ED 063 558AUTHORTITLEINSTITUTIONPUB DATP1NOTEEDRS PRICEDESCRIPrORSDOCU - PDF document

1 ED 063 558AUTHORTITLEINSTITUTIONPUB DATP
ED 063 558AUTHORTITLEINSTITUTIONPUB DATP1NOTEEDRS PRICEDESCRIPrORSDOCUMENT RESUMECG 007 258Deci, Edward L.; Cascio, Wayne F.Changes in Intrinsic Motivation as a Function ofNegative Feedback and Threats.Rochester Univ., N.Y.Apr 7224p.; Paper presented at the Eastern PsychologicalAssociation Meeting in Boston, Massachusetts, April19, 1972MF-$0.65 HC-$3.29*Behaviora1 Science Research; Learning Motivation;*Low Motivation; *Motivationl *Motivation Techniques;Positive Reinforcement; Reinforcement; RewardsABSTRACTRecent studies have demonstrated that externalrewards can affect intrinsic motivation to perform an activity. Moneytenis to decrease intrinsic motivation, whereas positive verbalreinforcements tend to increase intrinsic 1N)tivation. This paperpresents evidence that negative feedback a',1 threats of punishmentalso decrease intrinsic motivation. Subjo-zs solved puzzles duringthe first part of the experimental sessiafter which observationsrelevant to their intrinsic motivation wk%: made. Subjectsin thenegative feedback condition were given more (lifficult puzzles tosolve than were the controls so that they rcl*led more freplently thanthe control subjects. Those in the high fai,condition showed lessintrinsic motivation following their puzzle-Ilving

2 session than didcontrol subjects. Subje
session than didcontrol subjects. Subjects in the threat conion received anaversive buzzer each time they were unable to rolve a puzzle, whiI,the control subjects did not. Those subjects thtened with tnebuzzer showed less intrinsic motivation than cortrol subjects.(Author/9W) Changes in Intrinsic MotivationAs A Function Of NegativeFeedback And ThreatsEdward L. Dad andWayne F.CasclolUniversity of RochesterPaper presented at EasternPsychological Association Meeting,Boston, Mass., April, 1972.U.S. ^EPARTMENTOF HEALTH.EDUCATION & WELFAREOFFICE OF EDUCATIONTHIS DOCUMENTHAS BEEH REPRODUCED EXACTLY ASRECEIVED FROMTHE PERSON ORORGANIZATION ORIGINATING IT POINTSOF VIEW OR OPINIONS STATED DONOT NECESSARILYREPRESENT OFFICIALOFFICE OF EDUCATION POSITION ORPOLICY ABSTRACTRecent studies by Deci (1971, 1972a) have demonstratedthat externalrewards can affect intrinsic motivationto perform an activity.Moneyhas been shown to decrease intrinsicmotivation, whereas positive verbalreinforcements tend to increase intrinsicmotivation.The present paperpresents evidence that negative feedback and threatsof punishment alsodecrease intrinci:: motivation.Subjects in this study solved puzzlesduring the first part of the experimentalsession, and then observationsrelevant to t

3 heir intrinsic motivationwere made.Subje
heir intrinsic motivationwere made.Subjects in the nega-tive feedback conditionwere given very difficult puzzles to solve sothat they failed onmore puzzles than the control subjects who were giveneasier puzzles.Those in the high failure (negative feedback)conditionshowed loss intrinsic motivation followingtheir puzzle-solving sessionthan did control subjects.Subjects in tne threat condition receivedanaversive buzzer each time theywere unable to solve a puzzle, while thecontrol subjects did not receive the buzzer.Those subjects threatenedwith the buzzer showed less intrinsicmotivation than control subjects.The resultswere seen to be consistent with a cognitive evaluation theorypresented earlier by Deci (1972a).2 Deci aad Cascio1when a person engages in some activityfor no apparent rewardexcept the activity itself, he is saidto be intrinsically motivatedto perform that activity.Several recent studies (Decis 1971,1972a)have shown that a person's intrinsicmotivation for an activity willbeaffected when he receives externalreinforcements for performing theactivity.Money has been shown to decrease intrinsicmotivation, whereasverbal reinforcements tend to increaseintrinsic motivation.Deci hasproposed a cognitive evaluation theoryto account for these phen

4 omena.ThQ theory focuseson a person's co
omena.ThQ theory focuseson a person's cognitive interpretation of theactivity and his reasons for performingthe activity.It suggests thaLsince different reinforcements will beperceived and evaluated differentlyby a person, different reinforcementswill have different effectson hisintrinsic motivation.Specifically, the theory states that thereare two processes bywhich intrinsic motivation will beaffected.One process involvesachange in the person's perception oflocus of causality (Heider, 1958;de Charms, 19(38).When he is intrinsically motivated,the locus ofcausality of the activity is within himself;however, when externalreinforcements are introduced, hemay become dependent on those externalreinforcements.He is, therefore, performing for theexternal reinforce-ment rather than for the intrinsic enjoyment.The locus of causality hasshifted from within himself to theexternal reinforcement, and he wouldbe less likely to perform the activityin the absence of these reinforce-ments.Pis intrinsic motivation has decreased. Deci and Cascio2Deci (1971, 1972a) has reported studies in whichmoney has beenas external rewards to subjects who performed intrinsically moti-vated activities.Money is widely perceivedas an effective controlmechanism (Skinner, 1953; Opsah

5 lDunnette, 1966), and the experimentsind
lDunnette, 1966), and the experimentsindicate that the subjects did, infact, become dependenton the ex-ternal rewards and lose intrinsicmotivation.Kiesler and Sakumura(1966) havereported results thatare consistentwith these.They found that if people statedopinions which theybelieved and received largepayments ($5) for doing so, theywere moresusceptible to counterarguments thanpeople who stated the opinions forsmall payments ($1).The cognitive evaluationinterpretation of this isthat the larger the payment, themore the subjects came to believe that theywere performing for the payment,so the less strongly they held to theiroriginal attitudes because theyno longer perceive of their attitudesas thereason for their behavior.There are also numerous other studieswhichbear on the cognitive evaluationtheory; however, they willnot be reviewedhere as they have been discussedelsewhere (Deci, 1971; 1972b).The secondprocess by which intrinsic motivationcan change is relatedto information or feedback.When a person is performingan intrinsicallymotivated activity, thereare positive value properties (Koch, 1956)associated with the activity.If he then receives feedback,the valueproperties associated with this feedbackmay not be phenomenologically dis-tinguishable fr

6 om the value propertieswhich the person
om the value propertieswhich the person associates with theactivity itself, so there isa change in the total amount of posicive4 Doci and Cnscio3value.Therefore, if a person receivespositive feedback, the positivevalue properties from the feedbackincrease the total positivevalueproperties associated with theactivity, causingan increase in his in-trinsic motivation.That is, there isan increase in his tendency toperform the activity in theabsence of external reinforcementsbecause ofthe positive value that heexperiences from the activity.This positivevalue which theperson experiences from the feedbackmay be due to a con-firmation of hiscowpetence or his sense of self-determination.Deci (1971, 1972a) hasreported that subjects(especially males) whowere rewarded for solving puzzles withpositive verbal reinforcementsfrom the experimenter (e.g."That's very good; thispuzzle was a difficultone") showedan increase in intrinsic motivationfor working with thesepuzzles.Negative Feedback:If a person doingan intrinsically motivatedactivity receives negative feedbackabout his performance, this feedbackcould cause negative value propertiesto be associated with the activity.In the same way that positive feedbackand the resulting positive valuestrengthens intrinsic

7 motivation foran activity, negative feed
motivation foran activity, negative feedbackshould decreasea person's intrinsic motivation because the negativefeedback offsets some of the positivevalue of the activity.The negativefee(back would weaken his feelings ofcompetence and self-determination,making the activity less rewarding.As a reslat, he would be less likelyto do the activity in the absence ofexternal reiuforcements; that is,he would be less intrinsicallymotivated to do the activity.One would Doci and Cascio4expect, then, that if subjects were given very difficult tasks and if theyfailed badly at them, the feedback which they receive from failing at thetask would cause a decrease in intrinsic motivation.niethesis IPeople who receive self-administered negative feedback about theirperformance on an intrinsically motivated activity will show a decreasein intrinsic motivation.We know of no evidence which tests Hypothesis I; however, thereis some work by Feather (1966, 1968; Feather and Saville, 1967) which isrelated.He has shown that negative feedback about performance causessubjects to perform less well and have less confidence about subsequentperformance.While this is not evidence in support of Hypothesis I, itdoes seem to complement it.Threats of Punishment:Subjects who were paid fo

8 r performing aaintrinsically motivated a
r performing aaintrinsically motivated activity showed a decrease in intrinsic motiva-tion (Deci, 1971,1972a).It was suggested that this was due to a shiftin perceived locus of causality from within the person to the externalreward.By this same process, then, one would expect that if a personwho was performing an intrinsically motivated activity begins to receivethreats of punishment for not doing the activity satisfactorily, hisbehavior will become dependent on that external reinforcement (avoidanceof punishment), and he will lose intrinsic motivation.The threat of punislunent will pllenomellally, be a strong controlleror his behayior.Kite (1964) reports that punishments tend to be per- Doci and Cascioceived as more powerful external forces than rewards.Hence, we wouldexpect that the person who receives threats of punishment willcome toperceive that the reason for his behavior is the avoidance of theaver-sive stimulus.The perceived locus of causality of that activity willhave switched to the external control, and he will have less intrinsicmotivation.Wothesis IIIf a person receives threats of punishment for not performingsatisfactorily on an intrinsically motivated activity, his intrinsicmotivation for that activity will decrease.Although there

9 is no direct evidence which bearson Hypo
is no direct evidence which bearson Hypothesis II,there is some indirect evidence from researchon "insufficient justifi-cation" (Festinger, 1961; Aronson, 1966) which is related to the hypoth-esis.Aronson and Carlsmith (1963) have reportedthat children whore-frained from playing withan attractive toy because of strong threatsderogated the toy less than children whowere given mild threats.Chi1dronwho got strong threats refrained becauseof these threats (i.e., becauseof an external control), and they feltno need for internal justification.Those who refrainad because of mildthreats, however, did chang, theirattitudes toward the toyas a means of justifying their behavior.Whenthere was insufficient extrinsic control,they developed intrinsiccon-trol, namely, derogating the toyso they wouldn't "want" to play with it. Dad. and Cascio6Freedman (1965) has reported similar findings.Children who re-ceived mild threats (rather than strong threats)for refraining fromplaying with an attractive toy not only derogatedthe toy, butwere lesslikely to play with that toy ina free situation as much as two monthslater.These studies suggest that people will developintrinsic controlwhen there is insufficient external justification.Although these resultsneed not imply that

10 external controls (i.e. threatsof punis
external controls (i.e. threatsof punishment) willdecrease intrinsic motivation, theyare nonetheless consistent with thatprediction.METHODSubjects in this experimentwere 64 undergraduates at The Universityof Rochester who were randomly assignedto three conditions.Each subject participated fora cour session during which hespent much of his time working ona puzzle called Soma.The puzzle iscomposed of seven different pieces, eachof which is made to look likeit is three or four one-inch cubes.These pieces can be fitted togetherto form millions of configurations--onlya few of which were used forthe experiment.The first experimenter met each subject ina waiting room andescorted him to the experimentalroom where he was seated at a table.Theexperimenter then left th:ougha door at the back of the experimentalroomso that he would be outside the room observing througha one-way window.The subject knew, ofcourse, that the experimenter was observing him, and 7Deci and Casciohe commtmicated with the experimenter through an intercom.On the table in front of the subject were the seven puzzle pieces--each with a number on it so that the experimenter could refer to it overthe intercom.To the left of the subject was a stack with either three orfour configurations t

11 hat he would be asked to reproduce.To hi
hat he would be asked to reproduce.To his rightwere three other configurations.The top one of the thrce was a sample;the other two will be discussed below.On anal table to the subject'sright were the microphone, speaker, recent issues of three magazines (NewYorker, Time and Playboy), and an ashtray.When the experimenter got to his position behind the one-way window,he read the instructions to the subject.The subject was told that it wasan experiment to study certain problem-solving conceptst and hewould beasked to solve some puzzle problems.After the instructions were read,the experimenter told the subject to look at the s.nple to his right.hewas told how it could be solvedand was allowed about a minute to manipu-late the pieces and reproduce it.The subject then worked on the puzzleconfigurations in turn.During the session, the subject was asked to reproduce the configura-tions which had been drawn on paper for him.The time to complete eachconfiguation was measured with a stop watch, and if a subject were unableto reproduce a configuration within ten minutes, he was stopped and thenhe assembled it as the experimenter explained how.This let him know thatall the configurations ware possible. Decl and Cascio8The Soma was chosen as the experimental t

12 ask for severalreasons.First, pilot test
ask for severalreasons.First, pilot testing showed that college students do indec3 consider itan intrinsically interesting task; second, there are a variety of possibleconfigurations of the puzzle, ranging from very easy to impossible; andthird, by numbering each piece differently the experimenter could easilyexplain the correct solution over the intercomso he did not have to bein the room.This decreased the possibility of experimenter bias (Rosenthal,1966).There were 16 subjects in the controlgroup.They were asked toreproduce four configurations whichwere relatively easy configurations,although the difficulty of the four varied.To test Hypothesis I, that self-administered negative feedbackwould cause a decrease in intrinsic motivation, 24 subjects inoneexperimental group were given different Soma configurations to reproduce.The first two of these were very difficult to solve.All but three of thesubjects were unable to solve both, and those three were unable to solveone of the two.The third configuration was much easier and was thesameas the third one used for the control group and the other experimentalgroup.The manipulation then, was that the experimental subjectsweregiven much more difficult puzzles, which were expected to lead toa muchhigher

13 failure rate.To test Hypothesis II, the
failure rate.To test Hypothesis II, the 24 subjects in the second experimentalgroup were given the same four relatively easy puzzles as the controlsubjects were given.They were told that if they were unable to solve10 Deci and Cascio9any of the configurations within ten minutes, a buzzer would soundindicating that their time was up for that configuration.They werethen given a short (about I second) exposure to the extremely noxiousbuzzer.The subjects, therefore, realized that if they were unable tosolve a configuration, they would be punished with a noxious stimulus.The difference, then, between the second experimental group and thecontrol group was that the experimental group had been threatened withpunishment for not performing well.Consequent:1Y, they were performingthe intrinsically motivated activity partially to avoid a punishmert.To obtain the independent measure of motivation, the eximrimenterleft his position for a period of eight minutes following the puzzlesolving.The pretext was as follows:When a subject had completed thefour puzzles, tho experimenter told him that he had done all the problemsolving which he had to do, but there was one more thing which he wouldbe asked to do, and that was to complete a short questionnaire.Since itwas

14 an experiment in problem solving, the s
an experiment in problem solving, the sulk:ect would be asked a fewquestions about the way he had solved the puzzles.However, there werofour different sets of questions, only one of which would be most appro-priate for this subject and that would be determined by how he had doneon the puzzles.To select the appropriate set of questions, data fromthe se5sioa would be fed into a computer through a teletype.To dothis, the experimenter would have to leave for a short time, 5 to 10minutes.The subject was told that he could do anything that he caredto during that time, but he was asked to stay in the room.The experimenter Deci and Cascio10left his positii:n and entered the expermentalroom through the back doorand exited through the, front door. He then,very noisily, climbed a smallset of steps outside the room and.left the labarea through a door at thetop of the stairs.The subjectcould hear him climb the stairs andopenand close the door.The subject was then alone in theroom and was free to work onthe puzzles, read magazines, or do anything he liked.Therefore, theamount of time out of the eight minutes which he spent workingon thepuzzles was used as the &Ten:teatmeastAre of intrinsic motivation.Itwas reasoned that if he worked on the puzzles when hewas

15 alone for this"free-choice" time and whe
alone for this"free-choice" time and when hewas given an e-portunity to do other things,then he must be 5 4.ansically motivatedto do the activity.The amountof time out of the eight minutes which thesubject spent workingon thepuzzle was determined by a second experimenter whoo'aserved through theone-vrif window and used a stop watch to record the time.The secondexperimenter was blind to the condition andalso to the hypotheses of theexperiment.The first experimenter signaledto the second to assume hisposition just after the first experimenterleft the room.The secondexperimenter got to the outside of theone-way window through a differentdoor which the subjects did not know about.There is no indication thatthe subjects suspected that theywere being observed during this free-choice period.Since any subject who was unable to reproducea configurationwithin the ten minutes allowedwas shown the solution, the possibility Deci and Cascio11that the Zeigarnik (1927) effect would tnfluence whetheror not he workedon the puzzle in the eight-minute free choice period was minimized.The two configurations whose drawings were in the pile to his rightunder the sample during the entire experiment were impossible to do.Thisprecluded the possibility that a subject would

16 finisha configuration inthe eight-minut
finisha configuration inthe eight-minute period and have that be a causal factor in determiningwhether or not he continued working on the puzzle.After eight minutes, the first experimenter returned to theroom andasked the subject to complete the questionnaire.RESULTSBefore testing Hypothesis I, it is important to ascertain whetherthe manipulation was successful, namely, whether the experimentalgroupactually failed on more of the puzzles than the control group.Ile controlgroup uls unable to reproduce 22 out of 64 configurations (34.3 percent)whereas the experimental subjects missed 49 out of 72 (68.0 percent).The difference in the two percentages is highly significant, indicatingthat the manipulation was successful.It should be noted that the subjects in the first experimentalgroup uere given only three puzzles each, whereas those in the controlwere asked to reproduce four.The reason for this difference was stmplypragmatic, to keep the total experimental session less thanone hour.Any bias which this might introduce, however, would he expected to be inthe direction opposite to the prediction oC Hypothesis I for tworeasons.First, iC each experimental subject were given another difficult puzzle,_19tl Doci and Casein12then his failure would be even g

17 reater, which according to Hypothesis Is
reater, which according to Hypothesis Ishould produce an even greater decrease in intrinsic motivation.Secondly,a fourth puzzle for each of the experimental subjects would have increasedthe tendency for a satiation effect causing him to spend less free choicetime working on the puzzles.Hence, using only three, rather than fourpuzzles, for the experimental subjects means the experimental effectwould have to be stronger to be detected.Table 1 shows the results which test Hypothesis I.The controlgroup spent an average of 285.1 seconds of free choice time working onthe puzzles, whereas those who were given the difficultpuzzles spentInsert Table 1 about here.only 166.2 seconds working on the puzzles.The difference of 118.9seconds is significant at the five percent level (one-tailedt-test),thereby supporting Hypothesis I that self-administerednegative feed-back decreases intrinsic motivation.The test of Hypothesis II, that threats of punishmentwill causea decrease in intrinsic motivation is summarized in Table 2.The experi-mental subiects who were punished withan aversive buzzer for failure tosolve puzzles speut 193.5 seconds of free choice time workingon theInsert Table 2 about here.puzzles, while the control subjects as mentioned above,spent 285.114

18 Deci and Casa°13seconds.The difference (
Deci and Casa°13seconds.The difference (91.6 seconds) is significantat the .10level.While this does not math the customary significance level,it does nonetheless give some support to Hypothesis U.DISCUSSIONIt is clear from the experiment just described and from previousresearch that external reinforcements do affect intrinsic motivation,andthat the cognitive (valuation theory doesseem to account for those re-sults.As was mentioned in the introduction, research in thearea of in-sufficivnt justification is related to the cognitiveevaluation research.In essence, the insufficient justification research showsthat personswill develop internal justificationor control when there is insufficientexternal justification or control.In a classic study (FestingemetCarlsmith, 1959) subjectswere paid either $1 or $20 for telling anotherthat they had enjoyed a dull task.Those who received insufficientexternal rewards ($1) for lying grew to like the taskmore than those whoreceived sufficient rewards.Those who were insufficiently rewardeddeveloped internal justification, namelya more favorable attitude towardthe dull task.In the insufficient justification studies suchas the Pestingerand Carlsnith study, they begin witha dull, boring task, hence the lowreward subject

19 s have insufficient justification (rio i
s have insufficient justification (rio internal justification,and inadequate rewards) but the high reward subjects have sufficientjustification (no internal justification, but sufficient rewards).Theprediction is that the insufficient justification subjects will developintrinsic interest, while the sufficient justification subjects willnot.On the other hand, in the copitive evaluation studies (e.g.Peci, 1)71),an intrinsically interesting activity is mployed.Hence, the non-rewardedsubjects have sufficient justification (intrinsic interest, butno rmards) Deci and Cascio14and the rewarded subject; have "over sufficient" justification (intrinsicinterest, plus substantial rewards).The prediction is that the overjusttfication subjects will decrease intrinsic motivatlor.while thesufficieh' justification subjects will remain unchanped.The samereasoning applies to the Aronson and Carlsimth (1963) insufficient justi-fication study using threats, on the one hand, and the present Deci andCascio, over sufficient justification study using threats, on the other.Onecan sec then, that the insufficient justification phenomenon is a sort ofmirror image of the over sufficient justification or cognitive evaluationphenomenon.Nisbett and, Valins (1971) pointout, howeve

20 r, as did Deci (1971)that the processes
r, as did Deci (1971)that the processes underlying thetwo phenomena are not the same.Cog-nitive dissonance reduction (Festinger,1957) is suggestedas the processunderlying the insufficient justificationphenomenon; however, there isno reason to believe that too much justificationis dissonance producing.Nonetheless, although thetwo phenomena have different underlying psycho-logical processes, theyare still quite complementary (see Deci, 1972bfor a further discussion of this).Although the findings presentedin this paper give further supportto the cognitive evaluation theory, thereare still many unansweredquestions which need to be investigated.It was reported that positivefeedback can lead toan increase in intrinsic motivation.However, inone study (Deci, 1972a), one group of female subjectsshowed a substantial,though not significant, decreasein intrinsic motivation following theexperience with verbal reinforcements.It was suggested that the effectsof verbal reinforcements might bedependent on the amountor strength ofthe verbal reinforcements.There are two situations where verbal reinforce-ments might lead to a decrease in intrinsicmotivation.If there were agreat deal of positive verbal feedback, subjectscould become dependent on 'Doci and Casa()15it

21 (especially if the person giving the rei
(especially if the person giving the reinforcement were a significantevaluator).This would cause a decrease in intrinsic motivation throughthe process of change in perceived locus of causality.Alternatively, theverbal reinforcements or positive feedback might be perceived by thesubjects as attempts at ingratiation (Jones, 1964), in which case, theverbal messages would lose their reinforcing value.In this case,intrinsic motivation would not increase because the person does notexperience positive value associated with the ingratiating feedback, infact, he would probably experience negative value which wouldcause adecrease in intrinsic motivation. Aperson most often perceives that theother is ingratiating when there is a large amount of positive verbalfeedback, so both situations where we are suggesting that verbal reinforce-ment could cause a decrease in intrinsic motivation involve large amounts ofof verbal reinforcements.It is possible, therefore, that there may besome relationship such as an inverted U between amount of verbal reinforce-ment (or approval or positive feedback) and intrinsic motivation.Asverbal reinforcement increases, intrinsic motivation may increaseup to apoint and then begin to decrease.Of course, this is merely speculative,bu

22 t represents an interesting question for
t represents an interesting question for research.Similarly, the amount of negative feedbackmay not be monotonicallyrelated to intrinsic motivation.For example, a small amount of negativefeedback may serve as a challengeor a stimulation and perhaps increaseintrinsic motivation.In the puzzles given to the negative feedbackgroupin this experiment, the first twowere very difficult, and many subjectsreported after the experiment that they felt like "complete idiots"or"absolute fools" since they were unable to make muchprogress on these17 1)eci and Cascio16puzzles.Their competence (White, 1959) andtheir sense or personalcausality (de Charms, 1968),may have been strongly threatened causingnegative value to be associated withthis activity.This, in tarn, leadsto a decrease in intrinsic motivation.On the other hand, if there hadbeen a small amount of negativefeedback which did not leavethem feelinglike "absolute fools," theirsense of competence may not have beensothreatened, and this milder negativefeedback could have been perceivedasa challenge and led to an unchangedor possibly even enhanced intrinsicinterest in doing the activitywell.Again, this is speculation,but itmay be that the relationship between negativefeedback and intrinsicmotivation is also s

23 omethinglike an inverted U.In summation,
omethinglike an inverted U.In summation, the evidencepresented in this and preliiouspaperssuggests strongly that intrinsicmotivation can be affected byexternalreinforcements and controls; thecognitive evaluation theorydiscussedearlier seams useful in accountingfor these changes. Doci and CnscioREFERENCES17Aronson, E.The psychology of insufficient Justification.In S.Feldman (Ed.), gsmityLonsistoncy. Now York:AcademicPress, 1966.Pp. 115433.Aronson, E.,Carlsmith, J.M.Effect of severity of threaton thedevaluation of forbidden behavior.Journal of Abnormal and SocialPsycholow, 1963, 66, 584-588.deCharms, R.Personal causation: The internal affective determinants ofbehavior.New York: Academic Press, 1968.Deci, E.L.Effects of externally mediated rewards on intrinsic motivation.....JournaalorPersonalita.......dSocialPscholo, 1971, 18, 105-115.Deci, E.L.Intrinsic motivation, extrinsic reinforcement, and inequIty.Journal of Personalitand Social Ps cholt.om, 1974,12, 113_126:Deci, E. L.Intrinsic and extrinsic notivation:A cognitive evaluationtheory.Unpublished manuscript.University of Rochester, 1972.(b)Feather, N.T.Effects of prior success and failure on expectations ofsuccess and subsequent performance.Journal of Personality andSocial Psychology, 1966, 3, 2

24 87-298.Feather, N.T.Change in confidence
87-298.Feather, N.T.Change in confidence following success or failure as apredictor of subsequent performance.Journal of Personality and SocialEaslIplogx, 1968, 9, 38-46.Feather, N.T., & Saville, M.R.Effects of amount of prior success andfailure on expectations of success and subsequent task performance.Journal of Personalitand Social Psysholom5 1967,195, 226-232. Deci and Casein16Festinger, L.A, Iimmlaf_cianItive dissonance.Evanston, la.:Row Peterson, 1957.Festinger, L.The psychological effects of insufficientwwarr),s.All-icateilologist, 1961, 16, 1-11.Festinger, L., & Carlsmith, J.N.Cognitive consequencles of forcedcompliance.Journal of Abnormal and Socinl Plyehotu,1959,58, 203-210.Freedman, J.L.Long-term cognitive dissonance.Journal of ExperimentalSocial Psycyologb 1965, 1, 145-155.Heider, F.7he mych212mpf int2.0220212e1ations.New York: Wiley,1953.Jcmes, E.E.Ingratiation.New York: Appleton-Century-Crofts, 1964.Kiesler, C. A., & Sakumra, J.A test of a model for commitment.Journal of Personality p.nd SocialPsychology., 1966, 3, 349-353.Kite, W.R.Attributions of causalityas a function of theuse of rewardand punishment.Unpublished dissertation,Stanford University, 1964.Koch, S.Behavior as "intrinsically"regulated: Work notes towardapre-theory of p

25 henomena called"motivational."In M.R. Jo
henomena called"motivational."In M.R. Jones (3d.)Nebraska1956.Nisbett, R. E., & Valins, S.Perceiving the causes of one'sown1)havier.New York:General Learning Press,bdule Series, 1971.Opsahl, R.L, E1 Dunnette, M.D.The role of financialcompensation inindustrial motivation.Psycholooical Bulletin1966, 66, 94-118.Rosenthal, R.aerimenterNew York:Appleton-Century-Crofts, 1966. Deci and Cascio19Skinner, B.F.Science and human behavior:.New York: Froe Press, 1953.White, R.W.Motivation reconsidered: The concept ofcompetence.Eachalogical Review, 1959, 66, 297-333.Zeigarnik, B.Das behalten erledigter und unerledigter handlungen.Psychaogische Forschung, 1927, 9, 1-85. D'Ici and Cascio20FOOTNOTE1Theauthors would like to thank Larry Coff, Jim Gould,and StanKaplan for assistance with data collection. Deci and CascioTABLE 1NUMBER OFSECONDS OF "FREE-CHOICE" TIMESPENT BY CONTROL SUBJECTSANDSUBJECTSINTHENEGATIVE FEEDBACK CONDITION WORKING ON THE PUZZLES21ControlSubjectsNegative Feed-back SubjectsDifferenceSignificance285.1n = 16166.2n = 24...1118.9s.e. = 70.6.05 Deci and Cascio22TABLE 2NUMBER OF SECONDS OF "FREI3-CHOICE" TIMESPENT BY CONTROL SUBJECTS AND SUBJECIS IN THE THREAT aNDITIONControlSubjectsThreat,SubjectsDifferenceSignificance285.1n = 16193.5n = 2491.6s.e.