m I I < W I ! 'H I ‘lJl'IHIHIHIIJ ”HI {I 145 172 HTHS EEEF'ECFE Q? 'E'EM‘E. GEN GE'VE WAGE? RESPQVEQNG EN. FEGE'SiVEg EM“: {0? E'Em Deqme 0E My A. REECEEGEN SEEK“. UREVEESETY Thomas 3’. Kramer 1968 THESIS 'a u" L ‘3 (JBRARY' i‘»’lichignr first: University V”? f..- - ~! 7‘ MN} m. .- —._— . EFFECTS CF TIME CUT (N SFACED RESTCNDING IM PIG.CN3 L13 by Thomas J. Kramer This eXperiment was concerned with the temporal pat- tern and rate of responding of pigeons on a differential- reinforcement-of-low-rate-of-responding schedule (oil), where the subjects received reinforcement for all responses terminating interresponse times (IRES) greater than or equal to a certain time value. if Ihl refers to the time which intervenes between any two consecutive responses. The purpose of this experiment was to see if pigeons could be trained to adjust their rate of reSponding to the delay requirements of a DBL schedule of reinforcement by use of Special discrimination training involving time out from reSponding for all liTs greater than or equal to that time value. In addition, the effects of this discrimina- tion procedure in controlling reSponding on a URL schedule were compared to the effects of the punishment procedure using shock in other studies. Three hhite Carneaux pigeons were trained to peck a translucent disk in a standard Skinner box on a DRL ZO-sec schedule for twenty daily sessions. During alternate sessions for the next thirty sessions, every response with an IRT of less than 20 sec was followed with a time out of either 5, 10, or 20 sec during the entire session. A time out was accomplished by completely darkening the eXperinen al chamber. At the end of this BO-session period, all birds were run on URL ZO-sec for 15 additional sessions, after which the delay requirerent necessa y for reinforcement was increased to 30 sec (DdL 30-sec) for ten sessions. Following this, all responses with IRTs of less than 30 sec were followed by a time out during alternate sessions for the next thirty sessions, using the same time-out values and procedure as with URL ZO-sec. The results were analyzed using the Ide/OF statistic rhich estimates the probability that the subject will respond within a certain tine interval, given that he has reached the initial boundry of the interval and thus has an opportunity to respond within that interval. Inapec- tion of the IlTs/OP curves showed little difference in responding among the three tine-out values, both at DRL 20-sec and at URL BO-sec. While temporal discriminations did not emerge during the initial ZO-session eXposure on DRL ZO-sec, all birds formed a temporal discrimination while the tine-out procedure was in effect at all RO-sec, and maintained this discrimination after the termination of the time-out procedure, although not to the same degree. Hhen the delay requirement was increased to 0 sec, the rate of responding decreased and the temporal discrimina- tion was preserved. Reinstatement of the time-out pro- cedure further decreased the rate of responding while sharpening the discrimination. In general, the effect of the time-out procedure was similar to effects achieved in other studies when all responses were punished with low intensities of shock, but superior to it in that responding did not revert back to the initial level prior to discrimination training, as is the case with punishment. Datefil-éf 1953‘ Chairman WM/g‘ EFFECTS OF TIME OUT ON SPACED RESPONDING IN PIGECNS By ‘9‘) m. ,, afl‘l. inomas u. hramer A THESIS dubmitted to Fichigan state University in partial fulfillment of the requirements for the degree of hASTER OF ARTS Department of Psychology ACKNOWLEDGEMENTS The author wishes to xpress his sincere appreciation to Dr. hark Killing, chairman of his conmittee, for his instruction, guidance, and the use of his equipment through- out all stages of this research. Thanks are also due to 'W urs. Ralph Levine and h. Ray Denney for their helpful criticism and advice.. TABLE CE" CC:~1TEIETS II‘ETRCJUCTIfh. . . . . . . . . . . . ”Tao RES'LTLT.5.............. Ho Ho Ho *0 9) (7'). (>14 ‘3 Table T«£€ l. The response rate and reinforcement rate for each bird before, during, and after the time-out procedure (T0) at DRL ZQ-sec, and before and during the time-out proce- dure at URL 3U-sec. . . . . . . . . . . . . 28 Figure 1. LIST CF FIGURfiS Three methods of depicting the probabil- ity of interresponse times. . . . . . . . Ide/CP Values as a function of IRT cate- gories for eacn of the last three session combined at each of the three time-out (TC) values during URL ZO-sec and DEL 3C- sec for rigeon fi 45. . . . . .‘. . . . . IRTs/OP values as a function of the IhT categories fvr Tigeon ; 45 tn DBL ZO-sec before time out (3 F61; Tf), DRL ZO-sec during time out (DRL 20 TC), and DRL 30- sec during time out (DRL 30 TC). . . . . IXTs/CP values as a function of the IRT categories for Pigeon } 23 on DRL ZU—sec before time out (BLFCRE TC), DAL ZO-sec during time out (DRL 20 TC), and DRL 30- sec during tire out (JRL 30 TC); . . . . Ide/O} values as a function of the IR? categories for Eigeon j 21 on Jdt ZC-sec before tine out (B1?fl; Tt), DdL ZO-sec s during time out (JRL 2C TC), and DRL 30- sec during tine out (DdL 30 TO). . . . . IhTs/OF values a function of the IiT (a) 93 categories for Tigeon # #5 on JRL ZO-sec before time out (311 20 BLFCRE TC), DEL V S Page 16 18 ZO-sec after time out (URL 20 AFTER TC), and UdL BO-sec before time out (URL 30 BLFCRL TC). . . . . . . . . . . . . . . 21 IRTs/CP values as a function of the ldT categories for Tigeon # 23 on URL 2C-sec before time out (JRL 20 euros; Tr), URL 20-sec after time out (DRL 20 AFTER Tf), Ide/CP values as a function of the 11? categories for Pigeon f 21 on URL zU-sec before time out (DRL 2U Bannn TO), URL (— L-SE after time out (URL 20 AFTBd TC), IN) 0 and DRL 30-sec before time out (UdL 30 BbFL‘RE TC). 0 o c o o o o o o o o o o o 25 L 1.3T CF AF 1“ LED ICES . ' INTRODUCTION A schedule of reinforcement which has received increasing attention in the past few years is the differ- ential reinforcement of low rates of responding (URL), where the subject receives reinforcement for all responses terminating an interresponse time (IHT) greater than or equal to a certain time value. An IRT refers to the time which intervenes between two consecutive responses. For instance, if a subject is exposed to a URL ZO-sec schedule all IRTs of 20 sec or more are followed by reinforcement, whereas all IRTs of less than 20 see are not. It is important to note that there is no external stimulus specified by the schedule which is correlated with time. Little success has been achieved with pigeons in forming a temporal discr’mination on DdL schedules. Reynolds (1964a,b) has noted that pigeons perform at a very inefficient rate on a URL ZO-sec schedule while maintained at tO% of their free-feeding weight. He re- ported that the performance after over 100 sessions on the schedule was basically unchanged from that after six sessicns. The subjects emitted very few responses with long IRTs, and the data presented gave no indication that the subjects had formed a temporal discrimination. But Reynolds noted in a later article (1956) that pigeons performed poorly because they seemed to be unable to in- hibit responding, but nevertheless do form some sort of temporal discrimination. Staddon (1965), after exposing pigeons to a variety of schedules with varying delay requirements for 255 sessions, found that pigeons performed close to the max— imum rate of efficiency up to about DHL 20-sec and appar— ently formed a temporal discrimination, but that the rate of responding at DRL BO-sec increased over that at URL ZO—sec for two of the three birds. He concluded that few pigeons, when eXposed to URL BO-sec, could adjust their rate of responding to the schedule requirements in a way comparable to their adjustment to shorter values. Staddon suggested that this may be accomplished by special train- ing procedures. Punishment has_been used to reduce the rate of re— sponding of pigeons on a DRL schedule (H012 and Azrin, 1963; Holz, Azrin, and Ulrich, 1963). Punishment of all responses with shock decreased the rate of responding as a direct function of the shock intensity, and in particu- lar reduced the number of short IRTs, even at low inten- sities. But after punishment was discontinued, the rate of responding as well as the temporal pattern of respond- ing returned to normal. Thus, the punishment had no last- Iing effect once it was removed, and the subjects did not form a temporal discrimination. The purpose of the present experiment was to see if pigeons could be trained to adjust their rate of respond- ing to the delay requirements of a DRL schedule of rein- forcement by the use of Special discrimination training ‘ involving time out from reSponding for all IRTS less than the minimum reinforced value, and reinforcement for all IRTs greater than or equal to this same value. In addition, the effects of this discrimination procedure in controlling responding on a DRL schedule were compared to the effects of the punishment procedure in other studies using shock. The subjects were three eXperimentally naive male white Carneaux'pigeons maintained at 80% of their free- feeding weight. They were purchased from the Palmetto Pigeon Plant and were approximately five years of age at the start of the experiment. Apparatus \ The eXperimental chamber in which the subjects were tested was 12 X 14 K 13 in. contained within a larger ice chest. A fan, mounted on the door of the chest, provided ventilation and masked any extraneous sound. The front panel of the experimental chamber contained three Lehigh Valley Electronics plastic pigeon keys which were mounted 8 1/2 in. above the floor with a horizontal separation of 3 in. Each key, mounted behind a l-in. diameter hole, could be illuminated fror behind with light. A force of approximately 15 gm was required to close the key. A rectangular opening located below the center key permitted access to the grain magazine which was raised to the feeding position and lighted during a 6-sec period for reinforcement. Two house lights were located above the response keys on the front panel. The programing equip- ment consisted f a system of electromechanical switches and timers, and IRTs were printed out on a Lehigh Valley Electronics 9 channel serial-parallel entry printer, model 1660-9. The programing and recording equipment were located in a separate room. Procedure During session one the house lights were on and the center key was illuminated with white light. The food magazine was repeatedly presented to each bird until he ate promptly upon presentation. Immediately after a peck of at least 15 gm on the center key, the food magazine was presented. Hhile the food magazine was presented, the center key light was always off. Every reSponse on the center key was reinforced. The session was terminated with 25 reinforcements which occurred in rapid succession. dtarting with session tho the house lights were on and the center key illuminated with white light. The subjects were placed on a DRL t-sec schedule, where t represents the IRT from either (1) the last reSponse, (2) the last reinforcement termination, or (3) the start of the session, whichever occurred most recently. For session two t was 3 sec and was progressively increased by 1 sec each day until DEL 20-sec was reached. The criterion for termination of a daily session throughout the entire eXperiment was either (1) two hours of presen- tation of the illuminated center key, or (2) 25 reinforce- ments, whichever occurred first. The subjects were run on the average of six days out of seven. All birds continued on JdL 20-sec for twenty sessions. 0\ During alttrnate sessions for the next thirty sessions for all birds, every response to the center key with an IRT of less than 20 sec was followed with a time out of either 5, 10, or 20 sec during the entire session. A time out was accomplished by turning out the house and key lights which completely darkened the experimental chamber. All responses with IRTs of 20 sec or more did not produce a time out and were followed with reinforcement. The order of occurrence of the three time—out values was variable for each bird, five sessions being given at each of the three values. For those sessions in which time outs occurred, IRTs were measured from either (1) the start of the session, (2) the last reinforcement, or (3) the last time-out termination, whichever occurred most recently. For those sessions in which time outs did not occur, the procedure was as before at URL 2C-sec. At the end of this 30—session period, all birds were run on URL 20-sec with no time out for an additional 15 sessions, after which the delay requirement necessary for reinforcement was increased to 30 sec (DRL 30-sec) for ten sessions. Following this all responses with IRTs of less than 30 sec were followed by a time out during alternate sessions for the next thirty sessions, using the same time-out values and procedure as with JRL 20—sec. The results of a study involving responding on a DRL schedule are usually presented in one or several graphs with the abscissa representing categories of IITs in seconds (0—1, 2—3, 4-5, etc.) and the ordinate represent- ing either IRTs/CP (interresponse times/opportunities) or IRTs/Total. lRTs/UP is a statistic which estimates the probability of a response occurring in a certain time interval on the condition that the subject reaches the initial boundry of the interval, and hence has the Oppor- tunity for a response in the interval. The IiTs/OP sta- tistic is calculated by dividing the number of reSponses with IRTS which fall into a certain category by the number of responses with IiTs as long as or longer than the lower limit of the IRT category in question. Therefore, the I Ts/Ci statistic is a conditional probability statistic while Infs/To is a relative frequency measure. These two wt 1 sti 1cs can often show quite different pictures of the results leading to opposite conclus ions. Anger (1956, 1963) pointed out that there is good reason for believi ing that comparisons cf relative freauency reasures of l de between suLjects or among different conditions for the seas subject may oe inap,icxi {'te for many purposes, par- ticularly temporal discriminations. The opportunity for .1.- for responses to occur with short ihTs is much greater than for long since every respcnse is fcllered it}: the op; crtunity to er:it a re prnse in the shortest IxT cate- gory. This can 1e easily seen from Fig. l vhich shOhS three curves dratn frrm actual. data. The triangles retre- sent the percent of the available Opportunities for a response to occur in each 2-sec category. This curve shows the e:h;Ip ce r ease in the number of opportunities rt first and then the steady decrease thereafter. The relative frequency curve (squares) and the lhTs/CE curve (circles) are quite different. The relative frequency curve shows little or no evidence of a temporal discrim- ination, while the lils/LP curve shows an increasing probahility of responding as the length of the IRT in- creases. The points plotted on the IRTs/GE curve night he read as follows: given that the subject waited t sec or longer, what was the probability that he responded in the t - t+2 sec interval? Thus, the IRTs/CF statistic equ tes all IRT categories by using the Opportunities for reSponses to occur in.a category as the denominator. The desireability of this measure is frrther enhanced by the fact that, when no temperal discrimination exists in reSponding on a DhL schedule, rough equality can be ex- pected in the various liTs/(P v lues for the different IRT categories. Thus, a peak in the ATS/(' values at or near the minimum reinforced IRT is evidence for a ten poral discrimination. This discrimination is often obscured by the relative frequency graph due to the much greater Fig. 1. Three methods of depicting the probability of interresponse times. The IRTs/CP curve shows the number of Ide in each class divided by the total number of Opportunities for Ide in that class; the IHTs/N curve shows the number of IRTs in each class divided hy‘the total number of IRTS during the session; the third curve shows the percentage of the total IRTs that were greater than the lower limit of each IRT category. 10 misfit/z emfm z_ memoomzo F: A o_ Om Om z,ao I em. 0% exodus on sea mm ALFHBVBOHd F4 number of orportunities for short InTs. colsecuentlr it 1y, seems that fa lure to find evidence for a temporal dis- crirination in the relative ire~uency distribution of ILTs 1 V inccncl us ive; the IR”iS/C curve shou Id be studied. }_.'0 Visual inspection of th. lATs/(T curves plotted as a function of the IAT categories for each bird during the time-out procedure at DBL ZD-zec and URL BO—sec showed that there tas very little dii ference in respanding as a function of the time—out duration within any bird. Tie 2 10. shows tie ILTs/LI values as a function of the :1? category for each of the three time-c ut valu.es dLri g URL ZO-sec and JRL EO—sec for Iigeon # 45. The curves for the other two birds were quite sinilar in variability. Ti13s. 3-5 show the L Ts/(I values plotted as a func— Fr tion 6. ategories for days lb-ZO combined on DRL 20- so, the last three sessions of time out at each of the three time-out values coinineo at DEL ZO-sec, and the last three sessions of time out at each of the three time-out'values combined at DdL BO-sec for each bird. Althourh the curves thou some dissimilarities across birds, distinc‘t ive trends stazd out for all birds. The curves fwr DBL 2u-se eprior to any manipulations for each bird show a high probability of responding in the shortest category, followed by a sharp drOp for the 2-4 sec cate- gory, and then a steady rise. All reach an initial peak at about t sec and fluctuate about that value at longer Isz, aptro oximatin- equality. This is in marked contrast ,7 C) l2 Fig. I. IRTs/OP values as a function of IRT categories for each of the laSt three sessions combined at each of the three time—out (TC) values during DiL ZO-sec and JR BO-sec for Iigeo: # 45. he InTs/CI valu s were computed where the oyjortunities were less than 20. The numbers on the abscissa represent the lower limit of each 2-sec Ix? Category. 13 dO/Sial 3O INTE RVA LS IN Z’SEC IRT CATEGORIES 7’1 ' rig. ,. IRTs/OF values as a function of the IiT categories ), for Pigeon # #5 on DEL ZO-sec before time out (“?F 1“ TC 121.: I) ‘. DRL ZO-sec during time out (JJL 2C TC), and DAL BO-seo curing time out (DRL 30 TC). The numbers on the abscissa represent the lower limit of each 2—sec IRT cateeory. 15 m4<>KMFZW Ummlm 7: wmioomeqo em. / ti .,, \\v\\/\\ ON 0. 4 i a n . n _ ll I. k \\ \ \ \ \ t m? OH Om .EQ. . 9. ON nEQ- .. OH mmOmmmo o dO/Sléjl Fig. A. IiTS/CY values as a function of the IfiT categories for Pigeon # 23 on DBL BO-sec before time out (BEFORE TC), DRL ZU-se during time out (DRL 20 TO}, and DBL BO-sec during time out (DRL 30 TC). The numbers on the abscissa represent the lower limit of each 2-sec IfiT category. l7 m4<>m mez. umm-m Z. mmEOOMCNU .5: MN 0% On .EQ. . OH Om lie-Ill- Oh mmOummI N. m. dO/Siéji 18 is. 5. IRTs/CP values as a function of the IHT categories *n for Pigeon # 21 on URL ZO-sec before time out (BEFORE TO), DRL ZO-sec during time out (DRL 20 TC), and net BO-sec during time out (URL 30 T0). The numbers on the abscissa represent the lower limit of each 2-sec IxT category. l9 m4<>a mez. 7. .) C N Z _ mmfiioomtflku E. O. _ m K TIL. oe on sea Op. ON #5- I. OH mmoufimo dO/Slhl to the curves representing responding under the‘time—out condition at DEL 20-soc. The probability of a response in the shortest category is near zero for each bird, in— creasing monotonically to a peak probability just prior to the reinforced value for Pigeon # 21, and a peak for Pigeon # 45 aid Iigeon # 23 near 2h sec. The curves for W the time-out condition at JRL BG-sec follow the same general pattern as those for the time-out condition at URL ZO-sec. Each increases monotonically from a low initial value, but at a slower rate. Each curve peaks roughly at the same place relative to the minimum rein- forced value for each birc as it did at DdL ZO—sec. Figs. 6-8 show the Ide/OP values for each bird plotted as a function of the IRT categories combined for days 16-20 on URL ZO—sec (replotted as in Figs. 3-5), days 11-15 after the end of the time-out procedure at URL ZO-sec, and days 6—10 of DRL 30-sec, just prior to the start of the time-out procedure at that value. The curves for all three birds still show the presence of a temporal discrimination, although not to the same degree as when the time-out procedure was in effect. In addition, the discrimination is not as sharp as for the time-out cen- diticn, perticularly around the ar‘a of the reinforced values. The important fact is that responding did not revert back to the level which existed prior to the insti- tution of the time out. It can also be seen that all. three birds adjusted their rate of reSponding toward the G Pie. 6. IRTs/CP values as a function of the IRT categories 4 for Pigeon # 5 on DEL ZO-sec before time out (DRL 20 BEFORE TC), DEL ZO-sec after time out (JRL 20 ATTBH TC), and URL BO-sec before tire out (DAL 30 BEFORE TC). The numbers on the abscissa represent the lower limit of each 2-sec lRT category. 1 VI \- II ”I O K O v. m. 8 some on 4%] m. dO/Slcjl 23 Fig. 7. IRTs/OP values as a function of the IRT categories for P'geon # 23 on DEL 2Q-sec before time out (DRL 20 ,V I *FCRL TC), DhL ZO-sec after time out (DRL 20 AFTLE TO), ['13 f. and DEL BO-sec before time out (DRL 30 BEECH; TO}. The nurbers on the absccissa represent the lower limit of each 2—sec IaT category. 23 21+ IO CATEGORIES dO/Slhl INTE RVA LS SEC 2- IRT 25 Fig. 8. IRTs/CP values as a function of the IRT categories for Pigeon 4 21 on URL EO-sec before time out (URL 20 assess TC), JRL ZO-sec after time out (DRL 20 AFT,x TO), and fiL BO-sec before time out (URL 30 BEFCRE TO). The numbers on the abscissa represent the lower limit of each 2-sec IRT category. 25 Z . m mU<>mmeZ_ umm .. mmEOome/‘o em: 0_ O N) “O (\J IV _m 9. wmobm on as] E. $E< ON #571. OF mmodmm Om lie-Iii 7 J v. m. w. dO/SlHI 27 delay requirement at DAL BU-sec prior to the start of the time-cut proCedure at that value. Table 1 presents the rate of responding during the conditions shown in the previous graphs as well as the rate of reinforcement for the same conditions. All birds de- creased their rate of responding during the time-out sessions at DdL BU-sec over time- ut sessions at URL 20-sec. In addition, the percent decrease in the rate was quite comparable for all birds; 24.2% for Pigeon # 45, 23.8% for Pigeon 3 23, and 21.h~ for Pigeon # 21. The rate of responding after the time-out period at URL ZO-sec was lower than the rate prior to the time-out condition for all birds. The rate of responding on URL BO-sec prior to the time-out procedure was lower than the rate at URL ZO-sec after tie time out for all birds, further confirming the observation that all birds adjusted their rate of respond- ina to the schedule requirements. The reinforcement ratios included in Table l are in aareement with the data on respons rate. The reinforce- ment rate at DEL 20—sec during time-out sessions is higher than either before or after the time-out period for all birds, despite large individual differences. The reinforcement rates during time-out sessions at URL BO-sec are again uniformily higher than before that procedure. Finally, the rate of reinforcement at URL BO-sec during, time-out sessions is equal to or greater than the rate I" at 93L 2o—sec prior to exposure to time out. Table l. The reSponse rate and reinforcement rate for each bird before, during, and after the time-out (T0) procedure at URL 20'560’ EHd before and during the time-out Procedure at URL 30-sec. l Beeponses/min Reinforcements/min Condition # #5 # 23 d 2 § #5 # 23 # 21 _ URL 20 before T0 8.9 8.8 9.7 0.32 0.2 0.04 URL 20 during To 3.3 4.2 5.6 1.16 0.51 0.12 URL 20 after T0 5.4 7.8 6.8 0.97 0.38 0.06 URL 20 before T0 4.6 5.6 6.0 0.61 0.15 0.02 URL 20 during T0 2.5 3.2 0.4 0.68 0.25 0.04 DISCUSSICN ‘ When time outs followed responses with IiTs which were less than the minimum reinforced value, all birds formed a temporal discrimination. It is unlikely that the birds would have formed this discrimination within 55 sessions since Reynolds (1904b) has shown that pigeons do not perform any better after over 100 sessions than they did after 0 sessions on URL 20-sec. The effect of time out in controlling responding on a URL schedule is roughly similar to the effect achieved with low intensities of shock (Holz and Azrin, 1963; Holz, Azrin, and Ulrich, 1903). But the effect of time out and shock are not exactly the same. Then reSponding was no longer punished with shock, responding reverted back to the pre-shock level, whereas with time out the birds con- tinued to show a temporal discrimination, although there was some loss in discrimination. It should be pointed out that this difference in the rate of responding after re- moval of punishment as Opposed to the removal of time out could be due to procedural differences, since in the studies cited above all responses were followed with shock, whereas in the present experiment only those responses which were not reinforced were followed with a time out. The controlling effect of the time-out procedure is further stubstantiated by the decrease in the rate of re- 29 30 spondina and corresponding increase in reinforcement rate at JiL BO-sec from that at URL ZO-sec for all birds, both with and without the time-out procedure in effect. Staddon (1965) has presented IRTs/C? curves for three birds at DRL ZO-sec after approximately 220 sessions of responding at various DEL values which are very similar to those pre- sented here for DRL ZO-sec with time out. But after apprx- imately 255 sessions, only one bird showed a temporal dis- crimination at URL 30-sec while the rate of responding for the other two birds increased over the rate at URL ZO—sec. The present data show that all three birds formed a tem- poral discrimination at DAL BO-sec, even prior to the start of the time-out procedure at that value. Thus, pigeons do seem to be able to adjust their rate of responding to the delay requirements of a DEL BO-sec schedule, at least when special techniques are used. The fact that there was little difference in reSpond- inn under the three time-out Values has important theore- tical implications. It has been suifiested (helleher, Fry, .and Cook, 1959; Anger, 1903; Slough and Yillward, 1965) that the important factor involved in temporal discrim- inations is the interoceptive stimulation produced from the last response and nOt any exteroceptive stimulation, such as a chain of behavior. If this were the case then it wculd be eXpected that the birds in this experiment would show a difference in responding under each of the three time—out conditions, since the actual delay necessary 31 between an unreinforced response and the availability of reinforcement was either 25, 30, or 40 sec at URL ZO-sec and 35, 40, or 50 sec at URL BO-sec. Since prior research indicates that it is unlikely that pigeons could adjust their rate of reSponding to the delay requirements of a DRL 50-sec schedule, the subjects in this experiment seemed to be estimating the interval by the length of time that the lights were on and not by the length of time from the last response. Therefore, the stimuli that seem to be controlling the discrimination in this experiment do not appear to be the response produced stimuli from the last key peck. The high probability of responses in the shortest IRT category, which occurred during those conditions dien the time-out procedure was not in effect, was quite similar during all three conditions for two of the three birds. bidman (1956) referred to these short IfiTs as bursts, defined as any sequence of two or more responses in which no consecutive responses are separated by more than 2 sec. Sidman showed with rats that the probability of a burst occurring increased as the length of the previous 13? in- creased, reaching a peak probability just prior to the minimum reinforced value. But there is little published data which has substantiated this same relationship with pigeons, although hols and Azrin (1963) indicated that they obtained a partial relationship. In fact, those studies which have directly investigated the presence or 32 abscence of this relationship in pigeons, showed that it was absent (Blough, 1963, 1966). Inspection of the present data for each bird did not reveal any consistent relation- ship between bursts and the length of the previous 13?. In addition, the probability of a burst was practically unchanfied before and after time out at DRL 20-sec and before time out at URL BO-sec while the shape of the IRTs/O. curves were quite different. This suarests that bursts of responding have a separate function, _ossibly that of providing additional stimulus feedback on a DRL schedule. Two additional facts support this idea; (1) bursts rarely occurred during time-cut sessions, the time out providing immediate feedback, and (2) bursts never occurred if the response was reinforced, the response key light and magazine operation providing immediate feedback. Collateral behavior. It was observed that Pigeon # 45 developed a well-defined chain of resoonses which effectively filled in the time between responses on both the DEL 20 and URL 3U-sec schedule. This bird would move to the back of the experimental chamber after a response and pace back and forth four to six times. He then approached the left front corner of the chamber, from where he would mdic a complete turn, and then peck the response key. This behavior is very similar to that re- ported for one bird by holz, Azrin, and Ulrich (1963). The other two birds did not form any repeated chain, but both birds consistently turned away from the response key after pecking and moved toward the back of the chamber. The lack of a consistent pattern of behavior in all birds between responses, combined with the observation that all birds moved away from the response key after pecking, seems to indicate that at least some kind of competing behavior might be an important factor in the ability of pigeons to form a temporal discrimination on a DRL schedule. The recognition of this competing behavior, if it does exist in most birds, can only be achieved through direct observation over a long period cf time. This could account for the lack of similar observations in the liter- ature. Concludina_remarks. In general, the function of the time out as a special technique for bringing spaced respond- ing under temporal control seems to be that of making the passage of time more salient to the animal. The assumption that animals do not normally attend to time as the rele- vant variable on a URL schedule is supported by the find- ings of Bruner and Bevusky (1961), who found that human subjects, exposed to DBL 8.2-sec Limited Fold 2.25-sec had no idea that reinforcement depended in any way upon the passage of time, as judged from post-experimental inter- views. dtaddon (1965) also suggested that exposing pigeons to various delay requirements over a long period of time makes the passage of time more salient to the anima . a. Anger, n. The dependence of lRTs upon the relative rein- forcement of different interresponse times. J. comb. physiol. Psychol., 956,,2g, 146—161. Anger, D. The role of temporal discrimination in the rein- forcement of Sidman avoidance behaVior. J. exp. final. Behav., l963, §, #77-506. Blough, D. 5. Interresponse times as a function of con- tinuous variables: a new method and some data. g. 237-246. exo. Anal. Behav., 1963, o ....9 Blough, D. S. The reinforcement of least-frequent inter- reSponse times. J. exp. Anal. Behav., lQéo, 2, 581- 591. Blough, D. d and Fillward, h. B. Learning: operant con- ditioning and verbal learning. Annual Rev. Psychol., 1965, _1_;_, 63—94. Bruner, A. and Hevusky, S. H. Collateral behavior in humans. J. exp. Anal. Behav., 1961, g, 3&9-350. 'Holz, I. C. and Azrin, N. H. A comparison of several pro- ’ cedures for eliminating behavior. J. exg, Anal. Behav., 1963, Q» 399-h06. Holz, H. C., Azrin, N. H., and Ulrich, R. B. Punishment of temporally spaced responding. J. exp.‘Anal. hehav., 1963, _6_, 115-122. hellehar, n. 1., Fry, N., and Cook, L. lnterresponse times 34 \J) \ 3'1 distribution as o functinn of differehtial reinforce- ment of temporally spaced responding. J. exn. final. _A' Behav., 19 9, g, 91-106. Reynolds, G. 3. Accurate and rapid reconditioning of spaced responding. J. exp. Anal. Behav., 190A, 1, 273-276. Reynolds, G. d. Temporally spaced responding by pigeons: t development and erec s of deprivation and extinction. J. exp. Anal. Behav., 1964, Z, hlB-th. Reynolds, G. S. Discrimination and emission of to peril intervals by pigeons. J. expJ Anal. Behav., 1966 65-58. Sidman, N. Time discrimination and behavioral interaction in a free Cperant situation. J. comp. physiol. Psychol., 1956 , £62, 4569-473 . Stadcon, J. E. R. dome preperties of spaced responding in pigeons. J. eXp. final. Pebav., 1965, Q, 70-27. —-o/ A I I EN l) I}; A Pigeon ¥ 21 Reaponse Frequency IXT URL 20 DRL 20 DRL 20 DRL 30 DRL 30 cate- before during after before during series TC TC T0 T0 T0 (2-sec) (5 days) (7 days) (5 days) (5 days) (9 days) 0-1 1513 199 687 889 110 2-3 572 2A6 186 73 150 6-5 756 373 3h0 169 269 6-7 1053 631 470 263 #18 8-9 950 874 665 606 518 10—11 508 s65 674 468 515 12-13 209 760 518 Ah3 507 14-15 107 156 263 390 525 16-17 63 211 85 268 483 18-19 21 99 33 2d 380 20-21 12 47 17 75 295 22-23 2 19 5 39 139 54-25 5 11 5 12 99 26-27 2 5 h 11 75 25-29 0 6 a 9 35 30-31 0 1 O 5 lo 32-33 0 C C 3 12 34-35 0 2 O l- 7 36-37 0 3 O O 1 38+ 0 3 1 1 8 36 \o \1 Piaeon i 23 ReSponse Frequency 11 DRL 20 DRL 20 DRL 20 DRL 30 DRL 30 cate- before during after before during gories 1%) TC} TC) TC: T0 (2-sec) (5 days) (9 days) (5 days) (5 days} (9 irfs) C-l 1371 41 849 1176 79 2-3 127 55 169 134 78 4—5 433 100 299 212 88 6-7 816 9 191 159 127 8-9 590 148 190 141 128 10-11 308 190 191 158 199 12-13 187 257 167 162 203 14-15 107 276 141 191 215 16-17 92 294 122 178 236 18-19 59 173 81 166 276 20-21 46 110 58 180 277 22-23 19 6O 36 144 214 24-25 14 27 13 123 217 26-27 ll 7 9 75 177 28-29 8 2 7 55 120 30-31 4 1 3 32 81 32-33 2 2 2 16 47 4-35 4 O 1 10 33 ‘6-37 0 Q 1 ll 14 8 + 3 3 l 16 36 \JO ’\‘\ \N Pigeon # 45 I ReSponse Frequency IiT DRL 20 DRL 20 DRL 20 DRL 30 JRL 30 cate- before during after before during gories_ T0 T0 TC TC TC (2-sec) .(5 days) (9 days) (5 days) (5 days) (9 days) 0-1 2019 3 311 503 22 2-3 27 11 4 15 28 4-5 101 .18 4 9 6 6-7 515 10 3 4 10 8-9 825 11 3 12 11 10-11 538 21 15 ‘ 4 13 12-13 307 33 32 9 17 14-15 130 62 49 14 \ 16 16-17 97 125 67 . 12 26 18-19 50 140 78 21 39 20-21 39 104 48 32 54 - 22-23 37 55 27 65 67 24-25 14 18 9 88 85 26-27 7 2 9 76 88 28-29 5 4 9 77 88 30-31 3 l 5 95 65 32-33 6 2 3 33 72 34-35 1 O 3 25 ‘ 29 30-37 1 O 2 17 19 33 + 2 l 11 7 l2 FEB ~71968 LIB 1111'” 5 “1111111(11111