpretations, they suggest that per-

Table 4Proportion of Subjects Who Returned the Memo

Memo type

Condition

Demand

Request

Personal Impersonal

.60

.70

.50

.90

Note, n = 10/cell.

haps there has been misdirected emphasis onpeople as rational information processors. In-stead of viewing people as either rational orirrational, it would seem wise to at least con-sider the possibility that their behavior maybe arational and yet in some way systematic.These studies then raise questions about theinferential processes traditionally assumed bycognitive social psychology. This has beenalluded to by Bern (1972) and more recentlyby Dweck and Gilliard (197S). It may not bethat a person weighs information and thenproceeds but that he or she more often justproceeds on the basis of structural cues thatoccasion further regular participation in theinteraction. To the extent that this scriptdomination is typical of daily interaction, cor-rections must be made in our accounts of howindividuals behave.

When does this mindless activity takeplace? If the interpretation offered for thesestudies is correct, then it would suggest thatthe occurrence may not be infrequent norrestricted to overlearned motoric behavior liketypewriting. Instead, if complex verbal inter-actions can be overlearned, mindlessness mayindeed be the most common mode of socialinteraction. While such mindlessness may attimes be troublesome, this degree of selectiveattention, of tuning the external world out,may be an achievement (cf. Langer, 1978)and perhaps should be studied as such. Atleast it would seem that both the advantagesand disadvantages should be investigated, asthe boundaries of the phenomenon are de-limited. At present, however, we may be inthe uncomfortable position of ovcrgencralizingour laboratory findings for reasons not yetmentioned by laboratory-research critics.Once an individual is brought into the labora-tory he or she is likely to be self-conscious.

642 E. LANGER, A. BLANK, AND B. CHANOWITZ

This self-consciousness may be thought pro-voking and habit inhibiting. Thus, we maybe lef t with the situation where we are study-ing the responses of thinking subjects andthen generalizing to successfully nonthinkingpeople.

Author's nole. Since the Langer and Abclson (1972)paper was published, there have been diverging usesof the terra script which did not become apparentuntil after this manuscript was prepared. The clari-fication of the present distinction lies in the degreeof active information processing implied by theword script. Abelson's use of the term script seemsto allow a range of cognitive activity. In ourformulation, the use of script signifies only relativecognitive inactivity. To avoid confusion, the wordscript as it appears in this article should be read as"mindlessncss."

References

Abelson, R. P. Script processing in attitude formationand decision-making. In J. S. Carroll & J, W.Payne (Eds,), Cognition and social behavior. Hills-dale, N.J.: Erlbaum, 1976.

Abclson, R. P , Aronson, E., McGuirc, W. L., New-comb, T. M., Rosenberg, M. J., & Taunenbaum,P. H. (Eds.). Theories of cognitive consistency:A sowcebook. Chicago: Rand McNally, 1968.

Bern, 1). J. Self-perception theory. In L. Berkowilz(Ed.), Advances in experimental social psychology(Vol. 6). New York: Academic Press, 1972.

Berne, E. Games people play. New York: GrovePress, 1964.

Daricy, J. M., & Latanc, B. Bystander interventionin emergencies: Diffusion of responsibility. Journalof Personality and Social Psychology, 1968, 8, 377-383.

Dweck, C., & Gilliard, D. Expectancy statements asdeterminants of reactions to failure: Sex differences

in persistence and expectancy change. Journal ojPersonality and Social Psychology, 1975, 32, 1077-1084.

Feslingcr, L, A theory of social communication pro-cesses. Human Relations, 1954, 7, 117-140.

Goffman, E. Frame analysis: An essay on the orga-nization of experience. New York: Harper & Row,1974.

Heider, F. The psychology oj interpersonal relations.New York: Wiley, 1958.

Harrc, H., & Secord, P. F. The explanation of socialbehavior. Totowa, N.J.: Litllefield, Adams, 1973.

Jones, E. E., Kanouse, D. E., Kellcy, H. II, Nisbctt,R. E., Valins, S., & Weiner, B. Attribution: Per-ceiving the causes of behavior. Morristown, N.J.:General Learning Press, 1972.

Kclley, H. H. Attribution theory in social psychology.In D. Levine (Ed,), Nebraska Symposium on Mo-tivation (Vol. IS). Lincoln: University of NebraskaPress, 1967.

Langer, E. J. Rethinking the role of thought in socialinteraction. In j. Harvey, W. Ickcs, & R. Kidd(Eds,), New directions in attribution theory (Vol.2 ) . Hillsdalc, N.J.: Erlbaum, 1978.

Langer, E , & Abelson, R. P. The semantics of askinga favor ; How to succeed in getting help withoutreally dying. Journal oj Personality and SocialPsychology, 1972, 24, 26-32.

Miller, G. A., Galanler, E., & Pribram, K. H. Plansand the structure oj behavior. New York: Holt,Rinehart & Winston, 1960.

Ncisscr, U. Cognitive psychology. New York: Apple-ton-Century-Crofts, 1967.

Schachter, S. The. psychology of affiliation. Stanford,Calif.: Stanford University Press, 1959.

Shannon, C. E., & Weaver, W. The mathematicaltheory of communication. Urbana: University ofIllinois Press, 1949.

Thorngate, W. Must we always think before we act?Personality and Social Psychology Bulletin, 1976,2, 31-35.

Received January 25, 1978 •