Animal Learning & Behavior 19808(2)282-286 An investigation of ambigos-ce learning in pigeons GEOFFREY HALL University ofyork York YOJ 5DD England Two experiments demonstrated that pigeons can solve a simltaneos discrimination in which on half the trials the positive and an ambigos ce (A) are presented and on half the trials choice is between A and the negative stimls. n Experiment 1 where a relatively nondistinctive A ce was sed performance on the former type of trial was sperior to that shown on the latter. n Experiment 2 where a distinctive A ce was provided this pattern of reslts was reversed. These findings are interpreted in terms of an approach-avoidance explanation first proposed by Leary (1958). Experiment 3 tested and confirmed a central prediction of this explanation by showing that in an orthodox simltaneos discrimination occasional reinforcements of the negative stimls prodce less accrate performance than do nonreinforcements of the positive. Leary (1958) trained rhess monkeys on a simltaneos discrimination in which three stimli were sed. On half the trials the sbjects had to choose between the positive stimls (P) and a nonrewarded stimls (A); on the remaining trials choice was between the negative stimls (N) and a rewarded stimls which again was stimls A. This problem has been called an ambigos-ce problem since the vale of ce A depends pon the stimls with which it is paired. The monkeys solved the problem in that they developed a preference for the rewarded stimls on both types of trial. Performance was however better on NA trials than on PA trials a reslt which may be abbreviated NA>PA. This reslt has been fond in several other experiments sing primate sbjects (Boyer & Polidora 1972; Boyer Polidora Fletcher & Woodrff 1966; Fletcher & Garske 1972; Fletcher Grogg & Garske 1968). These reslts stand in marked contrast to those prodced by a nmber of other experiments on ambigos-ce learning in primates. Fletcher and Bordow (1965) and Thompson (1954) fond performance to be better on PA trials than on NA trials (PA>NA); Boyer and Polidora (1972) and Boyer et a1. (1966) also report experiments in which this pattern of reslts was fond. The chief featre distingishing these experiments from those prodcing the NA>PA reslt lies in the natre of the stimli sed. n Leary's experiment for example the stimli were the "jnk objects" often sed in the Wisconsin General Test Apparats. Thompson's (1954) experiment on the This work was spported by a grant from the U.K. Science Research Concil. thank E. Macphail and S. Channell for their comments. Reqests for reprints may be sent to G. Hall Department of Psychology University of York Heslington York YO! 5DD England. other hand gave sbjects a choice on each trial between two foodwells next to one of which was a stimls plaqe either P or N stimls A being simply the absence of a stimls plaqe. The other experiments that prodced the PA>NA reslt sed as stimli the white lids covering the foodwells; the P and N stimli were distinctively striped and colored bt the lid serving as the A stimls was left blank. n general those experiments which sed a separate and distinctive A stimls prodced the NA> P A reslt; those that did not prodced PA>NA. This generalization has been tested by Boyer and Polidora (1972) and Boyer et a1. (1966). These experiments first demonstrated the PA>NA reslt with a stimls set lacking a distinctive A stimls and then went on to show that NA>PA is fond when sch a stimls is provided. The aims of the experiments reported here are: first to show that pigeons can solve what seems at first sight to be a complex discrimination problem (Richards 1973 has provided a preliminary demonstration of this); second to try to generate in pigeons both the PA>NA reslt and the NA>PA reslt by maniplating appropriately the natre of the stimli; and third to provide an experimental test of the most poplar explanation for these findings an explanation which acconts for performance in this relatively complex sitation in terms of simple conditioning principles (Berch 1974; Leary 1958). EXPERMENT 1 n this experiment pigeons were trained on an ambigos-ce problem with stimli that it was thoght wold prodce the PA>NA reslt. A plain red field was sed as the ambigos ce; white stripes differing in orientation were sperimposed on this red backgrond to prodce the P and N stimli. The Copyright 1980 PsychonomicSociety nc. 282 00904996/80/020282-05$00.75/0
AMBGUOUS CUE PROBLEM 283 red backgrond therefore bore the same relationship to the featres distingishing the P and N stimli as did the white lid of the food well in the experiments with monkeys. Sbjects and Apparats. The sbjects were seven pigeons which had previosly been sed in a stdy of free-operant discrimination learning. They had been atoshaped to respond to a key illminated with white light and had formed a discrimination between a steady and a flashing hoselight. No hoselight was sed in the present stdy. The birds were maintained throghot at 80llfo of their freefeeding weights. They were trained in a pigeon apparats on one wall of which were three response keys each 2 ern in diameter. The central key positioned above a grain feeder and 20 em from the floor cold be lit from behind with white light. The side keys were at the same level and were 8 ern (center to center) from the central key. Behind each side key was an in-line projector which cold displaya red field (the R stimls) or the red field with a single white line (2.0 x 0.2 em) sperimposed. Lines rnning horizontally (H) vertically (V) and the 45 deg obliqe (0) were sed. This apparats was monted inside a sond-attenating chamber. Procedre. The birds were given daily sessions of 60 trials. The interval between trials was always 20 sec and reinforcement consisted ofaccess to grain for 4 sec. On the first day of pretraining each trial consisted of the presentation of the white center key; a single response to this key trned off the light and prodced reinforcement. On the next day the sbjects were pretrained to respond to the side keys. Each trial again began with the illmination of the center key; a peck to this -key prodced a stimls on one of the side keys and a response to this side key trned off the stimls and was followed by reinforcement. Responses to nlit keys were withot programmed conseqences. The stimls displayed on the side key was R H or V. Each was presented 20 times and appeared eqally often on the left and on the right. The schedling of the stimli was determined by modified Gellerman seqences (Fellows 1967). There followed a test session designed to assess the extent to which each of these three stimli was preferred with respect to some other. On 20 trials therefore the birds were given a choice between H and the 0 stimls on 20 trials between V and 0 and on the remaining trials between Rand O. These trials were organized in the same general way as previosly except that two side keys were presented simltaneosly and a response to either trned off both and reslted in reinforcement. The next 10 sessions comprised training on the ambigos-ce problem. For all the sbjects R was ce A H was ce P and V was ce N. Half of each day's trials were PA and half were NA; within each set of trials A appeared eqally often on the left and on the right. A response to the appropriate side key (P or A on NA trials) prodced reinforcement bt a response to the other key simply prodced the lo-sec dark interval. Response to either side key trned off both key lights. On the first session of training on the ambigos-ce problem the mean score for the grop was 50.4% correct responses; by the last session this score had risen to 59010 sggesting that learning proceeded only slowly. These overall means obscre a marked difference in performance on the two types of trial. This is shown in Figre 1 which makes it apparent that PA performance was sperior to NA performance throghot. An analysis of variance carried ot on the data shown in the figre prodced a significant 80 70 U e 0 60 c: CJ & 50 40 0---0 PA...--. NA.0..0--- 0 " ' 0 / ' 0... / / '0 "'0..."""0- _.()---o Figre 1. Mean percent correct on the ambigos-ce problem in Experiment 1. PA trials with positive and ambigos ces; NA trials with negative and ambigos ces. effect of trial type [F(l6) = 9.28 p <.05] and a marginally significant effect of sessions [F(954) = 2.0 p <.1]. The interaction between these two factors was not significant. t is also apparent that the main effect of training is to raise the level of performance on the NA trials which started at below the chance level (50%). On the first session of training for of the sbjects scored below 50% on NA trials n contrast all bt one of the sbjects scored above 50010 on PA trials dring this session. This preference for stimli inclding the white line (i.e. the H and V stimli) over the red stimls was not apparent on the test session given before training began. On the trials where choice lay between Rand 0 R was chosen on 46% of occasions. This grop mean score is representative of the performance of the individal sbjects. Three showed no preference on these trials and of the remainder the most extreme preference was shown by a sbject who chose R rather than o on 7 of the 20 trials with these stimli. The other test trials also failed to reveal any marked preferences; H was chosen overall on 48% of the H vs. 0 trials and V on 51% of the V vs. 0 trials. The failre to find a preference for the striped stimls over the plain red stimls on the test session (given that sch a preference was present on the first training session) may mean either that the test session provided an insensitive measre or that the preference developed very rapidly dring the first training session and was therefore not present dring the test session. At any rate it was the existence of this preference dring training that prodced the PA>NA reslt. The extent to which this and other PA>NA reslts can be flly 5 Sessions 10
284 HALL explained in terms of stimls preference effects will be discssed after the next two experiments have been described. EXPERMENT 2 The stimli sed in Experiment 1 were intended to parallel those sed in experiments with monkeys which prodced PA>NA. n the present experiment a distinctive A stimls was spplied in the hope of reversing the otcome of the previos stdy and prodcing NA>PA. Horizontal and vertical stripes were again sed as the P and N stimli bt there was no red backgrond and a white cross on a dark backgrond was sed as the A stimls. The sbjects were for pigeons that had previosly learned a free-operant sccessive discrimination between key lights differing in color. After pretraining to respond to the white center key and to H.V and the cross (x ) on the side keys they received 12 sessions of training on the ambigos-ce problem. For all sbjects H was the positive V the negative and x the ambigos ce. There were no test sessions. The x stimls was prodced by simltaneos presentation of the 45- and 135-deg obliqe lines. Procedral details not specified here were the same as in Experiment. With only for sbjects it is appropriate to present the performance of each on the two types of trial. Figre 2 shows that all sbjects rose above chance level on both the PA and the NA trials. On the NA trials scores started at abot 50010 and rose (rapidly for three birds) to 90% or better. Performance on the PA trials was less accrate; it started below 50% for three sbjects and never reached the high level shown on NA trials. These reslts show that pigeons can solve the ambigos ce problem and prodce the NA>PA effect. They confirm the findings of Richards (1973) who sed a training procedre very like that sed here and green red and orange key lights as the stimli. 100 cloo e ~ a.! '. ' PA NA?~.0- ~o--o- -o~ -o~/ 0 p.j Ses srons Figre 2. Percent correct for individal sbjects in Experiment 2. PA trials with positive and ambigos ces; NA trials with negative and ambigos ces. An explanation for reslts of this sort (i.e. those showing NA>PA) can be developed from a relatively simple set ofassmptions abot approach and avoidance responses (see Berch 1974; Leary 1958). f we assme that a rewarded response to a stimls prodces a tendency to approach (and in this case peek at) that stimls and that a nonrewarded response prodces an avoidance tendency it follows that sbjects will come to approach P and to avoid N. f the sbjects receive a roghly eqal nmber of rewards and nonrewards for responding to A it seems likely that the overall reslt will be an approach tendency. The effects of the rewards may be assmed to otweigh those of the nonrewardsanimals will after all learn new responses and maintain performance nder a 50010 partial reinforcement schedle. Ths on NA trials the animal mst choose between a negative stimls and one that it has some tendency to approach (stimls A); performance shold therefore be qite good. Performance on PA trials will be inferior since here a tendency to approach A will be a handicap. The animal is in effect faced on these trials with a choice not between positive and negative bt between positive and more positive. f we are to accept this explanation it will clearly have to be extended or modified to encompass those experiments that prodce the PA>NA reslt. Bt before considering this isse it seemed worthwhile as a first step to try to sbject the central assmption of the explanation to experimental test. EXPERMENT 3 This experiment was designed to test the critical assmption that nderlies the approach-avoidance explanation of the NA>PA reslt: the assmption that the positive effects of rewards in discrimination learning will otweight the effects prodced by the same nmber of nonrewards. The procedre adopted was to train birds on an orthodox (P vs. N) simltaneos discrimination ntil they were reliably responding to stimls P rather than stimls N. They were then given training with jst one of the stimli. Birds in one grop (Grop P) were allowed to respond to the P stimls bt responses were not rewarded. Birds in Grop N were reqired to respond to the N stimls for the same nmber of trials and responses to this stimls were rewarded. The effects of these procedres on performance of the simltaneos P-N discrimination was noted. t was anticipated that nonrewards for responding to P in Grop P wold detract little from the approach tendency governed by this stimls and that accrate performance wold ths be maintained by this grop on the P-N task. On the other hand rewarded responses to the N stimls shold according to the approach-avoidance theory markedly increase the approach tendency elicited by this stimls in Grop P rendering accrate performance difficlt on the P-N task.
AMBGUOUS-CUE PROBLEM 285 The apparats was that sed in the previos experiment. The sbjects were 12 pigeons again maintained at 80010 of their freefeeding weights. Since these birds had previosly learned a sccessive free-operant discrimination between lines differing in orientation colored key lights were sed as the stimli. These were plain red and orange fields. Richards (1973) has shown that the NA>PA reslt is fond when stimli of this sort are sed in an ambigos-ce problem. Preliminary training was carried ot as in the previos experiments except that on the 2nd day red and orange were displayed on the side keys. All animals then learned a simltaneos discrimination between red and orange with red as the positive stimls. Red and orange were presented on all trials the positive appearing eqally often on the left and on the right. n other respects the training procedre was identical to that sed in Experiments and 2. Each sbject was trained to a criterion of 10 sccessive correct responses at which point the session was terminated atomatically. Three frther (overtraining) sessions each of 60 trials were given. The sbjects were then divided into two grops. Both received eight frther sessions with 30 of their 60 daily trials being on the red-orange discrimination. The grops differed in the treatment they received on the remaining trials of the session. Grop P received 30 trials intermixed with the red-orange trials on which jst the red stimls was presented. t was shown 15 times on the left and 15 times on the right the other side key remaining nlit. Response to the red stimls was not rewarded. Grop N received similar intermixed single-stimls trials on which only the orange stimls was presented. Response to this stimls reslted in reward. Pilot work has shown that pigeons that have been trained on a simltaneos discrimination will always peck a single illminated side key and that they will do so (althogh often with a lengthened latency) even when the lit key has previosly been established as a negative stimls. Throghot training responses to nlit keys were withot programmed conseqences. t will be noted that this procedre of intermixing red-orange and single-stimls trials makes the present experiment a close parallel to the ambigos-ce problem itself. For Grop P nonrewarded experience with the red stimls on the single-stimls trials trns the red-orange discrimination into a task reqiring choice between a negative stimls and a stimls that is sometimes rewarded and sometimes not. n this way the red-orange task becomes eqivalent to an NA sbproblem. n an analogos fashion the red-orange task becomes the eqivalent of a PA sbproblem for Grop N. Figre 3 shows the performance of the two grops on the overtraining sessions on red-orange and their performance on this problem after the introdction of single stimls trials. The grops did not differ dring overtraining on red-orange. An analysis of variance carried ot on the overtraining data presented in the figre showed a significant effect of sessions [F(220) = 3.55 p <.05] bt no significant effect of grops [F(iO) = 3.05 p>.] and no significant interaction between these two factors (F < 1). Grop P maintained its level of performance when singlestimls trials were introdced bt Grop N showed a marked decline in performance. Over the eight sessions of training in these conditions the grops differed significantly [F(liO) = 12.59 p <.01]. There was a significant effect of sessions [F(770) = 5.06 p <.01] and the interaction between sessions and grops was also significant [F(770) = 4.42 P <.01]. The approach-avoidance accont of ambigos ce learning is spported by these reslts. Nonreward 100 - R-O training 90 0 a: c 0 80 U ~ 0 70 C < Q; 60 a.. 50 o O. f...0 Single stimls trials added o ---- Grop P 0----0 Grop N Sessions Figre 3. Mean percent correct on red-orange (R-O) discrimination. Grop P received added nonreinforced trials with the positive stimls; Grop N received added reinforced trials with the negative. for reponse to the positive (which occrs in this case with roghly the same freqency as reward) does not distrb performance on a simltaneos discrimination. Bt eqivalent reward for response to the negative does distrb performance. By this measre therefore reward proves to be a mch more effective (in this case disrptive) procedre than nonreward. GENERAL DSCUSSON The first two experiments establish that pigeons can solve an ambigos-ce problem. They frther show that PA performance is sperior to NA performance when there is no distinctive A stimls bt that NA performance is sperior to PA performance when a distinctive A stimls is available. The reslts of the third experiment lend spport to an accont of ambigos-ce learning that predicts the NA>PA reslt. t remains therefore to determine if this accont can be extended to deal with experiments prodcing the PA>NA reslt sch as Experiment 1 of this report. One possibility reqires the approach-avoidance theory to be modified hardly at all. f we adopt the plasible assmption that primates have a tendency to respond to a distinctive stimls object rather than to an nmarked foodwell it follows that PA performance wold be helped and NA performance hindered. This might be enogh to otweigh the factors that tend to prodce the NA>PA reslt. Boyer et a. (1966) who fond a PA>NA reslt tested some sbjects on PA alone and others on NA alone before transferring them to the ambigos-ce problem itself. They fond that althogh the PA problem was learned readily the NA problem was not. Ths a natral
286 HALL preference for a ced over a nonced foodwell seems enogh to explain their reslts. Other experiments prodcing PA>NA cannot be immediately explained in this way. Thompson (1954) tested stimls preferences by training a grop of sbjects on a discrimination between the presence and absence of a stimls card. Those trained with the empty stimls holder as the positive learned as readily as those given a stimls card as the positive. Fletcher and Bordow (1965) inclded a similar test and in this case too no preference was fond. To explain their findings Fletcher and Bordow pt forward what may be viewed as a modified version of the stimls preference accont jst described; they sggest that as a reslt of ambigos-ce training itself the sbjects may develop a preference for ced over nonced foodwells which otweighs the factors promoting the NA>PA reslt. Their sggestion ses the notion of stimls generalization. The NA>PA reslt is fond when three separate and different stimli are sed - jnk objects in Leary's (1958) experiment; and different patterns of stripes in the present Experiment 2. With these stimli there will be little generalization or at least there will be no more generalization from P to N than from P to A. Bt with the stimli sed by Fletcher and Bordow themselves by Thompson (1954) and in the present Experiment 1 we might expect differential generalization to occr. There will be generalization between P and N which are similar bt not between A and the other stimli. f we now make or sal assmption abot the preeminence of the approach tendency prodced by reward we can expect the N stimls to acqire a greater approach tendency by generalization than the avoidance tendency acqired by the P stimls by generalization from N. t cold ths occr that the difference in approach strength between P and A wold be greater than that between Nand A. n this way we can accont for the occrrence of the PA>NA reslt with certain stimli withot departing from the principles sed to explain the NA>PA reslt. The ambigos-ce problem seems at first sight to be a complex one that might tax the abilities even of primates. t is of interest therefore that pigeons can solve this problem and that an explanation can be derived both for their performance and that of the primates from a relatively simple theory of discrimination learning that departs very little from the principles sggested by Spence (1936 1937). REFERENCES BERCH D. B. A theoretical analysis of the PAN ambigosce problem. Learning and Motivation 1974 S 135-148. BOYER W. N. & POLDORA V. J. An analysis of the soltion of PAN ambigos-ce problems by rhess monkeys. Learning and Motivation 19723325-333. BOYER W. N. POLDORA V. J. FLETCHER H. J. & WOODRUFF B. Monkeys' performance on ambigos-ce problems. Perceptal and Motor Skills 196622883-888. FELLOWS B. Chance stimls seqences for discrimination tasks. Psychological Blletin 19676787-92. FLETCHER H. J. & BORDOW A. M. Monkey's soltion of an ambigos-ce problem. Perceptal and Motor Skills 1965 21115-119. FLETCHER H. J. & GARSKE J. P. Response competition in monkeys' soltion of PAN ambigos-ce problems. Learning and Motivation 19723334-340. FLETCHER H. J. GROGG T. M. & GARSKE J. P. Ambigosce problem performance of children retardates and monkeys. Jornal of Comparative and Physiological Psychology 1968 66477-482. LEARY R. W. The learning of ambigos ce problems by monkeys. American Jornal ofpsychology 1958 71 718-724. RCHARDS R. W. Performance of the pigeon on the ambigosce problem. Blletin of the Psychonomic Society 1973 1 445-447. SPENCE K. W. The natre of discrimination learning in animals. Psychological Review 193643427-449. SPENCE K. W. The differential response in animals to stimli varying within a single dimension. Psychological Review 1937 44 430-444. THOMPSON R. Approach verss avoidance in an ambigos-ce discrimination problem in chimpanzees. Jornal of Comparative and Physiological Psychology 1954 47 133-135. (Received for pblication Jne 281979; revision accepted October 301979.)