More This is to certify that the thesis entitled TRACE SUPRAORDINATE STIMULUS CONTROL OF DELAYED MATCHING-TO-SAMPLE PERFORMANCE presented by Donald F. Kendrick, Jr. has been accepted towards fulfillment of the requirements for M.A. degree in Psychology Major professor DateWfiw 0-7639 , - *. ) P n W L .L'. éCKm (‘1 VA" *7 H I-.- OVERDUE FINES: 25¢ per day per item RETURNIMS LIBRARY MTERIALS: Place in book return to remove charge from circulation records TRACE SUPRAORDINATE STIMULUS CONTROL OF DELAYED MATCHING-TO-SAMPLE PERFORMANCE By Donald F. Kendrick, Jr. A THESIS Su‘rmitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1980 CV N. . «no .Q\\ Hru ABSTRACT TRACE SUPRAORDINATE STIMULUS CONTROL OF DELAYED MATCHING-TO-SNMPLE PERFORMANCE By Donald F. Kendrick, Jr. Pigeons were studied in a delayed matching-to-sample (DMTS) procedure with trace supraordinate stimulus interpolated into the delay intervals to signal the occurrence or nonoccurrence of delayed stimulus events. In all three experiments one supraordinate stimulus, a C-cue, signalled presentation of the comparison stimuli and another supra- ordinate stimulus signalled the absence of the comparison stimuli. Control of mm performance by this stimulus was assessed in a probe procedure. 0n infrequent probe trials the comparison stimuli were presented at the end of the delay interval, contrary to training. Accurate choice responding may be considered remembering and inaccurate responding may be considered forgetting. The results demonstrated that when contingencies required similar behavioral output correlated with both supraordinate stimuli matching performance was accurate, indicating remembering. When contingencies required different behavioral output. correlated with each supraordinate stimulus matching performance was inaccurate, indicating forgetting. These experiments serve to estab- lish an empirical analysis of DMTS performance and offer a context- dependent retrieval theory of short-term remembering behavior. To the children, Chloe (age 9) and Zak (age 7) who patiently (and sometimes not so patiently) attended my long-winded discourses on this research-mot a word of which they understood. And to Emily (age ’4) whose childish wonderment of writing, typing, and editing provided a source of amusement and was the primary motivation for the finishing and binding of this thesis. ii ACKNOWLEDGMENTS I would like to thank Dr. Mark Rilling for his helpful guidance and advice on this thesis. Dr. Hiram Fitzgerald, Dr. Daniel Tranberg and Thomas Stonebraker offered invaluable comments and suggestions. I would also like to thank Lana Tackett for drawing Figure l and her invaluable assistance in the preparation of the other figures. Special thanks to Dr; M. Ray Denny without whom the bulk of this research would not have occurred. iii List of Tables . List of Figures Introduction Experiment 1 Method Results Experiment 2 Method Results Experiment 3 Method Results Discussion . . . List of References TABLE OF CONTENTS iv Page vi LIST OF TABLES Table ' Page 1 Outline of Experimental Procedures and Primary Results . . . . . . . . . . . . . . 7 2 Summary of data presented in Figure 1 showing the mean percentages correct and standard errors for each pigeon and each condition . 16 3 Delay interval keypecking and choice responses for each condition . . . . . . . . . . . . 18 Summary of'Matching Performance Data . . . . 25 Summary of Matching Performance Data . . . . 29 Correlations among trial events and behaviors 32 \J O\U\ 3' Summary of Matching PerfOrmance Data . . . . 37 Figure LIST OF FIGURES Page Matching performance as function of rein- forcement conditions . . . . . . . . . . . 15 Matching performance in the RI-cued condition. Open circles are from RI-cued probe trials. Closed circles are from baseline C-cued trials . . . . . . . . . . . . . . . . . . 24 Matching performance in the K-cued and the M-cued conditions. Open circles are from probe trials. Closed circles are from uselinetnalseeeeeeeoeeeee28 Matching performance in the MUG-cued condi- tion. Open circles are from probe trials and the closed circles are from the base- line trials . . . . . . . . . . . . . . . 36 vi INTRODUCTION Recently attention has turned to delayed matching-to-sample (DMTS) as a procedure for investigating animal short-term memory (D'Amato, 1973; Grant and Roberts, 1976; Maki. 1979; Medin. Reynolds, and Parkinson, 1980: Roberts and Grant, 1978: Shimp and Moffitt. l97#: Tranberg and Billing, 1980; Zentall, Hagan. Howard and Moore. 1978). In the typical DMTS procedure a sample stimulus is presented for a fixed duration and, after a brief delay interval. comparison stimuli are presented. A response to the comparison stimulus that matches the previously presented sample stimulus is reinforced, while a response to the nonmatching comparison stimulus terminates that trial without rein- forcement. Some researchers assume that a representation of the sample stimulus is maintained or rehearsed during the delay interval (e.g., Maki, 1979; Roberts and Grant, 1978). In contrast, Stubbs. Vautin. Reid and Delehanty (1978) offer a procedural definition of memory free of such theoretical encumbrances: "...memory refers to the control of behavior by prior stimuli" (p. 168). "Prior stimuli" are also termed trace stimuli. According to Stubbs', et al, definition, DMTS is a trace stimulus control procedure: the sample stimulus is a trace stimulus which is terminated prior to the occurrence of the behavior it controls (cf, Meehan. 1979; villas and wiison, 1977). Maki and Anundson (1979) modified the typical mrs procedure by including a post-sample stimulus that signalled the cancellation of the comparison stimuli on that trial. Once a steady baseline had been 1 2 achieved, they occasionally included ”probe trials" in which the comparison stimuli were presented after the post-sample stimulus, con- trary to training. On probe trials performance dropped to 73% correct responses compared to 93% on baseline trials. They assumed that the pigeons rehearsed the sample stimulus during the delay interval and that by "cuing" trials in which no comparison stimuli were presented rehearsal was discontinued. That is, they hypothesized that the cue gained stimulus control of rehearsal. They labeled this procedure ”directed forgetting" to reflect the notion that the post-sample cue. a forget cue, terminates rehearsal and results in the apparent forgetting of the sample stimulus. Stonebraker (1980) argued that the results of'Maki and Anundson's (1979) experiment could be attributed to generalisation decrement. Disruption of DMTS performance may have occurred because presentation of comparison stimuli on probe trials was inconsistent with previous training. In his.Master's Thesis, Stonebraker positioned remember cues (R-cues), signalling the presentation of the comparison stimuli, and forget cues (F-cues). signalling the cancellation of the comparison stimuli, at three locations within the delay interval: immediately after the Sample stimulus. in the middle of the delay. and at the end of the delay interval. His rationale was that presenting comparison stimuli after an F-cue in probe trials is equally novel regardless of the temporal positions of the cue. Therefore, if the lowered accuracy of’DMTS performance on probe trials is due to generalization decrement, then all cue positions should equally disrupt performance. The rehearsal position assumes that the amount of disruption is dependent on the point of interpolation since the later the cue the more time to rehearse and the less time to forget. Stonebraker (1980) found that 3 matching perfOrmance was improved the later the forget cue; a result he interpreted as support for the rehearsal position. Although rehearsal is an intriguing theoretical concept. Maki and Anundson's (1979) and Stonebraker's (1980) procedures may benefit from an empirical orientation which views their results as demonstrating trace supraordinate stimulus control of HITS performance. A trace supraordinate stimulus (i.e., an instructional stimulus) is a stimulus that is presented and is terminated prior to the occurrence of the specific procedural variation that it is uniquely correlated with (Mechan, 1979: Reynolds, 1975). For example, an F-cue is a trace supraordinate stimulus informing the subject that the comparison stimuli are cancelled for that trial. Similarly, the R-cue is a trace supraordinate stimulus informing the subject that the comparison stimuli are to be presented at the end of the delay interval. In delayed matching-to-sample. trace supra- ordinate stimulus control is exhibited when matching performance is shown to differ following the occurrence of different trace supraordinate stimuli. The advantages of this behavioristic approach are: integration of animal memory research with research on stimulus control. subjecting cognitive concepts to an experimental analysis. and extending the con- ditions within which trace supraordinate stimuli are manipulated. Using only two procedures, each associated with a unique trace supraordinate , stimulus, dichotomous terminology such as remember and forget cues may serve well enough, but they are not easily extended to cover the full range and richness of the possible variations and combinations of conditional and unconditional discrimination procedures. Therefore, it is advisable to employ a term with broader scope, a term such as trace supraordinate stimuli, and to refer to these stimuli or cues in reference 4 to the procedures they are correlated with rather than their intended effects. For'example, a trace supraordinate stimulus correlated with the occurrence of the comparison stimuli is a C-cue, rather than an R-cue. to reflect that it is correlated with the presentation of the‘gonditional discriminative stimuli. orggomparison Stimuli. The trace supraordinate stimulus correlated with the nonoccurrence of the comparison stimuli is a NRFT-cue, rather than an F—cue. to reflect that it is correlated with‘yonfieingorcememgx cancellation of the comparison stimuli. The data of Maki and Anundson (1979) and Stonebraker (1980) indicate stimulus control of’matching performance by trace supraordi- nate stimuli. Their "directed forgetting" effect may simply be another example of stimulus control; specifically, trace supraordinate stimulus control. Therefore. elaborate theoretical constructions (e.g.. rehearsal) may not be required to explain their finding. The objective of the following experiments was to determine to what extent factors influencing the development of stimulus control have similar influences in the development of trace supraordinate stimulus control of delayed matching- to-sample performance. Nevin (1973) and Billing (1977) define stimulus control in terms of the slope of stimulus generalization gradients. Differential reinforce- ment is a procedure that steepens the generalization gradients indicating enhanced stimulus control. Differential reinforcement has also been shown to enhance supraordinate stimulus control of'DMTS performance. Mintz, Mourer and Heinberg (1966) used an array of lights as supra- ordinate stimuli. Vhen the last lights of the array were illuminated reinforcement was available contingent upon a correct DMTS response. Correct responses on trials prior to the last one were not reinforced by food presentation, but sequentially illuminated another light in the 5 array. This procedure resulted in relatively more accurate matching toward the end of the sequence than at the beginning, i.e., accuracy increased as proximity to reinforcement increased. To determine whether the lights at the end of the array were controlling matching performance Mints, et a1, illuminated them out of sequence. at the beginning of probe trials, and found that matching performance improved over baseline. This result demonstrated that the light array exerted supraordinate stimulus control of DMTS performance. Nevin (1973) viewed this experi- ment as differential reinforcement of'matching behavior, i.e., those DMTS responses early in the light array were not reinforced and the DMTS response occurring at the end of thelight array sequence was reinforced. These data indicate that differential reinforcement is a factor in supraordinate stimulus control and suggest that differential reinforce- ment may also be a factor in trap; supraordinate stimulus control. Experiment 1 compares the effects of differential and nondifferen— tial reinforcement correlated with trace supraordinate stimuli on matching performance in procedures similar to Maki and Anundson's (1979) and Stonebraker's (1980). Experiment 1 required a keypeck to obtain rein- forcement after both trace supraordinate stimuli in the nondifferential reinforcement condition. Experiment 2 furthers the analysis of non- differential reinforcement by employing a procedure in which response— independent reinforcement is correlated with one trace supraordinate stimulus and response-dependent reinforcement is correlated with the other trace supraordinate stimulus. Response-independent reinforcement weakens stimulus control (Nevin, 1973) and therefore may also weaken trace supraordinate stimulus control. The results of Experiment 2 suggested that delay interval behaviors may be an important factor in the development of trace supraordinate stimulus control of'DMTS performance. 6 Conditions that produced similar delay interval behaviors also resulted in similar'matching perfOrmance. Experiment 3 examined the role of differential delay interval behavior'in a procedure in which different behaviors were required during the delay intervals, each correlated with a different trace supraordinate stimulus. and in which the same behaviors were required at the end of the delay intervals. It was shown that end-of-delay interval behaviors. rather than behaviors during the delay intervals are the critical factor in accurate matching performance. Table 1 presents the design, manipulations, and primary results of each of these experiments. 7 Table 1 Outline of Experimental Procedures and Primary Results Trace Supraordinate Matching Experiment Stimulus Procedure Performance A C-cue comparison stimuli accurate presented UC-cue unconditional accurate discriminative stimuli presented B C-cue comparison stimuli accurate presented 1' NRFT-cue comparison stimuli inaccurate cancelled A C-cue comparison stimuli accurate presented UC-cue unconditional accurate discriminative stimuli presented A C-cue comparison stimuli accurate presented RI-cue response—independent equivocal reinforcement B C-cue comparison stimuli accurate presented g, K-cue orient to keys or accurate peck center key for reinforcement C C-cue comparison stimuli accurate presented M-cue magazine approach inaccurate response delivers reinforcement A C-cue comparison stimuli accurate presented ‘3 MUG-cue magazine approach accurate response initiates presentation of the unconditional dis- criminative stimuli EXPERIMENT 1 Differential reinforcement of trace supraordinate stimuli is a characteristic of bothflMaki and Anundson's (1979) and Stonebraker's (1980) procedures. The negative trace supraordinate stimulus, a NRFT-cue, was correlated with the absence of the comparison stimuli and therefore with the absence of an opportunity for reinforcement. The conditional trace supraordinate stimulus, a C-cue, was correlated with the presence of the comparison stimuli and therefore correlated with an opportunity for reinforcement. The trace supraordinate stimulus control exhibited in their experiments may be due to differential reinforcement correlated with the C- and NRFT-cues. Experiment 1 employed a trace supraordinate conditional discrimina- tion procedure. On all trials one of two trace supraordinate stimuli was presented during the delay intervals immediately after the sample stimulus and was terminated prior to the end of the delay interval. In the first condition a trace supraordinate stimulus, a C-cue, was correlated with the presentation of the conditional discriminative stimuli, the comparison stimuli, and a second trace supraordinate stimulus, a UC-cue, was correlated with the presentation of the,gngonditional discriminative stimuli. This condition represents nondifferential reinforcement correlated with trace supraordinate stimuli. In the second condition a trace supraordinate stimulus correlated with the absence of the comparison stimulus, a NRFT-cue, replaced the UC-cue of the first condition. C-cued trials continued as in the first condition. This 8 9 second condition is a modified replication of Maki and Anundson's (1979) procedure and reflects differential reinforcement correlated with the trace supraordinate stimuli. This experiment pits the stimulus control interpretation against the rehearsal interpretation. According to the rehearsal interpretation the trace supraordinate stimulus that is correlated with the absence of the comparison stimuli, the NRFT-cue, terminates a rehearsal process. It seems plausible that rehearsal may also be terminated by a trace supra- ordinate stimulus correlated with the presentation of unconditional discriminative stimuli, the UC-cue, since memory of the sample stimulus is not required to gain reinforcement. Thus, a rehearsal position predicts inaccurate matching performance on probe trials in both the differential and nondifferential reinforcement conditions. According to the stimulus control interpretation nondifferential reinforcement correlated with the trace supraordinate stimuli weakens stimulus control and differential reinforcement strengthens stimulus control. Thus, a stimulus control position predicts accurate matching performance (weak stimulus control) on probe trials in the nondifferential reinforcement condition, C-and UC-cues, and reduced matching accuracy on probe trials (enhanced stimulus control) in the differential reinforcement condition, C-and NRFT-cues. user-d Subjects Four naive adult White Carneaux pigeons were maintained at 80%;:20g of their free-feeding weights and were individually housed in a temper- ature controlled, constantly illuminated, colony room. Water and grit were always available. 10 Apparatus A three-key Lehigh Valley Electronics pigeon chamber was used. The 2.5“ cm diameter plastic keys required a force of 15 g (.15 N) for activation. The three keys were positioned 5.5 cm apart, were 9 cm from the top of the chamber, 8 cm from the sides and 25.5 cm from the floor. The houselight (28 v dc, CE 757) was above the center key, 2. 5 cm from the top of the chamber» The grain magazine was 5 x 5 cm, 11.5 cm from the floor and positioned under the center key. The keys were illuminated by a display projector (IEE Series 10, 28 v dc, GE 74?). The walls of the chamber were white matte with an unpainted aluminum stimulus panel. Masking noise was provided by a small ventilation fan. Electro-mechanical programming equipment controlled by a paper-tape reader was in an ad- joining room. Procedure In all conditions reinforcement was 2.5 sec access to mixed grain. The maximum number of reinforcers attainable depended upon the stage of the experiments. After pretraining, the actual number of reinforcers delivered depended upon the birds' performance. Pretrainigg. All birds were first magazine trained on a VT 45 sec schedule following Catania and Reynold's' (1968) constant probability formula. when birds were approaching the magazine and eating reliably, autoshaping trials began. Autoshaping followed the VT 45 sec schedule, used in magazine training with six seconds of red or green illumination on the center key. Red and green were presented randomly such that one color did not appear more than three times successively and each occurred an equal number of times within a session. Autoshaping continued in approximately one hour'daily sessions until a minimum of 100 total responses were obtained in a single session. ll Acquisition of Delayed Matching-to—sample. The basic procedure was two-choice delayed matching-to-sample with red and green key lights as sample and comparison stimuli. Following a 30 sec intertrial interval (ITI) red or green sample stimulus transilluminated the center key for 12 sec. A keypeck then.terminated the sample stimulus and initiated a 1 sec supraordinate stimulus. A white 5 mm diameter circle on the center response key (C-cue) signalled the presentation of the comparison stimuli, and a white 3 mm equilateral triangle (UC-cue) signalled the presentation of the unconditional discriminative stimuli. Following the termination of the C-cue, both side keys were illuminated red and green and a single peck to the color that matched the sample resulted in reinforcement. A peck to the nonmatching color initiated the ITI. Following the termination of the UC-cue, one side key was illuminated by a white horizontal bar and the other side key by a white vertical bar, each 2 mm x 2.54 cm on a black surround. A peck to the vertical bar resulted in reinforcement. A peck to the horizontal bar initiated the ITI regardless of the color of the preceding sample stimulus. Red and green sample stimuli were semi-randomly presented such that the same color did not appear more than three times in succession and each occurred an equal number of times per session. C-cues and UC-cues were equally likely after red and green sample stimuli, did not appear more than three times on successive trials, and occurred an equal number of times per session. Left and right side keys were red and green an equal number of times with no more than three trials of the same color on the same side in succession. The two side key combina- tions, red-left, green-right, and red-right, green-left, appeared an equal number of times per session. 12 The birds were exposed to all stimulus events and contingencies from the first session. There were 36 C-cued and 36 UC-cued trials per session. Each session lasted approximately one hour. Delay intervals were 1 sec, the same duration as the cues, and were incremented by 1 sec upon the completion of two sessions with 80% or'more correct responses. All birds were thus advanced to and maintained at h sec delay intervals. Baseline and Testing. Two baseline conditions were arranged in an ABA format. In the A condition, the equilateral triangle projected on the response key was a trace supraordinate stimulus signalling the presentation of the unconditional discriminative stimuli, UC—cue, the same as during acquisition. In the B condition, the equilateral triangle was a trace supraordinate stimulus signalling the end of that trial without keylight stimuli and without an opportunity for reinforce- ment, NRFT-cue. Each of the three stages of the experiment lasted approximately three to five months. Completion of condition A was required prior to condition B, completion of which was required prior to reinstating condition A. Twenty sessions of each condition were required prior to test sessions with the last five at or above 80% correct choice performance. The 80% criterion was based on percentage correct responses on the 36 C-cued trials. The minimum requirement was 29 correct responses out of 36 responses. Performance on the UC-cued trials was not considered. Three blocks of test sessions, one during each of the three stages, were conducted. A test session contained four probe trials in which the comparison stimuli were presented instead of horizontal and vertical bar stimuli (condition a) or instead of the ITI (condition B). Probe trial stimulus events were controlled such that red and green sample 13 stimuli were equally likely to be followed by the two comparison stimuli combinations in each session. A response to the matching comparison stimulus produced reinforcement and an incorrect response ended the trial. Five test sessions were conducted for a total of 20 probe trials per stage. Baseline sessions alternated with test sessions as long as 80% or'more correct choice responding was maintained. 0n test sessions, the baseline percentage criterion was lowered to 75% (27/36) to allow for possible local disruptive effects due to the incon- gruency of probe trial contingencies. The experiment ended upon com- pletion of the fifth probe session of the last condition. Results Figure 1 shows percentage correct choice responses in baseline and probe test conditions for each of the four pigeons as a function of the cue contingency conditions. Table 2 shows the same data for each pigeon averaged over all sessions. In Figure 1, the solid circles represent the baseline data of 36 C-cued trials per session. Open circles represent the test data of four UC- and NRFT-cued probe trials per session. In the first A condition, the left panel, all birds maintained high percentages of correct choice responses. Performance on the 5-day baseline, test baseline and test probes were similar. The respective means are 88.5%, 87.6%, and 89.0% averaged over all birds. Nondifferential reinforcement associated with the trace supraordinate stimuli resulted in similar and accurate matching per- formance. The middle panel of Figure 1 shows that the B condition effectively reduced matching accuracy on NRFT-cued trials. The NRFT-cued test probe mean for all birds dropped to 53.3%, near chance, and baseline 11+ Figure 1. Matching performance as function of reinforcement conditions. 15 DAY BASELINE llllllll l A E 6 N YU 7 AE 7 OS 3 &MA P p P P »L m 0000 II 8642 I38 |33 |05 I00 56 5| |||||||| A B 5 3 P Phpr 00000 muv8642 |03 98 73 39 34 J 53 ”2 IO? 82 77 l00~ - L O O 8 6 Powmmoo szomwn— 40- 48 ||II|I|A 40» IS? I32 6| I00 I05 SESSIONS 56 l6 oo.o oo.ooa oo.n oo.nm oo.o oo.ooH oo.n oo.nm nooosa shoe ms.o om.mm oe.u os.mm sm.o om.sm as.H om.um commoner s once mm.o os.om an.o oo.~m nm.~ om.sm mm.a oo.oo ocaaonom saecn Hm.a oo.on n~.NH oo.ne nm.m oo.nm oo.m oo.mm noooaa once ma.~ om.nm ma.s or.wm om.m oo.~m rn.~ om.sm assesses m shoe sm.~ o~.me mr.a o~.mr mm.a os.mm eo.~ os.om condense soetn Nu.m oo.mm ma.o oo.nm oo.ou oo.nm oo.o oo.oou noooaa once am.~ oo.mm am.m o~.om su.m o~.nm m~.~ oe.em ossuenam « once mm.a o~.mr on.~ o~.mm as. om.Hm ma.a os.nm osauonsm are.“ mm eooohoo mm sooonoo mm escapee em penance soacdesoe R see: R see: R see: R see: smnna enema manna wanna .sogaecoo some use comma some mom macho “newsman use voounoo momdvceohoa some on... mode—one an enema a.“ economoha made me hues—gm N 93.3. 17 means increased to 91.3%, for the 5-day baseline, and to 92.4% for the test baseline. When trace supraordinate stimuli were correlated with differential reinforcement they gained stimulus control of matching performance. The A condition was repeated and the data appear in the right panel of Figure 1. Again, baseline means of all birds are comparable, 92.9%, for the 5-day baseline, and 93.7%ifor the test baseline. 0n probe trials performance was somewhat better at 97.5% mean correct choice responses. The nondifferential reinforcement con- dition again demonstrated loss of stimulus control by trace supra- ordinate stimuli. In summary, comparing the means of the probe trials of the A conditions with that of the B condition, 89.0% and 97.5% vs 53.3%, it is clear that differential reinforcement effectively reduced matching performance to near chance and that the nondifferential reinforcement condition resulted in matching accuracies similar to baseline accuracies. Position preferences were also noted in all four birds, 3776 and 3513 peeked the left key more than the right key and 3777 and 3524 peeked the right more. Table 3 presents side key position preferences and delay interval data for each pigeon across all conditions. Percentage correct choice responses were affected by the birds' side key preference as indicated by higher'matching accuracy on the pre- ferred key. Table 3 also shows that delay interval keypecking was low with the NRFT-cue and at a higher rate with the UC-cue. In the B condition C-cued delays supported more responding than NRFT-cued delays, 142 mean responses vs 25 mean responses, respectively. In the A conditions responding was similar during C-cue and UC-cued delay intervals, 123 18 .madana opona near mcoammom awe» one o» Hoaha macammom one moum our madvoa comm vmoe one .msoawmom n so women mason ohm dado .nuosaa eoso-earz .earz .sor sewer .m .sox secs .4 .naosaa eoso-o= .0: .nadsao eoso-o .o a an on we an an we we we on an on mm m mm an an mm mm an mm on mm was mos New A menace we mm mm as mm as nu ma as me as we r « ea mu as as as ed as as as am we mu A ooaose and ass sea and was mom and cam sow em on mm o: as am as are and cos are mad omm om m m o assoc as on mm as no mm mm em am om NN mm m as om we mm rm we om on am as was me a menace we as mm as as ea ea mm ms nu ea mu m m as as on mm as em as as as mu mm as A echoes and was arm or we or new emu was so on mos earz mm or m: mm as an an em on n m o e assoc no: son Hnn no ms we we we as mm mm mm r em mm mm on an as on mm mm was are and a nuance ma mm mm as as ea cm as ms nu we as m be as mm as we ma nu mm mm mm as mm mm a oososo was owe was com or am arm cum Ham am we aw 0: sum [lens was AHA am or man ummw new,»lmm mm limm o assoc code cache ones chem cocoa onam poem oooam chem onnm oooha once a couosseoo shoe Jlmmrmm‘ enoe_ soo.m once saute once scenm. smnma enema manna wanna soauaecoo some Mom mom:OAnoH ooaono use msaxooa50x Hd>nmpsa heaon m canoe 19 mean responses vs 133 mean responses, respectively: in the first A condition: and 126 vs 103, respectively, in the second A condition. This difference is primarily due to one bird, 3776, who did not recover its former high rate of UC-cued delay responding during the second A condition. Informal observation indicated 3776 increased left key peeks during the delay intervals and the increase in total left key peeks provide support (see Table 3). Pigeon 3776 increased left key peeks from a mean of 173 to 204 from the first A condition to the second, while the other three birds decreased the rate of response to their biased key from the first condition to the last. Pigeon 3513 decreased from a mean of 73 peeks on the left key to 56, pigeon 3777 from a mean of 70 to 60 and pigeon 3524 from a mean of 432 to 52. In general then, in the NRFT-cued condition delay interval keypecking was reduced and in the UC-cued condition delay interval keypecking was maintained at a rate similar to that maintained during C-cued delay intervals. Disruption of matching performance was noted in all birds when changed from condition A to condition B. The mean percent correct for the last five sessions of condition A was 87%, the mean correct for the first five sessions of condition B was 68%. These percentages are based on the C-cued trials only. When returned to condition A from condition B no disruption was evident. The mean for the last five sessions of candition B was 91% and for the first five sessions . of the second condition A the mean was 92%. EXPERIMENT 2 Experiment 1 demonstrated that differential reinforcement is an important factor in the acquisition of trace supraordinate stimulus control. This suggests that trace supraordinate stimulus control procedures may be viewed as a subset of more traditional stimulus con- trol procedures. Another important determinant of stimulus control is the response-reinforcer contingency. Nevin (1973) reviews literature demonstrating that stimulus control established by response-independent reinforcement is not as strong as stimulus control established by response—dependent reinforcement. Nevin's rationale is that response— dependent reinforcement "...is likely to establish some superstitious pattern of behavior as a result of accidental correlations of particular movements and reinforcemen " (p. 126) and that these behaviors are likely to be incompatable with keypecking, the traditional response used to measure stimulus control. It is possible that response- independent reinforcement may also reduce trace supraordinate stimulus control of DMTS perfoImance. Experiment 2 investigated the possibility that the response- reinforcer contingency is an important determinant of trace supra- ordinate stimulus control by comparing response-dependent, C-cue, and ‘fiesponsefilndependent, RI-cue, conditions in a.DMTS procedure. 20 21 £221.24; Subjects The four White Carneaux pigeons that served in Experiment 1 also served in this experiment. Two pigeons with previous and extensive DMTS experience were also included. 1 Apparatus See ExPeriment 1. Procedure The basic DMTS procedure with trace supraordinate stimuli described in Experiment 1 was employed with the following modifications. Immediately following Experiment 1, the UC-cued trials were altered from terminating with unconditional discriminative stimuli and response- dependent reinforcement to terminating with response-independent presentations of mixed grain. The white equilateral triangle keylight stimulus was therefore an RI-cue, a trace supraordinate stimulus correlated with response-independent reinforcement. Denny's (1967) elicitation theory suggests that pigeons may approach the magazine following the RI-cue rather than remaining at the response keys. This behavior may then be adventitiously reinforced and effectively compete with the keypeck response required on probe trials. Therefore magazine approach was defined as interruptions of a photobeam within the magazine enclosure and was recorded during C-cued and RI-cued delay intervals.. The initial results indicated that one bird, P3777, approached the magazine enclosure on response-independent trials and that the other birds did not. Therefore, a second condition was added in which P3777 was required to peek the darkened center key for reinforcement and P3513 was required to interrupt the magazine photobeam for rein- forcement. These behaviors replaced the response-independent 22 reinforcement contingency on RI-cued trials. Accordingly, the trace supraordinate stimulus correlated with these two conditions was redesignated K-cue and M-cue, respectively. Results Figure 2 shows percentage correct matching responses during test sessions on RI-cued trials, open circles, and on baseline C-cued trials, closed circles. As Figure 2 indicates response-independent reinforcement did not disrupt probe-trial matching performance. Matching accuracies on probe trials ranged from means of 85% to 95%. Table 4 presents delay interval keypeck data and magazine approach data. These pigeons did not develop magazine approach behavior and showed no disruption of matching performance. Figure 3 shows the matching performance of the two pigeons that completed both conditions of the present experiment. The RI-cued condition was divided into K—cues and M—cues to indicate the conditions in which the birds remained oriented at the‘geys and conditions in which the birds approached the flagazine. The upper left panel of Figure 3 shows that P3513 maintained probe trial matching accuracies of 100% on the probe trials of all five test sessions in the K-cued condition. P3777 was required to peek the darkened center key during K—cued delay intervals and accurate matching performance on probe trials resulted. The lower left panel of Figure 3 shows the matching performance of P377? in the K-cue condition. The overall mean correct responses for P3777 on these K-cued probe trials was 80%. The upper right panel of Figure 3 shows that when required to approach the maga- zine during H-cued delay intervals P3513 failed to maintain accurate matching performance on probe trials. The mean correct responses 23 Figure 2. Matching performance in the RI-cued condition. Open circles are from RI-cued probe trials. Closed circles are from baseline C-cued trials. 24 L 1 1 O 1 ..... 6 1 Q 2 J 4 1 O 5 1 P _ FL h a e a _ FL 1 J 1 I 1 1 tub—navy 1 O. -. % DI _ krfi L a L _ a 5L 0 O O O O O O 0 6 5 O 9 8 7 6 5 2345 l2345 SESSIONS 25 .eodaondm ewes scum haoomndaom sopnoooh poo one: moaad> Hm com o mononm amok .msoammom o>am no names one no women one mesa .osonHm .Hm .oaoto .o .Hohuo easesepm .mmt '1" ‘l‘ . - ms.sfl o.mm - - oo.n o.mm . - .oo.n o.me . - oo.n o.nm noooha shoe o o o~.~ s.ms o o mm.a 0.0m o o em.m o.~m o o ma.o s.mm commoner shoe 0 o so.H o.mm o o as.u s.em o o NH.H e.sm o o Hm.u :.sm credence septa He 9 mm oonuaoo Hm 1m.111mm scammowl am 0 11mm eoobuoo He 0 mm eooammu 11 .M one: immune: Inhomo: _R see: emeww 1. .1. owmwa smwmm:1.111. enema demo mosesuomuea wednovd: mo finessem : wand? 26 for P3513 on.M-cued probe trials was 55%. P377? developed magazine approach behavior in the RI-cued condition and matching performance was disrupted on probe trials, declining to 60%. This is shown in the lower right panel of Figure 3. Table 5 presents the mean matching accuracies, magazine approach behavior, and delay interval keypeck data for these two birds in both the K-cued and M-cued conditions. The combined results of the RI-cued condition and the K- and M-cued conditions demonstrate that remaining oriented at the response keys, whether adventitiously developed or experimentally manipulated, reduced trace supraordinate stimulus control. Matching accuracy averaged over the 4 pigeons in the RI-cued condition and P3513 and P3777 in the K-cued condition (remaining oriented at the response keys) was 89.3%. In contrast, orienting away from the response keys increased trace supraordinate stimulus control. Matching accuracy averaged over the two pigeons completing the M-cued condition (orienting away from the response keys), P3513 and P3777, was 57.3%. 27 Figure 3. Matching performance in the K-cued and the M-cued conditions. Open circles are from probe trials. Closed circles are from baseline trials. 28 D \b - 1 E .. - 1 Nu .....nw c o; 1 1 M {MO 1 1 Q\ I. I. p p p — E _ _ — fall a _ abaf w ‘0 w Q. \Q l I 1 a e n 5 1 . 7 1 3 -..O 3 D. 1 Q. P 1 a h e p a a . . »L( L a a _ _ a . If 0 O O O O O O O O O O O 0 O O O O m98765432 0987 5432 Howmmoo szomma 2 3 14 5 2 3 ‘4 £5 SESSIONS 29 .ocaaewem ewes sou.“ 33.9393 eegooou won one: nomads, : Ed x use 0 hadrons” amen. .2333 95m .«o means on» so women one made “0:01: .2 “emetx 1: «osoueso .o «Moshe guesses .mm .3. - - ~H.¢ o.oo - - sm.mu o.ns 1 nn.m o.om - oo.o o.ooa noooha shoe me o nm.m 0.0m em a nw.n m.sm o mn.a ~.Nm o os.a ~.sm commoner shoe n o as.o m.~m m o mm.m m.mm o mm.H N.Hm o sm.a o.am commoner sooth : o mm pooeaoo : 0 mm: escapee 0 mm penance x mm eoobaoo R comm IIRWWo: Lu ado: hm ado: enema «Hume enema manna eoso12 eoson mama oossfiotom mangoes: mo ransom m. 93.8. gmammr 3 Nevin's (1973) suggestion that response-independent reinforcement weakens stimulus control through the adventitious acquisition of competing responses was supported in Experiment 2, at least in terms of trace supraordinate stimulus control. In Experiment 2, the RI-cued condition apparently provided adventitious reinforcement or orienting towards the stimulus panel for most of the pigeons: the same behavior that occurred during C-cued delay intervals. Therefore, both trace supraordinate stimuli were correlated with similar behaviors and similar DMTS perforr mance was obtained. In P3777 magazine approach behavior was apparently adventitiously reinforced. Since that behavior differed from that occurring during C-cued delay intervals, the reduced matching performance on probe trials may have been due to response competition. That is, the M-cue may have set the occasion for’magazine approach responses which effectively competed with accurate keypecking responses required on probe trials. The second condition of Experhment 2 provided experi- mental support for this analysis by demonstrating that manipulation of conditions that reversed the delay interval behaviors resulted in a similar reversal of DMTS performance. The pattern of results obtained in these first two experiments suggest that when experimental contingencies produce differential delay interval behaviors trace supraordinate stimulus control is strengthened and when experimental contingencies produce nondifferential delay inter- val behaviors, trace supraordinate stimulus control is weakened. 30 31 Table 6 presents the behaviors occurring during each of the DMTS trial events as a function of the type of trial. Table 6 reveals that when behaviors that are present at the end of the delay interval or that are required for the delivery of reinforcement at the end of a trial, were similar after both trace supraordinate stimuli, DMTS performance was accurate on probe trials and when behaviors at the end of the delay intervals differed, DMTS performance on probe trials was reduced relative to baseline performance. Therefore, the present data do not differen— tiate between control of differential delay interval behaviors and con- trol of differential end of delay interval behaviors as the critical factor in the control of DMTS performance. The purpose of Experiment 3 was to determine whether behaviors at the end of the delay interval are the primary factor in the establish- ment of trace supraordinate stimulus control of DMTS performance. This was accomplished by arranging contingencies such that after a magazine- unconditional trace supraordinate stimulus, MUC-cue, interruption of the photobeam within the magazine enclosure produced unconditional dis- criminative stimuli and following the conditional trace supraordinate stimulus, C-cue, the comparison stimuli were presented. This procedure produces differential delay interval behaviors, i.e., magazine approach after the MUG—cue and orientation towards the response keys after the C-cue. This procedure also produces similar behaviors at the end of the delay intervals, i.e., peck the unconditional discriminative stimulus after the MUG-cue and peck the comparison stimulus after the C-cue. The behaviors generated by this procedure are also shown in Table 6. Chance performance on probe trials implicates differential delay interval behaviors as the critical factor in control of DMTS .90.?an codename ceasdwde mcaesaosa .mhex omsoamon on» son.“ herd mefieoflo $.23 .maox emcoawon eons... on» wages. modpcoano Ho mcgooa .meadroe t. 32 mousse"; macros s83 macros 33:02: m ACE madame"; masses madman. macros emsouo opasooch has; Adria hadron. «scoot: madame“: assures monarch hadron. vosonx N ACE Haves—sum adoozsom fleecing masses emsouHm 3.9.28": ensures age manage cosouo eagooeeH >82 has; hadron. 350152 evens—00¢ mnndroe wgoe nausea 83010: H axe madame“; monarch. masses egos 3:010 oosdfinomhem mariopsH heaom wamzeosH Adamo magnum meal? Hanna wsflovd: mo new weaken 333m mucosa SE 2.5 33223 was 3:96 Hers moose 203.3380 0 canoe 33 performance since delay interval behaviors differ and matching behavior differs as a function of the trace supraordinate stimuli. Similar matching accuracies between probe trials and baseline trials implicates end of delay interval behaviors as the critical factor in control of DMTS performance since behaviors at the end of the delay intervals are similar and matching behavior is similar as a function of the trace supraordinate stimuli. Leta-d Subjects The two pigeons, P3513 and P3777, that completed both conditions of Experiment 2 served. Appgggtus The apparatus of the previous experiments were used. Procedure A11 parameters, criterion, and testing procedures of the previous experiments were maintained for this experiment with the following modifications. Immediately after completion oanxperiment 2 both birds were required to break the magazine photobeam upon termination of the 4 sec delay interval. One bird, P3777, was handshaped to approach and interrupt the photobeam. P3513 was performing this behavior as a result of training in Experiment 2. Interruptions of the photobeam initiated the simultaneous presentation of the vertical and horizontal bar unconditional discriminative stimuli used previously. A peek to the vertical bar produced reinforcement, 2.5 access to mixed grain. On the UC-cued trials of Experiment 1, presentation of these unconditional discriminative stimuli was response-independent. In 31. this experiment presentation of the unconditional discriminative stimuli was contingent upon a magazine approach response after the 4 sec delay interval had timed out. Consequently, the trace supraordinate stimulus correlated with this condition, the white equilateral triangle, is referred to as a.MUC-cue, indicating that a,!agazine approach response produces the yngonditional discriminative stimuli. The C-cued trials were continued as in the previous experiments. Results Figure 4 shows the percentage correct matching responses on probe trials and baseline trials of the five test sessions. From Figure 4 it is apparent that P3513 performed accurately on probe trials, with an overall mean of 95% correct responses. P3777 performed somewhat less accurately with an overall mean of 85% correct, due primarily to 50% correct on test session four. These results suggest that DMTS performance was not disrupted, or disrupted only slightly, by delay interval magazine approach behavior. Table 7 presents summary data of matching percentages on the 5-day baseline measure, the test session baseline, and the probe trials and also presents the mean number of magazine approach behaviors per test session. Both birds increased magazine approaches from the pretest baseline to test sessions. P3777 did not approach the magazine as frequently as P3513, 19.8 mean ° times per session vs 43.4»mean times per session, respectively. 35 Figure 4. Matching performance in the MUG-cued condition. Open circles are from probe trials and the closed circles are from the baseline trials. mzo_mmmm 00. 1038800 1N3383d 37 won one: mosad> on: end 0 mononm pmoa .ecaaemem ewes some maeesneaew anemones .msoammom o>am no names one no women one dude “ozonoaz .03: nosooo .o .Houuo easecevm .mm .x. 1 1 oo.oa o.mm 1 1 oo.m o.ma noooaa once om e um.” m.~a m: o na.a e.mm someones enoa on o oo.a o.ma mm o ma.a ~.ow someones aoe1m on: o mm ooohhoo on: 0 mm soohoo R ado: R ado: aaama manna oven oocdenoauom weanovsz mo anesesm m wands Discussion Experiment 1 compared differential and nondifferential reinforcement correlated with trace supraordinate stimuli. When both trace supra- ordinate stimuli were correlated with reinforcement contingencies, C- and UC-cues, conditional matching performance after both stimuli was accurate. When the two trace supraordinate stimuli were correlated with nondifferential reinforcement, C- and NRFT-cues, conditional matching performance following the stimulus correlated with nonrein- forcement, NRFT-cue, was reduced to near chance. Conditional matching performance following the trace supraordinate stimulus correlated with reinforcement, C-cue, remained accurate. In the nondifferential rein- forcement condition, the UC-cue signalled the occurrence of unconditional discriminative stimuli. Memory of the sample stimulus was not required to gain reinforcement since a single peck to the 8+, the vertical bar, produced reinforcement regardless of the color of the sample stimulus. Yet, in this nondifferential reinforcement condition accurate matching performance was obtained, contrary to the prediction of a cOgnitive rehearsal position. A stimulus control interpretation of the results of Experiment 1 was supported. According to this position, nondifferential reinforce- ment weakens stimulus control and weakens trace supraordinate stimulus control. Consequently, accurate matching performance following the UC-cue was predicted and was obtained in the nondifferential reinforce- ment condition. The results of the differential reinforcement condition 38 39 were also consistent with the stimulus control position in that differential reinforcement strengthens stimulus control thereby producing inaccurate performance after the NRFT-cue, as was obtained. Experiment 2 extended the analysis of the similarity of stimulus control and trace supraordinate stimulus control by eliminating the response-reinforcer contingency in the nondifferential reinforcement condition. Experiment 2 also eliminated the unconditional discrimina- tive stimuli following the trace supraordinate stimuli and presented response—independent reinforcement after the RI-cue. Since presenta- tion of the comparison stimuli on probe trials was a novel event, a generalization decrement position based on the novelty of the presenta- tion of comparison stimuli predicts chance performance under this condition. However, accurate matching was obtained failing to support this generalization decrement position. The stimulus control position predicted accurate matching since response-independent reinforcement weakens stimulus control as was found. Informal observation indicated that all but one pigeon, 3777, remained oriented at the response keys, while 377? approached the magazine enclosure. This behavior is consistent with Nevin's (1973) position that stimulus control is weakened by response-independent reinforce- ment through the adventitious reinforcement of competing responses (also see Skinner and Morse, 1957). In all birds. except 3777. orienting towards the response keys was apparently adventitiously reinforced and was not a competing behavior since it was the same behavior’maintained on C-cued trials. For pigeon 3777 magazine approach behavior was adventitiously reinforced and was a competing behavior since it differed from that generated on C-cued trials. 40 Experiment 2 also explored this analysis of the role of behaviors generated by nondifferential reinforcement by comparing differential, C- and M-cues and nondifferential, C- and K-cues, delay interval be- haviors correlated with trace supraordinate stimuli. It was found that when pigeons remained oriented toward the response keys, matching accuracies remained high. When the pigeons oriented away from the response keys, matching accuracies were reduced to near chance, thus supporting the response competition analysis. This suggests that trace supraordinate stimulus control of DMTS performance is not strengthened by differential reinforcement per se, rather stimulus control is strengthened because differential reinforcement increases the probability of differential behavior correlated with different stimuli. Weisman (1970) presents evidence indicating that differential response rates, rather than differential reinforcement frequency, is the important factor of discrimination training that establishes inhibitory stimulus control. The present data suggest that differential behavior also is the primary factor in the establishment of trace supraordinate stimulus control and that differential reinforcement is but one method of establishing differential behavior. Experiment 3 demonstrated that the behaviors occurring at the end of the delay interval are critical for accurate matching performance, rather than behaviors during the delay interval. Matching performance ' was accurate on probe trials when similar behaviors at the end of the delay and different behaviors during the delay intervals were correlated with the differentially-cued delays. This result indicates that the consistent occurrence of differential behavioral chains is not always sufficient to establish stimulus control and suggests that the terminal link of behavioral chains is the behavior responsible for the 41 establishment of stimulus control. That is, it may be that behavioral chains with different topographies are equivalent to the extent that the terminal links are similar. The results indicate that when the behavior occurring at the end of the delay interval is similar in the presence of both stimuli accurate matching is obtained regardless of the specific tapographies of the behavior during the delay intervals. This finding illuminates the inconsistent correlaion found between sample-specific mediating behavior and DMTS performance (see Carter and Werner, 1978, for a review). Hunter's (1913) early finding that maintenance of bodily orientation towards the stimuli facilitates delayed responding sparked a controversy that has continued to the present. Recently, Zentall, Hogan, Howard and Moore (1978) have noted that sample-specific behavior is highly consistent in some subjects, but is unidentifiable in others, even though all subjects exhibited similar DMTS performance. Also, some studies (e.g., Blough, 1959) report clear evidence of sample-specific behavior in all subjects with accurate matching performance, while other studies (e.g., Stonebrakery 1980) report no such correlation. The present data may resolve this contro- versy since they indicate that perhaps only the behavior at the end of the delay interval is correlated with matching performance. It is suggested that previous failures to find direct correlations between mediating behavior and DMTS performance is because consistent behaviors at the end of the delay intervals are responsible for accurate matching performance, rather than specific chains of behavior throughout the delay. In other words, maintenance of bodily orientation is not necessary for accurate performance, but appropriate orientation‘ppipp‘pp testing is essential. 42 The present data may also enhance our understanding of underlying memory processes. Recently, notions of context-dependent retrieval of memories have been employed to describe a large body of both human and animal memory literature (Lett, 1978; Lewis, 1979: Spear, 1978). According to these notions accurate remembering occurs to the extent that conditions are similar between learning or acquisition (information input) and retention testing (information output). That is, retrieval of a target memory is more probable the more similar the conditions of training are to the conditions of testing. This analysis may also apply to the short-term remembering of the sample stimulus in DMTS. In the baseline condition of all experiments, C-cued trials, the pigeons maintained visual orientation towards the response keys and responded accurately to the comparison stimuli. Therefore, it may be that the presentation of the comparison stimuli was a retrieval cue for memory of the sample stimulus (see D'Amato and Worsham, 1974) only when the comparison stimuli were presented in the appropriate context; i.e., visual orientation towards the response keys. The test conditions manipulated the visual orientation of the pigeons and it was shown that when the birds were oriented towards the response keys matching performance was similar to matching performance on the C-cued trials, i.e., accurate. Whenever the comparison stimuli were presented in a context different from the C-cued context, i.e., visual orientation towards the keys, retrieval of the sample stimulus could not and did not occur because the supporting context was absent. This context retrieval hypothesis is also a generalization decrement hypothesis. The novel presentation of comparison stimuli, on probe trials, was viewed as a possible source of generalization decrement by Stonebraker (1980) and 43 was shown not to account for the data. The present position is that the behavior at the end of delay intervals is the primary source of generalization decrement, rather than the stimulus conditions. General- ization decrement does not occur in DMTS because of novel stimulus conditions, but because those novel stimulus conditions require responses different from those occasioned by prior stimulus conditions. In other words, when presented with two stimulus situations, each controlling opposing response tendencies, matching accuracy suffers. The present experiments have demonstrated the similarity of stimulus control and trace supraordinate stimulus control. The present experi- ments also failed to support a cagnitive, information-based, rehearsal position to account for animal memory behavior. The notion that stimulus control of DMTS is based on response competition at the time of testing rather than during the delay interval was supported. When trace supra- ordinate stimuli were correlated with different behaviors, performance suffered. But when they were correlated with similar behaviors, per- formance remained accurate. This finding suggests that end of delay interval behavior established during baseline must be maintained in the test conditions in order to obtain accurate DMTS performance. This finding also indicates that maintenance of specific behavioral chains is not necessary to obtain accurate DMTS performance. The present results also serve to extend recent formulations of memory retrieval processes. It is suggested that in the present experiments bodily orientation towards the response keys may have established the visual context necessary for retrieval of a representation of the most recent sample stimulus, thereby affecting accurate matching. LIST OF REFERENCES LIST OF REFERENCES Blough, D.S. Delayed matching in the pigeon. Journal of the EXpep- imental Analysis of Behaviop, 1959,‘2, 151-160. Carter, D.E. & Werner, T.J. Complex learning and information processing by pigeons: A critical analysis. Journal of the Experimental Analysis of Behavior, 1978,‘§9, 565-601. Catania, A.C. & Reynolds, 6.8. A quantitative analysis of the responding maintained by interval schedules of reinforcement. Journal of the Experimental Analysis of Behavigg, 1968, Supplement to Vol. 11, 327-383- D'Amato, M.R. Delayed matching and Short-term memory in monkeys. In G.H. Bower (Ed.), The psychology of learning and motivation (Vol. 7). New York: Academic Press, 1973. D'Amato, M.R. a Wordham, R.W. Retrieval cues and short-term memory in capuchin.monkeys. Journal of Comparative apd Physiological Psychology, 1974,‘§§, 274-282. Denny, M.R. A learning model. In W.C. Corning J: 3.0. Ratner (Eds.), Chemistpy of learning. New York: Plenum Press, 1967. Grant, D.S. & Roberts, W.A. Sources of retroactive inhibition in pigeon short-term memory. Journal of Ex rimental Ps cholo : Animal Behavior Processeg, 1976,‘g, 1-16. Lett, B.T. Long delay learning: Implications for learning memory theory. In N.S. Sutherland (Ed.), Tutorial essays in experimental psychology. New York: Academic Press, 1978 Lewis, D.J. Psychobiology of active and inactive memory. (Psycholggical BulletinI 1979, §§, 1054-1083. Maki, W.S. Discrimination learning without short-term memory: Dissociation of memory processes in pigeons. Science, 1979, 29-11" 83-850 Maki, W.S. & Anundson, D.K. Stimulus control of rehearsal in pigeons: An analysis of directed forgetting. Paper presented at the meeting of the Midwestern Psychological Association, Chicago, Ill., 1979. (Available from the first author at North Dakota State University.) Medin, D.L., Reynolds, T.J. & Parkinson, J.K. Stimulus similarity and retroactive interference and facilitation in monkey short-term memory. Jougnal of Expepimental Psychology: Animal Behavior Processes, 1980,.6, 112-125. Meehan, E.F. The acquisition of trace supraordinate stimulus control in pigeons. Animal Learnipg & Behavior, 1979:.2: 473-476. Mintz, D.E., Mourer, D.J. a Weinberf, L.S. Stimulus control in fixed ratio matching-to-sample. Journal of the Experimental Analysis of Behavior, 1966,‘2, 627-630. Nevin, J.A. The study of behavior: Learning, motivapion, emotion, and instinct. Glenview , 111.: Scott, Foresman and Company, 1973. Reynolds, 6.3. A prcimer of operant conditioning (Revised Edition). Glenview, 111.: Scott, Foresman and Company, 1975. Rillingh M. Stimulus control and inhibitory processes. In W.K. Honig1& J.E.R. Staddon (Eds.), Handbook of operant behavior. Englewood Cliffs, N.J.: Prentice-Hall, 1977. Roberts, W.A. & Grant, D.S. An analysis of light-induced retroactive inhibition in pigeon short-term memory. Journal of Experimental Psychology: Agimal Behaviongrocesses, 1978,‘fl, 219-236. Shimp, C.P. & Moffitt, M. Short-term memory in the pigeon: Stimulus- response associations. Journal of the Experimental Analysis of Behavior, 1974,122, 507-512. Skinner, B.F. & Morse, W.H. A second type of "superstition" in the pigeon. The American Journal of Psychology, 1957, 29, 308-311. Spear, N.E. The processing of memories: Forgetting and retention. Hillsdale, N.J.: Lawrence Erlbaum Associates, 1978. Stonebraker, T.B. The effects of cue delay on directed forgetting in the pigeon. Masters Thesis, 1980. (Available from Michigan State University.) Stubbs, D.A., Vautin, S.J., Reid, H.M. & Delehanty, D.L. Discriminative functions of schedule stimuli and memory: A combination of schedule and choice procedures. Journal of the Expephmental Analysis of Behavior, 1978, 2 , 167-180. Tranberg, D.K. a Billing» M. Delay-interval illumination changes interfere with pigeon short-term memory. Journal of the Expepi- mental Analysis of Behavigp, 1980,‘1, 39-49. Weisman, R.G. Factors influencing inhibitory stimulus control: Differential reinforcement of other behavior during discrimination training. Journal of the Experimental Analysis of Behavior, 1970, 14, 87-910 45 Wilkie, D.M. & Wilson, S.C. Control of pigeon's pecking by trace stimuli. Journal of the Experimental Analysis of Behavior, 1977, 221 293-299 - Zentall, T.R., HOgan, D.E., Howard, M.M. & Moore, B.S. Delayed matching in the pigeon: Effect on performance of sample-specific observing responses and differential delay behavior. Learning_and Motivation, 1978,‘2, 202-218. ’ "111111111111111111111111'111fl1111111'“