Performance to varied reward following continuousreward training in the runway*

RICHARD S. CALEF and DAYID C. HOPKINSWeIt J 'irginia wesleyan College. Buckhan non , W. Va. 26201

EARL R. McHEWITTBard College. Annandale-on-Hudson, N. Y. 12504

and

FREDERICK R. MAXWELLSouth west Missouri State College, Springfield, Mo. 65302

Depressed runway performance of rats to varied largeand small reward following consistent high-incentivetraining was found . Ss shifted to var ied reward displayeda de pression effect following both large' andsmall-reward trials, a finding particularly nonsupportiveof Capaldi's stimul us aftereffect theory and supportiveof Amsel 's fru stration theory .

Depressed performance to small reward followinghigh-incentive training is a reliable finding, occurringwith several parameters of preshift reward (cf. Black,1968). More recently, postshift reward conditions havereceived attention in an attempt to relate the depressioneffect to other literature (e.g., extinction) employingreward shifts in simple instrumental conditioning(Gonzales & Bitterman, 1969). The present studyprovides further information on the generality of thedepression effect with respect to postshift reward byemploying a previously neglected postshift incentivevalue, varied large and small reward. Theoretically,depressed postshift performance following a shift fromcontinuous to varied reward is expected by severalinterpretations of reward shift effects (e.g., Amsel , 1958:Capaldi, 1967).

According to Amsel's (1958) conditioned frustrationtheory , Ss shifted from continuous large magnitude ofreward to varied large and small magnitude of rewardshould run more slowly during postshift training than acontrol group maintained on varied large and smallmagnitude of reward. The experimental Ss should showdepressed postshift speeds as a consequence ofconditioned interfering responses elicited by frustrationstimuli, which, in turn , result from the small magnitudeof reward experienced following preshift training.Control Ss, however, should have learned to run in thepresence of frustration cues (rf-sf) that result fromrepeated experience with large and small magnitude ofreward during the preshift period.

Capaldi's (1967) stimulus aftereffect theory wouldalso predict a "depression" effect in the present study.According to Capaldi (1967). the reward magnitudereceived on any trial provides magnitude-specific stimulipresent when the organism responds on the next trial. Ssshifted from consistent large magnitude of reward to

'This paper is sponsored by James H . McHos e. who takes fulleditorial responsibility for it.

Bull. Psychon. Soc., 1973, Vol. 2 (2)

varied large and small magnitude of reward should runmore slowly than control Ss, because the novelaftereffect of the new reward for experimental Ss hasnot had as much habit strength accrued to it as has thesmall-reward aftereffect present for control Ss.

METHODThe Ss were 20 experimentally naive male albino rats.

approximately 90 days old at the beginning of the experimentand bred at West Virginia Wesleyan College.

The runway apparatus was essentially the same as thatemployed by Ludvigson & Gay (1966) , except that only one ofthe multiple parallel alleys was used. The alley compri sed a33.02-cm white startbox, a 66.04-cm white run section . and a33.48-cm white goalbox . The inner width and height of eachsection of the runway was 7.62 em. Photocell and clock circuitryprovided independent traversal times over the rust 20.32-cmsegment (start time) , the second 30.48-cm segment (run time) .and the third 20.32·cm segment (goal time) of the alley. Theapparatus above contained a solenoid-operated opaqueguillotine-type retrace door. separat ing goal from alley sections.and an opaque Plexiglas ceiling.

All Ss were given preliminar y training, which consisted ofexploration of the alley. 14 days on a 23-h food-deprivationschedule, and 1 g of 45-mg Noyes pellets, identical to thesubsequent reinforcement pellets. being incorporated into Ss'daily food ration . During experimental training, all Ss weredeprived of food for 23 h and received 60 presh ift and 30postshift trials at the rate of 2 trials/day . The experimentalgroup (C-Y) received constant large reward (16 pellet s) duringpreshift training and varied large and small (1 pellet) rewardduring postshift training, whereas the control group (Y-V)received varied large and small reward throughou t the study. Avaried reinforcement schedule, which allowed large and smallreward to be counterbalanced over 4 days, was employed.

Trials were administered to two squads of lOSs each.consisting of five Ss per group . The running order of Ss within asquad was random ized from day to day . The intertrial intervalwas approximately 7 min.

RESULTS AND DISCUSSIONAs the pattern of results did not differ across

measures, only reciprocated run times are reported.Group mean speeds as a function of blocks of six trialsare presented in Fig. 1. As may be seen. terminalacquisition performance (Block A) did not differ. andcontinuous and varied reward produced comparablelevels of performance. These preshift data areinconsistent with previous reports of faster or of slowerspeeds for varied relative to continuous reward. beingconsistent only with the similar observation that therelative performance levels for percentage of reward, insimple instrumental conditioning, also presently defydescription (McHose, 1970).

Looking at the postshift period, the shift fromcontinuous to varied reward (Group C-V) resulted inspeeds slower than those for the condition maintainedon varied reward (Group V-V). As may be seen in Fig. I.this difference occurred early in the postshift period.and. following the point of maximal depression at

103

BLO CKS O F 5 TRIALS

p < .05 , respectively . The analyses of the S I6 data alsoyielded significant preshift reward and Preshift Rewardby Blocks effects , F = 20 .83 , df = 1/18 , P < .01 , and F =6.54, df= 1/18 , P < .05, respectively .

The finding of depressed speeds early in the postshiftperiod following a shift to varied reward is analogous toprevious findings with shifts to small or nonreward(Gonzales & Bitterman, 1969) and is consistent withcontemporary interpretations of the effects attendingchanges in reward. Thus, these data seem to beconsistent with the notions of frustration-produceddisruptions in performance (Amsel, 1958) and with theview that the depression effect is produced by a stimulusgeneralization decrement (Capaldi , 1967). However, thepresent findings that a "depression" effect occurredfollowing both large- and small-reward trials isparticularly nonsupportive of Capaldi's stimulusaftereffect theory, which would predict a "depression"effect following small-reward trials on ly (SI), since habitstrength for experimental Ss has accrued only to thestimulus present following large-reward trials (SI6). Thisfinding of a pervasive "depression" effect is supportiveof frustration theory , since the alley cues should elicitfrustration-interfering responses regardless of theimmediately preceding reward . In essence, the present"depression" effect is best explained by Amsel'sfrustration theory . It should be mentioned that Calef(1972) also found evidence that the "depression" effectis a frustration phenomenon rather than a result of astimul us generalization decrement.

To conclude, Amsel's frustration theory also easilyincorporates the absence of a "depression" effectfollowing both large- and small-reward trials late in thepostshift period . In other words , experimental Ss shouldlearn to run in the presence of frustration cues (rf-sf)late in the postshift period, resulting in the attenuationof the "depression" effect.

(Received for publication June 1, 1973.)

REFERENCESArnsel, A . The role of frustrative nonreward in noncontinuous

reward situations. Psych ol o gical Bulletin, 19 58. 55, 10 2-11 9 .Black, R . W. Shi fts in magnitud e of reward and co ntrast effects

in instrumental an d selective learning: A reinterpretation.Psychological R eview . 19 68, 75,114-126.

Calef, R. S . T he effect of lar ge and small magnitude of intertrialreinforcement on successive cont rast effects. PsvchonornieScience, 1972, 29, 30 9- 3 1 2.

Capaldi, E. J. A sequential hypothesis of instrumental learning.In K. W. Spence and J . T . Spence ( Eds.), The psychology oflearning an d motivation . Vol. 1. New York: Academic Press,1967 .

Gonzales . R . C., & Bit ter ma n , M. E . Spaced-trials partialreinforcement effect as a function of contrast. Journal ofComparative & Physiological Psychology, 1969, 67, 94-103.

Ludvigson, H . W. , & Gay, S. Differential reward conditioning:5- contrast as a function of the magnitude of S+ .Psvchonomic Sicence, 1966, 5, 289·290.

McHose, J. H . Relative reinforcement effects: S./S 2 and S./S .paradigms in instrumental conditioning. Psychological Review,1970,77,135-146.

54

.-. C ' V

..~/

.----_.

,,_0__o~ 0

0" ~?O0,,/

•

---._.

A

o-ov-v

2 3TRI AL BL OCKS

Fig. 1. Group mean run speeds as a function of trial blocks.

'"~~ ~ 2 0~V

0

~ QC:C.

'"Z<:

s be

::::l0

'"c J:

Block 3, dissipated by Block 5. Analysis of variance,including preshift reward and blocks (1-2 vs 4-5) asfactors, yielded significant preshift reward and PreshiftReward by Blocks effects, F = 19.85, df = 1/18, P < .01,and F = 6.13, df = 1/18, P < .05, respectively. Acomparison (t test) of Groups C-Y and Y-Y overBlocks 4-5 was not significant. These analyses supportthe conclusion that a depression effect occurred forGroup C-Y and that this effect was markedly attenuatedat the end of the preshift period.

As can be seen in Fig. 2, a significant "depression"effect occurred and dissipated following both smallreward (SI) and large reward (SI6). Analyses ofvariance , including preshift reward and blocks (1 vs 3) asfactors for the SI data, yielded significant preshiftreward and Preshift Reward by Blocks effects, F =23 .21, df = 1/18, p< .01, and F = 7.23 , df = 1/18,

u- Iw

~ I

~ Io ·,ot

~ ~w I~ .;

a..:::>oer: ,:;;o

Fig. 2. Group mean run speeds as a function of trial blocksfollowing small-reward trials (left frame) and large-reward trials(right frame).

104 Bull. Psych on. Soc ., 197 3, Vol. 2 (2)