The Detrimental Effects ofExtrinsic Reinforcement · PDF fileTheDetrimental Effects...

15
The Behavior Analyst 1989, 12, 1-15 No. 1 (Spring) The Detrimental Effects of Extrinsic Reinforcement on "Intrinsic Motivation" Alyce M. Dickinson Western Michigan University Extrinsic consequences have been criticized on the grounds that they decrease intrinsic motivation or internally initiated behavior. Two popular rationales for this criticism, Lepper's overjustification hy- pothesis (1981) and Deci's motivational theory (Deci & Ryan, 1985), are reviewed and the criticism is then redefined behaviorally. "Intrinsically controlled" behavior is defined as behavior maintained by response-produced reinforcers, and the question concerning extrinsic consequences is thus restated as follows: When behavior is maintained by response-produced stimuli, does extrinsic reinforcement decrease the reinforcing value of those stimuli? The empirical support for this detrimental effect is summarized briefly, and several possible explanations for the phenomenon are offered. Research results that reflect on the effect's generality and social significance are discussed next, with the conclusion that the effect is transient and not likely to occur at all if extrinsic rewards are reinforcing, noncompetitive, based on reasonable performance standards, and delivered repetitively. Key words: extrinsic reinforcement, intrinsic motivation In applied settings operant condition- ing procedures often include the delib- erate manipulation of behavioral conse- quences such as money, praise, tokens, edibles, and access to preferred activities. For over 10 years such practices have been criticized on the grounds that al- though they may increase the frequency of behavior, they may also decrease an individual's "intrinsic motivation" to perform the behavior (Condry, 1977; Deci, 1971; Lepper & Greene, 1978). Thus, many have argued that behavior change programs involving the manip- ulation of consequences should be re- placed with more "intrinsically motivat- ing" systems in all applied settings, but particularly in schools, businesses, and clinics (Condry, 1977; Deci, 1975a; Deci & Ryan, 1985; Kohn, 1988; Levine & Fasnacht, 1974). The criticism that contingent conse- quences may decrease intrinsic motiva- tion is based on the belief that a sharp distinction can be made between behav- ior maintained by obvious environmen- Portions of this paper were presented at the an- nual meetings of the Association for Behavior Anal- ysis, May 1986, Milwaukee, Wisconsin, and the Northern California Association for Behavior Analysis, March 1987, San Mateo, California. I would like to express my appreciation to Wayne Fuqua, Jack Michael, and William Redmon for their helpful comments on earlier versions of this paper. tal consequences-extrinsically motivat- ed behavior-and behavior that occurs in the seeming absence of consequences- intrinsically motivated behavior. Accord- ing to this position, the impetus for ac- tion is primarily internal or intrinsic or primarily external or extrinsic (Pittman & Heller, 1987). Because intrinsically motivated behavior is believed to be more creative, spontaneous, and flexible than extrinsically motivated behavior, it is feared that by damaging intrinsic moti- vation, the use of extrinsic consequences will result in a decrease in highly valued human behavior (Amabile, 1982, 1985; Deci & Ryan, 1985, p. 35; McGraw & McCullers, 1979; Shapira, 1976). INTRINSICALLY VERSUS EXTRINSICALLY CONTROLLED BEHAVIOR Intrinsic motivation was originally proposed to account for exploratory and manipulative behavior in nonhuman an- imals (Harlow, 1950; Harlow, Harlow, & Meyer, 1950) and was later used to ex- plain human behavior believed to be caused by such motives as a need for achievement (McClelland, Atkinson, Clark, & Lowell, 1953), a need to be ef- fective when dealing with the environ- ment (White, 1959), a need to be self- determining (Deci, 1 975a), and a need to be a causal agent when dealing with the 1

Transcript of The Detrimental Effects ofExtrinsic Reinforcement · PDF fileTheDetrimental Effects...

The Behavior Analyst 1989, 12, 1-15 No. 1 (Spring)

The Detrimental Effects of Extrinsic Reinforcement on"Intrinsic Motivation"

Alyce M. DickinsonWestern Michigan University

Extrinsic consequences have been criticized on the grounds that they decrease intrinsic motivation orinternally initiated behavior. Two popular rationales for this criticism, Lepper's overjustification hy-pothesis (1981) and Deci's motivational theory (Deci & Ryan, 1985), are reviewed and the criticism isthen redefined behaviorally. "Intrinsically controlled" behavior is defined as behavior maintained byresponse-produced reinforcers, and the question concerning extrinsic consequences is thus restated asfollows: When behavior is maintained by response-produced stimuli, does extrinsic reinforcement decreasethe reinforcing value of those stimuli? The empirical support for this detrimental effect is summarizedbriefly, and several possible explanations for the phenomenon are offered. Research results that reflecton the effect's generality and social significance are discussed next, with the conclusion that the effect istransient and not likely to occur at all if extrinsic rewards are reinforcing, noncompetitive, based onreasonable performance standards, and delivered repetitively.Key words: extrinsic reinforcement, intrinsic motivation

In applied settings operant condition-ing procedures often include the delib-erate manipulation of behavioral conse-quences such as money, praise, tokens,edibles, and access to preferred activities.For over 10 years such practices havebeen criticized on the grounds that al-though they may increase the frequencyof behavior, they may also decrease anindividual's "intrinsic motivation" toperform the behavior (Condry, 1977;Deci, 1971; Lepper & Greene, 1978).Thus, many have argued that behaviorchange programs involving the manip-ulation of consequences should be re-placed with more "intrinsically motivat-ing" systems in all applied settings, butparticularly in schools, businesses, andclinics (Condry, 1977; Deci, 1975a; Deci& Ryan, 1985; Kohn, 1988; Levine &Fasnacht, 1974).The criticism that contingent conse-

quences may decrease intrinsic motiva-tion is based on the belief that a sharpdistinction can be made between behav-ior maintained by obvious environmen-

Portions of this paper were presented at the an-nual meetings ofthe Association for Behavior Anal-ysis, May 1986, Milwaukee, Wisconsin, and theNorthern California Association for BehaviorAnalysis, March 1987, San Mateo, California. Iwould like to express my appreciation to WayneFuqua, Jack Michael, and William Redmon for theirhelpful comments on earlier versions of this paper.

tal consequences-extrinsically motivat-ed behavior-and behavior that occurs inthe seeming absence of consequences-intrinsically motivated behavior. Accord-ing to this position, the impetus for ac-tion is primarily internal or intrinsic orprimarily external or extrinsic (Pittman& Heller, 1987). Because intrinsicallymotivated behavior is believed to be morecreative, spontaneous, and flexible thanextrinsically motivated behavior, it isfeared that by damaging intrinsic moti-vation, the use of extrinsic consequenceswill result in a decrease in highly valuedhuman behavior (Amabile, 1982, 1985;Deci & Ryan, 1985, p. 35; McGraw &McCullers, 1979; Shapira, 1976).

INTRINSICALLY VERSUSEXTRINSICALLY CONTROLLED

BEHAVIORIntrinsic motivation was originally

proposed to account for exploratory andmanipulative behavior in nonhuman an-imals (Harlow, 1950; Harlow, Harlow, &Meyer, 1950) and was later used to ex-plain human behavior believed to becaused by such motives as a need forachievement (McClelland, Atkinson,Clark, & Lowell, 1953), a need to be ef-fective when dealing with the environ-ment (White, 1959), a need to be self-determining (Deci, 1 975a), and a need tobe a causal agent when dealing with the

1

2 ALYCE M. DICKINSON

environment (deCharms, 1968). In allcases, behavior that could not be attrib-uted to external controls was attributedto intrinsic motivation. "'Intrinsic mo-tivation' was thus defined by default: per-formance in the absence of extrinsic re-wards" (Zimmerman, 1985, p. 118). Asillustrated in Franken's recent text onmotivation (1988), behavior is still op-erationally classified as extrinsically orintrinsically motivated depending uponwhether or not external controlling vari-ables can be readily identified:When the activity itselfprovides the reward, we saythat the activity is intrinsically rewarding; when anactivity is done in order to obtain a reward that isunrelated to it, then we say that the activity is ex-trinsically rewarding. A child who does well in schoolsimply to gain approval from his parents or to ac-quire a skill that he can then use to earn money oracquire fame is engaging in the activity for extrinsicreasons (rewards). If, however, he finds the activitymotivating [italics added] even in the absence ofapproval or some other form of gain, then we saythat he is engaging in the activity for intrinsic rea-sons (rewards). (p. 31)

The teleological "in order to," the pejor-ative "simply," and the equation of "re-wards" with "reasons" are all typical ofthe traditional approach to this distinc-tion.Horcones (1987) has formalized a sim-

ilar, but not identical, distinction be-tween intrinsic and extrinsic conse-quences from a more behavioralperspective:Intrinsic consequences are the natural and auto-matic results of responding (see Horcones, 1983;Vaughan and Michael, 1982). They are more or lessinevitably produced by the structural characteris-tics of the physical environment and the biologicalorganism; they are not programmed by others tooccur. In contrast, extrinsic consequences occur inaddition to any intrinsic consequences. They maybe programmed by our social environments, by ap-plied behavior analysts, by researchers, and byteachers and others, but they do not occur solely asa natural consequence ofresponding. (pp. 291-292)

From this perspective, then, intrinsicallycontrolled behavior is simply behaviormaintained by consequences that are thenatural and automatic results ofrespond-ing. Extrinsically controlled behavior isbehavior controlled by stimuli externalto the task. Furthermore, there is no ob-vious reason to view behavior controlled

by intrinsic consequences as functionallydifferent from behavior controlled bymore obvious forms of consequences, aposition contrary to the traditional dis-tinction.Another important difference exists

between the traditional and the behav-ioral accounts of intrinsically controlledbehavior. Intrinsic motivation has gen-erally been described as being innaterather than learned. From the presentperspective this would mean that theconsequences associated with intrinsicneeds-signs ofself-determination, com-petence, and control over the environ-ment-function as unconditioned rein-forcers. The behavioral interpretation ofintrinsically controlled behavior makesno such assumption. Intrinsic reinfor-cers, like extrinsic reinforcers, may beunconditioned, conditioned, or general-ized conditioned reinforcers. Skinner(1953), for example, suggested that be-haviors that occur in the absence of ob-vious rewards may be maintained bycontrol over the environment, and thatsuch control may function as either gen-eralized conditioned or unconditionedreinforcement.One kind of generalized reinforcer is created be-

cause many primary reinforcers are received onlyafter the physical environment has been efficientlymanipulated. One form ofprecurrent behavior mayprecede different kinds ofreinforcers upon differentoccasions. The immediate stimulation from suchbehavior will thus become a generalized reinforcer.We are automatically reinforced, apart from anyparticular deprivation, when we successfully con-trol the physical world. This may explain our ten-dency to engage in skilled crafts, in artistic creation,and in such sports as bowling, billiards, and tennis.

It is possible, however, that some of the rein-forcing effect of "sensory feed-back" is uncondi-tioned. A baby appears to be reinforced by stimu-lation from the environment which has not beenfollowed by primary reinforcement. The baby's rat-tle is an example. The capacity to be reinforced inthis way could have arisen in the evolutionary pro-cess, and it may have a parallel in the reinforcementwe receive from simply "making the world be-have." Any organism which is reinforced by itssuccess in manipulating nature, regardless of themomentary consequences, will be in a favored po-sition when important consequences follow. (pp.77-78)

Intrinsic reinforcers may also representa simple form of conditioned reinforce-ment, in which the stimuli associated with

EFFECTS OF REINFORCEMENT 3

the task have been correlated with ap-proval, praise, or some other form of re-inforcement.

In some cases behavior identified asintrinsically controlled may in fact becontrolled by infrequent extrinsic re-wards or by a combination of intrinsicand extrinsic rewards. For example, achild may play with tinker toys partlybecause such play results in control overthe environment, partly because it resultsin objects that resemble other objects thatare sources of reinforcement (e.g., a toycar), and partly because parents provideinfrequent intermittent praise and atten-tion. The more complex the behavior,the more difficult it becomes to deter-mine the source of control and to com-pletely eliminate the possibility of exter-nal control. Many examples exist ofimportant complex human behaviors thatoccur in the absence of obvious extrinsicrewards: the unappreciated artist or mu-sician, the scientist pursuing a line of re-search despite strong discouragementfrom colleagues, the business executivewho adopts a costly practice simply be-cause he or she "feels that it is the rightthing to do," heros and heroines who at-tempt to save someone's life and lose theirown. Such individuals have been im-mersed in a culture that provides manysources of indirect support for helpingothers and for contributing to the sur-vival of the group or the culture, and toignore this kind ofextrinsic control seemsunreasonable.As suggested above, determining the

actual provenance of any behavior thatoccurs in the absence ofobvious extrinsicconsequences requires considerable ex-perimental analysis. The results of suchanalyses, however, need not be incom-patible with a behavioral orientation.

EXPERIMENTALINVESTIGATIONS

The detrimental effects of extrinsic re-wards on behavior controlled by intrinsicrewards were first demonstrated experi-mentally by Deci and by Lepper and hiscolleagues (Deci, 1971, 1972a, 1972b;Greene & Lepper, 1974; Lepper &

Greene, 1975; Lepper, Greene, & Nis-bett, 1973), and subsequently by manyothers (e.g., Anderson, Manoogian, &Reznick, 1976; Calder & Staw, 1975;Harackiewicz, 1979; Kamiol & Ross,1977; Loveland & Olley, 1979; Pinder,1976; Pritchard, Campbell, & Campbell,1977; Ross, 1975; Weiner, 1980). In thetraditional research paradigm, activitiesassumed to be intrinsically motivating-assembling puzzles, solving anagrams,drawing with magic markers-are pro-vided to subjects. Experimental subjectsare promised a reward for performing theactivity and the reward is provided at theend of a single 30- or 60-min session.Control subjects perform the activitywithout being promised or ultimately re-ceiving any reward. Both groups are thenobserved during a nonreward period thatoccurs immediately following the exper-imental session or several weeks later. Ifexperimental subjects perform the taskless than control subjects, the extrinsicrewards are said to have decreased in-trinsic motivation.

Behavioral researchers, using a mul-tiple-trial within-subject research para-digm, began investigating the detrimen-tal effects ofextrinsic rewards in responseto the research cited above. Their resultsappeared to conflict with those reportedby cognitively oriented researchers (Da-vidson & Bucher, 1978; Feingold & Ma-honey, 1975; Mawhinney, Dickinson, &Taylor, in press; Vasta, Andrews, Mc-Laughlin, Stirpe, & Comfort, 1978; Vasta& Stirpe, 1979). In the behavioral para-digm, subjects are typically provided withtwo or three activities during baseline.The activity that is performed most isthen extrinsically reinforced for severalsessions. Extrinsic reinforcement is ter-minated during the final phase and per-formance during this post-reward phaseis compared to performance during thepre-reward baseline phase. Differences areattributed to the extrinsic reinforcement.As with the between-subject compari-sons, the effects of extrinsic rewards areassessed after the extrinsic rewards havebeen terminated. One exception to thispost-reward assessment is a proceduredeveloped by Mawhinney et al. (in press)

4 ALYCE M. DICKINSON

in which concurrent schedules were usedto assess the relative control of extrinsicand intrinsic rewards during administra-tion.Having demonstrated that extrinsic re-

wards do not always decrease intrinsi-cally controlled behaviors, behavioral re-searchers did not generally pursue whytheir results differed from those reportedby cognitively oriented researchers, nordid they investigate conditions underwhich decreases might occur. As othershave noted (Lepper, 1981; Mawhinney,1979; Scott, 1975, 1980), some of thedifferences in results can be attributed todifferent research strategies. However,careful examination ofthe research leadsto the conclusion that under some con-ditions extrinsic rewards do decreasesubsequent performance when the re-wards are no longer available, althoughthose conditions remain to be fully spec-ified.There are now well over 100 studies

that have examined the detrimental ef-fects of extrinsic rewards. Most of thestudies, as with much of the traditionalresearch in personality and social psy-chology, are theory-driven and as a resulttend to be highly contrived. Experimen-tal conditions are designed to test a spe-cific hypothesis or combination of hy-potheses, and conclusions are based oncomplex statistical comparisons betweensubtly different conditions. Further, ar-guments in support of favored hypothe-ses are often based on small butstatistically significant differences (Ma-whinney, 1979). Due to the large numberof studies, their diversity, and their com-plexity, this research will not be system-atically reviewed. Rather, nonbehavioraland behavioral explanations of post-re-ward decrements will be presented and,following those, a number of generalpoints about the research findings will bemade. These points identify some char-acteristic patterns in the research that willhelp readers respond to claims that ex-trinsic rewards have harmful effects andbetter enable the critical review of rele-vant experimental studies. Readers whoare interested in detailed reviews of theliterature are referred to Deci and Ryan

(1985), Lepper and Greene (1978), Mor-gan (1984), and Zimmerman (1985).

COGNITIVE EXPLANATIONS OFDETRIMENTAL EFFECTS

Claims that extrinsic rewards mayweaken intrinsic motivation were origi-nally derived from self-perception andattribution theories. According to thesetheories a person's perceptions about thecauses of ongoing behavior strongly in-fluence future motivation and perfor-mance. In the absence of obvious exter-nal controls, a person will attribute hisor her behavior to intrinsic interest ormotivation and will, therefore, continueto engage in the behavior when extrinsiccontrols are not present. But if salientextrinsic controls are present, behaviorwill be attributed to those controls and,as a result, will not readily occur in theirabsence in the future. Based on these te-nets, Lepper et al. (1973) proposed thefollowing:Self-perception theory has a number of heuristicimplications, one of the most intriguing of whichcould be termed the "overjustification" hypothe-sis-the proposition that a person's intrinsic inter-est in an activity may be undermined by inducinghim to engage in that activity as an explicit meansto some extrinsic goal. If the external justificationprovided to induce a person to engage in an activityis unnecessarily high and psychologically "over-sufficient," the person might come to infer that hisactions were basically motivated by the externalcontingencies of the situation, rather than by anyintrinsic interest in the activity itself. In short, aperson induced to undertake an inherently desirableactivity as a means to some ulterior end shouldcease to see the activity as an end in itself. (p. 130)

Stated more behaviorally (without, forthe moment, considering the validity ofthe assertion), when behavior that waspreviously controlled by intrinsic re-wards comes to be controlled by extrinsicrewards and the behaver reacts to thecontrolling relation itself as a stimulus(perceives it), the reinforcing value oftheintrinsic consequences is weakened.Therefore, in the future when extrinsicconsequences are no longer available, thebehavior occurs at a lower frequency.Lepper (1981) has explained that ex-

trinsic rewards and controls will decreaseintrinsic motivation only when the afore-mentioned perceptual shifts occur and has

EFFECTS OF REINFORCEMENT 5

criticized extreme advocates for extrap-olating "far beyond the available evi-dence to argue, in general, against the useofsystematic reward programs to modifybehavior" (p. 170). According to Lepper,perceptual shifts are likely when there issufficient initial interest in the activity toallow perceptual shifts ofcausality, whenthe relationship between performance andthe extrinsic rewards is salient, and whenrewards do not increase perceived com-petence.

Deci (1975a) also originally main-tained that perceptions of external con-trol decreased intrinsic motivation.However, he subsequently revised histheory, believing that this strictly cog-nitive explanation did not adequately ad-dress the underlying motivational pro-cesses (Deci & Ryan, 1985). Accordingto the revision, there are three distinctmotivational processes: intrinsic, extrin-sic, and amotivational. Extrinsic rewardsmay, by reducing self-determination and/or perceived competence, initiate eitherextrinsic motivational or amotivationalprocesses and permanently "co-opt" in-trinsic motivation. Perceptions of exter-nal control accompany, but do not cause,decreases in intrinsic motivation.The explanations offered by Deci and

Lepper maintain that once intrinsic mo-tivation or interest is lost, it may neverbe regained. This assumes that the cog-nitive and motivational processes re-sponsible for the weakening are irrevers-ible. Extrinsic rewards may causeperceptions of extrinsic control and ini-tiate extrinsic motivational or amotiva-tional processes, but performing in theirabsence following reward terminationapparently does not cause perceptions ofintrinsic control nor does it initiate in-trinsic motivational processes. Thisasymmetry has generally been ignored byboth theoreticians and researchers, al-though it seems to be an essential featureofthe explanations suggested by Deci andLepper.

BEHAVIORAL EXPLANATIONSOF DETRIMENTAL EFFECTSBecause operant psychologists main-

tain that, in general, behavior returns to

baseline levels following reinforcementtermination, Deci and Ryan (1985) haveclaimed that post-reward performancechanges challenge "several importantfoundations of operant psychology" (p.182). However, there are a number ofpossible behavioral reasons why behav-ior maintained by intrinsic reinforcersmight deviate from baseline levels fol-lowing the termination of extrinsic re-wards.

First, decrements in task performancemay result from a reduction in the rein-forcing effectiveness of the intrinsic con-sequences due to simple repetition. Ex-trinsic reinforcement of behaviormaintained by sensory reinforcers in-creases the frequency ofbehavior and thusalso increases exposure to the sensoryreinforcers. Repeated exposure to sen-sory reinforcers has long been known tomomentarily weaken their reinforcing ef-fectiveness, although the physiologicalmechanisms responsible for this processare unknown (Antonitis & Barnes, 1961;Berlyne, 1955; Kish, 1966; Kish & Bar-on, 1962; Montgomery, 1952; Welker,1956). Performance controlled by intrin-sic sensory reinforcers would, thus, beexpected to temporarily drop below base-line levels when extrinsic reinforcementwas discontinued.

Second, the performance ofan activitymay be subsequently affected if the con-trol procedures are aversive. Althoughrewards are generally positive events,there are at least three reward situationsthat might involve aversive stimulation.Performance standards are-often intro-duced during reward, and rewards aremade contingent upon meeting thosestandards. Ifindividuals fail to meet thosestandards, such failure could, as condi-tioned punishment, exert relatively morecontrol than the intrinsic consequences,temporarily decreasing the subsequentfrequency ofthe behavior, and could alsoweaken the reinforcing value of the in-trinsic consequences with which it waspaired. This analysis is supported by theresults of studies that have examined theeffects of failure on subsequent perfor-mance (Karniol & Ross, 1977; Rosen-

6 ALYCE M. DICKINSON

field, Folger, & Adelman, 1980; Salancik,1975; Zimmerman, 1985).Another way that rewards could be re-

lated to a form ofaversiveness is throughcoercion. In our society, rewards are oftenused to induce people (especially chil-dren) to engage in nonpreferred activitiesand, in addition, are sometimes com-bined with threats of punishment fornoncompliance. Skinner (1953) haspointed out that such control techniquesmay be aversive: "Techniques based uponthe use of force, particularly punishmentor the threat ofpunishment, are aversiveby definition, and techniques which ap-peal to other processes are also objec-tionable when, as is usually the case, theultimate advantage to the controller isopposed to the interest ofthe controllee"(p. 321). If individuals are offered re-wards primarily for engaging in nonpre-ferred activities, and/or concomitantlythreatened with punishment for non-compliance, offers of rewards may be-come conditioned aversive stimuli. Suchoffers may increase the frequency of thebehavior but, due to pairing, decrease thereinforcing value of the intrinsic conse-quences, resulting in post-reward decre-ments. Others have noted the relevanceof coercion but have approached it froma more cognitive perspective (Deci &Ryan, 1985; Lepper, 1981; Williams,1980).Performance may also decrease follow-

ing coercive control procedures becauseof what is commonly referred to as de-liberate noncompliance, refusing to en-gage in the task "in order to get even"with the coercer. If the coerced individ-ual is angry, signs of discomfort on thepart ofthe coercer will be reinforcing andany behavior that has in the past pro-duced such discomfort will be strong.During reward administration, the re-wards may maintain behavior; however,when they are no longer available be-havior that has irritated or inconveni-enced coercers may be momentarilystrong.A third possible reason for perfor-

mance decrements is based on subtle cul-tural reinforcing practices. In our society,

people are often admired and praisedwhen they engage in intrinsically con-trolled behavior, especially if it is viewedas artistic, original, creative, or intellec-tual. The very fact that the behavior isunrelated to any obvious extrinsic con-sequence is, in part, the basis for the ap-proval. Explicit promises of reward andpayment, on the other hand, are typicallyprovided for behaviors that are not con-sidered to be interesting or valuable inand of themselves. Promises of rewardand payment may, therefore, become dif-ferentially correlated with the absence ofpraise for creativity, originality, and soon, and, for those who are strongly rein-forced by such praise, counteract someof the reinforcing effectiveness of the in-trinsic consequences.

Finally, Mawhinney (1979) has sug-gested that post-extrinsic-reward perfor-mance decrements may be explained interms of optimal duration theory (Dun-ham, 1977). This theory is too complexto be adequately explained in the.presentpaper but, roughly speaking, proposes thatvarious behaviors, when not artificiallyconstrained, have an optimal duration ofoccurrence. Reinforcement may increasethe duration above that optimal valueand thus the decline following reinforce-ment may represent a momentary ad-justment for the excessive time spent onthe activity.The effects described previously need

not be strong in order to produce the post-reward changes reported in the literature.Experimental studies have been designedin a way that enables the detection ofsmall changes. During post-reinforce-ment phases, subjects have typically beenprovided with several alternative tasks,including the previously reinforced task.Because it is unlikely that subjects willrefuse to perform any of the tasks, thepreviously rewarded task need only exertrelatively less control over behavior thanthe others.Not all of the post-reward decrements

reported in the literature can be ex-plained by one ofthe preceding analyses.However, they illustrate that post-rewarddecrements can be explained behavior-

EFFECTS OF REINFORCEMENT 7

ally, and, in addition, suggest that somesituations might well result in such dec-rements.

GENERAL CHARACTERISTICSOF THE RESEARCH ONDETRIMENTAL EFFECTS

Five aspects of the research findingswill be discussed in the following section.They do not represent an exhaustive listofthe issues and controversies generatedby this body of research, but they areissues that should be considered whenevaluating the generality and significanceof post-reward decrements.

TransienceThe main concern of cognitive psy-

chologists and practitioners is that ex-trinsic rewards may permanently destroyintrinsic motivation (Deci, 1978; Deci &Ryan, 1985; Lepper & Greene, 1978). Itis feared that a child, having lost intrinsicmotivation for math or art, for example,may never regain it. But if post-rewarddecrements were transient, they wouldlose much of their theoretical and prac-tical significance.

Cognitive researchers have consistent-ly reported that performance decrementspersist as long as two to four weeks afterrewards have been terminated (Greene &Lepper, 1974; Harackiewicz, 1979; Lep-per & Greene, 1975; Lepper et al., 1973;Morgan, 1983; Ross, 1975), while be-havioral researchers have reported thatwhen decrements occur, which is rare,they are transient and disappear withinone to two sessions (Davidson & Bucher,1978; Deal & Madsen, 1980; Dickinson,1985; Feingold & Mahoney, 1975; Hom& Maxwell, 1980; Vasta & Stirpe, 1979).The differences in these results are due,at least in part, to the ways in which "per-sistence" has been experimentally inves-tigated and thus defined by these twogroups of researchers.

In the between-group research para-digm adopted by cognitive researchers,subjects perform a task during a single30- or 60-min session and are promisedand given a reward for doing so. Two orfour weeks later the previously rewarded

task and several new tasks are madeavailable to subjects in a nonreward ses-sion. During the intervening period sub-jects do not have access to the previouslyrewarded task. The amount of time sub-jects spend performing the previously re-warded task during the nonreward ses-sion is then compared with the amountof time nonrewarded control subjectsspend performing the task. With thismethod, researchers have found that sub-jects who were previously rewarded spendless time performing the task than nonre-warded subjects. Persistence, then, isbeing defined as the occurrence of a dec-rement following a passage oftime duringwhich subjects have not had any contactwith the task.

Behavioral researchers, on the otherhand, have examined post-reward taskperformance given repeated exposure tothe task. In these within-subject researchdesigns, a subject is usually given severaltasks to perform during an initial baselineperiod that lasts five to ten sessions. Thetask performed the most is then extrin-sically reinforced for several sessions. Af-ter reinforcement is terminated, task per-formance is observed for another five toten sessions. Performance during the post-reward phase is compared with perfor-mance during the pre-reward baselinephase. Behavioral researchers, then, havetaken repeated measures of task perfor-mance following reward termination and,under these conditions, decrements havedisappeared within one or two sessions.These two assessments of persistence

are very different: One represents persis-tence over time without repeated expo-sure to the task; the other represents per-sistence with repeated exposure to thetask. The quick recovery observed whenindividuals continue to perform the tasksuggests that the reinforcing value of in-trinsic consequences is not permanentlyaltered or, in less behavioral terms, that"intrinsic motivation" is not perma-nently damaged.Type ofReward Contingency

In research studies rewards have beenprovided for simply participating in the

8 ALYCE M. DICKINSON

experiment, for performing the experi-mental task, for meeting specified per-formance standards and, by informingsubjects that their performance met orexceeded bogus standards, for success.Researchers have labeled these rewardsystems differently, provided differentinstructions to subjects with respect tothe performance-reward relationship, andgiven the rewards to subjects at varyingtimes (e.g., before the experimental ses-sion, after the experimental session butbefore the nonreward period, or after thenonreward period). These differencesmake it difficult to compare the relativeeffects of the various reward procedures.Nonetheless, when terminological incon-sistencies are resolved and proceduraldetails ignored, some important trendsemerge.

First, extrinsic rewards are most likelyto decrease post-reward performancewhen they are provided for engaging inthe activity; that is, when subjects arepromised and given rewards for perform-ing the activity irrespective of quality orquantity. This type of reward contingen-cy is referred to as a task-contingent pro-cedure in contrast to participation-con-tingent procedures in which subjectsreceive rewards for attendance and par-ticipation, performance-contingent pro-cedures in which rewards are based onperformance standards, and success-con-tingent procedures in which subjects aretold they have received the rewards be-cause of good performance. Post-rewarddecrements have been reported in themajority of studies that have examinedtask-contingent rewards (e.g., Andersonet al., 1976; Calder & Staw, 1975; Danner& Lonky, 1981; Greene& Lepper, 1974;Lepper & Greene, 1975; Lepper et al.,1973; McGraw&Fiala, 1982; Ross, 1975;Ross, Karniol, & Rothstein, 1976). Incontrast, decrements have not been re-ported in the majority ofstudies that haveexamined participation-contingent re-wards (e.g., Deci, 1972a; Pinder, 1976;Swann & Pittman, 1977) or perfor-mance-contingent rewards (e.g., Farr,1976; Farr, Vance, & McIntyre, 1977;Feingold & Mahoney, 1975; Reiss &

Sushinsky, 1975; Vasta & Stirpe, 1979).Further, when compared within the sameexperiment, task-contingent rewards havesignificantly decreased post-reward per-formance relative to performance-con-tingent rewards (Boggiano & Ruble, 1979;Enzle & Ross, 1978; Luyten & Lens,1981) and relative to success-contingentrewards (Boggiano & Ruble, 1979; Kar-niol& Ross, 1977; Rosenfield et al., 1980;Zimmerman, 1985). An exception is thestudy by Ryan, Mims, and Koestner(1983) in which post-reward respondingwas not differentially affected by task- andperformance-contingent rewards. In spiteof the fact that the detrimental effects oftask-contingent rewards appear well-documented, Bandura (1987) has arguedthat they are "of no great social importbecause rewards are rarely showered onpeople regardless of how they behave"(p. 246). Of greater social significancewould be post-reward decrements due torewards based on standards of perfor-mance, the type of rewards considerednext.When consequences depend upon

meeting performance standards, a com-plication arises from the fact that somesubjects may not meet the standards andthus may not receive the rewards. Theproper analysis of their data is problem-atic. In many studies, this problem is ig-nored and the data for subjects exposedto performance standards is combined,irrespective of performance, and com-pared with the performance of subjectswho were not provided with performancestandards. This procedure may accountfor the conflicting results of studies thathave examined performance-contingentrewards (Deci, 1971, 1972a, 1972b;Enzle & Ross, 1978; Farr, 1976; Farr etal., 1977; Harackiewicz, 1979; Pinder,1976; Porac & Meindl, 1982; Weiner,1980). The problem has been avoided insome studies by informing subjects ofthestandards and then manipulating theirperformance so that it met or exceededthe standards, or by informing subjectsat the time of reward, but following theirperformance, that they had performedabove the norm. When these types ofbo-

EFFECTS OF REINFORCEMENT 9

gus performance standards are studied,the reward procedure is referred to as suc-cess-contingent.

Success-contingent rewards have sus-tained or increased post-reward perfor-mance (Anderson et al., 1976; Arkes,1979; Boggiano & Ruble, 1979; Kamiol& Ross, 1977; Martin, 1977; Rosenfieldet al., 1980; Weiner & Mander, 1978;Zimmerman, 1985). This effect is robustand consistent, although often ignored indiatribes against the use of performance-contingent rewards. Results of threestudies, further, suggest that success-based rewards are preferable to task-basedrewards for high performers, results thatshould be of considerable interest tobusiness executives and educators. Inthese studies, the post-reward perfor-mance of high success subjects increasedfollowing success-contingent rewards butdecreased following task-contingent re-wards (Karniol & Ross, 1977; Rosenfieldet al., 1980; Zimmerman, 1985).Even strong opponents of contingent

rewards recognize that success-based re-wards do not have harmful effects. Forexample, when arguing against perfor-mance-contingent rewards in work set-tings, Deci and Ryan (1985) repeatedlyrefer to the beneficial effects of success-based or, in their terminology, "infor-mational" rewards:The important point is that rewards, like feedback,when used to convey to people a sense of appre-ciation for work well done, will tend to be experi-enced informationally and will maintain or enhanceintrinsic motivation, but when they are used to mo-tivate people, they will surely be experienced con-trollingly and will undermine intrinsic motivation.(p. 300)

Whenever rewards are used to motivate people-in other words, to control them-it is probable thatthey will have a negative effect on the people's in-trinsic motivation.... However, rewards that areappropriately linked to performance, representingpositive feedback in an informational context, oughtnot to be detrimental. The cost to the system, how-ever, in signifying good performance through theuse ofperformance-contingent rewards is that manypeople end up receiving the message that they arenot doing very well and this is likely to be amoti-vating. (p. 310)

As suggested by Deci and Ryan (1985),the effects ofperformance-contingent re-

wards certainly depend upon how wellpeople perform in relation to specifiedperformance standards, that is, whetherrewards provide positive or negativefeedback. Studies have documented thatrewards provided for success and successirrespective of reward increase post-re-ward performance, but failure to performwell enough to receive rewards and taskfailure in general decrease post-rewardperformance (Kamiol & Ross, 1977; Ro-senfield et al., 1980; Salancik, 1975;Zimmerman, 1985). If, under perfor-mance-contingent reward systems,"many people end up receiving the mes-sage that they are not doing very well"and, as a result, lose interest in the task,the fault lies not with the contingent re-wards, but with the performance stan-dards upon which the rewards are based.Considering the detrimental effects oftask-based rewards and the enhancing ef-fects of success-based rewards, the use ofperformance-contingent rewards shouldnot be discouraged, but rather the de-velopment of objective, attainable per-formance standards upon which rewardsare based should be encouraged. Appliedbehavior analysts have, of course, tra-ditionally advocated the development ofsuch peerformance standards.

Rewards Versus Reward ProceduresIn some studies decrements attributed

to rewards, whether task-contingent orperformance-contingent, may have beendue to other features of the reward pro-cedures. A number of factors have beenshown to decrease the subsequent per-formance of intrinsically rewarding ac-tivities irrespective of reward: failure tomeet specified performance standards(Boggiano & Ruble, 1979; Karniol &Ross, 1977; Salancik, 1975); instructingsubjects that they "should" perform thetask (Ryan, 1982; Ryan et al., 1983; Wil-liams, 1980); close monitoring of perfor-mance (Lepper& Greene, 1975; Pittman,Davey, Alafat, Wetherill, & Kramer,1980); the experimenter's ignoring per-formance (Anderson et al., 1976); eval-uation of performance (Amabile, 1979;

10 ALYCE M. DICKINSON

Smith, 1975); temporal deadlines (Am-abile, DeJong, & Lepper, 1976); andcompetition (Deci, Betley, Kahle,Abrams, & Porac, 1981). Reward pro-cedures used in this type ofresearch haveoften contained one or more ofthese fea-tures, which could have been responsiblefor or at least contributed to reportedpost-reward decrements. For example, inone frequently cited study competitionand failure to receive the reward wereconfounded with financial rewards, yetsubsequent performance decreases wereattributed to the rewards (Pritchard et al.,1977). Seventeen subjects in the mone-tary reward condition were divided intothree groups and informed that the topperformer of each group would receive$5.00. Thus, in this competitive situationonly three subjects actually received a re-ward. In spite of that, the performancesofall 17 subjects were combined and des-ignated as the financial incentive groupdata. When the performance of thesesubjects was compared with the perfor-mance of nonrewarded control subjects,decrements were found. It is entirely pos-sible that the decrements were not dueto the "financial incentives" per se, whichmost subjects did not receive, but tocompetition and/or failing to performwell enough to receive the incentives.As illustrated by the preceding study,

when the results ofstudies suggesting thatextrinsic rewards have harmful effects arereviewed, the reward procedures shouldbe carefully scrutinized to determinewhether the reported decrements mayhave been due to factors other than therewards. While some researchers havemade efforts to control for these factors,those efforts have not always been suc-cessful.

Reward Versus ReinforcementAlthough nonbehavioral researchers

usually refer to their critical manipula-tion as a reward procedure, they use theterm reinforcement frequently enough tomake it clear that they consider theseterms to be synonymous. In behavioralpsychology, however, the term reinforcerrefers only to a stimulus change that in-

creases the frequency of the behavior itfollows. In addition, reinforcement pro-cedures typically involve the repeatedpresentation of the consequent stimuluscontingent upon the relevant behavior.Reward procedures adopted by manynonbehavioral researchers have differedfrom a typical reinforcement procedurein two ways: instructions and promisesofrewards have usually been substitutedfor repeated contingent delivery, and theevents used as rewards have not beenshown to increase the frequency of taskbehavior.

Instructions and promises versus re-peated contingent delivery. Telling peoplewhat to do and offering them a rewardfor doing it will often result in the oc-currence of the relevant behavior, but itmay not show all ofthe dynamic featuresof behavior maintained by reinforce-ment. Similarly, the aftereffects ofprom-ises and ofactual repeated contingent de-livery of reinforcement may be quitedifferent. Skinner (1969) discusses thisissue in detail, referring to such prom-ised-reward procedures as the circum-vention of the independent variable:

The manipulation ofindependent variables appearsto be circumvented when, instead of exposing anorganism to a set ofcontingencies, the contingenciesare simply described in "instructions." Instead ofshaping a response, the subject is told to respondin a given way. A history of reinforcement or pun-ishment is replaced by a promise or threat....

Descriptions ofcontingencies are, ofcourse, ofteneffective.... Verbal communication is not, how-ever, a substitute for the arrangement and manip-ulation of variables.There is no reason why a description of contin-

gencies of reinforcement should have the same ef-fect as exposure to the contingencies. A subject canseldom accurately describe the way in which he hasactually been reinforced. Even when he has beentrained to identify a few simple contingencies, hecannot then describe a new contingency, particu-larly when it is complex. We can scarcely expecthim, therefore, to react appropriately to descrip-tions by the experimenter. Moreover, the verbalcontingencies between subject and experimentermust be taken into account. Instructions must insome way promise or threaten consequences notgermane to the experiment ifthe subject is to followthem. (p. 114-115)

Task performance evoked by instruc-tions and promises of reward can be in-fluenced by a number of factors, includ-

EFFECTS OF REINFORCEMENT 11

ing the sophistication of the subject'sverbal repertoire, the subject's historywith respect to whether promised re-wards were actually received, the natureof the subject's prior exposure to the ob-ject being offered as the reward, whetherthe particular wording of the request toperform the task has been correlated withpunishment for noncompliance, andevents that occur during the delay be-tween the promise and reward delivery,such as the way the experimenter inter-acts with the subject. As mentioned ear-lier, some of these factors have beendemonstrated experimentally to producepost-intervention decrements irrespec-tive of reward, which makes the attri-bution ofperformance decrements to ex-trinsic reinforcement quite controvertible.Most behavioral researchers have im-

plemented procedures that conform toconventional reinforcement practices,and thus some ofthe differences observedbetween behavioral and nonbehavioralresearch may be due to the differencesbetween reward and reinforcement pro-cedures. A few behavioral researchers,however, have also used instructions andpromised-reward procedures rather thanactual reinforcement but, unlike nonbe-havioral researchers, they have conduct-ed multiple reward sessions (Deal &Madsen, 1980; Feingold & Mahoney,1975). This difference is important fortwo reasons. First, during initial rewardsessions subjects come into contact withthe reward contingencies, so performancemay come under their control and be lessinfluenced by factors related to instruc-tions and promises. Second, subjects whoreceive delayed rewards have a favorablehistory at least within the context of theexperiment in which promised rewardswere actually delivered, eliminating onepossible confound due to differences insubject reinforcement histories.

Nonreinforcing versus reinforcing re-wards. Lepper (1981) has indicated thatnonbehavioral researchers have inten-tionally used rewards that were not likelyto increase task performance in order toeliminate factors such as boredom andsatiation as possible explanations forpost-reward decrements. Consistent with

this observation, many nonbehavioralresearchers have presented data indicat-ing that their rewards did not increasethe performance of rewarded subjects(Calder & Staw, 1975; Deci, 1975b;Greene, Sternberg, & Lepper, 1976;Kruglanski, Alan, & Lewis, 1972; Krug-lanski, Riter, Amitai, Margolin, Shabtai,& Zaksh, 1975; Lepper & Greene, 1976;Ross; 1975; Ross et al., 1976). When re-wards do not affect task performance, itis completely inappropriate to attributepost-reward decrements to reinforce-ment: rather, they should be attributedto nonreinforcing rewards.

Results from a study by Williams(1980) suggest that nonreinforcing re-wards may be more likely to result inpost-reward decrements than reinforcingrewards, emphasizing the importance ofthe distinction. The study consisted ofthree one-session phases: baseline, re-ward, and post-reward. The relative "at-tractiveness" of several rewards was as-sessed in a pre-baseline session. Duringthe reward phase, subjects in one groupwere promised attractive rewards for per-forming a task while subjects in anothergroup were promised unattractive re-wards. Because rewards were promised,and delivered only once at the end of thesession rather than repeatedly, the inter-pretation ofthe results is problematic, asdiscussed previously. However, thepromise of the attractive rewards did in-crease performance relative to the sub-jects' own baseline performance and rel-ative to the performance of subjects whowere promised unattractive rewards, sug-gesting that the attractive rewards wouldprobably have functioned as reinforcersunder a repetitive-delivery procedure.Results indicated that the post-rewardperformance of subjects who were prom-ised unattractive rewards decreased rela-tive to their baseline performance andrelative to the performance of subjectswho were promised attractive rewards.The post-reward performance ofsubjectswho were promised attractive rewards,however, did not decrease relative tobaseline performance. In a post-hoc anal-ysis, Williams examined the relationshipbetween the degree of reward attractive-

12 ALYCE M. DICKINSON

ness and post-reward responding. Sub-jects were divided into four categories de-pending upon the extent to which theirperformance increased during the rewardphase. Post-reward increases were ob-served for subjects whose performanceincreased the most during the rewardphase while post-reward decreases wereobserved for the other subjects. For thesubjects who displayed post-reward dec-rements, the size of the decrement wasinversely related to the size of the in-crease during reward: the greater thepost-reward decrement, the smaller theincrease during reward. These data sug-gest that highly reinforcing rewards mayresult in post-reward performance in-creases, while less reinforcing or non-reinforcing rewards may result in post-reward decreases.

Additional research and replication arecertainly warranted to determine the va-lidity of the differential effects of rein-forcing and nonreinforcing rewards.However, ifthe preceding results are val-id, they may explain why most behav-ioral researchers have failed to find post-reward decrements while nonbehavioralresearchers have consistently reportedsuch decrements: behavioral researchershave insured that their rewards have beenreinforcing, while most nonbehavioralresearchers have used nonreinforcing re-wards. Reinforcing rewards may be lesslikely to generate countercontrol, delib-erate noncompliance, and feelings ofbeing controlled than nonreinforcing re-wards, as suggested by Skinner (1974):The fact that positive reinforcement does not breedcountercontrol has not gone unnoticed by would-be controllers, who have simply shifted to positivemeans. Here is an example: A government mustraise money. If it does so through taxation, its cit-izens must pay or be punished, and they may escapefrom this aversive control by putting another partyin power at the next election. An an alternative, thegovernment organizes a lottery, and instead ofbeingforced to pay taxes, the citizen voluntarily buys tick-ets. The result is the same: the citizens give thegovernment money, but they feel free and do notprotest in the second case. Nevertheless they arebeing controlled. (p. 218; italics in original)

CONCLUSIONMuch of the controversy surrounding

the detrimental effects of extrinsic re-

wards is spurred by the philosophical as-sumption that some, though not all, ofan individual's behavior is self-initiated.Extrinsic rewards are believed to reduceintrinsic or internally initiated motiva-tion, thereby decreasing much highly val-ued human behavior. As long as the in-dividual is viewed as the initiator ofaction and the behavior thus induced isbelieved to be qualitatively superior tobehavior caused by the external environ-ment, the controversy surrounding ex-trinsic rewards will no doubt continue.From a behavioral perspective, all be-

havior is ultimately initiated by the ex-ternal environment. Because internal andexternal sources ofcontrol are not placedin opposition, decrements following ex-trinsic reward lose much of their philo-sophical importance, although they re-main empirically interesting.

Experimental investigations reveal thatextrinsic rewards may, under some con-ditions, decrease post-reward respond-ing. For example, rewards based on taskengagement, failure to meet performancestandards upon which rewards are based,and competitive reward systems may re-duce post-reward performance. On theother hand, such decrements are tran-sient if the individual continues to per-form the task following reward, and arenot likely at all if individuals meet orexceed specified performance standards,or if rewards increase the frequency ofbehavior and are delivered repetitively.The transience and restricted general-

ity noted in the research may explain whyreward-induced loss ofinterest- seems in-congruous with behavior outside of thelaboratory. Adults rarely object to beingpaid for engaging in interests, and oftenseek careers that enable their continuedpursuit. Further, such individuals ex-press considerable pleasure with the factthat they get paid for doing what theylike to do. Bandura (1987) has noted thisincongruity with respect to those whoclaim that extrinsic rewards decrease in-trinsic motivation: "Social commenta-tors who decry the use of extrinsic in-centives rarely foreswear such rewards forthemselves when it comes to salary in-creases, book royalties, and performance

EFFECTS OF REINFORCEMENT 13

fees, for fear the currency of the realmwill sap their interest. Valued rewards areaccepted as though innocuous to oneselfbut harmful to others." (p. 250)Because post-reward decrements are

transient and limited, broad assertionsthat extrinsic rewards adversely affect theperformance of intrinsically rewardingtasks are unwarranted, as are recommen-dations to abandon performance-contin-gent reward systems in applied settings.Nonetheless, the research reminds read-ers that some reward procedures can beaversive and provides some guidelinesfor the development of reward systems:e.g., rewards should be noncompetitive,reinforcing, and contingent upon perfor-mance standards rather than task en-gagement; and performance standardsshould be objective and attainable. Mostof the guidelines, if not all, are ones thathave been recommended by applied be-havior analysts for many years, irrespec-tive of whether the target behavior wasintrinsically or extrinsically controlled.Thus, to avoid the harmful effects of ex-trinsic rewards, behaviorists need only tocontinue past practices. In other words,given current behavioral practices, thecontroversy surrounding the detrimentaleffects of extrinsic rewards is "much adoabout nothing."

REFERENCESAmabile, T. M. (1979). Effects of external eval-

uations on artistic creativity. Journal ofPerson-ality and Social Psychology, 37, 221-233.

Amabile, T. M. (1982). Social psychology of cre-ativity: A consensual assessment technique.Journal ofPersonality and Social Psychology, 43,997-1013.

Amabile, T. M. (1985). Motivation and creativ-ity: Effects of motivational orientation on crea-tive writers. Journal of Personality and SocialPsychology, 48, 393-399.

Amabile, T. M., DeJong, W., & Lepper, M. R.(1976). Effects of externally imposed deadlineson subsequent intrinsic motivation. Journal ofPersonality and Social Psychology, 34, 92-98.

Anderson, R., Manoogian, S. T., & Reznick, J. S.(1976). The undermining and enhancing of in-trinsic motivation in preschool children. Journalof Personality and Social Psychology, 34, 915-922.

Antonitis, J. J., & Barnes, G. W. (1961). Groupoperant behavior: An extension of individualmethodology to a real life situation. Journal ofGenetic Psychology, 98, 95-111.

Arkes, H. R. (1979). Competence and the over-justification effect. Motivation and Emotion, 3,143-150.

Bandura, A. (1987). Socialfoundationsofthoughtand action. Englewood Cliffs, NJ: Prentice-Hall.

Berlyne, D. E. (1955). The arousal and satiationofperceptual curiosity in the rat. Journal ofCom-parative Physiological Psychology, 48, 238-246.

Boggiano, A. K., & Ruble, D. N. (1979). Com-petence and the overjustification effect: A devel-opmental study. Journal ofPersonality and So-cial Psychology, 37, 1462-1468.

Calder, B. J., & Staw, B. M. (1975). Self-percep-tion ofintrinsic and extrinsic motivation. Journalof Personality and Social Psychology, 31, 599-605.

Condry, J. (1977). Enemies of exploration: Self-initiated versus other-initiated learning. Journalof Personality and Social Psychology, 35, 459-477.

Danner, F. W., & Lonky, E. (1981). A cognitive-developmental approach to the effects ofrewardson intrinsic motivation. Child Development, 52,1043-1052.

Davidson, P., & Bucher, B. (1978). Intrinsic in-terest and extrinsic reward: The effects of a con-tinuing token program on continuing noncon-strained preference. Behavior Therapy, 9, 222-234.

Deal, R. M., & Madsen, C. H., Jr. (1980). Tokensystems and intrinsic motivation: An experimen-tal analysis. Unpublished manuscript, FloridaState University, Tallahassee, FL.

deCharms, R. (1968). Personal causation. NewYork: Academic Press.

Deci, E. L. (1971). Effects of externally mediatedrewards on intrinsic motivation. Journal ofPer-sonality and Social Psychology, 18, 105-115.

Deci, E. L. (1972a). Intrinsic motivation, extrinsicreinforcement, and inequity. Journal ofPerson-ality and Social Psychology, 22, 113-120.

Deci, E. L. (1972b). The effects ofcontingent andnoncontingent rewards and controls on intrinsicmotivation. Organizational Behavior and Hu-man Performance, 8, 217-229.

Deci, E. L. (1975a). Intrinsic motivation. NewYork: Plenum Press.

Deci, E. L. (1 975b). Notes on the theory and meta-theory of intrinsic motivation. OrganizationalBehaviorandHumanPerformance, 15, 130-145.

Deci, E. L. (1978). Applications of research onthe effects of rewards. In M. R. Lepper & D.Greene (Eds.), The hidden costs of reward: Newperspectives on the psychology ofmotivation (pp.193-203). Hillsdale, NJ: Lawrence Erlbaum.

Deci, E. L., Betley, G., Kahle, J., Abrams, L., &Porac, J. (1981). When trying to win: Compe-tition and intrinsic motivation. Personality andSocial Psychology Bulletin, 7, 79-83.

Deci, E. L., & Ryan, R. M. (1985). Intrinsic mo-tivation and self-determination in human behav-ior. New York: Plenum Press.

Dickinson, A. M. (1985). The effects of rewardand reinforcment on intrinsic interest. Disserta-tion Abstracts International 46, 3202B. (Univer-sity Microfilms No. DA8524457.)

14 ALYCE M. DICKINSON

Dunham, P. J. (1977). The nature of reinforcingstimuli. In W. K. Honig& J. E. R. Staddon (Eds.),Handbook ofoperant behavior (pp. 98-124). En-glewood Cliffs, NJ: Prentice-Hall.

Enzle, M. E., & Ross, J. M. (1978). Increasing anddecreasing intrinsic interest with contingent re-wards: A test of cognitive evaluation theory.Journal ofExperimental Social Psychology, 14,588-597.

Farr, J. L. (1976). Task characteristics, rewardcontingency, and intrinsic motivation. Organi-zational Behavior and Human Performance, 16,294-307.

Farr, J. L., Vance, R. J., & McIntyre, R. M. (1977).Further examinations ofthe relationship betweenreward contingency and intrinsic motivation. Or-ganizational Behavior andHuman Performance,20, 31-53.

Feingold, B. D., & Mahoney, M. J. (1975). Re-inforcement effects on intrinsic interest: Under-mining the overustification hypothesis. BehaviorTherapy, 6, 367-377.

Franken, R. E. (1988). Human motivation. PacificGrove, CA: Brooks-Cole.

Greene, D., & Lepper, M. R. (1974). Effects ofextrinsic rewards on children's subsequent in-trinsic interest. Child Development, 45, 1141-1145.

Greene, D., Stemnberg, B., & Lepper, M. R. (1976).Overjustification in a token economy. Journal ofPersonalityand SocialPsychology, 34, 1219-1234.

Harackiewicz, J. M. (1979). The effects ofrewardcontingency and performance feedback on in-trinsic motivation. Journal ofPersonality andSo-cial Psychology, 37, 1352-1363.

Harlow, H. F. (1950). Learning and satiation ofresponse in intrinsically motivated complex puz-zle performance by monkeys. Journal of Com-parative and Physiological Psychology, 1950, 289-294.

Harlow, H. F., Harlow, M. K., & Meyer, D. R.(1950). Learning motivated by a manipulativedrive. Journal of Experimental Psychology, 40,228-234.

Hom, H. L., Jr., & Maxwell, F. R. (1980, April).Individualpatterns in intrinsic motivation. Paperpresented at the annual meeting of the South-western Psychological Association, OklahomaCity, OK.

Horcones. (1983). Natural reinforcement in aWalden Two community. Revista Mexicana deAnalisis de la Conducta, 9, 141-143.

Horcones. (1987). The concept of consequencesin the analysis of behavior. The Behavior An-alyst, 10, 291-294.

Karniol, R., & Ross, M. (1977). The effects ofperformance-relevant and performance-irrele-vant rewards on children's intrinsic motivation.Child Development, 48, 482-487.

Kish, G. B. (1966). Studies of sensory reinforce-ment. In W. K. Honig (Ed.), Operant behavior:Areas ofresearch and application (pp. 109-159).New York: Appleton-Century-Crofts.

Kish, G. B., & Baron, A. (1962). Satiation of sen-sory reinforcement. Journal of ComparativePhysiological Psychology, 55, 1007-10100.

Kohn, A. (1988, January). Incentives can be badfor business. Inc., pp. 93-94.

Kruglanski, A. W., Alan, S., & Lewis, T. (1972).Retrospective misattribution and task enjoy-ment. Journal of Experimental Social Psychol-ogy, 8, 493-501.

Kruglanski, A. W., Riter, A., Amitai, A., Margolin,B., Shabtai, L., & Zaksh, D. (1975). Can moneyenhance intrinsic motivation? A test of the con-tent-consequence hypothesis. Journal ofPerson-ality and Social Psychology, 31, 744-750.

Lepper, M. R. (1981). Intrinsic and extrinsic mo-tivation in children: Detrimental effects of su-perfluous social controls. In W. E. Collins (Ed.),Aspects of the development of competence (pp.155-214). Hillsdale, NJ: Lawrence Erlbaum.

Lepper, M. R., & Greene, D. (1975). Turning playinto work: Effects of adult surveillance and ex-trinsic rewards on children's intrinsic motiva-tion. Journal ofPersonality and Social Psychol-ogy, 31, 479-486.

Lepper, M. R., & Greene, D. (1976). On under-standing "ovedjustification": A reply to Reiss andSushinsky. Journal ofPersonality andSocial Psy-chology, 33, 25-35.

Lepper, M. R., & Greene, D. (Eds.). (1978). Thehidden costs of reward: New perspectives on thepsychology ofhuman motivation. Hillsdale, NJ:Lawrence Erlbaum.

Lepper, M. R., Greene, D., &Nisbett, R. E. (1973).Undermining children's intrinsic interest withextrinsic reward: A test ofthe "overjustification"hypothesis. Journal of Personality and SocialPsychology, 28, 129-137.

Levine, F. M., & Fasnacht, G. (1974). Token re-wards may lead to token learning. American Psy-chologist, 29, 816-820.

Loveland, K. K., & Olley, J. G. (1979). The effectof external reward on interest and quality of taskperformance in children ofhigh and low intrinsicmotivation. Child Development, 50, 1207-1210.

Luyten, H., & Lens, W. (1981). The effect of ear-lier experience and reward contingencies on in-trinsic motivation. Motivation and Emotion, 5,25-36.

Martin, J. A. (1977). Effects of positive and neg-ative adult-child interactions on children's taskperformance and task preferences. Journal ofEx-perimental Child Psychology, 23, 493-502.

Mawhinney, T. C. (1979). Intrinsic X extrinsicwork motivation: Perspectives from behavior-ism. Organizational Behavior and Human Per-formance, 24, 411-440.

Mawhinney, T. C., Dickinson, A. M., & Taylor, L.A., III. (in press). The use of concurrent sched-ules to evaluate the effects of extrinsic rewardson "intrinsic motivation." Journal of Organi-zational Behavior Management.

McClelland, D. C., Atkinson, J. W., Clark, R. W.,&Lowell,E.L. (1953). Theachievementmotive.New York: Appleton-Century-Crofts.

McGraw, K. O., & Fiala, J. (1982). Underminingthe Zeigarnik effect: Another hidden cost of re-ward. Journal of Personality, 50, 58-66.

McGraw, K., & McCullers, J. (1979). Evidenceof a detrimental effect of extrinsic incentives on

EFFECTS OF REINFORCEMENT 15

breaking a mental set. Journal ofExperimentalSocial Psychology, 15, 285-294.

Montgomery, K. C. (1952). Exploratory behaviorand its relation to spontaneous alternation in aseries ofmaze exposures. Journal ofComparativePhysiological Psychology, 45, 50-57.

Morgan, M. (1983). Decrements in intrinsic in-terest among rewarded and observer subjects.Child Development, 54, 636-644.

Morgan, M. (1984). Reward-induced decrementsand increments in intrinsic motivation. ReviewofEducational Research, 54, 5-30.

Pinder, C. C. (1976). Additivity versus nonad-ditivity of intrinsic and extrinsic incentives: Im-plications for work motivation, performance, andattitudes. Journal ofAppliedPsychology, 61,693-700.

Pittman, T. S., Davey, M. E., Alafat, K. A., Weth-erill, K. V., & Kramer, N. A. (1980). Infor-mational versus controlling verbal rewards. Per-sonality and Social Psychology Bulletin, 6, 228-233.

Pittman, T. S., & Heller, J. F. (1987). Social mo-tivation. Annual Review ofPsychology, 38, 461-489.

Porac, J. F., & Meindl, J. (1982). Underminingovejustification: Inducing intrinsic and extrinsictask representations. Organizational Behavior andHuman Performance, 29, 208-226.

Pritchard, R. D., Campbell, K. M., & Campbell,D. J. (1977). Effects of extrinsic financial re-wards on intrinsic motivation. Journal ofAppliedPsychology, 62, 9-15.

Reiss, S., & Sushinsky, L. W. (1975). Overjusti-fication, competing responses, and the acquisi-tion of intrinsic interest. Journal of Personalityand Social Psychology, 31, 1116-1125.

Rosenfield, D., Folger, R., &Adelman, H. F. (1980).When rewards reflect competence: A qualifica-tion ofthe overdustification effect. Journal ofPer-sonality and Social Psychology, 39, 368-376.

Ross, M. (1975). Salience of reward and intrinsicmotivation. Journal of Personality and SocialPsychology, 32, 245-254.

Ross, M., Karniol, R., & Rothstein, M. (1976).Reward contingency and intrinsic motivation inchildren: A test of the delay of gratification hy-pothesis. Journal ofPersonality and Social Psy-chology, 33, 442-447.

Ryan, R. M. (1982). Control and information inthe intrapersonal sphere: An extension of cog-nitive evaluation theory. Journal of Personalityand Social Psychology, 43, 450-461.

Ryan, R. M., Mims, B., & Koestner, R. (1983).Relation ofreward contingency and interpersonalcontext to intrinsic motivation: A review and testusing cognitive evaluation theory. Journal ofPer-sonality and Social Psychology, 45, 736-750.

Salancik, G. R. (1975). Interaction effects of per-formance and money on self-perception of in-

trinsic motivation. Organizational Behavior andHuman Performance, 13, 339-351.

Scott, W. E., Jr. (1975). The effects of extrinsicrewards on "intrinsic motivation." Organiza-tional Behavior and Human Performance, 25,311-335.

Scott, W. E., Jr. (1980). The hidden costs of re-ward remain largely hidden. Contemporary Psy-chology, 25, 106-107.

Shapira, Z. (1976). Expectancy determinants ofintrinsically motivated behavior. Journal ofPer-sonality and Social Psychology, 34, 1235-1244.

Skinner, B. F. (1953). Science and human behav-ior. New York: Free Press.

Skinner, B. F. (1969). Contingencies ofreinforce-ment: A theoretical analysis. Englewood Cliffs,NJ: Prentice-Hall.

Skinner, B. F. (1974). About behaviorism. NewYork: Vintage.

Smith, M. C. (1975). Children's use of the mul-tiple sufficient cause schema in social perception.Journal ofPersonality and Social Psychology, 32,737-747.

Swann, W. B., Jr., & Pittman, T. S. (1977). Ini-tiating play activity of children: The moderatinginfluence of verbal cues on intrinsic motivation.Child Development, 48, 1128-1132.

Vasta, R., Andrews, D. E., McLaughlin, A. M.,Stirpe, L. A., & Comfort, C. (1978). Reinforce-ment effects on intrinsic interest: A classroomanalog. Journal of School Psychology, 16, 161-168.

Vasta, R., & Stirpe, L. A. (1979). Reinforcementeffects on three measures of children's interest inmath. Behavior Modification, 3, 223-244.

Vaughan, M. E., & Michael, J. (1982). Automaticreinforcement: An important but ignored con-cept. Behaviorism, 10, 217-227.

Weiner, M. J. (1980). The effect of incentive andcontrol over outcomes upon intrinsic motivationand performance. The Journal of Social Psy-chology, 112, 247-254.

Weiner, M. J., & Mander, A. M. (1978). The ef-fects of reward and perception of competencyupon intrinsic motivation. Motivation andEmo-tion, 2, 67-73.

Welker, W. I. (1956). Some determinants of playand exploration in chimpanzees. Journal ofCom-parative Physiological Psychology, 49, 84-89.

White, R. W. (1959). Motivation reconsidered:The concept of competence. Psychological Re-view, 66, 297-333.

Williams, B. W. (1980). Reinforcement, behaviorconstraint, and the overjustification effect. Jour-nal ofPersonality and Social Psychology, 39, 599-614.

Zimmerman, B. J. (1985). The development of"intrinsic" motivation: A social learning analy-sis. Annals of Child Development, 2, 117-160.