CANADIAN PSYCHOLOGICAL REVIEWPSYCHOLOGIE CANADIENNE

Vol. 20, No. 3, July 1979

FRUSTRATION, EXPLORATION, AND LEARNING

PAUL T. P. WONGTrent University

ABSTRACTD. E. Berlyne's principle that uncertainty leads to exploration is extended tosituations involving frustrativc nonreward. Both empirical evidence and logicalanalysis support the frustration-exploration hypothesis, which posits that uncertainfrustration leads to exploration, whose primary function is to broaden the scope ofresponse selection. The facilitative effects of frustration-motivated exploration inlearning are documented, and their implications for education are discussed.

One of Daniel Berlyne's abiding interests was incuriosity and exploration. Perhaps, among hismany significant contributions to psychology,his work in this area has made the greatestimpact. His 1960 monograph Conflict, Arousaland Curiosity is among the 100 books most citedby social scientists (Garfield, 1978). Few willdisagree that it is mainly due to Berlyne's influ-ence that exploration or information seeking hasbeen elevated to the same status as hedonism asone of the primary sources of motivation. Inview of the prevalence of information seeking(Berlyne, 1966), a case may be made that it is amore important motivation than the pursuit ofhedonic goals.

In paying tributes to Berlyne, I cannot help butreminisce on the good old days of being a grad-uate student at the University of Toronto, whenboth Daniel Berlyne and Abram Amsel were onthe psychology faculty. To be taught by thesetwo eminent psychologists was to be influencedby them in a profound way. Although differentin many ways in their approaches to research,both of them demonstrated a remarkable degreeof independence of thought in the midst of shift-ing paradigms and changing fashions in psycho-logical research. While Amsel concentrated his

Portions of this paper were presented in a symposium inhonour of the late D. E. Berlyne at the Canadian Psycho-logical Association Meeting in Vancouver, July, 1977.The paper was written while the author was a visitingscholar at the University of California, Los Angeles. Itwas supported by Grant 3A0701 from the NationalResearch Council of Canada, and a Canada CouncilLeave Fellowship. Requests for reprints should be ad-dressed to P. T. P. Wong, Department of Psychology,Trent University, Peterborough, Ontario, K9J 7B8.

research efforts in well defined problem areas,Berlyne's own curiosity led him to explore newfrontiers in psychology. One learned from Amselhow to rigorously and thoroughly investigate aphenomenon; one learned from Berlyne how tosee old problems in new lights and how to detectconnections in seemingly unrelated areas ofresearch. Both of these approaches have influ-enced my own research career, and these influ-ences have contributed to the present paperwhich attempts to interface frustration andexploration.

In the past ten years, I have done several studieson the problem of persistence and frustration,even though such research no longer attracts theattention it once did. Having carefully observedhundreds of rats in a great variety of frustratingsituations, I cannot help but notice the preval-ence of exploratory behaviour. For example, in arunway situation, partial reinforcement increasessniffing, exploration of irrelevant holes on therunway walls, and variation of routes leading tothe goalbox (Wong 1977a, 1978a). In this article,I will document the existing evidence supportingthe frustration-exploration hypothesis. Morespecifically, I will attempt to demonstrate that(1) uncertain frustration leads to exploration,(2) frustration-motivated exploration primarilyconsists of response-variation, and (3) this typeof exploration plays an important role in instru-mental learning. However, before we examinethe evidence and arguments in support of thefrustration-exploration hypothesis, we shouldfirst examine exploration as a concept and pastresearch on this topic.

133

134 Paul T.P. Wong

Uncertainty, Conflict, Curiosity and Exploration

A great variety of behaviours come under therubric of exploratory behaviour. In rats, orient-ing responses, locomotion, manipulation, rearingand sniffing etc. all may be classified as explora-tory behaviour. Berlyne (1960) has broadlydefined all "responses that alter the stimulusfield" as exploratory behaviour (p.78) and addedthat "their principal function is, in fact, to affordaccess to environmental information that wasnot previously available . . . they widen the scopeof stimulus selection enormously" (p.79). Hinde(1970) has pointed out that "even within onespecies the types of behaviour which come with-in the broad category of exploration or investiga-tion are diverse, and it is difficult to give a moreprecise definition than that they are such as tofamiliarize the animal with its environment orwith a source of situation." (p.351).

From the above definitions, it is clear thatexploration is typically defined in terms of itsfunction, which is information seeking. Thus,exploration and information seeking behaviourhave often been used interchangeably. It is alsoclear that the focus of the above definitions is onthe stimulus dimension, and only behaviour thatincreases the animal's contact with the externalenvironment is considered as exploration. Givensuch a stimulus-orientation, it is only natural thatpast research on exploration has centred on stim-ulus properties as determinants of exploration.However, information seeking need not necessar-ily be directed to external sources of stimulation.When an animal is placed in a puzzle box (seeGuthrie & Horton, 1946), it typically exploresboth the immediate physical arrangements of thebox as well as its own response repertoire until itperforms the correct response that releases thelatch. In this case, exploration has the functionof widening the scope of both stimulus selectionand response selection. Response-oriented ex-ploration may be conceptualized as trial anderror learning, hypothesis testing, or the variable,appetitive component of instinctive behaviour,depending on one's theoretical framework. Ineach case it is an important integral part of ex-ploration in situations where the necessary infor-mation for the solution of the problem resides inthe animal's own response repertoire. Therefore,a more comprehensive definition of exploratorybehaviour should include all responses that in-crease the scope of stimulus selection and/orresponse selection.

Berlyne has differentiated two major classes ofexploration:

On the one hand, when an animal is dis-turbed by a lack of information, thus left aprey to uncertainty and conflict, it is likelyto resort to what we may call specific ex-ploratory responses. These supply or inten-sify information from particular sources —sources that can supply the preciseinformation that the animal misses. Thecondition of discomfort, due to inade-quacy of information, that motivates spe-cific exploration is what we call "curi-osity". In other circumstances, an animalseeks out stimulation, regardless of sourceof content, that offers something like anoptimal amount of novelty, surprisingness,complexity, change or variety. For thiskind of behavior the term diversive explor-ation has been proposed. (Berlyne, 1966,P-26).

The main thrust of Berlyne's work has beenon curiosity-motivated specific exploration.According to Berlyne, the two antecedent condi-tions for curiosity, namely, uncertainty and con-flict, are very closely related. He states that "situ-ations in which uncertainty is of psychologicalimportance are situations of conflict" (Berlyne,1960, p.29). In his last writing on this subject, hecontinued to stress the affinities between thesetwo antecedent conditions. "We can speak of'subjective uncertainty', whenever somebody isinclined towards a number of competing expec-tations, beliefs, images or more generally, repre-sentations of a particular situation, withoutbeing in a position to commit himself to one ofthem and to reject or suppress the others"(Berlyne, 1978, p.143).

Most of the work done on specific explorationhas focused on perceptual curiosity, which isaroused by certain properties of external stimu-lus patterns, and which is directed toward stimu-lation capable of reducing or removing uncer-tainty (e.g., Berlyne, 1950, 1955, 1958). How-ever, Berlyne has extended his analysis of specificexploration to epistemic curiosity, where curi-osity is motivated by conceptual conflict, andthe responses are aimed at stimulation capable ofdispelling uncertainty as well as providing infor-mation for storage in symbolic structures(Berlyne, 1965).

Two defining characteristics may be abstractedfrom curiosity-motivated exploratory responses.First, they are stimulus-oriented rather than

FRUSTRATION AND EXPLORATION 135

response-oriented. They are directed to somesources of external stimulation rather than tovarious alternatives in the organism's responserepertoire. They are best described as stimulusseeking behaviours. Secondly, they seem unre-lated to various biological drives such as hunger,thirst, and pain, and there is no evidence thatthey are dependent on secondary reinforcement.In short, they are simply motivated by curiosity,which should be conceptualized as a primarymotive in its own right.

In spite of his emphasis on the intrinsic motiva-tion of stimulus seeking behavior, Berlyne alsorecognizes the ecological importance of explora-tion. Commenting on various kinds of explora-tory activities, he wrote:

In all these activities, sense organs arebrought into contact with biologically neu-tral or "indifferent" stimulus patterns —that is, with objects or events that do notseem to be inherently beneficial or nox-ious. Stimulus patterns encountered in thisway are sometimes used to guide subse-quent action aimed at achieving some im-mediate practical advantage. An animallooking and sniffing around may stumbleupon a clue to the whereabouts of food.(Berlyne, 1966, p.25).

At this point, one is tempted to raise a numberof fundamental questions. What will the organ-isms do when uncertainty and conflict are pres-ent in events that are inherently beneficial ornoxious? Is specific exploration in this contextdifferent from the kind of curiosity-motivatedexploration described earlier in the paper? Thefollowing represents an initial attempt to answerthese difficult and very involved questions, byextending Berlyne's analysis of specific explora-tion to schedules of reinforcement. More specifi-cally, I will examine exploratory behavior inthree kinds of schedules which clearly involveuncertainty and conflict

Frustration-motivated vs.Curiosity-motivated Exploration

On the basis of logical analysis, one can readilyidentify three kinds of schedules in which someelements of uncertainty and conflict are present:early extinction, successive acquisition and ex-tinction, and partial reinforcement. During theearly stages of extinction, the organism can notbe sure whether reinforcement is still attainable,and should experience an approach-avoidanceconflict. In the case of successive acquisition and

extinction, uncertainty and conflict should becreated by unpredictable changes in the sched-ule. Shaping is a good example of successiveacquisition and extinction. The shaping proce-dure typically involves successive steps orapproximations to a target behavior. Once anorganism has successfully acquired the first step,it is no longer reinforced until it emits a response(hat meets the requirement of the second step.This procedure is continued until Ihe organism isable to emit the target behavior. It is apparentthat whenever the criterion for reinforcement ischanged, the organism has no way to know whyreinforcement is discontinued, nor does itpossess the information to predict the reinstate-ment of reinforcement Finally, in partial rein-forcement situations where reinforcement isirregular and unpredictable, the organism isnaturally uncertain about the outcome and hasconflicting expectancies.

It is not difficult to recognize that all threesituations involve frustrative nonreward, definedas the occurrence of nonreward in a context thathas been associated with reward (Amsel, 1958,1962). It is the presence of frustrative nonrewardthat distinguishes these situations from other sit-uations of uncertainty and conflict. I will nowturn to some evidence that is consistent with thehypothesis that uncertain frustration leads toexploration.

Extinction. Psychologists have known forsome time that the onset of extinction increasesresponse variability (Antonitis, 1951; Kuo,1922; Gutfirie & Horton, 1946; Skinner, 1938).For example, Skinner (1938) observed that varia-tion in the force and duration of lever pressingwas greater during extinction than during condi-tioning, and commented that extinction serves toincrease the scope of responses for selective rein-forcement. In other words, during extinction theorganism will explore alternative response topog-raphies to increase the likelihood of reinforce-ment. With prolonged extinction, the elementsof uncertainty and conflict should no longerexist, because consistent and repetitive presenta-tions of nonreward should eventually convinceeven the most optimistic creature that reinforce-ment in the situation is no longer available. Atthis stage, the rate of instrumental respondingshould decrease to zero. Recently, it has beenpointed out that unrelenting frustration couldeventually lead to "helplessness" and "depres-sion" (Wong, 1978a, b; Wong, Note 1).

136 Paul T.P. Wong

Shaping. Success in shaping importantlydepends on the discrepancy between successivesteps of approximation. If the discrepancy is toogreat, the organism may cease to emit instrumen-tal responses and become "helpless". On theother hand, if the discrepancy is very small tominimize frustration, shaping might be slow anddifficult It is conceivable that if the successivecriteria for reinforcement are too lenient, organ-isms may become so "spoiled" that they fail tolearn the target behavior. In the past few years, Ihave observed several instances of failure toshape a rat by students due to this reason. Forshaping to be efficient, there should be an opti-mal discrepancy between successive steps for theexploratory tendency to reach its maximumstrength.

Partial reinforcement There is now sufficientevidence to suggest that partial reinforcementproduces greater response variability than con-sistent reinforcement. Supporting evidence hasbeen obtained in children (McCray & Harper,1962), college students (Newberry, 1971), andrats (Wong, 1977a, b). It has been suggested thatmore response elements r are sampled from thegeneric response class or a population R underpartial reinforcement than under consistent rein-forcement conditions (Schoenfeld, 1968).

Response space vs. stimulus space. In all threesituations of frustration, the exploratory re-sponses observed are primarily response-oriented. It may be argued that in some casesresponse variation is accompanied by stimulus-oriented exploration. For example, by increasingthe variability of routes leading to the goal box,the animal inevitably increases its contact withthe environment. However, there are also situa-tions where response variation, such as varyingthe vigor, duration, and latency of responses(Skinner, 1938) and transition of responses(Newberry, 1971), is not accompanied by a par-allel increase in the scope of stimulus selection.

On the basis of logical analysis, there is noa priori reason why exploration should bedirected to external stimulation exclusively. Inthe case of perceptual curiosity, where uncer-tainty clearly stems from the properties of ex-ternal stimulus patterns, it is only logical thatexploration be directed to the stimulus space.However, in the case of frustrative nonreward,the source of uncertainty may be due to a changein the environment or the inadequacy of theresponse; therefore, it is most adaptive to explore

both the stimulus and response dimensions. Infact, when the same response in the same situa-tion no longer produces the same outcome, thedominant tendency may be response variation.For example, when the same set of responses failsto start a car, the dominant tendency is usually tovary the strength and duration of responses.Therefore, on the basis of review of the literatureand logical analysis, it may be hypothesized thatfrustration-motivated exploration is primarilyresponse-oriented. This hypothesis is not centralto the proposition that frustration leads to ex-ploration, but is an interesting hypothesis thathelps differentiate two types of exploration.

Information Seeking vs. Hedonic Pursuit

Another difference between curiosity and frus-tration-motivated exploration is that in theformer organisms appear to engage in informa-tion seeking for its own sake, whereas in thelatter information seeking serves the dual func-tion of reducing uncertainty and securing anexternal reward. One may wonder whether frus-tration-motivated exploration will still persist, ifit has the effect of reducing uncertainty, butincreasing frustration. In other words, is informa-tion seeking mainly subservient to hedonic pur-suit in situations of ecological importance? Thisquestion deals with a very fundamental issue inthe psychology of motivation, namely, the rela-tive importance of information seeking as com-pared to hedonism. Berlyne (1966) seems tosuggest that information seeking aimed at uncer-tainty reduction may be more important thanhedonism. He observed that a hungry rat willspend time investigating a novel feature of theenvironment before turning to food, and a birdmay approach a strange object at the risk of itsown life. Of course, one can also readily citeinstances where curiosity not only kills the cat,but also has taken the hie of many a journalistFrom the perspective of attribution theory,Weiner suggests the possibility that informationseeking may be manifest even when it conflictswith the pleasure principle (Weiner, 1979). Tosettle this issue, one needs to pit informationseeking against pleasure seeking, such that anygain in information is counteracted by a loss inthe fulfillment of hedonic needs.

In real life, it is not difficult to think of situa-tions where people would seek out informationat the risk of receiving bad news. For example,patients want to find out whether their illness is

FRUSTRATION AND EXPLORATION 137

curable, and aspiring students want to knowwhether they really have what it takes to besuccessful in certain professions. There are alsonumerous instances in which information isvalued more than pleasures and "truth" is placedabove life itself. In the laboratory, it is not diffi-cult to set up situations where exploration hasthe effect of gaining information (e.g., increasingthe predictability of an outcome), but reducingthe likelihood or magnitude of reinforcement.Similarly, reduction of uncertainty may be offsetby an increase in the likelihood or intensity ofpunishment Clearly there are boundary condi-tions within which information seeking overrideshedonic pursuit If research can demonstrate thatin a variety of situations the need for informationoutweighs hedonic needs, it would bring to atriumphant conclusion the revolution spear-headed by Berlyne over two decades ago in thepsychology of motivation.

Reward Reduction and Exploration

From the perspective of frustration theory,reward reduction is also an antecedent for frus-tration (Daly, 1969; DiLollo&Beez, 1966). Onestudy with college students demonstrated thatreduction in reward magnitude increased thevariability of responding as compared to the con-trol group that did not undergo reward reduction(Boroczi & Nakamura, 1964). In an early studyof shifts in quality of reward, Elliott (1928)observed that a downward shift resulted in anincrease in blind alley entries in the rats, andsuggested that these wrong entries might repre-sent a search for the now missing preferredreward. In other words, reward reduction, likefrustrative nonreward, may also promote explor-ation as long as there is an element of uncertaintyin the situation. Prolonged or predictable rewardreduction should not lead to exploration.

Recently, in an interesting study, Flaherty,Blitzer, and Collier (1978) studied openfieldbehaviours elicited by a reduction of sucroseconcentration in drinking water. Their purposewas "to determine whether any increases in activ-ity that occurred subsequent to the shift mightbe interpreted as frustration or as exploratoryrelated responses." (p.430) By pitting frustrationagainst exploration as alternative interpretationsof increased activity, they apparently assumedthat frustration leads to only general arousal butnot exploration. Their hypothesis was likelyguided by past research on frustration effects,

which focused on general arousal or the invigora-tion effect (Amsel & Roussel, 1952; Scull, 1973).In contrast to Flaherty et al the present frustra-tion-exploration hypothesis assumes that frustra-tion induces general arousal as well as explora-tion; in fact, exploration may be considered asone of the several behavioural manifestations ofarousal induced by frustration. Flaherty et al.reported that reward reduction significantly in-creased ambulation and wall rearing, but did nothave any significant effect on the sampling ofalternative water tubes. They concluded thatreward induced frustration-related behaviouralarousal, but not exploration. However, theauthors recognized that "wall rearing and ambu-lation could also be interpreted as reflecting ex-ploratory tendencies on the part of the animals".Thus, their findings could also be regarded assupporting the frustration-exploration hypothe-sis. The absence of frustration induced responsevariation (i.e., licking alternative water tubes)may be due to the fact that licking is a consum-matory response. In most studies, frustrationinduced response variation is observed in instru-mental responses such as lever pressing orrunning, although frustration induced displace-ment pecking in the Barbary dove has also beenreported (McFarland, 1965, 1966). More re-search is needed to determine whether frustra-tion-motivated response variation can be ob-tained in a number of consummately responses.

The Role of Frustration-MotivatedExploration in Learning

Berlyne always took a keen interest in thegeneral question concerning the role of explora-tory behaviour in learning. In fact, his latest andunfinished book was entitled Curiosity andLearning (Berlyne, 1978). The importance ofcuriosity-motivated exploration (e.g., investiga-tory or orienting reflex) in learning has alreadybeen established (Pavlov, 1927; Sokolov, 1963).The role of frustration-motivated exploration inlearning is much less known. In this section I willpresent some recent findings demonstrating theimportance of frustration-motivated explorationin a number of learning situations.

Frustration and shaping. As mentioned earlier,the shaping procedure typically involves succes-sive acquisition and extinction, hence, frustra-tion. Recently, an experiment was conducted todetermine whether frustration indeed facilitatesshaping (Wong, Note 1). Two groups of rats

138 Paul T.P. Wong

served as subjects. Controls received food-magazine training in their home cages, while Ex-perimental subjects received the same training ina Skinner box. Subsequently, all subjects weretested in the Skinner box for autoshaping of alever pressing response. Since only the Experi-mental subjects had received free food pellets inthe Skinner box, they should experience frustra-tive nonreward when food pellets were no longerfreely available during lever training. Frustration-motivated exploration should result in fasterlearning. This prediction was clearly supported inthat Experimental subjects had a shorter latencyin their first lever response and acquired the leverpressing response at a faster rate than the Con-trols. While all the Experimental rats acquiredthe response very readily, several Controls failedto learn even after several sessions of exposure tothe Skinner box. Their failure to learn was partlydue to the fact that their exploration was pri-marily stimulus-oriented; these rats sniffed andinspected different aspects of the Skinner boxincluding the lever, but rarely pressed it strongenough to activate the feeder. One could furtherassess the role of frustration in shaping and auto-shaping by manipulating prior reward exper-iences such as the magnitude and the frequencyof food pellets during magazine training.

Frustration and higher-order operants. Accord-ing to Wong (1975), higher-order operants arecapable of modifying or controlling schedules ofreinforcement. This class of operants may be bestdescribed as controlling operants, because theycontrol the contingencies for other responses(Catania, Note 2). In a recent experiment (Wong,1977c), if the rat switched to an adjacent lever onthe nonpreferred side, the ratio requirement forthe centre lever was reduced. Rats acquired thecontrolling operant only when they encountereda period of extinction which increased thetendency to explore adjacent levers.

Frustration and creativity. In a very interestingstudy by Pryor, Haag and O'Reilly (1969), aporpoise was trained to be creative. It was rein-forced only when it emitted a novel behaviour. Inother words, the porpoise would be repeatedlyfrustrated until it emitted a behaviour that hadnever been displayed before. As a result of thisshaping procedure, the porpoise greatly ex-panded its response repertoire. The success ofthis creativity training procedure is clearlydependent on frustration-motivated exploration.Selective reinforcement of novel behaviour is

possible only when nonreinforcement or frustra-tion forces the porpoise to stretch its responsespace and exhaust its response repertoire. Morerecently, Wong (1977b) was able to train rats tovary their response routes from trial to trial byusing a similar procedure. Additional evidencethat frustration may increase creativity wasreported by Frost (1976). He reported that mildfrustration seemed to increase creative behaviourin all components except originality as measuredby the Torrance Tests of Creative Thinking. Un-fortunately, the author did not use the multivari-ate analysis of variance to test for significantdifferences.

Frustration and concept learning. The facilita-tive effects of frustration-motivated explorationhave also been demonstrated in a series of experi-ments on concept learning with public schoolchildren as subjects (Wong, Note 4). The learningtask was to abstract different relational rulesfrom different sets of exemplars. The rules in-cluded identity (all elements between two stimu-lus patterns being identical), non-identity (allelements between two stimulus patterns beingdifferent) and various conditions of partial iden-tity (e.g., same color, but different shapesand different orientations). An abstraction-production paradigm of concept learning wasused. That is, the youngster was first asked toabstract the relational rule from a set of exem-plars, and then asked to produce more instancesof this same rule. The children were assigned toeither the Demonstration condition or the Strat-egy condition. Under the former condition, if thechild made an incorrect response, the experimen-ter would demonstrate and explain step by stephow to produce a correct instance of the rule,thus protecting the child from further failure andfrustration. Under the latter condition, when amistake was made, the child was encouraged andguided to explore other responses until a correctproduction was made. The intent of the latterprocedure was to provide opportunities for thechild to learn a constructive strategy of copingwith failure and frustration. The results showedthat concept learning was superior under theStrategy condition where frustration-motivatedexploration played an important role in thelearning process.

Frustration and stimulus-oriented exploration.I have just documented empirical findingsdemonstrating the constructive role of response-oriented exploration in a variety of learning situ-

FRUSTRATION AND EXPLORATION 139

ations. The picture would not be complete with-out reporting that frustration may also facilitatelearning through stimulus-oriented exploration.In several experiments, Sutherland (1966) andMcFarland (1966) have shown that rats anddoves learned more about incidental or irrelevantcues under partial reinforcement conditions, andexhibited faster learning in a subsequent situa-tion where previously irrelevant cues were maderelevant. They hypothesized that when no stimu-lus analyser consistently predicts reward theanimal should try out many different analysers.In other words, frustrative nonreward shouldswitch attention to different aspects of the envir-onment. Sutherland (1966) has suggested thatpartial reinforcement increases the breadth oflearning. Thus, frustration not only compels theanimal to explore alternative responses, but alsodraws their attention to hitherto unnoticed fea-tures of the environment.

Persistence and learning. The importance ofpersistence in learning can not be overempha-sized. Englemann (1969) has rightly pointed outthat "persistence is particularly important in anew learning situation.... A child who has notbeen taught that he will succeed if he continuesto try will tend to interpret his first failure asproof that he cannot succeed.... The child whohas been taught that persistence pays off viewsthe failure in a different way. . . . It is extremelyimportant to teach children the persistence con-cept." (p.94)

However, it is important to distinguish be-tween response persistence and goal persistence(Wong, 1977a). The former refers topersevera-tion in a response that is no longer appropriate.The latter refers to perseverance in the pursuit ofa goal by whatever means. Response persevera-tion is frequently self defeating, whereas goalpersistence often pays off. In partial reinforce-ment situations, occasional frustration elicits agoal-oriented exploratory response as illustratedby the observation that rats vary their routesleading to the goalbox (Wong, 1977a). Rein-forcement of this reponse tendency results in itbeing conditioned to frustrative cues. Thereforewhat is learned in a partial reinforcement situa-tion is primarily goal persistence. More specifi-cally, what is conditioned to frustrative cues maybe called a try and vary coping strategy (Wong,1977a) because it increases goal persistence in avariety of situations (Wong, 1977d; Wong &Amsel, 1976). The word strategy does not imply

higher cognitive processes, but rather denotesthat what is conditioned is not a specific responsetopography but a general response rule (Wong,1977b). This kind of learning is consistent withAnsel's general theory of persistence (Amsel,1972), which posits that the countercondition-ing of an ongoing behaviour to any disruptiveevent will increase the persistence of any ongoingbehaviour in subsequent disruptive situations.Guided by this notion, we have demonstratedtransfer of persistence across different situationsand different response systems (Amsel, Wong &Scull, 1971; Amsel, Glazer, Lakey, McCuller, &Wong, 1973; McCuller, Wong & Amsel, 1976;Wong, 1971, 1977d). Such transfer of persis-tence suggests to us that what is conditioned tofrustrative cues is a general response rule ratherthan a specific response topography; in otherwords, the kind of persistencethat is generalizedis goal persistence rather than response persis-tence.

The only study that appears to contradict theabove rule learning notion is that reported byRoss (1964). He found that rats given partialreinforcement training for climbing subse-quently continued to exhibit the climbing re-sponse when running was being extinguished.One may resolve this apparent discrepancy bypointing out that as the try and vary responserule is being activated, alternative responses thathave been previously reinforced should have ahigher probability of occurring; therefore, it isnot surprising that the previously partially rein-forced rats should exhibit the climbing response,which was a subset of goal approach responses inthe runway, when they tried various alternativesin the face of frustration. One may also arguethat if Ross provided measures of goal persis-tence (i.e., frequency of goal entry or number oftrials to goal avoidence), these rats should showgreater goal persistence than their continuouslyreinforced counterparts. To really settle the issuewhether rule learning will override responselearning in the transfer of persistence, we need asituation where a specific response and the gen-eral persistence rule are pitted against each other.For example, we can first give rats partial rein-forcement training for barpressing, then continu-ous reinforcement training in a runway. Whenthe running response is being extinguished, a baris made available in the startbox. Will the ratspersist in barpressing or in approaching the run-way goalbox in this situation? If rule learning

140 Paul T.P. Wong

predominates, rats should persist in varying theirbehaviour patterns that lead to the goalbox.

Certain vs. uncertain frustration. Berlyne(1960) suggested that conflict and frustrationmay motivate thinking, epistemic curiosity, andthe acquisition of knowledge. I have shown thatthe beneficial effects that accrue to frustrationare primarily attributable to frustration-motivated exploration. One may wonder whyfrustration has been generally viewed in a nega-tive light (Lawson, 1965). This may be, in part,due to the negativity bias of psychologists whoseem to be particularly tuned to the dark side ofthe human nature or events (see deCharms &Muir, 1978). For example, agitated behaviourduring extinction has been described as neuroticor pathological (e.g., Miller & Stevenson, 1936).From my perspective, extinction induced beha-vioural arousal may represent the mobilization ofenergy and resources to cope with frustrationand conflict. A second reason may be that thenegative affective consequence of failure is oftennot balanced by the positive affective conse-quence of overcoming failure. Finally, it may bedue to the problem of not differentiating be-tween certain and uncertain frustrating events. Inthe former case, the causes of frustration areboth stable and uncontrollable. In the latter case,there is the likelihood of success if one tries longenough. When frustration is unrelenting and un-mitigable, as in prolonged extinction or a longhistory of failure, one may become a victim ofhelplessness. However, when frustration is occa-sional and/or controllable, it provides challenge(Berkowitz, 1964; McClelland, 1969) and pro-motes learning.

The beneficial effect of frustration-motivatedexploration on learning has been primarily ob-served in instrumental learning situations.Whether uncertain frustration also has facilita-tive effects on a wide range of learning situations,such as problem solving, memory retrieval, infor-mation processing, remains an empirial question.As well, the generality of the frustration-exploration hypothesis remains to be tested indifferent species.

It should be noted that since uncertainty is anintegral part of frustrating situations that arecapable of promoting exploration, one maywonder to what extent the internal motivationalstate of curiosity as instigated by uncertaintycontributes to exploration. It is possible toexperimentally assess the relative contribution of

frustration and curiosity to exploration byindependently manipulating the degree of uncer-tainty through varying reward and nonrewardsequences (Habu & Ono, 1976) and by manipu-lating the magnitude of frustration through vary-ing the magnitude and frequency of reward aswell as drive level. Apart from the issue of overlapbetween two motivational states, there is also theinteresting issue with respect to the similaritiesand differences between frustration and curi-osity — a question that also awaits furtherresearch.

The Frustration-ExplorationHypothesis and Education

Berlyne's research on curiosity-motivated ex-ploration has already been applied to educationto enhance creativity and intrinsic motivation(Day, Note 3; Day, Berlyne, & Hunt, 1971).Frustration-motivated exploration has perhapseven wider implications for education, especiallyin educational systems, where failure is viewed asan unnecessary evil to be circumvented wheneverpossible. Skinner's (1959) contention that errorresponding is an unnecessary accompaniment inthe learning process has certainly contributed tovarious educational practices, such as pro-grammed instructions and individualized learn-ing, which are geared to the elimination of failureexperiences. Other practices, such as anecdotalreporting and social promotion, are also designedto shield students from the experience of failureand frustration. I shall therefore first discusssome of the adverse effects of a failure freeenvironment before discussing some of theeducational applications of the frustration-exploration hypothesis.

First, a failure free environment means theabsence of challenge. McClelland (1969) hasdemonstrated that achievement-oriented peopleprefer moderately difficult tasks with a fairchance of failure. Berkowitz (1964) has made thesame point that optimal challenge requires occa-sional frustration. It is difficult to think of situa-tions where challenge exists without the possibil-ity of failure. Secondly, the absence of failurealso means the absence of true success. It isdubious whether one can experience the satisfac-tion of success in a task in which one could notpossibly fan. There is evidence that people do notexperience pride when they succeed in an easytask, and that the incentive value of success isinversely related to the probability of success

FRUSTRATION AND EXPLORATION 141

(Weiner & Kukla, 1970). Thirdly, the absence offailure means the absence of the needed stimu-lant to intellectual growth. According to Piaget(1966), some degree of discrepancy betweenexisting schemata and environmental demands isessential for cognitive growth, although too greata discrepancy may create learning problems(Adelman, 1978).

In addition to the above adverse effect onachievement motivation and intellectual growth,a failure free environment fails to equip studentswith the necessary skills to cope with failure andfrustration in real life. The consequences couldbe far reaching and serious, if students have notlearned how to tolerate and cope with frustrationconstructively. For example, attrition rateswould be high when these students move on toinstitutes of higher learning or the world of workwhere one is no longer protected from the harshreality of competition and assessment. Their in-ability to cope with frustration might also makethem vulnerable to mental disorder, since there isevidence that "proneness to mental disorder in-creases directly with anticipation of, or actual,failure to reach desired goals (i.e., frustration)"(Parker & Kleiner, 1966, p. 10). Lerner haspointed out that the "revolution of rising expec-tations" may touch off a "revolution of risingfrustration" (Evinger, 1971). While educationand mass media continue to raise expectations,economic realities make it increasingly difficultto fulfill these expectations, resulting in mount-ing frustrations. Inability to cope with thesefrustrations constructively may have serioussocial repercussions.

In view of the above considerations, it shouldhave become apparent that the complete absenceof failure could be just as harmful as too muchfailure, and the school has the responsibility ofteaching students how to cope with failure andfrustration. Unfortunately, the importance ofteaching coping skills and the constructive role offrustration in the learning process have not re-ceived due attention from educators. On thebasis of past research on frustration and explora-tion, I will venture some practical suggestionsas to how one may apply the frustration-exploration principle to education.

D'Amato (1970) points out that schools mustbe geared to the "real" world where no one canexercise complete control over error responding,and suggests that an individual's frustration toler-ance might be increased by gradually exposing

him/her to an increasing amount of frustrativenonreward. There is evidence that such a gradualprocedure increases tolerance of aversive situa-tions (Miller, 1960; Peters, Wong, &Traupmann,1971). We also have evidence that when theamount of frustrative nonreward is graduallyincreased as in the case of progressively in-creasing ratio requirements, animals becomemore resistant to extinction (McCuller, Wong, &Amsel, 1976; Wong & Amsel, 1976). One shouldtake into consideration individual differencesto ensure that the amount of frustration to beovercome should not exceed an individual'stolerance threshold, and there is an optimalamount of frustration which generates thestrongest exploratory tendency. How much frus-tration one could tolerate depends on theamount of frustration one has successfully over-come in the past Therefore, the procedure ofgradually increasing the period or amount offrustrative nonreward builds up one's frustrationtolerance, as long as such a procedure is notcarried to the "breaking point"'. The experienceof overcoming a very difficult task after pro-longed frustration and many trials and errorscould greatly increase one's sense of competenceand achievement motivation. For example,Silber, Hamberg, Coelho, Murphy, Rosenberg, &Pearlin (1961) interviewed academically out-standing adolescents and found out that animportant means whereby these students main-tained a high sense of competence was by re-calling the way in which they had successfullycoped with difficult situations in the past.

Another procedure is to mete out failure occa-sionally. This may be accomplished by inter-spersing some difficult items in instructionalmaterials or examination questions. As we haveseen earlier, such a procedure not only makes thetask more challenging, but also increases one'stendency to persist and explore in subsequentfrustrating situations. The persistence effect ofpartial reinforcement training is well establishedin the animal literature (Robbins, 1971). We alsohave evidence that a mixture of success and fail-ure increases persistence in the school setting(e.g., MacArthur, 1955). The optimal level ofpartial reinforcement in promoting persistenceand exploration may depend on one's age as wellas one's past reinforcement history. This proce-dure is especially useful in programmed instruc-tions where the problem of maintaining students'interest and task persistence is exacerbated in theabsence of an instructor.

142 Paul T.P. Wong

Finally, the frustration-exploration principlecan be readily applied by using the shaping proce-dure. Traditionally, this procedure has been usedin behaviour modification programs, and typi-cally the successive steps of approximation aremade small enough to minimize frustration andfailure. In contrast, I propose that this procedurebe applied to all kinds of academic behaviours,and the successive steps should be sufficientlylarge to permit the operation of frustrationinduced exploration. For example, once a certainlevel of performance has been reached by thestudents, a higher level of performance should berequired. This new level should be sufficientlydifficult to attain so that the students have to putforth serious efforts to explore more efficient,more sophisticated, or more creative ways toperform the task. Once this new requirement ismet, a still higher level of performance is insti-tuted. By constantly stretching the students'"growing edge", this procedure may help toshape them into competent and creative individ-uals who know how to cope with frustrationconstructively.

Conclusions

It seems warranted to conclude that Berlyne'suncertainty-exploration principle can be ex-tended to situations involving frustrative non-reward. The evidence supporting the frustration-exploration hypothesis is at this moment at leastcredible if not compelling. The generality of thishypothesis remains to be tested in differentlearning siutations, involving different species atdifferent levels of development.

I have also marshalled evidences attesting tothe beneficial effects of frustration-motivatedexploration in learning, and proposed proceduresof implementing the frustration-explorationprinciple in educational practices. Since the verynature of human existence involves frustration,we might as well learn how to turn it toour advantage by following the frustration-exploration principle.

RESUMELe principe formula par D.E. Berlyne exprimant que l'incertitude mene a l'explora-tion, s'applique a des situations non suivies de recompense frustiantes. Cettehypothese de frustration-exploration est soutenue a la fois par des observationsempiriques et des analyses logiques, qui posent que la frustration eprouvee devantl'incertain entraine l'exploration, dont le but primaire est d'61argir le rayon deselection de responses. Cette exploration ainsi motivee peut faciliter l'apprentissage,II en suit une documentation et une discussion sur les implications possibles de cepheiiomene dans le domaine de lMducation.

Reference Notes1. Wong, P. T. P. Extinction induced facilitation and

helplessness. Presented in a symposium on LearnedHelplessness and Frustration in the XlXth Inter-national Congress of Applied Psychology, Munich,August, 1978.

2. Catania, A. C. Personal communication. May 3,1977.

3. Day, H. I. Curiosity, creativity, and attitude toschooling in open and traditional schools(Grades 2-4). York University, Department of Psy-chology Report No. 31,1976.

4. Wong, P. T. P. Developing productive persistence inconceptual problem solving: The constructive roleof failure in the teaching-learning process. OntarioMinistry of Education Research Report, 1976.

ReferencesAdelman, H. S. The conept of intrinsic motivation:

Implications for practice and research with the learn-ing disabled. Learning Disability Quarterly, 1978,1,43-54.

Amsel, A. The role of frustrative nonreward in noncon-tinuous reward situations. Psychological Bulletin,1958,55,102-119.

Amsel, A. Frustrative nonreward in partial reinforce-ment and discrimination learning: Some recent his-tory and a theoretical extension. Psychological Re-view, 1962,69, 306-328.

Amsel, A. Behavioural habituation, countercondition-ing, and a general theory of persistence. In A. H.Black & W. F. Prokasy (Eds.), Classical condition-ing II: Current research and theory. New York:Appleton-Century-Croft, 1972.

Amsel, A., & Roussel.J. Motivational properties of frus-tration: I. Effect on a running response of the addi-

FRUSTRATION AND EXPLORATION 143

tion of frustration to the motivational complex.Journal of Experimental Psychology, 1952, 43,363-368.

Amsel, A., Wong, P. T. P., & Scull, J. Transfer ofpersistence in the domestic chick: Imprinting, pun-ishment, and partial reinforcement. PsychonomicScience, 1912,25,174-176.

Amsel, A., Glazer, H., Lakey, R., McCuUer, T., & Wong,P. T. P. Introduction of acoustic stimulation duringacquisition and resistance to extinction in the nor-mal and hippocampal-damaged rat Journal of Com-parative and Physiological Psychology, 1973, 84,176-187.

Antonitis, J. J. Response variability in the white ratduring conditioning, extinction, and reconditioning.Journal of Experimental Psychology, 1951, 42,273-281.

Berkowitz, L. The development of motives and values inthe child. New York: Basic Books, 1964.

Berlyne, D. E. Novelty and curiosity as determinants ofexploratory behavior. British Journal of Psychology,1950,41, 68-80.

Berlyne, D. E. The arousal and satiation of perceptualcuriosity in the rat Journal of Comparative andPhysiologicalPsychology, 1955,48, 238-249.

Berlyne, D. E. The influence of complexity and noveltyin visual figure on orienting responses. Journal ofExperimental Psychology, 1958,55, 289-296.

Berlyne, D. E. Conflict, arousal and curiosity. NewYork: McGraw-Hill, 1960.

Berlyne, D. E. Structure and direction in thinking. NewYork: John Wiley, 1965.

Berlyne, D. E. Curiosity and exploration. Science, 1966,153, 25-33.

Berlyne, D. E. Curiosity and learning. Motivation andEmotion, 1978,2, 79-175.

Boroczi, G., & Nakamura, C. Y. Variability of respond-ing as a measure of the effect of frustration. Journalof Abnormal and Social Psychology, 1964, 68,342-345.

Daly, H. B. Learning of a hurdle-jump response toescape cues paired with reduced reward or frustra-tive nonreward. Journal of Experimental Psychol-ogy, 1969, 76,146-157.

D'Amato, M. R. Experimental Psychology. New York:McGraw-Hill, 1970.

Day, H. I., Berlyne, D. E., & Hunt, D. E. (Eds.),Intrinsicmotivation: A new direction in education. Toronto:Holt, Rinehart and Winston, 1971.

deCharms, R., & Muir, M. S. Motivation: Social ap-proaches. Annual Review of Psychology, 1978,29,91-113.

DiLollo, F. D. & Beez, V. Negative contrast effect as afunction of magnitude of reward decrement. Psy-chonomic Science, 1966,5, 99-100.

Elliott, M. H. The effect of change of reward on themaze performance of rats. University of CaliforniaPublications inPsychology, 1928,4,19-30.

Englemann, S. Conceptual learning. San Rafael: Dimen-sions Publishing Co., 1969.

Evinger, J. Sociologist predicts "Revolution of risingfrustrations". East-West Center Magazine, Spring, 1,1971.

Flaherty, C. F., Blitzer, R. & Collier, G. H. Open fieldbehaviors elicited by reward reduction. AmericanJournal of Psychology, 1978,97, 429-443.

Frost, K. B. The effects of frustration on the figuralcreative thinking of fifth grade students. Journal ofExperimental Education, 1916,44, 20-23.

Garfield, E. The 100 books most cited by social scien-tists, 1969-1977. Current Contents, 1918,10,5-16.

Guthrie, E. R., & Horton, G. P. Cats in a puzzle box.New York: Rinehart, 1946.

Habu, Y., & Ono, S. Resistance to extinction as afunction of uncertainty of reinforcement sequence.Psychologia, 1916,19, 149-156.

Hinde, R. A. Animal behaviour: A synthesis of ethologyand comparative psychology (2nd ed.). New York:McGraw-Hill, 1970.

Kuo, Z. Y. The nature of unsuccessful acts and theirorder of elimination in animal learning. Journal ofComparative Psychology, 1922,2, 1-27.

Lawson, R. Frustration: The development of a scientificconcept. New York: McMillan, 1965.

Mac Arthur, R. S. An experimental investigation of per-sistence in secondary schoolboys. Canadian Journalof Psychology, 1955,9, 42-54.

McClelland, D. C. The role of educational technologyin developing achievement motivation. EducationalTechnology, 1969, October 7-19.

McCray, C. L., & Harper, R. S. Some relationships ofschedules of reinforcement to variability of re-sponses. Journal of Comparative and PhysiologicalPsychology, 1962,55, 19-21.

McCuller, T., Wong, P. T. P., & Amsel, A. Transfer ofpersistence from fixed-ratio barpress training to run-way extinction. Animal Learning & Behavior, 1976,4, 53-57.

McFarland, D. J. Hunger, thirst, and displacementpecking in the Barbary dove. Animal Behaviour,1965,75,293-300.

McFarland, D. J. The role of attention in the disinhibi-tion of displacement activities. Quarterly Journal ofExperimental Psychology, 1966,18, 19-30.

Miller, N. E. Incremental training to resist stresses suchas pain and fear. Journal ofComparative and Physio-logicalPsychology, 1960,55,59-62.

Miller, N. E., & Stevenson, S. S. Agitated behavior ofrats during experimental extinction and a curve ofspontaneous recovery. Journal of Comparative Psy-chology, 1936,21, 205-231.

Newberry, B. H. Response variability and the partialreinforcement effect Journal of Experimental Psy-chology, 1971,89, 137-141.

Parker, S., & Kleiner, R. Mental illness in the urbanNegro community. New York: The Free Press, 1966.

Pavlov, I. P. Conditioned reflexes. Oxford: Oxford Uni-versity Press, 1927.

Peters, J., Wong, P. T. P., & Traupmann, K. L. Increasein quinine tolerance with incremental training. Psy-chonomic Science, 1971,25, 134.

Piaget, J. Origins of intelligence in children. New York:International University Press, 1966.

Pryor, K. W., Haag, R., & O'Reilly, J. The creativeporpoise: Training for novel behavior. Journal ofthe Experimental Analysis of Behavior, 1969, 12,653-661.

144 Paul T.P. Wong

Robbing, D. Partial reinforcement: A selective review ofthe alleyway literature since 1960. PsychologicalBulletin, 1971, 76,415-431.

Ross, R. R. Positive and negative partial reinforcementextinction effects carried through continuous rein-forcement, changed motivation, and changed re-sponse. Journal of Experimental Psychology, 1964,68,44-52.

Schoenfeld, W. N. On the difference in resistance toextinction following regular and periodic reinforce-ment Journal of the Experimental Analysis ofBeha-vior, 1968,11, 259-261.

Scull, J. W. The Amsel frustration effect: Interpreta-tions and research. Psychological Bulletin, 1973, 79,352-361.

Silber, E., Hamberg, D. A., Coelho, G. V., Murphy,E. B., Rosenberg, M., & Pearlin, L. I. Archives ofGeneral Psychiatry, 1961,5, 354-365.

Skinner, B. F. The behavior of organisms. New York: D.Appleton-Century, 1938.

Skinner, B. F. Cumulative record. New York: Appleton-Century-Crofts, 1959.

Sokolov, E. N. Perception and the conditioned reflex.New York: Permagon Press, 1963.

Sutherland, N. S. Partial reinforcement and breadth oflearning. Quarterly Journal of Experimental Psy-chology, 1966,18, 289-301.

Weiner, B. A theory of motivation for some classroomexperiences. Journal of Educational Psychology,1979, 71, 3-25.

Weiner, B., & Kukla, A. An attributional analysis ofachievement motivation. Journal of Personality andSocial Psychology, 1970,75, 1-20.

Wong, P. T. P. Coerced approach to shock and resistanceto punishment suppression and extinction in the ratJournal of Comparative and Physiological Psychol-ogy, 197'1,75,82-91.

Wong, P. T. P. The concept of higher order opeiant: Apreliminary analysis: Bulletin of the PsychonomicSociety, 1975,5, 43-44.

Wong, P. T. P. A behavioral field approach to instrumen-tal learning in the rat: I. The partial reinforcementeffects and sex differences. Animal Learning andBehavior, 1977,5, 5-13. (a)

Wong, P. T. P. Strategy learning and the skill partialreinforcement effect in the rat American Journal ofPsychology, 1977,90, 663-674. (b)

Wong, P. T. P. Extinction facilitates acquisition of thehigher order operants. Bulletin of the PsychonomicSociety, 1977,9, 131-134. (c)

Wong, P. T. P. Durable partial reinforcement effects andsocial dominance in the rat Learning and Motiva-tion, 1977,5,275-283. (d)

Wong, P. T. P. A behavioral field approach to instrumen-tal learning in the rat: II. Training parameters and astage model of extinction. Animal Learning andBehavior, 1978,6, 82-93. (a)

Wong, P. T. P. A behavioral field approach to operantconditioning: Extinction induced sanddigging.Bulletin of the Psychonomic Society, 1978, 12,203-206. (b)

Wong, P. T. P., & Amsel, A. Prior fixed-ratio training anddurable persistence. Animal Learning and Behavior,1976,4,461-466.

CORRECTIONS ENDLER ARTICLE

Several corrections should be noted in NormanEndler's article, "Where the 'Stars' are: The 25most cited psychologists in Canada(1972-1976)" (CPR, 20,12-21). The correctionsare as follows: l ) in Table 1, the subheadings"1975 1974 1973 1972" were improperlyaligned with the columns, and Kimura's 1976citation rank should have been 3 instead of 2;2) in Table 2, the number 1 calling attention tofootnote 1 was omitted from the title; 3) inTable 3, an additional heading "Citations"should have appeared over the columns "19751974 1973 1972" on the same line as "Publica-tions;" 4) in Table 3, the probability levelsshould have read "*p < .05" and "**p < .01";5) on page 19, first paragraph left column, line

14, the number of psychologists should have read4,070 instead of 4.070; and, 6) the volumeand/or page numbers were omitted from thebibliographic entries for Endler's 1978 CPRarticle (Vol. 19, pages 152-157) and Endler et al's1978 American Psychologist article (Vol. 33,pages 1064-1082). CPR also extends apologies toProfessors Endler, Melzack and Tulving for typo-graphical errors in the spelling of their names asfollows: on the front cover, Professor Endler'smiddle initial should have been listed as S; onpage 16, in the last paragraph, in the left handcolumn, Melzack's name was misspelled; on page19, in the second paragraph in the right column,Tulving's name was misspelled.