THE EFFECT OF INCENTIVE SIZE 0N RESPONSE AMPLITUDE DURING ACQUISITION AND EXTINCTION Dissertation for the Degree of Ph. 0. MICHIGAN STATE UNIVERSITY DANIEL F. TORTORA 1973 IIILIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 62506 «z 7!; LIBRAR Y Michigan State University '7' " _ “fly-'- », This is to certify that the thesis entitled THE EFFECT OF INCENTIVE SIZE ON RESPONSE AMPLITUDE DURING ACQUISITION AND EXTINCTION presented by Daniel F. Tortora has been accepted towards fulfillment of the requirements for Ph .D . degree in Psychology %%4€2%7 Majop/ rofessor 0-7639 '5' I BINDING BY A IIIIAE & SIINS' BIIIIK BINDERY INC. LIBRARY BINDERS gummy}. Ital“! Wee-1i- a t6 Daniel F. Tortora Approved 22 fl” Date Dr. M. Rafi Denny Thesis Director Dr. Stanley C. Ratner Dr. Mark Rilling Dr. Lester M. Hyman Dr. Robert L. Raisler Committeemen (LL, .2; /77.5 (7 / I ABSTRACT THE EFFECT OF INCENTIVE SIZE ON RESPONSE AMPLITUDE DURING ACQUISITION AND EXTINCTION BY Daniel F. Tortora Research during the past twenty one years, using the double alley paradigm, has demonstrated that an instrumental response preceded by nonreinforcement is performed with greater vigor than a response preceded by reinforcement (i.e., the Frustration Effect, FE). Unfortunately, this paradigm is subject to a major confounding, demotivation,which pre- cludes the assessment of the effect of size of incentive upon the PB. The purpose of Experiment 1 was to investigate the effect of nonreinforcement on performance as a function of size of incentive during training, using a paradigm that was not subject to demotivational confounding. Thirty male hooded rats were run in an apparatus designed to measure force of panel pushing. Each g was trained to push two panel doors in succession shch that pushing the first panel door (Operant response) allowed gpto obtain reinforcement (a 45 mg. pellet) behind the second panel (goal response). The ITI was 30 sec and the maximum duration of a trial was 2 min. Once this sequence was established gs were assigned to one of five different groups; they differed in the size of the single Daniel F. Tortora pellet § received per trial:a 45, 97, 190, 300 or 500 mg pellet. All gs were run under the appropriate size of reward for 10 days (10 trials per day). .At the end of training gs were extinguished, half of each group under a 30 sec. ITI and half under an 8 sec. ITI. The results of the acquisition stage indicated a signif- icant increase in force of responding over days but no sig- nificant relation was found between size of incentive and force of responding. It was concluded that pellet size may not be a main incentive variable and that other variables uncontrolled in the present study such as ingestion rate may be more important. A type of goal gradient effect was also found during acquisition as reflected in greater force on the goal than on the Operant panel. There was a suggestion that the develOpment of this effect is retarded by shifts in incentive and facilitated by large sizes of reward. The results of the extinction stage indicated a signif- icant decrease in force of responding which was not function- ally related to incentive size. It was also found that an 8 sec. ITI significantly retarded the development of inhibition of the operant response. The opposite was true for the goal response. It was suggested that this differential effect of ITI on force of responding was a function of the response invigorating properties of the traces of frustrative nonreward which preceded the Operant response. The purpose of Experiment 2 was to test an implication of the results of Experiment 1, namely, that the force of Daniel F. Tortora the response following nonreinforcement will increase if the time between nonreinforcement and the performance of the next response is short enough. A second aim was to deter- mine if a relation exists between size of incentive and force of responding during acquisition and extinction using a more sensitive within subjects design. Ten male hooded rats were trained to panel push using a procedure analogous to Amsels double runway paradigm. Each g had to push two panels in sequence with food reinforcement (i.e., a single 45, 97, 190, 300, or 500 mg pellet) behind the first panel (61) and water reinforcement behind the second panel (62). Incentive manipulations such as shifts in incentive and nonreinforcement occurred at 61. The effects of these manipulations were assessed at G2. All §3 received training and extinction with all five incentive sizes. The order was counterbalanced using a Latin square design. The results indicated a significant increase in force on GZ and a significant decrease in force on 61 due to nonrein- forcement. These performance changes reached an asymptote after 11 nonreinforced trials. Consistent with Experiment 1 size of reward was not significantly related to performance during acquisition or extinction on either 61 or G2. It was concluded that size of incentive, as menipulated in these experiments, is not related to the FE. It was further sug- gested that ingestion rate, independently manipulated, might be used to great advantage to elucidate the relation between incentive value and FE. THE EFFECT OF INCENTIVE SIZE ON RESPONSE AMPLITUDE DURING ACQUISITION AND EXTINCTION BY rut! L15 Daniel F. Tortora A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1973 DEDICATION To Angie, Dawn and Danny ii ACKNOWLEDGEMENTS I would like to express my sincere appreciation and thanks to Dr. M. Ray Denny, my advisor and thesis director, for his sound direction and guidance for the 5 years of my graduate career. His example has served as a model of both an Experimental Psychologist and Scientist. I would also like to thank Dr. Stanley Ratner for his guidance and advice. He is primarily responsible for developing my interests and expertiSe in Comparative Psychology. Together these men have developed in me a specific orientation for the study of psy- chology in general and to my chosen area of interest, learn- ing. I will probably maintain this orientation throughout my career as a psychologist. Thanks are also due to Dr. Mark Rilling, Dr. Robert Raisler, and Dr. Lester Hyman for their service on my disser- tation committee. iii TABLE OF CONTENTS Page LIST OF TABLES v LIST OF PLATES vi LIST OF FIGURES _ vii INTRODUCTION 1 Magnitude of Reward and FE 3 FE Demonstrated in Other Types of Apparatus 7 EXPERIMENT I 11 Hypotheses 12 Method 13 Results 27 Discussion 48 EXPERIMENT II 56 Introduction 56 Method 59 Results 64 Discussion 75 REFERENCES 79 APPENDIX A 82 APPENDIX B 84 iv LIST OF TABLES Table Page 1. Summary table of studies investigating the 4 frustration effect (FE) as a function of size of incentive experienced in Gl or G2. All the studies summarized in this table have used rats as subjects, a double alley apparatus, and a between subjects (BS) or within subjects (WS) design. 2. Summary table for the analysis of variance of 33 force of responding for the reward shift stage of Experiment 1 with size of reward, days of training, and type of response (operant vs. goal) as independent variables. The dependent variable is the mean force for the first half of each day of training. 3. Summary table for the analysis of variance of 37 force of responding for the first day of ex- tinction of Experiment 1 with size of reward, intertrial interval (30 vs 8 sec.), type of response (Operant vs goal), and trials as independent variables. 4. Summary table for the analysis of variance of 42 force of responding for the extinction stage of Experiment 1 with size of reward, intertrial interval, (30 vs 8 sec.), type of response (operant vs goal), and fifths of extinction (blocks) as independent variables. 5. Summary table for the analysis of variance of 67 force of responding for the extinction sessions of Experiment 2 with size of reward stage of testing (Er vs Enr), and type of response (G1 vs G2) as independent variables. 6. Summary table for the analysis of variance of- 70 force of responding for the extinction sessions of ExperimentZ with size of reward, quarters of extinction (blocks), and type of response (G1 vs G2) as independent variables. LIST OF PLATES Plate Page 1. The outside View of the apparatus used in 14 this experiment. In the foreground are two pellet dispensers. On the side of each dispenser are wooden frames used to hold blowers that served to reset the force transducing wheels. Directly in front of each dispenser are the three sided enclosures which contained the recessed plexiglass food cups. The S was placed into the apparatus by opening a plexiglass door mounted on the tap. 2. An inside view of the push-panel-wall as 15 photographed from the rear of the chamber. Part of the three sided enclosure is visible on the right side of the photograph where a panel door has been prOpped Open. The light fixture illuminating the vicinity of the panel doors can be seen in the top left side of the photograph. 3. Plate three depicts a close up view of the force 16 transducer (tOp left corner) and three sided enclosure (bottom middle). There is a clear view of the panel door with the metal rod used to translocate the force to the wheels (center). As depicted the force transducer is reset and ready to be activated. When 8 pushes the panel an electromagnet (top centerT would draw the rod back allowing the wheel freedom to move. 4. The front view of the sound attenuating chamber 21 used in this experiment. At the tOp center of the photograph can be seen the plexiglass window and the mirror which allowed an unobstructed view of S during running. vi LIST OF FIGURES Figure Page_ 1. Mean output of the transducer (T-units) as a 19 function of applied force in grams for the left and right panel doors. The hatch marks above and below the points express the var— iance of measurement at each point. The line was fitted visually. 2. Relative frequency distributions of the force 29 of the Operant (closed circles) and goal (closed triangles) responding. Columns represent pro- gressive stages of the experiment and rows re- present different groups. The median of each distribution is depicted by the Open points on the graphs. 3. Mean force of operant (A) and goal (B) respond— 34 ing during reward shift as a function of size of incentive and days of training. The mean force on the last day of preshift and the first day of extinction is represented by the floating points on the left and right of the figures. 4. Mean force of Operant (A) and goal (B) respond- 40 ing during extinction as a function of size of incentive and fifths of extinction. The floating points on the left side of the graph represent mean force of responding on the last day of reward shift. 5. Mean force of responding as a function of fifths 45 of extinction, type of response and intertrial interval. Each point represents the mean for all five sizes of reward. 6. Mean median force of responding as a function size 65 of incentive, type of response (G1 vs G2) and stage of testing (i.e., reinforced (Er) vs non- reinforced trial (Enr). vii Figure Page 7. Mean median force of responding during 71 extinction as a function of type of response (G1 vs 62), size of reward and quarters of extinction. The floating points on the left of the figure represent mean median force for the 6 reinforced baseline trials given during the extinction session. 8. Mean median force of responding during ex— 73 tinction as a function of type of response (G1 vs GZ) and quarters of extinction. Each point represents the mean for all five sizes of reward. A-l Force of operant responding during Experiment 82 l as a function of trials in one day collapsed over all ten days of reward shift with reward size as a parameter. B-l Mean latency of responding to both G1 and GZ 84 during the acquisition and extinction stages of Experiment 2 as a function of size of reward. viii INTRODUCTION Research Specifically related to the effect of individual nonreinforcements on instrumental responding was initiated by Amsel and Roussel (1952). It was already known that an in- strumental response followed by a series of nonreinforcements would decrease in vigor. Their study demonstrated that a response preceded by nonreinforcement or frustration can be performed more vigorously than a response preceded by rein- forcement. This frustration effect (FE) has been investigated rather extensively for the last 21 years using various types of apparatus and several experimental designs. The results of these investigations have led to some superficial incon- sistencies and some major methodological problems which will be described later. Amsel's theory of frustration (1958; 1962) which empha- sizes the energizing function of frustrative nonreward hasbeen asignificant contribution to the field. Thus it is not sur- prising to find that the apparatus Amsel first used to study FE (Amsel & Roussel, 1952) has become as standard for research in frustration as the Skinner box is for research on schedules of reinforcement. This apparatus is the double runway. It was designed so the rat can perform an instrumental response (alley running) immediately following frustrative nonreward. 1 The double runway consists simply Of two straight alley runways constructed so that the goal box (G1) Of the first alley (R1) also serves as the start box (52) Of the second alley (R2). The general procedure is as follows: S (i.e., a rat) is placed in the start box (81) Of the first runway. The start box (51) door is Opened allowing S to traverse the first runway (R1) and enter the first goal box (G1). Reinforcement versus nonreinforcement in G1 is the mani- pulation generating the frustration effect. After S has consumed the reinforcement or has been detained in G1 for a period Of time(during nonreinforced trials) the door in G1 is Opened allowing S to traverse the second runway (R2) and enter the second goal box (G2). Reinforcement is always available in G2. An increase in running Speed in R2 after nonreinforcement in G1 has been considered by Amsel (1958, 1962) as a demonstration Of FE. According to Amsel, (1952, 1958, 1962) nonreinforcement (after a series Of reinforce- ments) elicits primary frustration which adds to the general motivational complex increasing the vigor (speed) Of the re- sponse it precedes (i.e., running in R2). Unfortunately the analysis Of the frustration effect using the double runway is subject to major confounding, es- pecially if size Of incentive in G1 is manipulated. Seward SE 31- (1957) were the first to point out that faster speeds in R2 after nonreinforcement may be due to decreased speeds in R2 after reinforcement. This interpretation Of the"appar- ent frustration effect" was supported by a significant decrease in R1 speeds when his subjects were prefed 1000 mg or 500 mg of food before traversing the first runway, meaning that a faster R2 time when G1 is empty than when it contains food, Operationally defined as the frustration effect, may be due to demotivational factor. This result is a significant criticism Of any work with the double runway since it is difficult if not impossible tO unconfound demotivational and frustration effects when large incentives in G1 are used. Thus, since the amount Of demotivational confounding is posi- tively related to the reward size, this confounding prevents a clear determination Of the effect Of size Of reward upon the magnitude Of FE. In fact, Amsel (1958) has suggested and Wagner (1959) has demonstrated that small reward sizes (100 mg) do not produce detectable demotivational effects. This restricts double alley research to small sizes Of re- ward. Magnitude Of Reward and FE Given the preceding criticism it is not surprising that investigations Of the effect Of size Of reward on frustrative nonreward using the double runway have yielded inconsistent results. Unfortunately almost all Of the research relating reward size to frustration have used the double runway. Table 1 summarizes the procedures and results Of the double runway studies that have investigated magnitude Of reward. Out Of the nine studies presented, only three studies (Peckham & III-III, IIIII II III I.II1.I.I|~‘II.I .III .mamfiuu mz Oo mooomm mm OH OOOOHOMMHO OO mmB mo Ofl OHOOB .O m moumm OOO» .O m Hmumo umBOHm muwaaom me me m who: mooomm m macaw» m OO ..o.H .Ooflum> 0 OH muoaaoa Ammmav xoam>mo IHDOEOO OD oouoamu on on UOOOm mm: =mm= mm a m3 me me m no H a omomoz .mamfluu O OO muflm ou m0 Ofl muwaaom ooumaou maowfluflmmom mowedm mm .mHmHHu mz me me oa HOHm .H A OO wooedm m OOm Omen o>HuOOOOH Ooo3uon 0 OH An homav UOOOM mm3 OOHDOHOH OO pan OOsOm mm3 =mm= m3 muoaaom OE mv m .Ho no .HOOQOHHM .mm .m m Own» Ho30Hm m0 OH ouoslmm .m OH mamaup O OO .wamflmu mz Oo uwaaom me mv H mm umaama m was amen umpmmm mm umaama u an mumaamo in assay oa OOu spas Ouam ouoson ou oopmaou =mm= mm me me OH HO m .Hm um .OOOOOHHO .HO OH OOOHOMOHOH m0 OH um>OO mm3 umOu HOHDOOO m on ooumm muoaaom OS mm m 1800 mums mmsouw .mmsonm uoaaom H HO HO OH muoaaom 0 may no: use ma men an ounce was =ma: mm as em ma no G .H imomav mama No as muwaama me me m .OOHuHmflswom mOflHso Oo>flm o>HuOmOOH H0 Ofl muoaamm Amomav mo muwm Op omumamuOs mm; :mm: mm m8 mv oa no w .v .Hm no .uumuumm womb woNHm muOoEEOU OOo muasmom Omflmmo o>HuOOOOH mo omOmm Houomflumo>OH .Omflmoo Amzv muoomnsm OH IODHB no Ammv muooflnsm Omo3umn m OOm .mspmummmm modam OHQOOO m .muOOnnOm mm mumu pow: o>mO magma menu OH OONHHOEEOm mmflosum ecu Had .mw HO 0 Ofl UOOOOHuomxo O>HDOOOOH mo muflm wo OOHDOOOM o mo Ammv Doommo Oofiumnumsuw mg» mprmmfiumm>OH weepsum mo oanmu huoEEdm .H mqmde .OOOOO OOHHOO m on» How OOHum>HuoEoo mo OOOOOH>O mos OHOOH pan muHm o>HuOOOOH O» .mumo OEmm mOu uOomoumOH mmE moEHu DOOEOMMHO um OOm OHMOHOOH DOOHOMMHO OH OOOOHHQOO mOHvsum ozu omega « muwHHoO m me AM N O OH muOHHOO O EH, «AnomHv HomEO ooumHou mHuOOHHo.onou OOOOm mmz =mm= m3 OE hm m no N w Emnxomm mo EH .OOOHO OOHHOO m on» mom OOHum>HuoEoo HO muwHHom OE hm m OOOOOH>O mos OHOOD pan oNHm O>HuOOOOH on O OH muOHHom «AvmmHv HomEO nmumHmu mHuoouHo on on venom was gum. me me Hm m‘uo m a emexomm .coHu No EH um>HuoEOO 0» one mHnmnoum mums muHsmom muwHHom OE mv m AmomHv .O>HuOOOOH mo ONHm Op OODOHOH wHomuo>OH O OH mDOHHom OOmOH>OOH mm3 mm mo ONHm pan OOOom mos zap: mm OE mv OH no N a mmomoz I No EH .muOHme m OuHs muwHHom OE mv m N uOoEHuomxm cusp mDmHHma v gums “mummm mum; mHmHuu o EH mumHHmO AmmmHv mz OO mooomm m usn OOOOm mo3 :mm: m3 OE mv O HO .¢ .Ho um .omomoz mo aH meHHOQ OE mv N H DOoEHHOOxm .OOmHuomEoo HHm you mHMHuu mz OO Hmeo O OH muoHHom Hmmev mums mooomm mm .OOOOm HOO mos :mm: m3 OE me m no v .H .Ho no .Omomoz coma mONHm muOmEEOU OOO muHsmmm OOHmOo O>HuOo0OH wo OOOmm HoumOHpmo>OH .HUOSCHUCOOV H mqmfifi Amsel, 1964, 1967 and Krippner, et a1., 1967 a) have Obtained results demonstrating a positive relationship between size Of reward in G1 and the amount Of frustration measured in R2 on nonreinforced trials. It must be pointed out that this re- lationship is critical for the Amselian interpretation Of frustrative nonreward. The intensity Of the frustration ef- fect is postulated by Amsel to be related tO the intensity Of the fractional anticipatory goal response (rg-sg) which in turn is a function Of the size Of reward. The lack of covariation between FE and incentive size found in the majority Of studies presented in Table 1 can be best understood by a detailed look at McHose and Gavelek's (1969) study. This study has been chosen for this analysis since: 1. their results are representative Of the six out Of nine studies (see Table 1) not demonstrating a functional relationship between size Of reward and the frustration ef- fect; 2. all Of the studies in Table 1 have used either between-- or within- subject design; their study employed both designs. In MCHOSe and Gavelek's within subjects design, one group received differential reward conditioning in the first alley (R1 and G1) Of the double runway. Large reward (8 45 mg food pellets) was consistently associated with one stimulus, S+ (i.e., a black or white alley), and small reward (1 45 mg food pellet) with the other stimulus, S- (i.e., a white or black alley), prior to the omission Of reward, on test trials, in both S+ and S-. In their between subject design, different groups of S3 were trained on 8 45 mg (large) or 1 45 mg pellets. Their data demonstrate that the difference between large and small reward is not on the nonreinforced (N) trials but on the reinforced (R) trials where the 8 pellet condition yielded slower running speeds in R2 than the 1 pellet condi- tion. However, this difference cannot be attributed to frus- tration. The reduced speeds in R2 after large reward in G1 may be due to temporary satiation or to the other confounding aspect Of the double alley i.e., the inhibitory effects Of frustration in R that Obtain because S is reinforced with 2 only 2 pellets in G after being reinforced with 8 pellets 2 in G1. FE Demonstrated in Other Types Of Apparatus Increased vigor Of responding due tO nonreinforcement (FE) has been demonstrated using types Of apparatus Other than the double runway. Many investigators (Amsel, 1962; Amsel & Ward, 1965; Goodrich, 1959; Haggard, 1959; Wagner, 1961; Weinstock, 1954) have studied the effect Of nonre- inforcement using a single runway. In these studies rats receiving 50% reinforcement ran slower than a group Of rats receiving 100% reinforcement early in acquisition (12-30 trials). However, the partially reinforced group eventually caught up and reached a higher asymptotic running speed than the continuously reinforced group. This "cross-over" effect (PRAE) has been interpreted by Amsel (1958) and Others (Spence, 1960, Ch. 6; Wagner, 1961) as‘a demonstration of the invigorating effects Of frustration. They have suggested that the frustrative nonreward under partial reinforcement adds to the general drive level Of the organism. Whether this in- creased drive level will be inhibitory or excitatory relative to the continuously reinforced Ss depends on the response' elicited by frustration. Early in training the response pro- duced stimuli Of conditioned frustration tend to elicit re- sponses which are antagonistic to the approach response. At this point the partially reinforced Ss will run slower than the continuously reinforced Ss. As training continues, the instrumental running response becomes conditioned to these same anticipatory frustration cues (sf). Thus frustration no longer elicits antagonistic responses but continues to produce an increase in drive level. The net effect, late in acquisition, is an increase in the performance Of the par- tially reinforced Ss greater than that Of the continuously reinforced Ss. These investigators have also shown that the "cross-over effect" Occurs earlier as the distance from the site Of frustration (goal box) is increased. Thus, start measures are the first to show this cross-over (12 trials). Running times cross-over later and later in acquisition as they are measured closer to the site Of reinforcement. Goal times do not cross-over (i.e., the continuously reinforced group maintains its superiority over the partially reinforced group throughout acquisition). Notterman and Mintz (1965), using a free operant situation have demonstrated that the force of bar pressing increases when rats are switched to a decreased density of reinforce- ment. Thus increasing the value Of an F1 schedule or switch- ing to extinction may yield an increase in peak force of bar pressing. They present interesting but untestable evolu- tionary explanations for this phenomenon but appear gener- ally disinterested in investigating it. KOk (1971) investigated the effect Of size Of reward and nonreinforcement on the force and latency Of a panel push response. She found that nonreinforcement decreased the force and increased the latency Of this response. These changes in force and latency were not clearly related to the size Of reward experienced during acquisition. Such a de- crease in vigor Of responding is not in accord with many Of the previous investigations cited; however, it can be ex- plained. One difference between Kok's study and the previous studies reported relates tO where the measurements were taken. In the double alley studies increased vigor of re- sponding following nonreinforcement is Obtained in R2. The rat in this situation is running away from the site Of non- reinforcement and running toward stimuli consistently asso- ciated with reinforcement. In the Notterman & Mintz (1965) studies one finds an increase in peak force following non- reinforcement on a bar horizontally displaced 8 inches from the site Of reinforcement and nonreinforcement (i.e., the fOOd cup). In the single runway, an increased vigor Of 10 responding is Observed in the start measure which is maxi- mally distant from the site Of reinforcement and nonreinforce- ment (i.e., the goal box cues) and occurs early in the re- sponse chain. In the same apparatus, measures Of responding in the presence Of the cues directly associated with rein- forcement (i.e., goal speeds) show a decrement in vigor Of responding due to nonreinforcement. The empirical general- ization appears to be that nonreinforcement can either in— crease or decrease the vigor Of responding depending upon where the measure is taken. Measures Of responding close to the site Of reinforcement may show a decrement in vigor due to nonreinforcement. Measures distant (or distinct) from the site Of reinforcement may show an increment in vigor due to nonreinforcement. EXPERIMENT I The present experiment combined the measures Of Operant force (Notterman & Mintz, 1965) and goal force (KOk, 1971) in a single study. This allowed for the energizing and inhi- bitory effects Of frustrative nonreward to be studied simul- taneously. TO facilitate such comparisons the force Of topographically identical responses was measured. This was accomplished by using an apparatus similar in conception to that Of Kok (1971) which employed a panel push response. This apparatus consisted Of a chamber with two panel doors located on one wall. An S had to push one door Open (the Operant re- sponse) in order to Obtain fOOd behind the other door (the goal response). The independent variable:was the size Of incentive used during acquisition Of the Operant and goal response. Of primary interest was the relation between the size Of reward and the changes in the force Of responding during extinction. It must be pointed out that the problem Of demotivation pre- viously described for the double runway apparatus should not exist for this experiment. The apparatus is analogous to a straight alley runway with the Operant response functionally similar to start measures and the goal response similar to goal measures. 11 Hypotheses Qperant and Goal Responses It has been previously pointed out that Operant re- sponse force increases (Notterman & Mintz, 1965), and goal response force decreases (KOk, 1971) due to nonreinforce- ment. Since the present study combined both measures, it was hypothesized that response force should increase from acquisition to extinction for the Operant door and decrease from acquisition to extinction for the goal door. The amount Of predicted change in the operant and goal responses from acquisition to extinction should vary directly with the magnitude Of reward used during acquisition. This hypothesis is based on Amsel's Theory (1962) Of frustrative- nonreward since the magnitude of reward is postulated to be directly related tO the intensity of the fractional antici- patory goal response (rg) which in turn is related to the amount Of frustration or Rf elicited by the removal Of the incentive (nonreward). 12 Method Subjects The subjects were 30 naive male, hooded rats from Windsor Biological Supplies, approximately 150 days Old at the be- ginning Of training. They were housed individually through- out the experiment in a cOlony room adjacent to the running room. This room was illuminated 24 hours a day by overhead fluorescent lights. Apparatus The apparatus consisted Of a testing chamber constructed Of .635 cm plate aluminum, 30.48 cm high, 53.34 cm long, and 30.48 cm wide. When Opened, two horizontally sliding alu- minum doors on one wall Of the chamber 22.86 cm apart, gave access to two identically constructed 8.89 cm by 13.34 cm galvanized sheet metal panel doors hinged at the top. Plates '1 and 2 depict the outside and inside view respectively Of this apparatus. Behind each panel door was a recessed plexiglass fOOd I cup 3.81 cm in diameter and 2.22 cm deep mounted on the floor Of a three sided enclosure (See Plate 3). This enclosure prevented egress from the apparatus when the panel was Opened and limited the swing Of the panel to 60 degrees from vertical. Inside the chamber were two light fixtures, one over each door, designed to direct illumination over the general area Of the door. There was also one light fixture behind each panel mounted on the far end Of the panel enclosure. In the 13 l4 I I I Blower- I I I I I I Plate 1. The outside view of the apparatus used in this ex- periment. In the foreground are two pellet dispensers. On the side of each dispenser are wooden frames used to hold blowers that served to reset the force transducing wheels. Directly in front of each dispenser are the three sided en- closures which contained the recessed plexiglass food cups. The.s was placed into the apparatus by Opening a plexiglass dOOf'mounted on the top. '15 {Enclo- Light Fixture :sure Enclo- sure~ Plate 2. An inside view of the push~panel-wall as photo- graphed from the rear of the chamber. Part Of the three sided enclosure is visible on the right side of the photo- graph where a panel door has been propped Open. The light fixture illuminating the vicinity Of the panel doors can be seen in the top left side Of the photograph. 16 Force Tranducer Electromagnet Housing Enclosure Plate 3. Plate 3 depicts a close up view of the force transducer top left corner) and three—sided enclosure (bottom middle). There is a clear view of the rear Of the panel door with the metal rod used to translocate the force to the wheels (center). As depicted the force transducer is reset and ready to be activated. When S pushes the panel the electromagnet (top center) would draw the rod back allowing the wheel free- dom to move. 17 center Of the wall Opposite the wall containing the doors was a hole 5.08 cm from the floor Of the chamber from which pro- truded a glass drinking tube that was attached tO a 100 m1 ‘ graduated cylinder. The floor Of the chamber was made Of .635 cm plate aluminum. On all four walls Of the chamber there was a 1.91 cm ledge, 5.4 cm from the plexiglass tOp. The force applied to the panels was transduced tO a digital output and recorded on Hunter Klockounters (Model number 120 A). This transduction was accomplished in two steps. First, the initial force applied to the panels was translocated (by metal rods) tO aluminum wheels mounted I above the panels causing these wheels to rotate (See Plate 3). Great care was taken to construct two identical wheels, axels and axel housings. The tolerance for all parts Of the apparatus connected with the measurement Of force was .0002 in. This resulted in two wheels that rotated at the slightest applied force (F411 gram) and whose output was identical. The wheels contained 20 evenly spaced .635 cm holes on their periphery and four evenly spaced .635 cm holes near the center. Above these holes were mounted two photo— cells, a selenium photogenerative cell on the periphery and a cadnium sulfide photoresistive cell above the central holes. Below these holes were mounted two lights. The photocells generated a stream Of electrical impulses as the wheels ro— tated(20 pulses per rotation) which were counted by the klockounters. This accomplished the transduction Of force tO a digital output. This transduction allowed the 18 Imeasurement of the initial force applied to the panels irrespective Of the distance through which the panel was moved. The functional relation between the amount Of applied force and the output Of the transducer is depicted in Figure 1. This function was generated by taking 1200 measurements, 200 at each Of six known forces and recording the output Of the transducer. The output of the transducer (T-units) is a linearly increasing function Of applied force. There was no more than a 2 per cent error across all levels Of applied force tested. Latency was measured by the activation of the Hunter Photorelays (model number 330 S) connected to the central photocells. Through appropriate electromechanical pro- gramming these photorelays started and stOpped Hunter Klock- ounters which measured latency to 1/100 Of a second. The apparatus was fully automatic. The vertical sliding aluminum doors were Operated by electric motors. FOOd pellets were dispensed automatically to the food cups by means Of solenoid Operated feeders. The force wheels were reset automatically by means Of blowers and electromagnets. The entire operation as the apparatus was programmed through electromechanical circuitry. The testing chamber was closed in a sound attenuating box constructed Of 1.91 cm plywood and two layers Of 1.27 cm Celotex. This box was 85.09 cm wide, 82.55 cm deep and 105.41 cm high and had a plexiglass window near the tOp (See 19 Figure 1. Mean output Of the transducer (T—units) as a func- tion Of applied force in grams for the left and right panel doors. The hatch marks above and below the points express the variance Of measurement at each point. The line was fitted visually. Output of Transducer (T-units) 240 220 200 ISO ISO I40 I20 IOO 80 60 4O 20 20 A Left Panel 0 Right Panel 1 L l L 5 IO I5 20 Applied Force in grams (F) Figure 1 25 30 21 Plexiglass window & mirror Plate 4. The front view Of the sound attenuating chamber used in this experiment. At the tOp center Of the photo- graph can be seen the plexiglass window and the mirror which allowed an unobstructed view of S during running. 22 Plate 4). Above the experimental chamber was a mirror mounted at a 45 degree angle which allowed a clear top view Of the inside Of the chamber through the plexiglass window. A speaker was mounted at the tOp Of the sound attenuating box for the presentation Of 80 db white masking noise from a Grason Stadler Noise Generator (Model number 910 B). Further masking and ventilation was supplied by a blower mounted on one Of the walls Of the box. A 78 watt red light bulb was mounted in back Of the 45 degree mirror 45.72 cm from the tOp Of the experimental chamber. This light provided diffuse indirect illumination during the intertrial interval. Procedure Upon receipt from the distributor, Ss were placed on ad libitum fOOd and water for at least two weeks before being placed on deprivation. During this time the animals were weighed and handled daily. At least three weeks before the start Of training Ss were placed on fOOd deprivation. Water was continuously available both in their individual cages and in the experi- mental chamber throughout the experiment. FOOd deprivation consisted Of giving Ss 3.5 grams Of Wayne Mouse Breeder Blox per day for the first week and from 10-15 grams thereafter. The fOOd given was adjusted daily to maintain S3 at 80% of their ad libitum weight. During the experiment all Ss were weighed and fed their daily ration approximately one hour after being run in the chamber. Three days prior to any experience with the experimental chamber Ss were given 22-45 23 mg Noyes pellets (990 mg) each day in their home cage along with an adjusted ration Of Wayne bloxs. The rest Of the procedure was divided into four distinct phases: Preliminary Training, Training, Reward Shift and Extinction. Preliminary Training This phase consisted of habituating S5 to the Operation of the experimental chamber, training S3 to Obtain food (Noyes pellets) from the fOOd cups behind the panel doors, and to press the panel doors open when they were presented. During preliminary training each S received an equal amount of exposure to each panel door. This phase was terminated and training begun when S successfully completed one full day (22 trials) with the panel doors completely closed i.e., S performed 22.panel pushing responses (11 per door) and Ob- tained 22 reinforcements on one day. This phase was com- pleted for all Ss within nine days. Training Once a consistent approach and panel push response was established to both panels, training was started. The object Of this procedure was to establish a sequence or chain Of responces in which S must push one panel in order to Obtain food (one 45 mg Noyes pellet) behind the second panel. The panel the S pushed first was analogous to the bar Of a Skinner box and was designated the Operant panel. The second panel, where food was Obtained, is analogous to the food cup 24 (or magazine) Of a Skinner box and was designated the goal panel. The goal panel was illuminated by the overhead light and the enclosure light. The Operant panel was not illumi- nated. In order to insure that intensity Of illumination was the only relevant cue signalling the apprOpriate response, the position of the goal panel (and the position Of the Operant panel) was varied from right to left using the fol- lowing 2 series: 1) LRLLRRRLLRLLRRLLLRRLRR 2) RLRRLLLRRLRLR: LLRRRLLRL. A different series was employed on alternate days. 22 trials were given each day. Reinforcement was one 45 mg Noyes pellet per trial. This chaining procedure ensured that Ss made consistent approach and panel-push responses to the Opening Of the slid- ing doors. Thus on any trial the sliding door in front Of the Operant panel was Opened first. When S pushed this panel, the sliding door in front Of the goal panel was Opened and a pellet of fOOd was delivered to the goal fOOd cup. The trail ended approximately 15 sec after the goal response or 2 min after the beginning Of the trial by the closing Of both sliding doors and turning Off the door and enclosure lights. The intertrial intervals was 30 sec. The next trial started with the Opening Of the Operant sliding door and the turning on Of the lights over the goal door and inside the goal en- closure. This procedure continued until S reached a latency criterion Of no more than two seconds for both the Operant and goal response two trials in a row, with no failure to respond on that day. When this criterion was reached Ss 25 were assigned tO one Of five reward shift groups, using trials-to-criterion as the basis for assignment such that each group was matched on mean trials-to-criterion. The elaborate procedure just described Was designed to prevent the occurrence Of nonreinforcement during acquisi- tion and can be considered analogous to errorless-discrimi- nation in an Operant situation. This procedure also ensured that the present experimental situation was analogous to straight alley runway situations, thus allowing at least rough comparisons with runway data. Reward Shift There were five independent groups (6 S3 per group). Each group was gradually shifted to a different size Of re- ward, except for one group which continued to be given one 45 mg Noyes pellet per trial. The other four groups were shifted tO a single pellet reinforcement weighing either 97, 190, 300, or 500 mg (per trial). The gradual shift tO larger sizes Of reward was accomplished over a single session Of 22 trials. This procedure was adopted because it was found with pilot Ss that an abrupt shift especially to the larger sizes Of reward (190, 300, and 500 mg) was inhibi- tory. The S5 simply had not learned the appropriate behaviors necessary for the removal Of the larger pellets from the food cup. The ITI, the maximum duration of a trial (2 min), and the time between Obtaining the pellets and the end Of the trial (15 sec) remained the same as in chaining. All Ss were run for 10 days, 10 trials per day. 26 Extinction During extinction no reinforcement was given. All five independent groups were further subdivided, half Of each group was extinguished under the original 30 sec ITI while the other half was extinguished under an 8 sec ITI. Other- wise there was no difference between reward shift and ex- tinction procedures. All Ss were run in the chamber until they reached a criterion Of 10 consecutive trials in a row on any one day without a response (i.e., Operant or goal) within a 2 min period. Results Acquisition The data were initially examined by plotting the rela- tive frequency (per cent) distributions Of force Of respond- ing for each Of the major stages Of the experiment. This measure has also been used by Notterman & Mintz (1965) in depicting force Of responding. It involves expressing the frequency Of each force category relative tO the total fre- quency of responding for each group at each stage. Figures 2-1 to 2-20 summarize this data on force Of responding for the major stages Of the experiment. First, one can see that force Of responding as measured in this experiment is approximately normally distributed.' This is probably due to the fact that the force measure was not sub- ject to the ceiling and floor effects that hold for probabili- ty and latency, respectively. When these two pages Of figures are scanned from left to right along any row (group) the change in the distribution Of force as a function of training can be seen. Viewed this way there appears tO be an increase in force across all groups as a function Of training. The distributions become more negatively skewed as training pro- ceeds from the last day Of preshift to the last half of reward shift. The effect Of prolonged training alone without a shift 27 28 in reward size can be seen by examining the 45 mg (nonshifted) group (see figure 2-1, 2-6, and 2-11). For this group train- ing served tO increase the difference between Operant (Md = 65.41) and goal (Md = 81.36) responding having its greatest effect on the goal responding. This could be considered an example of the development Of a goal gradient effect where the most vigorous responding occurs closest to the site Of reinforcement. This goal gradient effect is not as evident for the remaining groups. Although the goal response is usually pro- duced with greater force than the Operant response, the dif- ference between the distributions is not as striking as seen in the 45 mg group. Comparing the medians for Operant and goal force for each Of the four remaining groups across experimental stages, VH3 see a somewhat greater difference between these medians with the larger reward sizes (300 and 500 mg) than with the smaller sizes (97 and 190 mg) on the last half Of reward shift i.e., Figures 2-11 tO 2-15. This suggests that a shift in reward retarded the development Of a goal gradient but that this retardation was overcome more readily with the larger sizes Of reward. The relation be- tween size Of reward and force Of responding can best be seen in the last half Of reward shift, (see Figure 2-11 to 2-15). Examination Of the medians suggests an inverse re- lation. This inverse relation appears more pronounced for goal than for Operant force. . A more detailed statistical analysis Of the speculations 29 Figure 2. Relative frequency distributions Of the force of the Operant (closed circles) and goal (closed triangles) responding. Columns represent progressive stages Of the experiment and rows represent different groups. The median of each distribution is depicted by the Open points on the graphs. .- 5 __.4. .. Relative Frequency (Percent) 25 20 25 20 25 20 25 20 30 Last Day of Preshift (Chaining) [ 45mg group lid-69.07 | _ Md I 70. l2 1 1 1 l 1 l l l 1 F 97 mg group 2 ,_ MCI-66.54 Md .5425 I- I- ISO mg 3 9”” Mdl50.96 I' ind-56.93 l l l l l l l l i” 300 mg group 4 .. lid-56.65 l l l 1 65.5 85.5 I055 1 .1 l 5.5 25.5 455 l25.5 ' Force (T-units) Figure 2 Reward Shift Days l-5 F 45 mg group 6 .- ' MCI-58.25 F 97 mg group 7 I' Md - 7410 I ' m-easo 1 I l l l l l l l l I- l90 mg group 8 '- Md I 64.6 ”4.68.83 1 l L l l l 1 l l l l l T" 500 mg group " MCI-56.7 Md'55.5 o—o' OPERANT FORCE H can. FORCE I 5.5 25.5 45.5 65.5 55.5 I055 l255 l i = IO i I Relative Frequency (Percent) 31 Reward Shift Daye 6- IO First Day of Extinction 25 - 45mg group H I' 45 me group I6 20 ~ “.65.“ , I- rad-61.7 '5 _ _ rad-43.2: I0 ~ — 5 ~ . as f 97 mg group '2 E 97 mg group [7 Md I 57.50 _ l90 mg group I8 Md ' 47.43 20 - Md I 70.02 Nd I 74.46 Md I 50.50 Md 3 59." 20" r- 1 l 1 l l l J J l 1 L l I I I I 1 l 5.5 25.5 45.5 55.5 85.5 l05.5 I255 55 25.5 45.5 65.5 55.5 l05.5 l25.5 Force (T-unlte) it IO Figure 2 32 prOposed by the inspection of figures 2-1 to 2-15 was per- formed by 1OOking at the reward shift stage alone. This analysis is presented in Table 2. The mean force of respond- ing for the first half (5 trials) Of each day was used in the analysis to minimize the confounding Of demotivational and incentive size effects. Figure 3 A and B depicts these data for the Operant and goal responses, respectively, as a function Of reward size and days Of training. The effect Of reward size On force Of respOnding was not statistically significant (F = .3208). However, force in- creased significantly for both Operant and goal responses over days Of training (F = 6.6571, df - 9/225, pI‘uOOl). The goal force was significantly greater than the Operant force throughout reward shift (F = 9.2167, df = 1/25, p44.005) con- firming the suggesting Of a goal gradient effect proposed earlier. The significant reward size by type Of response by training interaction (F = 1.6657, df = 25/225, p.4.05) is probably due to the interaction Of the suppressive effects of a reward shift Egg E2 and the facilitative effects Of large reward upon the development Of the goal gradient. NO other interactions approached significance. Extinction The distribution Of force Of responding on the first day Of extinction can be seen in Figures 2-16 to 2-20. Goal force remained greater than Operant force the first day Of extinction with the 45 mg group showing the greatest dif- ference between these responses. There is also a decrease 33 mes MHmm.movoo cHauHs uouum mmoo.me m- eeme.mmmmH no Ame m uouum mo.eeH mam smoo.mommm O Ame m uouum ammo.HHe mm ~meo.msmoH o Ame m uouum mo.a. emoe.H mmem.me em omHo.mmeH com me sms~.H Hmem.mm m mmem.~mm no me ommm. eeom.meH mm omem.mmmm mm .m: mmse. mmom.emH e oeH~.mss om Hoe..v Heme.o mnem.~oHH m mHHm.mme0H low msmo moo..u osH~.m momH.omsm H memH.mmsm luv uncommom oem omHe.ooemm mm sHeuHs omee.memm mm omH~.ememm Ame m scene we moan. Hmmn.me~H e ~omo.mmom “my ouHm ouezom mm mmm~.oosmOH mm somsumm mam Nome.OOHmmH Hmuoa a O we no mm mosmHum> mo oousom may mH oHanuo> DOOOOOOOO OOH .OOHOHmuu mo moo some mo MHmn umHHm OOH Mom oouow OooE .OOHanHm> uOOOOOOOOOH mm HHOOO m> uOouomOv omOOOmou mo Oahu OOm .OOHOHmHu mo mmmo .oumsou mo ONHm OuH3 H uOoEHHOOxO mo OOmum “MHOm chosen on» How OOHOOOOmou mo oouOm mo OOOmHHo> mo mHmmHOOm on» How OHnmu mHmEEOm .m OHOOB 34 Figure 3. Mean force Of Operant (A) and goal (B) responding during reward shift as a function Of size Of incentive and days Of training. The mean force on the last day Of preshift and the first day Of extinction is represented by the float- ing points on the left and right Of the figures. Mean Operant Force (T-units) Mean Goal Force (T-units) 80 7O 60 50 4O 80 7O 60 50 4a 35 first day of extinction _ D .. 2 ' 3 5 O - o I 1 1 l 1 1 1 J 1 1 J A _ U A 3 8 H 45 mg group O—qa 97 mg group " o-—-O ISO mg group b—e 300 mg group ’ I———-l 500mg group 1 l l l l 1 l l l I last day of 2 4 6 8 preshift Reward shift, the first half of each day of training Figure 3 36 in force Of responding from reward shift to extinction. This is evident by the positive skewness of all the distributions during extinction. There does not appear to be any syste- matic relation between size Of reward during reward shift and any changes in the shape Of the frequency distribution during extinction. In order to Obtain a more detailed analysis Of the im- mediate changes in force Of responding the mean force Of responding for progressive blocks Of 4 trials on the first day Of extinction was Obtained for each subject. The first trial was treated separately since Ss had not experienced nonreinforcement before making responses on this tria1.Table 3 presents an analysis Of variance on these data. Consistent with the acquisition data reward size did not significantly affect force Of responding on the first day Of extinction (F = .2734). Surprisingly the effect of the duration Of the intertrial interval was also not significant (F = .0538). The interaction between reward size and ITI was also nonsignificant (F = 1.0714). The goal gradient effect seen during acquisition per- sisted through the first day Of extinction. This is evident from the fact that the goal force was significantly higher than the Operant force (F = 11.8975, df = 1/20, p L.005) and that the type Of response by trials interaction (F = 5998) was not significant. There was also a significant decrease in force Of responding over blocks Of trials (F = 16.1688, df = 5/100, p“.001) for both responses over all groups. 37 we seem. mws.om m aeHe.moe so H.A Hmsm.H emmm.mHm m oeem.emm~ 9H ms seem.H eeme.m~e H eems.m~e oH H..A mmem.H mMHm.~om om NmH~.mmOOH am we mMHN. eNeH.mm e emem.em~ om Hoe..u mmeH.oH eeem.mm~e m mmmm.Ha~H~ lac mHeHua moo:v memm.HH Hoo~.Hm~m H soo~.Hm~m loo emeoamom omm HmHe.meeem mm eHeon mooe.o~m~ om emoo.meeme HHmv m scene we eHse.H emeH.eme~ e mmme.eemm Hm me mmmo. omem.emH H omem.emH HHS HeH me emsm. ~eee.eme e memm.emm~ Ame esHm cusses mm mmmH.mHomm mm emmsumm mmm eHHm.ememmH Heuoe GUCMHHM> a a we on we we mousom .meHneHue> quOOemeOOH me mHeHuu OOe .AHeOO m> quuemov emOommeH mo emu» .Aoem m m> om. He>ueuOH HeHuuueuOH .oneseu mo eNHm OuHs H quEHuere mo OOHDOOHuxe mo moo umuHm eOu now OOHOOOOmeu mo eouOm mo eOOeHue> mo mHmmHeOe en» How eHneu wueEEsm .m eHneB 38 mmo.mo~ 0mm monm.mmmmv OHEHH3 Houum MHmm.OMH 00H mmNH.mvaH BO AHmv m m Houum hmmm.mwm OOH mmmm.mmmmm B AHmV m N uouum wmmm.mhm om mmhm.mmmm U AHmv m H Hounm mO memo.H hvmm.HvH om vvmm.mmmm BUHm moo..V mmm¢.v HmmN.mmm m mmvm.whmm BwH mO mmmm. mbmH.NHH on mhmh.mvmm 90m mO mom. mmmm.¢m o mmmh.mmm on moo..‘ ON¢5.N mmmm.mmh om mHhm.mmmv BHm eocmHHe> m m m2 NU mm m0 eouaom lemseHusooc m mHnee 39 This is not in agreement with the original prediction that nonreinforcement would lead to an increase in force on the Operant and decrease in force on the goal response. It ap- pears that, at least early in extinction, nonreinforcement- induced a general inhibitory effect on both responses. The only othersignificant effects were the reward by ITI by trials interaction (F = 2.7429, df = 5/100, p 4.001) and the ITI by type Of response by trials interaction (F = 4.4228, df = 5/100, p 4.005). Visual inspection Of the data related to these interactions produced no psychologically meaningful interpretation. The force Of responding over the entire extinction stage Of the experiment was also examined. It should be re- called that each S was run to an extinction criterion of 10 Itrials without a response. This is a stringent criterion requiring an S to go 20 minutes making a single response. Days to extinction criterion varied from a low Of 2 days tO a high Of 24 days. Thus, for each S the total trials to the extinction criterion was divided into fifths and then the mean force for each S at each Of these fifths was calculated. These means were used as the data points for subsequent ana- lysis. Figures 4 A & B depict force Of the operant and goal response respectively over progressive fifths Of extinction as a function of size of reward. Table 4 presents an analysis Of variance Of these data. Neither size Of reward (F = .46) nor intertrial interval (F = .0189) produced significant results. As can be seen 40 Figure 4. Mean force of Operant (A) and goal (B) responding during extinction as a function of size Of incentive and fifths Of extinction. The floating points on the left side Of the graph represent mean force Of responding on the last day Of reward shift. Mean Operant Force (T- units) Mean Goal Force (T-units) 41 Ole A A I O 70*- Cl 60*- A 50- 40 - 30 l l I I l J B 0 80 I- H 45 mg group A H 97mg group o O-———O ISO mg group I A-—-A 300 mg group 70 I— A H 500 mg group I 60 r- ‘ O . - 50 - a ‘ ‘3‘ _ 40 I- 30 l 1 l 4 l A last day cry st 2nd 3rd 4th 5th reward shift Fifth: of Extinction Figure 4 42 Hoo.u_ m~H~.HH mmem.H~H e emme.eme me we omoe. SHmo.He e meom.eeH mH OH.A Hm~o.~ ommo.ms~ H. ommo.msm oH me omm~.H emee.mmH SH eemm.HHom mm me swam. emHm.HMH e mmem.m~m om Heo.nu HeHm.H~ momH.mH- e ~mee.mmmm .lmv.mxoon mmo..u oemH.e mess.emm H msee.emm lee mmsoammm osm smee.eHmmm mm sHeuHs oeHH.HHMH om momm.~eme~ leV m sense we eHme.H Heme.smmH e mmem.mmme Hm me mmHe. Hmeo.e~ H Hmeo.em HHV HHH me some. Hess.mme e mome.mmm~ Ame mem.euesmm mm mHeH.mmmsm mm awesome mam OEHe.ee~ee Hence eOOeHHe> m m m: mo mm mo eOHOOm m> omv .He>ueuOH HeHHuueuOH .meHneHue> quOOeOeOOH me AmeOHQV OOHOOOHuxe mo mOuwHw OOm .HHeOO w> quuemov emOommeH mo emmu .Aoem m .oueseu mo eNHm OuHs H quEHHere mo eOmum OOHDOOHuxe eOu How OOHOOOOmeH mo eOHOm mo eOOeHHe> mo mHmeeOe eOu How eHOeu wueEEsm .v eHneB 43 .omH mmmm.mmmHH OHOuH3 uouum mmnm.0H om mHno.o>m mo AHmv m uounm mmmv.HOH om OHmm.mHHm m AHmV m Houum mmmm.VMH om «www.mmmm w AHmV m Honum Hoo.u. bmmm.m vmm>.mm OH evMH.omm mem mo.u. mmmm.m mMHn.>m v mmmm.0HH me mmo..v mmem.m momm.mm mH mnmm.mov mum mO momm. enmm.omH e hmmo.Hmv on mO mva.H mmeH.mHH OH ovmm.mmmH mHm eOOeHue> m m m2 mo mm mo eousom AemeeHusoov e mHsee 44 ‘from the figure, however, goal force was significantly greater than Operant force (F = 6.184, df = 1/20, p 4.025). This dif- ference decreased over progressive fifths of extinction with both responses decreasing in force tO the same asymptote (F = 21.8141, df = 4/80, p L.001). It is no surprise therefore, that response type interacted significantly with progressive fifths of extinction (F = 11.2128, df = 4/80, p.4.001). With Duncan's multiple range tests, goal force was found to be significantly different from Operant force on only the first and second fifths Of extinction (df = 80, p.A.05). The initial prediction Of a differential change in force Of responding as a function Of type Of response was parti- ally supported by the significant ITI by response type by fifths Of extinction interaction (F = 2.598, df = 4/80, pt. .05). This interaction is depicted in Figure 5. As can be seen from the figure, force Of Operant responding decreases more rapidly with a 30 sec ITI than an 8 sec ITI. With a Duncan's multiple range test, the difference between Operant force at the 8 and 30 sec ITI was significant at the fourth fifth of extinction (df = 80, pIL.05). For the goal re- sponse the reverse is true. The 8 sec ITI caused a faster decline in goal force than the 30 sec ITI. With a Duncan's multiple range test the difference between goal force at the 8 and 30 sec ITI was significant at the second and third fifths of extinction (df = 80, p L.05). This interaction suggests that frustrative nonreward has Opposing effects de- pending upon where it occurs. 45 Figure 5. Mean force Of responding as a function Of fifths of extinction, type of response and intertrial interval. Each point represents the mean for all five sizes Of reward. 46 60 — —o—— Operant- 8 sec III + Operant- 30 sec III -—O-— Goal - 8 sec 111 + Goal -30sec III .‘2 'E 3 I l.— ” I o h- 0 LL c: 8 2 40%- l l l l J lsf 2nd 3rd 4th 5th Fifths of Extinction Figure 5 47 There were two other significant interactions found. They were the size Of reward by type of response by fifths Of extinction interaction (F = 2.3465, df = 16/80, p L.025) and the fourth order interaction (F = 3.5637, df = 16/80, p 4.001). Graphic inspection Of the data related to these interactions produced nO interpretable relations. The latency for both Operant and goal responses were also recorded throughout the experiment. The mean latency and speed (i.e., the reciprocal Of the latency) for each stage Of the experiment was graphed in the same way as the force measure. Visual inspection of the graphs did not show any systematic differences. Thus, the latency measures were not subjected tO further statistical analysis. Discussion Acquisition Pubols (1960), after intensively reveiwing the litera- ture on the effect of incentive magnitude manipulations on performance, concluded that "Asymptotic performance is an increasing function Of incentive magnitude." A more recent review (Dunham, 1968) leads one to the same conclusion. Thus there appears to be ample evidence to justify Pubols conclu- sion. In the present experiment, however, there was a non- significant inverse relatiOnship between reward size and per- formance. One possible explanation Of this result is that the grad- ual shift in incentive allowed the S to "adapt" to the new reward size, eliminating any performance changes, but a study performed by Wike (1970) makes this suggestion untenable. Using an L—shaped runway he increased reward size gradually over trials from one to four pellets. This resulted in an increase in running speed which was monotonically related to the size Of reward. Another possibility is that the required response (panel pushing) is not sensitive to the incentive manipulations used in this experiment. This possibility can not be completely ruled out, but since other expected effects were demonstrated using this measure, it is an unlikely explanation. Ss showed 48 49 a significant increase in performance over days as well as a significant goal gradient effect. They also showed a signi- ficant decrease in performance over trials during extinction. This would lead one to suggest that panel pushing was sensi- tive to incentive manipulations. A third possibility is that the massing Of trials (ITI = 30 sec) caused a certain degree Of satiation. This suggestion tOO, can not~ be completely eliminated but is made less plausible by two facts. First, the use Of the mean force for only the first five trials Of each day should have reduced the influence Of demotivation. When mean force Of responding for all ten days of Reward Shift is plotted against trials within a day (see Appendix A-l) one can see that there is no demoti- vation for all groups except for the last 5 trials Of a day in the 500 mg group. Second, when the force on the first trial Of each day is used as a measure essentially the same relations are seen. This measure, however, was deemed too unstable for further analysis. The last possibility relates tO the interpretation Of changes in the performance measure. Perhaps a decrease in force of responding represents an increase in performance. This argument has been presented by DilOllO, Ensminger and Notterman (1965) to explain results similar to those Of the present study. Using a small range of sizes Of reinforce- ment (20 to 100 mg) they found that force Of bar pressing was inversely related to reward magnitude. This result and inter- pretation was replicated by Notterman & Mintz (1965, p. 207). 50 The crux Of their argument is that "the greater amount Of reinforcements, the better the animal learns to make the requisite cutaneous and kinesthetic discriminations" (Notterman & Mintz, 1965, p. 210). Their force criterion for reinforcement was 8 grams and their Ss were usually re— sponding considerably above this level (eg. 18 grams for 20 mg group). Thus increased reward magnitude tended tO cause their Ss to perform more in line with the force criterion. The problem with the application Of this "discriminative" interpretation to the data in the present study is that force Of responding significantly decreases over extinction. For the discrimination hypothesis to be tenable: 1. force Of re- sponding should decrease over training as the cutaneous and kinesthetic discriminations are learned and 2. force Of re- sponding should increase during extinction as these same dis- criminations break down. Notterman and Mintz (1965) have Obtained results in accord with the two above predictions. It must be concluded that force Of panel pushing is not directly related to the magnitude Of the incentive as mani- pulated in this study. Perhaps the size Of a single incen- tive is not the important variable. In the past (Pubols, 1960), most studies have manipulated incentive magnitude by varying the number Of pellets Of a constant size, and Kling (1956) pre- sented some correlational evidence which suggests that with rats, rate Of ingestion rather than incentive size is the sig- nificant variable. He found that with rats in a straight alley, fast ingestors ran faster than slow ingestors to the 51 same amount Of water. [Deaux (1973) is the first published study to actually manipulate ingestion rate. Rats were given equal volumes Of water, half given the incentive at a fast rate (over a period of 1 sec) and half given the incentive at a slow rate (over a period Of 5 sec). The high rate group was significantly superior to the low rate group (i.e., higher asymptotic per- formance) in a classical conditiOning paradigm as well as in an instrumental runway paradigm. Ancillary to these findings was the fact that performance was inversely related to the number Of consummatory responses (licks) made. In the pre- sent study, ingestion rate was uncontrolled. Perhaps larger pellets induce a slower rate Of ingestion. This hypothesis deserves further investigation. Extinction It was stated in the introduction that decreases in the density Of reinforcement generally lead to increases in the force Of Operant responding. This had been amply demonstrated by Notterman and Mintz (1965), but the question remains as to why this increase occurs. A motivational theory such as Amsel's (1968) which relies on the prOposed response invigo- rating properties Of frustrative nonreward could potentially explain this phenomenon. A discrimination hypothesis such as the one prOposed by Notterman and Mintz (1965) which stresses the response shaping properties Of reinforcements can explain this result equally well. The main emphasis Of this discussion is to see whether 52 these hypotheses can be reconciled with the apparently con- tradictory results of the present study. It should be re- called that nonreinforcement produced a decrease in force of responding on the first day of extinctionas well as through- out the entire extinction stage. Contrary to prediction, this decrease in force was apparent for both the operant and goal responses. The discrimination or response shaping hypothesis sug- gests that reinforcement causes S to produce responses in line with the criterion for reinforcement by sharpening the apprOpriate kinesthetic and prOprioceptive discriminations. Nonreinforcement causes the breakdown of these discrimina— tions formed during acquisition. For this discrimination hypothesis to hold in this present experiment the decrease in force during extinction representing the breakdown in kinesthetic discriminative control should necessarily be preceded by a progressive increase in force during acquisi- tion, where the kinesthetic discriminations are formed. Both the increase in force during acquisition and the decrease in force during extinction occurred in this experiment. Although the force requirement to open a panel was minimal (Fl-l gram) there could have existed another contingency that caused S5 to progressively increase their force. Since Ss were Opening the panels and receiving reinforcement early in training with a much lower force it would be untenable to suggest that Ss were approaching a high force criterion during acquisition. The only other hypothesis comes from observing Ss making the 53 response. To attain the reinforcement Ss must not only push the panel Open (this they usually do with their nose) but also hold the panel Open with the top of their head while retrieving the pellet. One might speculate that the neces- sity to sustain the panel push response over this period of time could be responsible for the progressive increase in force. This suggestion is supported by results reported by Notterman & Mintz (1965, p. 36). They found that reinforcing S for a longer than normal duration (.8 sec) of bar pressing caused a concommittant increase in force of responding as well as an increase in duration. In addition they found that any criterial manipulation that caused S5 to progressively increase their force of responding during acquisition yielded a decrease in force during extinction (Notterman & Mintz, 1965, Ch. 3). Thus it could be concluded that the decrease in force of responding during extinction in the present experi- ment was caused by the breakdown of discriminative control due to nonreinforcement. The discrimination hypothesis does not however explain another significant finding of the present experiment, that is, the interaction between the type of response, fifths of extinction and intertrial interval. It should be recalled that the Operant response force decreased much faster when S5 were extinguished under 30 sec ITI than under an 8 sec ITI, while force for the goal response decreased much faster for the 8 sec ITI than for the 30 sec ITI. This is understand- able if it is realized that the operant response directly 54 follows nonreinforcement at the goal during extinction. That is, on a trial during extinction S first performs the operant response then is nonreinforced at the goal and then, 8 or 30 sec later, must again perform the Operant response. Presumably the 8 sec ITI allows for less decay of the effect of frustra- tive nonreward than the 30 sec ITI. This has two effects, a momentary invigorating effect of primary frustration and an increase in the build up of conditioned frustration which is generally inhibitory. According to Amsel (1968) if the stim- ulus trace of primary frustration is present while S is per- forming an instrumental response this would lead to the invig- oration of that response. It can be postulated that for the 8 sec ITI enough excitory primary frustration was remaining to retard the development of inhibition for the Operant response. However, since the goal response is not directly preceded by nonreinforcement, frustration would be dissipated by the time the goal response is performed. Thus, only the build up of in- hibition can effect the GR, producing the more rapid drop in force with the 8 sec ITI. This suggestion is partially sup* ported by the results of a study reported by Scobie and Fallon (1972). Using the reinforcement ommissions procedure of Staddon (1966), they found that increases in rate of responding oc~ curred up to an 8 sec ITI, decreases in rate of responding oc- curred with longer ITIs. The implication of the present moti- vational analysis is that roughly a 0 sec ITI should produce an increase in force on operant responding and decrements in force on goal responding. Thus the increments of force of responding found by Notterman and Mintz (1965) using a free 55 operant situation could be due to the very short interval between nonreinforcement and the next bar press which is characteristic of the free operant situation. The two interpretations of incentive manipulations can only be separated by further research, manipulating the time between the occurrence of nonreinforcement and the occurrence of the next response. Summary of Results and Conclusions 1. No significant relation was Obtained between magnitude of incentive and force of responding. It was concluded that pellet size may not be a main incentive variable and that other variables uncontrolled in the present study such as ingestion rate may be of more importance. 2. A type of goal gradient effect was found as reflected in greater force on the goal than the operant panel, and there was a suggestion that the development of this effect is re- tarded by shifts in incentive level and facilitated by large sizes of reward. 3. The significant increase in force found during acqui- sition and decrease in force during extinction was supportive Of the discriminative interpretation of Notterman & Mintz (1965). However, the differential effects of different inter- trial intervals on the force of Operant and goal responding during extinction was more easily interpreted with a motiva- tional hypothesis (Amsel, 1962). EXPERIMENT II Introduction The results of the previous experiment indicated that force of responding decreases with nonreinforcement, and that the rate of this decrease seems to be related to the type of response being measured and the length of the inter- trial interval. The purpose of the present experiment was to test an implication of these results, namely, that the force of the response following nonreinforcement will in- crease if the interval between nonreinforcement and the per- formance of the next response is short enough. A second aim of the present study is to determine if a relation exists between the size of incentive used during acquisition and the force of responding during extinction. A procedure analogous to Amsel and Roussel's (1952) double runway paradigm was adopted. As in the double runway and in Experiment I Ss performed two responses (panel pushes) in sequence. This experiment was different from Experiment I in that reinforcement was presented behind each panel. Food reinforcement (Noyes pellets) was presented behind the first panel (G1). Water reinforcement was presented behind the sec- ond panel (G2). It was also different from Experiment I in that incentive manipulations such as shifts in incentive size and 56 57 nonreinforcement occurred at the first panel. The effects of these manipulations, i.e., invigoration of responding due to nonreinforcement, were assessed on the second panel. This allowed for roughly a zero sec interval between incentive manipulations and the performance of the next response instead of an 8 or 30 sec interval as in Experiment I. This experiment differed from the double runway paradigm in two ways. First, water instead of food reinforcement was used at G2 to reduce the effects of demotivation present in double runway paradigms. This procedure was adopted from a study similar to the present study (Levine & Loesch, 1967). As discussed in the introduction to Experiment I, demoti- vational confounding occurs in the double runway since per- formance post reinforcement is compared with performance post nonreinforcement. With the double alley paradigm, it is usu- ally difficult to untangle the suppressive effects of previous reinforcement from the later invigorating effects of nonrein- forcement (Seward, et a1., 1957). Given the positive cor- relation between food and water intake (Adolf, 1947; Bolles, 1961) it can be predicted that the larger the size of food reinforcement behind the first panel (G1) the greater is the response to water behind the second panel (G2). This is opposite to demotivational counfounding. Second, size of incentive was manipulated within subjects. All Ss received all sizes of incentive. This was done since it was observed in Experiment I that there existed large but stable individual difference in force of responding. This 58 large variability might have masked of the effects of the size of incentive in the first experiment. Method Subjects Ten naive, male hooded rats from Windsor Biological Supplies, approximately 120 days old at the beginning of training,served as subjects. The Ss were housed individ- ually in a colony room adjacent to the running room. The room was illuminated 24 hours a day by overhead fluorescent lights. Apparatus The apparatus used was the same as used in Experiment I with only two modifications. First, the panel doors and the enclosures behind them were covered with Contact, an adhesive shelving material. The right panel and the interior of the right enclosure were covered with smooth white Contact. The left panel and the interior of the left enclosure was covered with black textured Contact. Second, a water dispenser was added to the apparatus. The water dispenser consisted Of a solid cylinder of plexiglass 3.75 cm in diameter and 2.1 cm deep. This cylinder fitted snugly into the food cups. A graduated L-shaped hole was drilled into the cylinder termi- nating at the top and center of the cylinder with a small V—shaped depression .40 cm in diameter and .15 cm deep. The rest of the water dispensing devise consisted of a two gallon 59 60 resevoir placed two meters above the water dispenser; surgical tubing to deliver the water to the dispenser; Skinner solenoid, valves and adjustable stopcocks to meter the flow of water. With the apprOpriate arrangement of these parts a single drop of water could be dispensed to the depression at the center of the cylinder. The apparatus was arranged so that a Noyes pellet was delivered to the food cup behind one panel and a single drop of water was dispensed to the cylinder placed in- side the food cup behind the other panel. The glass tube used to deliver water in Experiment I was remOVed. Procedure All Ss were maintained on ad libitum food and water and placed on food deprivation as in the first experiment. After Ss had stabilized at 80 per cent Of their ad lib. weight and ten days before the start of training Ss were also placed on 23.5 hours water deprivation. All Ss were maintained at these levels of food and water deprivation throughout the experiment. The rest of the procedure was divided into three phases: Preliminary Training, Training, and Reward Shift. Preliminary Training At the start of preliminary training Ss were haphazardly 5). One group was trained to assigned to two groups (n find food behind the black panel and water behind the white panel. These conditions were reversed for the other group. Initially all Ss were magazine trained with the 61 apprOpriate type of reinforcement in each food cup. The Ss were then shaped to Open the appropriate panel doors for both food (one 45 mg Noyes pellet) and water (one 1 m1 drop) re- inforcement, respectively. This procedure was continued until Ss made 10 reinforced panel pushes (i.e., 5 reinforced by food and 5 reinforced by water) in one 30 min session. Care was taken to insure that all Ss had equal eXperience with the food and water panels throughout this procedure. Training Once consistent approach and panel pushing responses were established to both food and water reinforcement, training was started. The purpose of this training procedurewas to establish a sequence or chain of responses in which S would first Open the food panel and then open the water panel. The food panel was analogous to the first runway and goal box of a double runway apparatus and was designated G1. The second panel reinforced by water was analogous to the second runway and goal box of a double runway apparatus and was designated G The enclosure light was illuminated behind each panel. 2. However, the overhead light was on only over G1. A typical trial started with the opening of the sliding door in front of G1 simultaneous with the delivery of'a single 45 mg Noyes pellet to the food cup and the illumi- nation of the light above G1. When S pushed this panel, the sliding door in front of G2 was Opened and a single drOp of water was delivered to the water dispenser. The trial ended 62 approximately 15 sec after the G2 response was made or two min after the beginning of the trial by the closing of both sliding doors and the extinguishing of the overhead and en- closure lights. The intertrial interval was 30 sec. All Ss were given 11 such trials per day for five days before reward shift was started. Reward Shift During this phase of the experiment all Ss received a series of five acquisition-extinction sessions under each of five sizes of reward: a single 45, 97, 190, 300, or 500 mg. Noyes pellet. An acquisition session consisted of three days, ten trials per day under the appropriate size of reward in G Each acquisition session was followed on the next day 1' by an extinction session. This consisted of six reinforced trials (Er) followed by 22 nonreinforced trials (Enr) on G1 in one day. Water (one 1 m1 drOp) was continuously avail- able behind G during both acquisition and extinction ses- 2 sions. The order of administration of the acquisition- extinction sessions under the five sizes of reward was counterbalanced using a Latin square design. 'Two subjects were haphazardly assigned to each of 5 different orders, one S trained with a black G1 and the other trained with a white G1. The intertrial interval (30 seconds), the maximum dur- ation Of a trial (2 minutes) and the time between performing a G2 panel push and the end of a trial (15 seconds) remained the same as in training. 63 As in the first experiment, the force for both G1 and G2 responses was measured throughout the experiment. The latency Of the G response was the time between the 1 beginning of the trial and the performance of the first (G1) }panel push response. The latency of the G2 response was the time between the performance of the G1 response and the per- formance of second (G2) response. Both latencies were auto- matically recorded to 1/100 of a second on Hunter Klockounters. It must be pointed out that the G latency was more indica- 2 tive of time it took S to ingest the Noyes pellet obtained behind G than of actual response time. Thus latency was not 1 analyzed in this experiment. Results Figure 6 depicts the mean median force of responding over the six baseline reinforced trials (Acquisition-Er) and the 22 nonreinforced trials (Extinction-Enr) given on each extinction session as a function of size of reward and type of response. It was expected that performance on these six reinforced trials which immediately preceded the extinction trials would serve as the most accurate baseline from which to compare changes in performance due to nonreinforcement. Mean median force was used as the measure of performance because it appeared to be more stable than mean force. However, essentially identical results were Obtained using either measure. Mean median force was calculated by obtaining for each subject the median force of responding for each acqui- sition and extinction session. Data points plotted represent the mean of these medians. An analysis of variance of the results plotted in Figure 6 is presented in Table 5. As can be seen, force of respond- ing on the first panel in the sequence (G1) reinforced by food was significantly greater than force of responding on the second panel reinforced by water (G2) both on reinforced (Er) and nonreinforced (Enr) trials (F = 38.4720, df = 1/9, p 4.001). Unfortunately, the confounding of type of rein- forcement and sequence Of responding does not allow any 64 65 Figure 6. Mean median force of responding as a function size of incentive, type of response (G vs 62 ) and stage Of testing (i.e., reinforced (Er) vs nonreiniorced2 trial (Enr)). Mean Median Farce (T- units) IOO 90 80 7O 60 5O 4O 66 H G. Acquisition (Er) H 62 Acquisition (Er) .-—Q G. Extinction (Enr) A—-‘ 62 Extinction (Enr) e \\ , \ / \ / \\\ / ‘————J A \\ \ f \ / \ / 15-“""‘T.N\h ./ ‘~‘\ /' \1 L l l L i_] 45 97 I90 300 500 Size of Reward (mg) Figure 6 67 Hmom.om Gm mmoo.Hmm~ e x um x m x m names as mHae. meme.am e ~omm.smH o x um x m aams.amH a ~ame.~meH o x um x m uouum Hoe..u mHme.mm mmam.e~mm H mmam.e~ma o x um «Hmo.~e~ em mHm~.mmea o x m x m names as Heam. MHHs.mmH e ~mem.-e o x m emmm.amH om mmmH.emem um x m x m poses we seem. H~m~.~a e ~moo.amm um x m aeam.oom a mamm.mOHm o x m nouns Hoe..q owae.mm mmem.eeeem H mmee.seaem HOV emsoamem moaH.~o~ a -s5.aHmH um x m uouum OH..A masm.m momm.m~e H momm.m~s lame mmmum aHmH.eem em meme.sseo~ m x m uouum as macs. mmmH.moe e ~aoa.~meH HOV eNHm nausea osmm.mGMH a NMHo.~e-H Amy muoeHnam aaH oome.emmMHa Hauoe eOOeHHe> a a we we me no mousom .meHbeHHe> quOOemeOOH me ANU m> HOV emOommeu mo emwu OOe .HuOm m> umv OOHueeu mo eOeum .oueseu mo eNHm OuHs m quEHHere mo mOOHmmee OOHuOOHuxe eOu How OOHOOommeu mo eOH0m mo eOOeHHe> mo memHeOe ecu Mom eHneu hueEEsm .m eHbea 68 definitive interpretation of this difference, though from Experiment I it would seem that type of reinforcement is more important. Consistent with the first experiment no significant re- lation between force of responding and incentive size was found (F = .7072). When G1 and G2 forces were analyzed separately the relation still remained nonsignificant (F = .5499 and .7505 respectively). Thus size of incentive as manipulated in this study produced statistically nonsigni- ficant results. This is of particular interest since "incentive size" was manipulated within subjects. This type of design should be especially sensitive for two reasons. First, it removes stable individual differences as a source of error variance. Second, the exposure to many sizes of incentive should maximize any effects present, such as, con- trast effects due to incentive shifts (Ehrendfreund, 1971). Nonreinforcement in the present study yielded opposite performance changes depending on where the effects were measured. When frustrative nonreward immediately preceded the response to the G2 panel push there was a significant increase in force of responding (FE) above the reinforced G2 baseline (F = 32.5411, df = 1/9, p L.OOl). When frustra- tive nonreward did not directly precede the response, as with the G panel push, there was a significant decrease in force 1 Of responding below the reinforced Gl baseline (F = 16.0849, df s 1/9, p 4.005). These effects yielded the significant goal by stage interaction (F = 58.4218, df = 1/9, p 4.001). 69 To examine the progressive effects of extinction, the nonreinforced trials of each extinction session (approxi- mately 22 trials) were divided into four quarters for each animal. The median force at each quarter of extinction was used as a measure. Figure 7 depicts the force Of responding at G1 and G2 over progressive quarters of extinction with reward size as a parameter. Table 6 presents an analysis of variance of these data. As can be seen from this table, there was a significant goal (G1 vs G2) (F = 9.9291, df = 1/9, p L.025) and stage by goal interaction (F = 9.5144, df = 3/27, p L.001). These results can be more clearly seen in Figure 8. From this figure it can be seen that G2 force is higher than G force throughout extinction. The goal by stage inter- 1 action was probably due to the drOp in force on G1 and the increase in force on G2 from the first to the second quarters of extinction. The functions appear to be parallel beyond the second quarter. Using the Duncans multiple range test the difference between operant and goal force is significant at the first quarter but is not significant at subsequent quarters (df = 27, p.4.05). Thus it appears that the invi- gorating (G2) and inhibitory (G1) effects of frustrative nonreward reach asymptote at about 11 trials. There was also a significant reward size by stage inter- action (F = 2.1054, df = 12/108, p 4.05). Visual inspection of this interaction yielded no psychological meaningful re- lation. All other effects produce S's less than one. 70 emea.mo~ meH Hamm.mmm- m x m x m mouse as NHHa. Hmmm.s~H NH HHme.ommH u x m x m. omma.omH Hm mama.eame o x m x m uouum Hoe..u eaHm.a moem.mHeH m momm.amHm o x m meom.ome am memo.HaemH o x m x m house as Hmoe. aame.eem a mama.~aoH o x m Hemm.em~ mOH maeo.mosom m x m x m house mo..u emOH.N emmm.mam NH mmoe.maHH m x m eaoa.aemH a aomm.om~HH o x m house mmo. u Hama.a ease.mam~H H eaaa.mammH loo maeoammm Hmaa.aom Hm aesa.mHmm m x m house as emaa. aame.aom m mmHH.mHa Amy axoon mamm.memH em omas.eHmme m x m scene as mmam. omHm.ams e ooem.aHa~ Ame aNHm euasam a meea.aemom Amy muooHnsm mam amma.meameeH Hauoe eOOeHHe> m m m: mo mm mo eousom .meHceHue> quOOemeOOH me Hum w> HOV emOommeH mo emhu OOe .AmxooHcv OOHuOOHuxe mo mueuueOO .Oueseu mo eNHm cuH3 HH quEHuere mo mOonmem OoHuOOHuxe ecu How OOHOOommeH mo eouom mo eOOeHue> mo mHmeeOm ecu How eHceu mueEEOm .m eHces 71 Figure 7. Mean median force Of responding during extinction as a function Of type of response (G1 vs G2), size of reward and quarters of extinction. The floating points on the left of the figure represent mean median force for the 6 rein- forced baseline trials given during the extinction session. 72 Cl 45 mg A 97 mg 0 ISO mg ’5‘ A 300 mg -l- I 500 mg Ck 80 - QS‘ IIR 70- \\ .3. \ , h :11 ' . \ 60 ,_ \ i )\ I Mean Median Force (T-units) I ‘. . / \’ e O O .4}. 50 - 41' A _°_ + :2: 4a - 1 1 1 1 4 Acquisition (Er) lst 2nd 3rd 4th \L J/ r Extinction (Enr) by quarters Figure 7 73 Figure 8. Mean median force of responding during extinction as a function of type of response (G vs G ) and quarters of extinction. Each point represent; the fiean for all five sizes of reward. Mean Median Force (T- units) 90 80 7O 60 50 4O 74 ~11 6, Ext (Enr) Ga Ext (Enr) 6. Acquisition (Er) 62 Acquisition (Er) \+_. l 1 l 1 it Acquisition (Er) \lLst 2nd 3rd 4th. I Extinction (Enr) by quarters I Figure 8 Discussion The force of responding during acquisition on G1 in the present experiment was similar to the goal force measured during acquisition in the first experiment. Thus the results of the first experiment which employed a between subjects design have been replicated using a more sensitive within subjects design. There was no significant relation during acquisition between response force and incentive magnitude in either experiment and the nonsignificant decreasing re- lation between force and incentive magnitude seen in the first experiment was not seen in Experiment II. Thus it must be concluded, at least for force, an amplitude measure of response, that the actual physical size of the incentive is not relevant variable. I It has already been pointed out that most studies that manipulated incentive size have done so using number of pel- lets (Pubols, 1960; Dunham, 1968). For the most part, these studies have found an increasing monotonic function relating incentive magnitude to runway performance. Using force of bar pressing as a dependent measure Notterman and Mintz (1965) found a decreasing monotonic function relating incentive mag- nitude (i.e., number of pellets) to performance. The differ- ence in the direction of the relation found when using bar pressing instead of alley running as an instrumental response 75 76 may be related to differences in the reinforcement criteria implicit in each response. In the straight alley faster running leads to a decrease in delay of reinforcement. In the bar pressing situation increased force of responding only leads to a greater expenditureof effort. Thus improving performance in an operant situation may mean a decrease in force of responding, to conform to the reinforcement criterion. Recent work with ingestion rate (Deaux, 1973) suggests that number of pellets may only indirectly cause performance changes. Deaux (1973) found that when equal volumes of water were given to rats at different rates, high ingestion rate gs were superior in performance to low ingestion rate gs. This occurred even when low ingestion rate §s produced more con- sumatory responses (licks) than high rate gs. The suggestion is that ingestion rate, an organismic variable, is the neces- sary condition for performance changes. Other incentive manipulations may have their effect indirectly by influencing ingestion rate. Preliminary data from our laboratory indicate that rats will consume 11-45 mg pellets (495 mg) approximately twice as fast as they consume a single 500 mg pellet. There is also a krger amount of intersubject variability in speed of consumption when gs consume a single large pellet than many small pellets. Thus, it is possible that neither size nor number of incentives directly affects performance, but do so indirectly by affecting ingestion rate to lesser or greater degrees. The second major finding relates to performance changes 77 due to nonreinforcement. In the first experiment it was found that extinction significantly decreased force of responding. This effect was replicated in the present experiment for G1 force.~ It was also suggested in Experiment I that changes in‘ performance during extinction were related to the interval between nonreinforcement and the performance of the next re- sponse. For the operant force (i.e., the response that directly followed nonreinforcement during extinction) an 8 sec ITI appeared to produce less decrement in force than a 30 sec ITI. This was interpreted as being due to the invi- gorating effects of the trace of frustrative nonreward. The implication was that a very short time between nonreinforce- ment and responding might lead to an increase in force of responding i.e., an FE, per Amsel (1952). Experiment II tested this hypothesis. It was found that G2 force increased from acquisition to extinction and that this increase asymptoted in about 11 trials. This is analo- gous to the frustration effect (FE) found traditionally in the double runway paradigms (Amsel, 1962). In accord most other studies there was no functional relation found between the FE and the magnitude of incentive (see Table 1). However, in previous studies any potential functional relation between incentive size and the FE was obscured by demotivational con- founding (eg. McHose & Gavelek, 1969). In the present study, no strong evidence could be found to support this demotiva- tional hypothesis (Seward et a1., 1957). To support this hypothesis, force of responding on G2 should have been a 78 monotonically decreasing function of incentive size at G1. That is, the larger the pellet consumed before pressing the G2 panel the greater the presumed demotivation and the less the vigor of responding (Force) on G2. This did not occur in the present experiment. There was no significant rela- tion during acquisition between size of incentive and force of responding on G2. It has been previously pointed out that a positive re- lation between size of incentive and the FE is critical to Amselian frustration theory. To date no study, including the present one, has demonstrated this relation. Previous stud- ies (see Table l) have manipulated incentive size by varying the number of.a constant sized pellet and the present study varied the weight of a single pellet. It has been suggested earlier that both these procedures may only indirectly affect incentive motivation by their effect on ingestion rate. Deaux (1973) has already demonstrated that ingestion rate can be independently manipulated. A procedure similar to his would have significant advantages over the previously mentioned procedures. First, since this variable is directly manipu- lated by E, the between subject variability in ingestion rate would be controlled. Second, when ingestion rate is varied the amount consumed can be held constant. This should com- pletely remove any confounding due to demotivation. Thus, it appears that future research varying ingestion rate di- rectly and holding amount consumed constant should have the greatest success in elucidating the relation between incentive magnitude and the FE. LIST OF REFERENCES LIST OF REFERENCES Adolf, E. F. Urges to eat and drink in rats. American Journal of Physiology, 1947, 151, 110-127. Amsel, A. The role of frustrative nonreward in noncontinuous reward situations. Psychological Bulletin, 1958, 2;, 102-119. Amsel, A. Frustrative nonreward in partial reinforcement and discrimination learning: Some recent history and a theoretical extension. Psychological Review, 1962, 69, 306-328. , . Amsel, A. Advances in frustration derived from-a variety of within-subjects experiments, in Spence, K.W. & Spence, J. T., (Eds.) The Psychology of Learning and Motivation. Advances in Research and Theory. Vol. 1, New York: Academic Press, 1967. Amsel, A. & Roussel, J. Motivational perperties of frustra- tion: I. effect on a running response to the addition of frustration to the motivational complex. Journal of Experimental Psychology, 1952, 1;, 363-268. Amsel, A. & Ward, J. S. Frustration and persistance: resistance to discrimination following prior experience with the discriminanda. Psychological Monographs, 1965, 19, No. 597. . Barrett, R. J. Peyser, C. S. & McHose, J. H. Effects of complete and incomplete reward reduction of a subsequent response. Psychonomic Science, 1965, 3, 277-278. Bolles, R. C. The interaction of hunger and thirst in the rat. Journal of Comparative and Physiological Psychology, 1961, E1, 580-584. Daly, H. B. Excitatory and inhibitory effects of complete and incomplete reward reduction in the doubel runway. Journal of Experimental Psychology, 1968, 76, 430-438. Deaux, E. Ingestion rate as a reward-magnitude variable in classical and instrumental conditioning in rats. Journal of Comparative and Physiological Psychology, 1973, E2, 308-315. 79 80 DiLollo, v. Ensminger, w. D. & Notterman, J. M. Response force as a function of amount of reinforcement. Journal of Experimental Psychology, 1965, 19, 27-31 Dunham, P. J.. Contrasted conditions of reinforcement: A "selective critique, Psyghological Bulletin, 1968, 69, 295-315. Ehrendfreund, D. Effect of drive on successive magnitude shifts in rats. Journal of Comparative and Physiological Psychology, 1971, 1g, 418-423. Goodrich, K. P. Performance in different segments of an instrumental response chain as a function of reinforce- ment schedule. Journal of Experimental Psychology, 1959, 57, 57-63. Haggard, D. F. Acquisition of a simple runningresponse as a function of partial and continuous schedules of reinforcement. Psychological Record, 1959, 3, 11-18. Kling, J. W. Speed of running as a function of goal-box behavior. Journal of Comparative and Physiological Psychology, 1956, 19, 474-476? Kok, G. L. Frustration Effect as a function of reward magnitude. Unpublished Masters Thesis, MSU, 1971. Krippner, R. A., Endsley, R. C. & Tacker, R. S. Magnitude of G reward and the frustration effect in a between-subjects désign. Psychonomic Science, 1967, 2, 385-386. Krippner, R. A., Endsley, R. C. & Tacker, R. S. Magnitude of 62 reward and the frustration effect in a between-subjects design, Psychonomic Science, 1967, 2, 399-400. Levine, G., and Loesch, R. Generality of response-intensity following nonreinforcement, Journal of Experimental Psychology, 1967, 1§,'97-102. McHose, J. H. & Gavelek, J. R. The frustration effect as a function of training magnitude: Within-and-between-gs designs. Psychonomic Science, 1969,11, 261-262. McHose, J. H. & Ludvigson, K.W. Role of reward magnitude and incomplete reduction of reward magnitude in the frustra-' tion effect. Journal of Experimental Psychology, 1965, 10, 490-495. McHose, J. H. & Meyer, P.A. & Maxwell, F. R., Jr. Frustration effect as a function of training magnitude in a within- S design. Psychonomic Science, 1969, ii! 137-138. 81 Notterman, J. M. & Mintz, D. E. Dynamics of Response. New York: John Wiley & Sons, Inc., 1965. Peckham, R. H. & Amsel, A. Magnitude of reward and the frustration effect in a within-subject design. Psychonomic Science, 1964, 1, 285-286. Peckham, R. H. & Amsel, A. Within-subject demonstration of a relationship between frustration and magnitude of reward in a differential magnitude of reward discrimi- nation, Journal of Experimental Psyghology, 1967, 13, 187-195. Pubols, B. H. Incentive magnitude: learning and performance in animals, Psychological Bulletin, 1960, 51, 89-114. Scobie, S. R. & Fallon, D. After effects of reinforcement and its ommission: ~A fresh look at time between trials: 13th Annual Meeting of Psychonomic Society, St. Louis, Missouri, 1972. Seward, J. P., Pereboom, A. C. Butler, B. & Jones, R. B. The role of prefeeding in apparent frustration effect. Journal of Experimental Psychology, 1957, 54, 445-450. Spence, K. W. Behavior Theory and Learning. Englewood Cliffs, N.J.: Prentice—Hall, 1960. Staddon, J. E. R. & Innis, N. K. An effect analogous to "frustration" on interval reinforcement schedules. Psychonomic Science, 1966, 4, 287-288. Wagner, A. R. The role of reinforcement and nonreinforcement in an "apparent frustration effect." Journal of Experimental Psychology, 1959, 51, 130-136. Wagner, A. R. Effects of amount and percentage of reinforce- ment and number of acquisition trials on conditioning and extinction. Journal of Experimental Psychology, 1961, 62, 234-242. Weinstock, S. Resistance to extinction of a running response following partial reinforcement under widely spaced trials, Journal of Comparative and Physiological Psychology, 1954, 31, 318—322. ' Wike, E. L. & Atwood, M. E. The effect of sequences of reward magnitude, delay, and delay—box confinement upon runway performance, Psychological Record, 1970, 20, 51-56. APPENDICES APPENDIX A 82 Figure A-l. Force of operant responding during Experiment I as a function of trials in one day collapsed over all ten days of reward shift with reward size as a parameter. MEAN OPEBANT FORCE (T-units) 83 80 - [3 45 mg 15 9? mg 0 190 mg 75 .. , 4 300 mg I 500 mg A ’ o o ‘ 3 7 _ o 2" A O 65 - A at \’ AM 60 - 55 - 50¢- // ’1’ L 1 1 g 1 1-2 3-4 5-6 7-8 9-10 TRIALS Figure A-l APPENDIX B 84 Figure B-l. Mean latency of responding to both G1 and G2 during the acquisition and extinction stages of experiment 2 as a function of size of reward. Mean Latency (sec) 60 50 4O 30 20 IO 85 o—-o iIF--Hli o--o A- —A 6. Acquisition (Er) 6; Acquisition (Er) G. Extinction (Enr-I) lst half 62 Extinction ( Enr- I) lst half 6. Extinction (Em-2) 2nd half 62 Extinction (Em-2) 2nd half 97 l90 300 500 Size of Reward (m9) Figure B-l nrcnxcnn STATE UNIV. LIBRARIES iiilHWIWiiiHlNilll"llWI]iiillllllilllilillllil 31293103962506