TEMPORAL CONDITEONING EN HUMANS AS A FUNCTION OF INTERTRIAL INIERVAL AND STEMULUS INTENSITY Thesis ‘0» ”to Degree of DE. D. MICHEGAN STATE UNIVERSITY Robert F. Morgan i965 .‘ 'kJ-.u.. \\\\§§\\\\}\\2\\§\\\3§\\\\\1\\(\)\\\\\\\\\1\\\\\;\\}}\\5\\\\\\\% [ML ’1’“? A :1 University This is to certify that the thesis entitled TEMPORAL CONDITIONING IN HUMANS AS A FUNCTION OF INTERTRIAL INTERVAL AND‘ STIMUEUS INTENSITY presented by Robert F. Morgan has been accepted towards fulfillment of the requirements for Ph. D ___degree in Psxchology @293 @‘W Major professor Date November 29, 1965 0-169 ABSTRACT TEMPORAL CONDITIONING IN HUMANS AS A FUNCTION OF INTERTRIAL INTERVAL AND STIMULUS INTENSITY by Robert F. Morgan Past decades of Soviet research on the classical conditioning of a temporal response to a periodic stimulus led to the formation of several testable theories and atti- tudes. Pavlovians consider temporal conditioning to be a respectable member of the conditioning family, subject to the laws of inhibition and differential stimulus intensity; without upper limit as to size of intertrial interval (ITI). Over the years, accumulated Soviet evidence indicated ITls below three to five minutes were difficult or impossible to condition. Sporadic American research came to an opposite conclusion: temporal conditioning became more difficult as ITIs increased to one minute or larger. Major purposes of the following experiments were to resolve the apparent American-Soviet ITI contradiction and to study temporal conditioning phenomena in depth through the use of the larger subject (S) samples, statistical evaluation, and experimental control common to less neglected areas of Arnerican experimental psychology. Robert F. Morgan Male and female volunteer college student SS vocally anticipated a periodic photic stimulus lUS: once each trial in all experiments. Anticipations occurring in the last fifth of the ITI were defined as conditioned responses (CRs). Experiment I gauged conditioning efficacy over 20 trials at ITIs of 45, 135, or 240 seconds and at high (100 watt) or low (7 watt) US intensity. With 3 S5 at each ITI- intensity condition, conditioning level at 135 seconds trailed the shorter and longer ITIs. High US intensity depressed conditioning level at the shortest ITI, raising it at the highest ITI. Introspective reports of methods of synchrony, estimation of ITI length, and stimulus unpleasant- ness were analyzed in this and the following experiments. Experiment II replicated Experiment I‘s conditioning results in 15 trials at ITIs of 30, 60, 90, 120, 150, 180, 210, 240, 270, 300 seconds with high or low US intensity. With 6 SS at each ITI-intensity condition (120 SS total), conditioning level at ITIs of 60 seconds or less and of 210 seconds or more excelled that of ITIs in between. High US intensity again depressed performance at short ITIs (30, 60, sec.) but enhanced performance significantly only at middle range ITIs (150, 180 sec.). Independent conditioning measures of CR frequency and error magnitude showed similar results, demonstrating general increase in conditioning level with blocks of trials. Galvanic skin response (gsr) measures yielded a highly significant positive relationship Robert F. Morgan kxetween pre-experimental relaxation and subsequent perfor- Inance. Gsr readings taken at the mideJnt of every ITI showed a significantly increa51ng level of relaxation at the midpoint over blocks of acquisition trials in the presence of temporal conditioning which was absent or in— consistent when such conditioning was absent. This held true even for SS having overall gsr decrease in relaxation in response to the entire experiment. Gsr findings sup- ported neo-Pavlovian inhibition theory prediction and a complementary discrimination-production hypothesis. Experiment III demonstrated conditioning for 12 S5 at high US intensity ITIs of 60 or 240 seconds to signifi- cantly improve with a second day of 15 acquisition trials and to significantly extinguish with 15 subsequent trials lacking the periodic US. Experiment IV failed to demonstrate significant Pavlov-predicted temporal conditioning over 15 high US intensity acquisition trials with ITIs of 30 and 60 minutes. The 3 S5 at each ITI did show higher estimation accuracy and less mean per cent time counting than S3 at the lower ITIs of the earlier experiments. These experiments generally reconciled American and Soviet thrusts towards the demonstration and understanding of the parameters of temporal conditioning, a gateway to the psychology of time. TEMPORAL CONDITIONING IN HUMANS AS A FUNCTION OF INTERTRIAL INTERVAL AND STIMULUS INTENSITY BY -\ ‘ r '._ Robert F? Morgan A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1965 ACKNOWLEDGMENTS The author gratefully acknowledges the five years of assistance, interest, and training given him by his major professor, Dr. Stanley C. Ratner. This thesis directly evolved from the last of these years but the fundamental technique upon which it was built was more the result of Professor Ratner's full influence than any other natural force. The author is also indebted to the other long time members of his guidance committee for continuing collabora- tion along lines of inquiry some of which converged on this thesis: Professors Paul Bakan, Abram M. Barch, M. Ray Denny of the Department of Psychology, and Professor John A. King of the Department of Zoology. In addition, the author would very much like to acknowledge the role of his wife, Dianne, and his three children, Robby, Barry, and Julie, in initially bringing home the great relevance of temporal methods. ii TABLE OF CONTENTS Page ACKNOWLEDGMENTS . . . . . . . . . . . . . ii LIST OF TABLES . . . . . . . . . . . . . iv LIST OF FIGURES . . . . . . . . . . . . . Vii INTRODUCTION . . . . . . . . . . . . . . l EXPERIMENT I . . . . . . . . . . . . . . 22 Method Results Discussion EXPERIMENT II . . . . . . . . . . . . . . 33 Method Results Discussion EXPERIMENT III . . . . . . . . . . . . . 66 Method Results Discussion EXPERIMENT IV 0 O O O O 0 e O O O O O O O 7 5 Method Results Discussion GENERAL DISCUSSION AND SUMMARY . . . . . . . . 90 REFERENCES O O O O O O O O O 0 O O O O O 9 5 APPENDICES . . . . . . . . . . . . . . . 103 iii Table 10. 11. LIST OF TABLES Some abbreviations and definitions used in Experiments I, II, III, IV . . . . . . . . Data summary for ITI-Int. conditions in Experiment I: sex, age, Sb Un, occurrence of CRBO, estimation of ITI, counting per S Analysis of variance of occurrence of CR80 by ITI, Int., and trial blocks in Experi- ment I O O O O O O O O O O O O O O O O O O Analysis of variance of per cent A. Est. by ITI and Int. in Experiment I . . . . . Occurrence of CR80 as a function of H gsr at 45 sec. ITI and H Int. in Experiment I Number of S5 at each method of synchrony used in Experiment I . . . . . . . . . . . Data summary for ITI-Int. conditions in Experiment II: sex, age, H gsr, Sb Un, occurrence of CRBO, E. Ant., A Est., CEDE, DEE, I gsr, OE gsr, counting per S . . . . Analysis of variance of H gsr by ITI and Int. in Experiment II . . . . . . . . . . Analysis of variance of occurrence of CR80 by ITI, Int., and trial blocks in Experi- ment II O O O O O O O O O O O O O O O O O Tukey significant gap test for selected differences in level of conditioning of ITI-Int. conditions in Experiment II . . . Analysis of variance of I gsr (all 120 83) by ITI, Int., and trial blocks in ExperI— ment II O O O O O O O O O O O O O O O O O iv Page 104 106 107 107 108 108 109 112 112 113 114 Table Page 12. Analysis of variance of I gsr (for 83 Ss having a + OE gsr) by trial blocks in Experiment II . . . . . . . . . . . . . . 114 13. Analysis of variance of OE gsr by ITI and Int. in Experiment II . . . . . . . . . . 115 14. Analysis of variance of per cent time per S spent counting by ITI and Int. in Experi- ment II . . . . . . . . . . . . . . . . . 115 15. Analysis of variance of per cent A. Est. by ITI, Int., and estimate order in Experi— ment II . . . . . . . . . . . . . . . . . 115 16. Correlation analyses in Experiment II . . 116 17. Mean per cent time spent per S on each method of synchrony used in Experiment II 117 18. Median CRd and number of Ss by majority method of synchrony in Experiment II . . . 117 19. E-test analyses in EXperiment II . . . . . 118 20. Chi square analyses in Experiment II . . . 119 21. t-test analyses of occurrence of CR80 with ITIs pooled by trial block and stage of learning in Experiment III . . . . . . . . 121 22. Analysis of variance of occurrence of CR in Acq. at H Int. by ITI (60 and 240 sec. and by Experiment II vs. Experiment III (D1) 122 23. Analysis of variance of occurrence of CR in Acq. (D1) by ITI and trial blocks in Experiment III . . . . . . . . . . . . . . 122 24. Analysis of variance of occurrence of CR30 in Acq. (D2) by ITI and trial blocks in Experiment III . . . . . . . . . . . . . . 122 25. Analysis of variance of occurrence of CR80 in Ext. (D2) by ITI and trial blocks in Experiment III . . . . . . . . . . . . . . 123 Table 26. 27. 28. 29. 3OO Analysis of variance of per cent A. Est. by ITI, successive estimate, and successive day in Experiment III . . . . . . . . . . . Analysis of variance of CE gsr by ITI and successive days of Acq. in Experiment III . Analysis of variance of I gsr by ITI, trial blocks, and successive days of Acq. in Experiment III . . . . . . . . . . . . . . Data summary for ITI conditions in Experiment IV as opposed to Experiment II: age, sex, H gsr, Sb Un, occurrence of CRBOI E. Ant., A. Est., CEDE, DEE, I gsr, OE gsr, counting per S . . . . . . . . . . . . . . . . . . . Sample data sheet from all experiments . . vi Page 123 124 124 125 128 Figure 1. 10. LIST OF FIGURES Mean per cent occurrence of CR at both stimulus intensities over all 28 acquisi— tion trials as a function of ITI in Experiment I . . . . . . . . . . . . . . . Mean per cent occurrence of CR80 at both stimulus intensities for each ITI as a function of 5-trial acquisition block in Experiment I . . . . . . . . . . . . . . . Photograph of entrance to double experi- mental rooms used in Experiments II, III, and IV O O O O O O O O O O O O O O O O O O Photograph of experimenter at data collection desk as situated in Experiments II and III Photograph of experimenter collecting data from gsr as situated in Experiments I, II, and III O O O O O D O O O O O O O O O O O Photograph of subject in gsr electrodes facing photic stimulus as situated in Experiments I, II, III . . . . . . . . . . Mean per cent occurrence of CR at both stimulus intensities over all 1% acquisition trials as a function of ITI in Experiment II Mean per cent occurrence of CR 0 at both stimulus intensities by 5-tria1 acquisition block for each ITI in Experiment II . . . Mean per cent occurrence of CR80 at both stimulus intensities over all 15 acquisition trials as a function of ITI length group in Experiment II . . . . . . . . . . . . . . Median error of anticipation at both stimulus intensities over all 15 acquisition trials as a function of ITI in Experiment II . . . . vii Page 27 29 34 34 36 36 40 43 44 46 Figure Page 11. Median per cent error of anticipation at both stimulus intensities over all 15 acquisition trials as a function of ITI length group in Experiment II . . . . . . 47 12. Median CRd over all 15 acquisition trials as a function of pre—treatment H gsr in Experiment II . . . . . . . . . . . . . . 50 13. Mean I gsr, averaged over both stimulus intensities by each 5-trial acquisition block, as a function of ITI length group in Experiment II . . . . . . . . . . . . 51 14. Mean OE gsr, averaged over both stimulus intensities, as a function of ITI length group in Experiment II . . . . . . . . . 51 15. Acquisition and extinction of a temporal CR: Mean per cent occurrence of CR8 , averaged over both ITIs, by 5-trial glocks in Experiment III . . . . . . . . . . . . 69 16. Photograph of experimenter setting stimulus timer under gsr unit as set up in outer room in Experiment IV . . . . . . . . . . 77 17. Photograph of subject in gsr electrodes holding water cup as situated in inner room in Experiment IV . . . . . . . . . . 77 18. Photograph of subject in gsr electrodes holding urinal and demonstrating freedom of movement allowed in inner room in Experi- ment IV . . . . . . . . . . . . . . . . . 78 19. Mean per cent occurrence of CR8 at high stimulus intensity over all 15 acquisition trials as a function of ITI length in Experi- ments II and IV . . . . . . . . . . . . . 80 20. Median per cent error of anticipation at high stimulus intensity over all 15 acquisi- tion trials as a function of ITI length in Experiments II and IV . . . . . . . . . . 80 21. Mean per cent occurrence of CR at high stimulus intensity by 5-trial acquisition block over all ITIs in Experiment II and for both ITIs in Experiment IV . . . . . 32 viii Figure Page 22. Median per cent error of anticipation at high stimulus intensity by 5-trial acquisition block over all ITIs in Experiment II and for both ITIs in Experiment IV . . . . . . . . . . . . . 82 23. Mean I gsr as a function of 5-trial acquisition block over all ITIs in Experiment II and for both ITIs in Experiment IV . . . . . . . . . . 83 24. Mean OE gsr at high stimulus intensity as a function of ITI length in Experiments II and IV 83 25. Mean per cent time per subject spent counting (for method of synchrony) as a function of ITI length in Experiments II and IV . . . . . . . . 85 26. Mean per cent accuracy of lst and 2nd estimate of ITI length as a function of ITI length in Experiments II and IV . . . . . . . . . . . . . 85 ix INTRODUCTION Time was one of experimental psychology's first in- dependent variables. Subsequent decades of research have not diminished this central role. Quite recent research, for example, supported temporal discrimination as critical to operant avoidance learning (Anger, l963)and as criticalto response latency in repetitive vigilance tasks (Hardesty and Bevan, 1965). This dissertation will concentrate the focus on this ubiquitous variable to its importance as a stimulus in the context of classical conditioning. A recent summary, translated from the Russian, re- viewed over half a century of such temporal conditioning research (Dmitriev and Kochigina, 1959). Beginning in 1907 “with Zelenyi, a student of Pavlov, 68 studies have subse- quently been completed. ,Subjects ranged from bats to hedge- hogs to collies to humans; unconditional stimuli varied from shock to food powder to temperature; measured responses included salivation, leg flexion, ear temperature, and even white blood cell count. Thus, over a wide range of species and techniques the Russian investigators have demonstrated respondent conditioning to a time interval to be an accom- plished fact. Psychology here in the United States has, however, barely acknowledged the phenomenon of temporal conditioning. Learning texts list it in a sentence or two but usually without discussion. Conditioning studies vary their inter- trial intervals (ITIs) or otherwise control for temporal conditioning but typically without much conviction or overt rationale. Only two temporal conditioning studies could be located in American psychology journals.l Brown (1939) shocked rats at 12 second intervals, measuring the force of the jump response on a postage scale. After 35 trials, rats were shocked at 3, 6, 9, 12, 15, 18, 21, or 24 seconds after the last shock. Rats jumped with most force at the 12 second test shock with decreasing force as the interval grew smaller or slightly larger. Bugelski and Coyer (1950) trained rats to jump a barrier in response to periodic shock and found quicker temporal conditioning at a 15 second ITI than at a 60 second ITI. On the other hand, American psychology has dealt more fully with temporal response in the context of fixed interval schedules in operant learning (Ferster and Skinner, 1957). Nevertheless, of 79 studies classically conditioning a response to a temporal CS (as located in Psychol. Abstracts up to August 1965), 72 originated in the USSR with others in France, West Germany, and Japan. The Russian studies were generally strong on detailed observation and imagination but too often weak on experimental controls and statistics even of descriptive nature. 1This does not include Grant, McFarling, and Gormezano (1960) or Prokasy and Chambliss (1960) since an external CS remained in their "temporal conditioning" studies. Neither human eye-blink study found fixed ITIs (15 and 25 sec.) to ex- cel variable ITIs but Prokasy (1965) later cited unpublished data supporting significant temporal cuezeffects with an alternate measure. Number of subjects (SS) rarely exceeded five, often with one 'S per condition. Furthermore, these very Ss were traditionally :passed down over the years from experiment to experiment much as precision timers might be here in the USA. A final draw- back prevalent until very recently has been a relatively rigid adherence to only those areas of exploration the early Pavlovians demonstrated as viable. Within this ground there was much imagination,but beyond it studies never reached the printed page. One case in point was the long exclusive Russian ITI range of five to 30 minutes, below or above which no Pavlovian went. One of Kochigina's more modern studies (Dmitriev and Kochigina, 1959) 113a good example of the Russian technique. Conditioning the leg flexion of dogs to periodic shock, she observed three successive stages of temporal anticipation. Stage 1, the "generalized reflex to time," was characterized by increasing, then decreasing, intersignal errors through- out-the ITI. (The intersignal error refers to premature leg flexions anticipating the shock.) Stage 2, the "differ- entiated reflex to time," begins when intersignal errors occur only during the last half of the ITI. Over further .trials errors converge in time on the end point of the ITI. Stage 3, the successfully conditioned temporal response, is Operationally defined as occuring when any and all inter- signal errors fall within the last 20% of the ITI or to a criterion of 80% accuracy. Kochigina's ITI in this case was the current Russian mode of 5 minutes; four dogs took from 90 to 135 trials to reach criterion. Temporal conditioning in its classical framework has much relevance for American psychology today. Research is needed,if for nothing else, than to justify controls against it2 as well as to explain "sensitization" effects where periodic US presentations elicit an enhanced response level. Russians use this latter effect to increase livestock yield: Optimal periodic feeding and milking rhythms are determined experimentally and universally applied (Dmitriev and Kochi- gina, 1959). Some of the aspects of temporal conditioning demanding basic well controlled research are the effects of ITI, stimulus intensity (Int.), concurrent temporal estimation, concurrent interoceptive process or methods of synchrony, and the role of inhibition. When Russian psychologists had first determined that a fixed rhythm of stimulation enhanced response magnitude over that elicited by random or haphazard stimulation, animal husbandry specialists set up immediate searches for optimal species-specific ITIs. Unfortunately, the academic temporal conditioners did not follow suit so extensively. Today the differential effect of ITI length on level of conditioning has not yet been broadly investigated. Bugelski (1956) acknowledged this need: "The area of temporal in- tervals is still largely unexplored. We do not as yet know the most effective intervals for such conditioning." (p. 131) 2. . .Whether fixed or varied, the time functions adoPted can influence behavior (i.e., can produce some form of temporal conditioning) and are worthy of analysis in their own right." (p. 121 , Prokasy, 1965) An illustrative "pseudo- conditioning" study is Kimble, Mann, and Dufort (1955). Prokasy rejected temporal influence as critical to "pseudo- conditioning" in 1960 but reversed himself in 1965 on the basis of new data. The findings to date on this point are, at first glance, among the most directly contradictory of all the explored aspects of temporal conditioning. There is what might be called a Washington-Moscow ITI controversy. Bugelski and Coyer (1950) in the study already cited found conditioning in rats at the 15 second ITI to be faster than at the 60 second ITI. Bugelski concluded that tem- poral conditioning became more difficult with increasing ITI length. The other American study operated at a 12 second ITI (Brown, 1939) while the French contribution (Fraisse & Jampolsky , 1952) successfully conditioned human gsrs to shock at an ITI of 8 seconds. On this side of the globe then, ITIs have been in terms of seconds backed by the cited opinion that intervals beyond a minute would yield diminishing or non-existent returns. The Russians came to the Opposite conclusion. Zelenyi (1907) launched temporal conditioning in Pavlov's laboratories with a 10 minute ITI. The following year it was replicated at an ITI of 20 minutes (Krzhish- kovskii, 1908). Four years later came the much discussed work of Feokritova (1912) with an ITI of 30 minutes. Under Pavlov's supervision (Pavlov, 1927) she brought salivary conditioning well within Kochigina's third stage at this ITI with no canine salivation until the last minute before stimulation. But with the 1930's came-moderation. From then up to the present, Soviet ITIs have clustered about five minutes with a general range of three to seven minutes. Nevertheless, there was a consensus that no upper ITI length limit need apply. Pavlov decreed: " . . . any length of time interval can be employed. No experiments, however, were made with intervals longer than half an hour" (Pavlov, 1927, p. 42). On the other hand, a lower limit soon crept in. In 1937 Baiandurov found it impossible to condition pigeon activity to periodic shock at ITIs of from 5 to 15 minutes. He had to push as far as 300 trials before 'even' a 30 minute ITI allowed criterion to be reached. Dogs, regarded as capable of conditioning at a somewhat lower ITI than pigeons, still seemed to have a lower limit of their own. A good illustrative study is that of Bolotina (l952a)who conditioned canine time flexion to either a 10 minute or 3 minute ITI. All Ss conditioned at 10 minutes with a mean 180 trials to criterion. Only 1/3 of his dogs were able to condition at the 3'minute ITI and these Ss needed a mean of 520 trials to criterion. Testing the possibility that this finding was a function of his choice of species, he replicated his study with monkeys (1952b) achieving substantially the same results. Bolotina concluded that ITIs of 3 minutes or less were nearly impossible for temporal conditioning since neural excitation was too arhythmic at short intervals to allow the neural traces of inhibition to concentrate. The next year Bolotina (1953) attempted to artificially set aside this neural difficulty by administering bromides to both his dogs and his monkeys. The relaxed animals were able to go as low as ITIs of 2 minutes with level of condi- tioning improving with increasing bromide strength. A possibility of course was that the bromides were aided by continued use of the same Ss from experiment to experi- ment. Dmitriev and Grebenkina (1959) demonstrated this possibility when, unable to temporally condition leg flexion in any of six dogs directly to ITIs of 1 or 2 minutes, they trained down S5 at successively lower ITIs (starting with 5 minutes) and eased their dogs into the difficult ITIs. The moral was clear: temporal condition- ing became more difficult with decrea51ng ITI length; any ITI at 3 minutes or less would yield diminishing or non- existent returns. Bugelski (1956), aware of these conflicting per- spectives, recommended further research. None has as yet appeared. One possible result of such research would be to support both sides. The Soviet-oriented and Washington- oriented studies are standing on different geographical ranges of ITI. Perhaps there is something involved in the exposure of complex mammals such as dogs and humans to -that 1 to.3 minute range that depresses performance. Or perhaps the two ranges reflect the dominance of separate methods of temporal synchrony. What was needed was a com- parative ITI study to explore the gap. Among the Russians only a recent few have crossed that gap at all (Dmitriev and Kochigina, 1959; Elkin, 1964). With Kochigina, Dmitriev temporally conditioned a verbal anticipatory response to an auditory stimulus in children aged 8 to 14. The ITI was 30 seconds. Elkin has very recently conditioned humanstXJITIs as low as 3 to 10 seconds. His study will be discussed more fully in another section. JThe intensity of the unconditional stimulus has Special implications for temporal conditioning although it has been a central variable in classical conditioning since its inception. Pavlov (1927) reported the speed of conditioning as well as resistance to extinction to vary with the intensity of the unconditional stimulus. American research has sup- ported this over the years. Passey (1948), for example, found that conditioning the eye-blink reflex to an air puff 'varied significantly with the pressure of the puff: both rate and final level of conditioning increased with increas- ing air puff intensity (cf. Spence, Haggard, and Ross, 1958; Ratner and Denny, 1964). In general it has been found that performance in the learning situation improves with increa- sing amount of positive or negative reinforcement (Kimble, 1961)./ Therefore, in that temporal conditioning is a member of the category of classical conditioning, similar results should be expected. But not necessarily. same t, of con: which set us ‘ PM CS' pa: ‘Q 1.le Y‘ i“( Temporal conditioning has a property which at the same time distinguishes it from the other more complex forms of conditioning (of which it often is a basic component) and which leads to opposite predictions. That property is the internal nature of the CS; interoceptive rhythms must be set up as cues to achieve the CR. It is relatively easy to set up a range of US intensities which does not prohibit the Ss.discrimination of an external CS. In temporal con- ditioning, however, a stimulus intense enough to disturb the S disturbs both the discrimination of internal rhythms as well as the regularityof these rhythms. Pavlovians have long demonstrated level of temporal conditioning to be in- versely related to S's level of arousal. Feokritova (1912) noted that somnolent dogs excelled normal animals. Stukova (1914) found "excitable" dogs to be more prone to distraction from extraneous stimuli than dogs with "well developed inhi- bition.processes." The tranquilizing effect of bromide in- jections has, as previously mentioned, been used to facilitate temporal conditioning (Stukova, 1914; Deriabin, 1916; Bolotina, 1953; Dmitriev and Kochigina, 1959) while the excitant of caffeine retarded the process (Dmitriev and Kochigina, 1959). Presumably, the increased arousal somehow interfered with the production and/or discrimination of the interoceptive CS. Pavlovians prefer to discuss this effect in inhibition terminology; the discrimination interpretation is my own. c1 10 It might be expected then that increasing intensity would retard level of temporal conditioning where the ITIs are so short as to allow S insufficient recuperation or re- laxation time before the next ITI occurs. On the other hand, when. ITIs are quite long enough for relaxation to occur in spite of a strongly arousing US, one would expect the traditional superiority of the more intense US to emerge. Data from fixed interval studies in operant condi- tioning contexts support this prediction. While stronger reinforcement at moderate or lggg ITIs has excelled perfor- mance at weaker reinforcements (Collier and Myers, 1961; Guttman, 1953; Collier, Knarr, and Marx, 1961; Dufort and Kimble, 1956), the opposite was found at very short ITIs (Collier and Myers, 1961; Conrad and Sidman, 1956). One final aspect of the intensity variable worth mentioning is the too often neglected obligation of the re- searcher to demonstrate that cranking up physical intensity produces any substantial subjective increment as well. Often a Russian study would designate a certain acid dosage on a dog's tongue as noxious without any proof; without appropriate and controlled avoidance measures the US might well only have been "tangy." Temporal conditioning with humans, especially, offers an easy opportunity for S to vocally identify the intensity as to pleasantness or un- pleasantness. The interaction of [IS intensity and ITI was once predicted and observed (ITIs of 15, 45, 135 sec.), but not significantly supported, in a human eyelid conditioning study where photic CS overlapped air puff US (Prokasy, W., Grant, D., and Myers, N., 1958). The simpler temporal conditioning procedure, omitting the external CS, may clarify the issue. gene have lite main 11 Temporal research with humans has been well culti- vated along another area parallel to our interests: tem- poral estimation. Taking advantage of the unique phylo- genetic link between S and S, humans for nearly a century have been asked to produce, reproduce, estimate, and describe intervals of time. During temporal conditioning, human Ss maintain some conscious impression of the magnitude of the ITI. That this process, if not directly related to the conditioning process, is at least affected by it has been shown by Elkin (1964). Elkin temporally conditioned 11 human S5 to ITIs of 3, 5, or 10 seconds. A temporal esti- mation of all 3 ITIs was made before and after training. Each S showed improvement at estimating the ITI he had been conditioned to with no generalization in increment of estimation accuracy at the other ITIs. This Was the only study to ever tie together the two processes of temporal estimation and conditioning. Further investigation of the aspects of their inter-relations is needed. American, French, and German psychologists have turned out volumes of temporal research when it comes to estimation. Among the best current review sources are Fraisse (1963), Woodrow (1951), and Wallace and Rabin (1960). Some of the aspects of estimation might well be investigated in a temporal conditioning context since the large body of research, despite its bulk, is contradictory on many points. Many studies associate over estimation 3mg 1' 5532! ”5‘ «ac with high stress and arousal (Gulliksen, 1927; Cutler, 1952; Eson and Kafka, 1952; Anliker, 1963) while many others as- sociate over estimation with low stress and relaxation (Rosenzweig and Koht, 1933; Bakan, 1955; Hare, 1963; Geiwitz, 1964). Johnson (1962} found no relationship between over estimation and stress concluding that "available methods in time perception experiments do not prov1de an adequate test." Less pessimistically, Zelkind and Spilka (1965) found over estimation to correlate pOSitively With an optimistic out- look for the future. Even general non-directional accuracy of estimate has been found to increase with high anxiety (Burns and Gifford, 1961) as well as decrease with high anxiety (Weybrew, 1963). But if the literature is far from clear as to arousal effects, the influence of ITI length is generally clear- Accuracy of estimation has been found to increase as the interval increases. Furthermore, the ranges of interval tested have been quite broad. Gilliland and Humphreys (1943), for example, found Ss estimated a 14 second interval at a mean accuracy of 72% as Opposed to a mean accuracy of 82% for a 117 second interval. Stimulus intensity ties in to estimation in so far as it heightens anxiety or arousal. Yet, as has been mentioned, the con- sequences of this are far from clear. If Benussi (1907), using an auditory stimulus, is any guide, increasing in- tensity shortens the perceived interval. A final point of temporal estimation worth exploring is that of S's ret the a... po: to en are on; syr ral COL Eh} Koc Pr< til SYI l3 retrospective Opinion of that estimation; an estimation of the estimation. Bakan (1962) found this retrospection to add information or accuracy to the estimate. Ss in a tem- poral conditioning context then might be generally expected to identify the direction and possibly the magnitude of the error of their estimate of the ITI. All these S reports are worth brief investigation. Verbal reports also Offer a look at some of the ongoing interoceptive processes used to accomplish temporal synchrony. The question of which process or processes are basic to time perception has long teased psychology. Long range process« change (blood sugar count, white blood cell count, and hormonal secretions)have been linked to circadian rhythms (Kayser, 1952) and even conditioned (Dmitriev and Kochigina, 1959). Along these lines, Kleitman (1939) deter- mined the existence of a diurnal temperature cycle in humans and conditioned it to different cycle lengths in an assistant adeitman, Titelbaum, and Hoffman, 1937). This kind of ap- proach with slow moving processes has demonstrated the rela- tive ease with which interoceptive rhythms can be made to synchronize with exogenous ones. Waking at a specific time each morning, for example, has been traced to individual empathy with degree of bladder distension (Fraisse, 1963). But when synchrony with intervals of minutes or seconds is involved, some more immediately periodic processes must be examined. Fraisse (1963) lists breathing, pulse, and EEG 19 th 11. Sh: wit ex; 14 activity as the three most important such rhythms, giving priority to the last. In the area of temporal estimation, choice of the critical interoceptive rhythm has again led to a wide variety of findings. Schaefer and Gilliland (1938) ruled out pulse rate, breathing rate, and blood pressure change as reliable indicators of estimation ac- curacy for intervals of 4 to 27 seconds. Bell and Provins (1963) also ruled out pulse and changes in room temperature as well. However, body temperature has been shown to be related to temporal estimation and production (Frangois, 1927, 1928; Hoagland, 1933, 1934, 1936a, 1936b, 1936c, 1936a, 1943«). As to EEG activity, Adrian (1934) has demonstrated that the alpha rhythm can be synchronized with intermittent light. Werboff (1957) brought EEG into the estimation of short intervals and found alpha activity highly related. With brain activity as a possible cue rhythm, one might expect research designed to locate its physiological source. Dimond (1964) has concisely summarized the ablation ap- proaches to locating a temporal center in the cortex. Over the last thirty years, parietal lobes, prefrontal lobes, dorsomedial thalamus, mammallary bodies, and other slices of cortex have been removed in this search (Ehrenwald, 1931; Harrison, 1940; Remy, 1942; Hyde and Wood, 1949; Spiegel and Wysis, 1949; Partridge, 1950; Spiegel, Wysis, Orchinik, and Freed, 1955) with "temporal confusion" (as Dimond put. it) occuring in all cases. The search for an exact location goes on although Dimond preferred the prefrontal areas as the most likely suspects. at) 0: ’rl 15 Theory has not lagged behind research; in attempts at pinning down this internal programming. ‘POpOV (1948, 1950a, 1950b), taking Off from Adrian's EEG work Of 1934 evolved the general temporal theory of "cyclochronism."- "Cyclochronism" bases itself on the nervous system which at all levels supposedly reproduces external excita- tions in the same order and at the same temporal intervals as when they first acted on the organism (cf. Fraisse, 1963). This theory of nervous mimicry, currently accepted by French psychology (Fraisse, 1963), does not directly Oppose Pavlo- vian doctrine since Pavlov (1927, 1928) insisted that all interoceptive processes had their effects on the cortex. However, Popov de-emphasizes the role of the middle man: the interoceptive processes relaying the periodicities to the cortex. Pavlov, for example, attributed temporal con- ditioning of the salivary reflex primarily to the slow rhythms of the digestive organs. Thus, ITIs were long to allow for the substantial latencies of digestion; it was not until the 1930's, when motor reactions were first in- vestigated (Beritov, 1932), that Soviet ITIs dropped to the 5 minute range. Furthermore, Pavlov felt that excitations were not immediately reproduced internally but rather gradually took on this form with repeated exposures. The process which ultimately separated responses from one another for the correct time interval was, according to Pavlov, inhibition. 16 Before discussing inhibition, a final word might be said about the search for critical methods of synchrony. Recently human Ss were asked to identify their methods of time judgement in a well controlled study. Spivack and Levine (1964) identified "visual clock" (S sees an imaginary clock and judges time by it) and "feel" (S has a strong intuition) methods as well as some form of counting when short intervals of l to 64 seconds were estimated. An in- ventory of the conscious methods used by human Ss as well as their differential effectiveness should be done for tem- poral conditioning. Recent evidence suggests that breathing rate, for example, may be as important as Munsterberg (1889) once thought it was. Stolz (1965) found vasomoter condie tioning impossible when breathing was "controlled" by syn- chronizing it to a metronome. The relative frequency and effectiveness of methods of choice for a large number Of humans temporally conditioned would, if nothing else, say something about human species behavior in a temporal situation. , Whatever the rhythm, another process is needed to hook up the external stimulus with an internal response specific in time. To Pavlov (1927, 1928, 1957) this process was the concentration of cortical inhibition in the right temporal areas. Extinction or unlearning involved irradia- tion of this inhibition such that concentration was destroyed. Although Pavlov thought of inhibition as a strictly cortical 17 event, it was analogized with other physiological events from the beginning. Frolov (1937), for example, said: . . . since nearly all activity of the musculature con- sists in the alternating flexion and extension of the extremities at the joints the . . . fact of reciprocal excitation and inhibition by means of reflex action acquires almost universal significance. All interfering movements are inhibited as soon as they become unneces- sary (p. 107). Although inhibition retained its central role for Pavlovian conditioning (Prokasy, 1965), its definition broadened over the years. Inhibition soon came to include relaxation of the musculature in any form up to and including sleep. Rather than demonstrating nervous tissues' concentration or irradiation of inhibition, Russian studies now only differen- tially manipulate observable relaxation. Operationally this approach has been quite productive, especially in the area of temporal conditioning. Feokritova, as far back as 1912, noted that her dogs took better account of the passage of time during sleep than during a period of activity (Frolov, 1937). Bromide and caffeine studies, manipulating relaxation by drug, have found temporal conditioning superior for those Ss best able to inhibit their responses at the right time: those relaxed artificially by drug (Stukova, 1914; Deriabin, 1916; Bolotina, 1953; Dmitriev and Kochigina, 1959). Pavlovian theory rests on the assumption that con- ditioning is learning when (and being able) to relax or (see Denny and Adelman, 1955, for "Secondary Elicitation," a similar emphasis on relaxation in an American learning theory.) 18 learning when not to respond. Naturally, it is represented in other more complex terminology: It appears that in the establishment of a conditioned reflex to time, internal inhibition becomes stronger, as a result of which it can at some point and under certain conditions become prepotent over the stimula- tion of the dominant response and delay the . . . reaction . . .. From this point of View, the condi- tioned reflex to time is the result of the imminent relative insufficiency of internal inhibition during interaction Of the dominant response (Rozin, 1959). Inhibition supposedly grows from the midpoint of the ITI, since that is the farthest point in time from stimulation, expanding to concentrate over the entire non-stimulated interval with sufficient trials (Pavlov, 1927; Dmitriev and Kochigina, 1959). How could this be measured? Birman (1953) classified human Ss as "excitables" and "relaxeds" on the basis of their observed waiting room behavior before the experiment began. Trace conditioning was far superior for the "relaxeds." A better way to get at pre-treatment arousal as well as the development Of relaxation during conditioning would be by gsr recording. A good deal is known about this measure (Woodworth and Schlosberg, 1954) and quite recently gsr changes have been found to be assoCiated with CRs in a human operant learning situation (Doehring, Helmer, and Fuller, 1964) and URs in an earthworm responding situation (Morgan, Ratner, and Denny, 1965). 19 In summary, the literature has delineated the in- dependent variables Of ITI, stimulus intensity, and prior relaxation as potentially important to temporal conditioning. The concurrent dependent variables of temporal estimation, interoceptive process or methods of synchrony, and relaxation during the ITI also seem relevant for inclusion in any para- metric study of temporal conditioning. It was decided to attempt such an investigation of these variables through a series of temporal conditioning experiments with humans in a common methodological context. Pilot research, including a variety of conditioning methodologies, ultimately led to the adoption of Dmitriev's (1959) vocal anticipation of the periodic stimulus as most fruitful method for pursuing the temporal variables suggested by the literature. Human Ss selected for the pilot research performed well in the vocal anticipation paradigm of temporal conditioning. A photic flash in a dark room was adopted as the periodic US. This reduced the problem of manipulating stimulus intensity to switching light bulbs of differential wattage Theoretically, US intensity might also have been changed by varying the intensity level during the ITI with absence Of light as the US. In either case, the change in photic inten- sity should have subsequent unconditional response consequences for S if such a change can in fact be taken as a US. Pilot work indicated that both high and low wattage flashes, at ITIs of 30 seconds or more, elicited consistent gsr arousal in humans. This gsr arousal had not habituated after 20 succes- sive flashes. The gsr drops were only part of a generalized response complex including slight head withdrawals, eye blinks, and general skeletal movement. Thus the US consistently produced a UR or URs. However, in the pursuit of a discrete and more readily identifiable response, the vocal anticipation was chosen. This choice departed from most American classical conditioning research in that the vocal response, in the absence of prior instructions, would not be a consistent 20 (unconditional) response to the photic flash. The prior instruc- tions, however, allow the vocal response to function as such a UR within the specific experimental context of the temporal condition- ing situation. Since it's significance as a consistent stimulus is thus acquired, we are realy dealing here with a higher order con- ditioning of the temporal response. Fortunately, through the use of the human "second signalling system" of language, such condition- ing is eminently feasible. One objection that might have been raised by American psychologists of an earlier era would be to the "voluntary" nature of the response in what is purported to be a classical con- ditioning paradigm. However, since in recent years "involuntary" responses have been conditioned in operant situations and even con- trolled by voluntary procedures (Stolz, 1965; Kimble, 1961), American definitions of the classical procedure have somewhat liberalized. Kimble (1961) says: The original distinction between instrumental and classical con- ditioning is made purely on operational grounds. The two desig- nations refer respectively to training procedures in which the response of the subject does and does not determine whether the US appears (p. 78). . . . the conditioned response is a combina- tion of voluntary and involuntary processes (p. 108). Pavlov was sensitive to the controversy of volition, which he regarded a pseudo-controversy, and for years the word "voluntary" was forbidden in the Pavlov laboratories (Frolov, 1937). Pilot research led to the establishment of experimental condi- tions designed to reduce, as much as plausible, all competing or distracting extraneous stimuli from the environment to maximize the chance of temporal conditioning's occurrence. Since such conditioning at the ITI range contemplated had long been regarded as difficult or impossible, such maximizing steps seemed warranted. The final pilot study is included in this dissertation as it was the prototype of the experiments that followed. On the basis of the relevant literature discussed here, as supported by pilot research, certain hypotheses seem tenable: A. Temporal conditioning in a classical framework can be demon- strated for adult humans. 21 B. When a temporal response is acquired it will show improvement with continued amounts of periodic stimulation and extinction when anticipations are elicited in the absence of that periodic stimulation. C. Performance observed in acquisition will be a func— tion of ITI with performance at ITIs of more than 3 minutes and at 1 minute or less excelling perfor- mance of ITIs in between. D. Performance Observed in acquisition will be en- hanced by increasing stimulus intensity at high ITIs and retarded by increasing stimulus intensity at low ITIs. E. Post-conditioning estimation of ITI length will be affected by some of the factors affecting temporal conditioning. Accuracy of estimation, for example, will increase with increasing ITI. F. The more relaxed a subject before conditioning begins, the higher the subsequent level of condition- ing Obtained. Gsr drops Obtained directly before acquisition will therefore be reliable predictors of subsequent performance. G. As temporal conditioning occurs over trials, in- creased relaxation during the ITI will also be observed with trials. Gsr changes from the center of one ITI to the next will therefore show increased relaxation over trials in the presence of temporal conditioning. H. Reported interoceptive process will vary with the subject and the ITI. Specifically, counting methods of synchrony will decrease in percentage of use as ITI length increases. The following experiments attempt to gather sufficient basic data to support or reject these hypotheses and to lay the framework for a better understanding of respondent condi- tioning to an interval of time. EXPERIMENT I Experiment I, as the last pilot study, stood as basic prototype to the experiments following it. Its purposes were twofold. One was to demonstrate the feasibility of temporal conditioning at the stimulus conditions contemplated within the context of the experimental procedure and setting evolved for that purpose. Secondly, a first look at the differential effects of ITI and stimulus intensity was to be taken. Method Subjects.--The Ss were 18 volunteer college students, or their wives, ranging in age from 19 to 25. There were 13 males and 5 females. Apparatus.--The experiment took place in a relatively light- tight single room. S sat at a desk directly behind S and collected data by the light Of a red 60 watt bulb. The photic US was a bulb flash controlled by S_with a Lafayette Stimulus Timer. The bulb was 7.5 watts white-frosted for the low intensity condition and was 100 watts white-frosted for the high intensity condition. The bulb was at eye level or below on a lamp 18 in. in front of S. The bulb and lamp rested on a table which S faced. S's left hand rested on this table within the gsr finger electrodes. The gsr 22 23 electrodes were connected to a Lafayette D.C. gsr unit placed over the Stimulus Timer at S's table. S also re- mained within reach of the room overhead lights. S sat in a comfortable wooden swivel chair with arm rests facing the lamp and US bulb which had a blank white wall behind it. S timed S's vocal responses with a Meylan stop watch, checking these readings at the shorter ITIs against tape recordings made during the experimental session. Dittoed data sheets (see Table 30) were used for uniform data re- cording; typed procedure and instructions for S were taped to Sis desk. Procedure.--Before each S, S warmed up gsr and timer ap- paratus for 10 minutes. At the end of this time, S was allowed to enter the experimental room, minus any wrist watch, and settled in chair and gsr finger electrodes. At this time S recorded Sis name, age, sex, and any other descriptive data that seemed relevant. S next handed S a carbon copy of the acquisition instructions. S followed this copy visually while S read the original out loud. The acquisition instructions were as follows: This experiment is designed to measure your gsr or the electrical resistance of the skin. The finger elec- trodes are for measurement only and will not shock you. A brief explanation of the gsr's purpose will follow the experiment; any questions about our purpose will be answered at that time. 1. Because of the delicate balance of the electri- cal equipment please keep your left hand perfectly still throughout the experiment. Ir“ ('3 r'f {'1 Pf) 24 2. Find as comfortable a sitting position as you can, facing straight ahead. 3. The light in front of you will flash on and Off very quickly every so Often. The time between these flashes is a FIXED INTERVAL of time. There will be the same amount of time between each flash. 4. Your job is to say "NOW" whenever you think the bulb is about to flash. Try to say "NOW" as closely as possible to the actual flash. I will be scoring your accuracy. The closer you come to the flash, the more accurate the score. 5. Say "NOW" only once between flashes. 6. If you don't beat the flash; if the bulb flashes before you can say "NOW," then say "NOW" while the bulb flashes and try to beat it next time. 7. Except for saying "NOW" please do not talk dur- ing the experiment. 8. When I turn Off the lights the experiment will begin. When I turn them on it will be over. There will be a short wait of a few minutes before the first flash while the gsr warms up. Are there any questions? S answered any questions by rereading the relevant portion of the instructions. Then S reclaimed the carbon copy of the acquisition instructions, shut the overhead lights, and turned on the red desk light. The stop watch was started. For some Ss, gsr readings were made at 0.5 minutes and at 2.5 minutes during the 3.0 minutes of habituation which now followed. At the end of these 3.0 silent minutes S set off the first photic flash. Twenty-one subsequent flashes seperated by a common ITI followed (20 trials) with the time of S's “NOW" anticipation recorded by S in every case. various gsr measures were made by S throughout the trials. After the last flash, the following instructions were read to S: 25 You've done very well. Before I turn on the lights I'd like you to guess how much time there was between each pair of flashes. (Bakan, 1955, has shown that lack of response set does not significantly affect temporal estimation.) After these estimation instructions, S recorded S's answer, and turned on the overhead lights. S then recorded S's introspections on methods of synchrony used, percent of time devoted to each method, and subjective unpleasantness Of the flash. Finally, S was given a brief lecture on the history and uses of the gsr and released from the experimental situ- ation. Ss were not told the exact ITI they had been run at until Experiment I was completed for all Ss. _ Experimental Desigp.--ITIs of 45, 135, and 240 sec. were used at either high or low US intensity. This formed six ITI-Int. conditions and 3 Ss were randomly assigned by card draw to each condition. Results Table 1 summarizes the most important abbreviations used in this and subsequent experiments. Table 2 summarizes the important S data for Experiment I. Looking first to temporal conditioning as a function of ITI and Int., mean occurence of the temporal CR for these variables is depicted in Figure 1. As Table 1 indicates, a temporal CR or CR80 is any response anticipating the US within the 26 last 20% of the ITI (Dmitriev and Kochigina, 1959). With this criterion, analysis Of the binomial probabilities (Siegel, 1956) indicated that a frequency of occurence of the CR of 55% or higher in a 5-trial block or 38% or 80 higher in 20 trials would be significant at p < .05. Figure 1 shows the predicted near chance level dip at the (135 sec.) ITI falling between previously successful short American ITIs and long Soviet ITIs. An analysis of variance (Table 3) showed this ITI effect to be highly significant (F = 7.10, df = 2, 12, p a .01). Int. also looked as pre- dicted with L Int. excelling H Int. at the short ITI and vice versa at the long ITI. However, neither the Int. nor its interaction with ITI were significant in the analysis of variance (Table 3). Figure 2 illustrates temporal con- ditioning over blocks. The over-all increase in level of conditioning with blocks of 5 trials was highly significant as gauged by analysis of variance (F = 1038, df = 3, 36, p < .005). Note in Figure 2 that this increase was only evident for the short and long ITIs after the second block of trials. Blocks at the 135 sec. ITI remained at or below chance level. Note also that conditioning at the long 240 sec. ITI levels Off or drops after the third block of trials. This may have been the result of fatigue. The curves generally demonstrated that temporal conditioning in the experimental setting of Experiment I was demonstrably present or absent by the end of 10 to 15 trials. Analysis trials CR80 occurs Mean % 27 Figure 1.--Mean % occurrence of CR80 at both stimulus intensities over all 20 acquisition trials as a function of ITI in Experiment I 100 90 x____x High stimulus intensity .----. Low stimulus intensity 80 7O \ 6O 50 40 3O 20 10 0 30 60 90 120 150 180 210 240 270 300 ITI in seconds 28 of variance showed none of the interactions between blocks of trials and ITI or Int. to be significant (Table 3). As for estimation of ITI length, a product-moment correlation between A. Est. and CR80 scores for all blocks was +.26 or not significant at the 5% level for the number of Ss involved. Those Ss showing the best conditioning did not necessarily show the most accurate estimation. An analysis of variance of per cent A. Est. as a function of ITI and Int. (Table 4) showed significant effects for ITI (F = 8.63, df 2, 12, p < .005), Int. (F = 55.38, df = l, 12, p < .005), and their interaction (F = 12.12, df = 2, 12, p < .005). Estimation grew in mean accuracy as ITI length grew, had greater mean accuracy at H Int. than L Int., and showed a more dramatic increase with ITI for the L Int. Ss than for the H Int. Ss. These results from this minia- ture experiment suggested that the estimation process is sensitive to the same variables as the conditioning process. Three Ss had their gsr changes gauged during the habituation period. The change from 0.5 minutes to 2.5 minutes, designated H gsr (for habituation), is plotted in Table 5 Opposite the respective and subsequent CR80 score over all trial blocks in acquisition. The product-moment correlation between H gsr and CR score for these 3 Se was 80 -.985 (P < .02). This finding was suggestive enough to give it more thorough consideration with the substantial number of Ss in Experiment II. Various methods of gsr OCCUI‘S Mean Z trials CR 100 80 60 40 20 100 80 60 40 20 100 80 60 40 20 Figure 2.--Mean Z occurrence of CR80 29 at both stimulus intensities for each ITI as a function of S-trial acquisition block in Experiment I ITI = 45 sec. ITI = 240 as 1.2 3 4 Blocks of 5 trials C. 4 X—-——X High stimulus intensity Low stimulus intensity 30 reading were attempted during acquisition. Of these, the most feasible appeared to be the taking of gsr readings at the midpoint of every ITI. Although ITIs as large as 300 seconds would leave that amount of time between readings there is evidence that even such 5 minute readings are reliable measures (Morgan and Bakan, 1965). All gsr results were analyzed in terms of square root conductance units since it has been shown that these units have a more normal distribution than any other method of depicting gsr data (Schlosberg & Stanley, 1953). On the H Int. condition, 7 out of 9 Ss reported the ,stimulus flash,"unpleasant" as opposed to "neutral" or' "pleasant." On the L Int. condition, only 2 of 9 Se re- ported the stimulus flash "unpleasant." This finding was significant (p < .05) according to Fisher's test of proba- bility (Siegel; 1956). S3 introspected that a mean 93.5% of their time was spent counting by one method or another at the short 45 sec. ITI while this decreased to a mean 68% and 70% for the two longer ITIs. There was approximately 76% mean time per S spent counting at each stimulus intensity. While counting was the favored method, as Woodrow (1951) would have pre- dicted, it was far from the only One. Fifteen Ss of the 18 counted at least one trial but only 4 Ss counted numbers divorced from any internal rhythm. Some Ss tapped their finger; others listened to their pulse or breathing; others 31 just guessed. Some Ss used a "Feel" or intuition method which, as opposed to blind guessing, gave them some definite physical sensation of anticipation prior to the US. One S, scoring 6 out of 20 possible CR visualized a clock with 805’ a second hand in motion at a constant rate of speed. He relaxed and let his thoughts wander at random until his "Feel" indicated a flash was close at which time he checked his imaginal clock to see how much time was left. Here was a real biological clock! (See Table 6.) Discussion Experiment I demonstrated that temporal conditioning could occur as a result of the experimental procedure and surrounds evolved from prior pilot research. Furthermore, the occurence of the conditioning was significantly related to the stimulus conditions. ITI showed the predicted dip in the gap area be- tween Soviet and American explored ITI ranges. Int. was as predicted in effect but was not significant for the number of Ss tested. Level of conditioning significantly improved with trial blocks. Since there was the suggestion of fatigue at the long 240 sec. ITI after three 5-trial blocks, subsequent experiments will restrict themselves to three blocks of acquisition trials. Classical conditioning studies with humans have usually domonstrated significant effects within 20 trials (Kimble, Mann, and Dufort, 1955; Prokasy, Grant, and Myers, 1958). 32 Introspections were suggestive and seemed worthy of expansion. It was decided that the next experiment would also collect methods of synchrony and per cent of time spent on each, as well as estimates of ITI length with additional estimates of the direction and magnitude of error of that estimate, and, again, introspective proof that the H Int. US was more subjectively unpleasant than the L Int. US. Experiment I allowed for the evolution of the most experimentally feasible, reliable, and theoretically mean- ingful gsr measures for subsequent experiments. Besides the general refinement of measures and procedures, critical variables demanding further scrutiny had been delineated. EXPERIMENT I I Experiment II was an extended replication of Experi- ment I. It was conducted under improved, refined, and ex- panded conditions, across a wider spectrum of critical ITIs, and, most important, with a much more substantial number of Ss. Method Subjects.--The Ss were 120 volunteer college students from an introductory psychology course. Males outnumbered females 79 to 41. Age was restricted to from 18 to 22 years; nearly half the sample was 18 (58 Se) with a mean age of 18.7 years for the full sample. All Ss received course credit for their time in the experiment. Apparatus.--This and subsequent experiments took place in a new light-tight room which was also sound-proofed to out-- side noise by virtue of being the insulated inner chamber of a double room. Figures 3 and 4 illustrate this. Other- wise the apparatus was exactly as described in Experiment 1., Proggdure.--The procedure was substantially identical to that of Experiment I with a few additions and changes which will be noted here. The positions of S and S in relation 33 34- FfiLgure 3.--Photograph of entrance to double experimental rooms used in Experi- ments II, III, and IV Figure 4.--Photograph of experimenter at data collection desk as situated in Experiments II and III H-__.-. ,_ _ ,_ 35 to each other and the apparatus remained the same (see Figures 5 and 6). The acquisition instructions read to S by S (still followed by S on a carbon) were not re-worded. Again, after the overhead lights were turned off, there was a 3 minute habituation wait before the first US was pre- sented. In this experiment all Ss had their gsrs recorded at the 0.5 minute and 2.5 minute marks of the habituation period. This H gsr change represented an increase or de- crease in arousal as a function of the habituation wait. After the first flash, l6 subsequent US flashes separated by a common ITI (15 trials) followed with the time of S's "NOW" anticipation recorded by S in every case. S recorded S's gsr at the midpoint of every ITI up to and including the midpoint of the ITI following the last US flash. The mean gsr change from midpoint to midpoint for each block of 5 trials was termed the I gsr after inhibition (since Pavlovian inhibition theory predicted this mean change would be towards greater relaxation over trials).. From these readings, S was also able to measure an overall experimental gsr or CE gsr by subtracting the pre-treatment 2.5 minute habituation reading from the very last or post- treatment ITI midpoint reading. This gave a before and after measure on arousal of the effect of the experiment as a whole. After the last midpoint gsr reading, S asked S for a verbal estimate of the ITI length. Following this estimate, additional retrospective information was collected. 36' Figure 5.--Photograph of experimenter collecting data from GSR as situated in Experiments I, II, and III Figure 6.--Photograph of subject in GSR electrodes facing photic stimulus as situated in Experiments I, II, and III 37 g was asked to indicate the direction and magnitude of the error in his initial estimation. The overhead lights were then put on and further introspective information gathered. g was again asked what methods of synchrony were used, what‘ per cent of the time they were used and when they were used. g was again asked to classify the US as "pleasant," un- pleasant," or "neutral." Following any final comments, g was lectured on the gsr. All gs received credit slips and were then released. Experimental Design.--ITIS of 30, 60, 90, 120, 150, 180, 210, 240, 270, 300 sec. were used at either high or low US inten- sity. This formed 20 ITI-Int. conditions and 6 §$ were randomly assigned by card draw to each condition. Results3 Table 7 summarizes the data for Experiment II. Again, abbreviations are defined in Table l. A first consideration was the pre-treatment composi- tion of the sample in terms of the major independent variables. Chi square analyses (Table 20) did not show a distribution by sex, age, or pre—treatment H gsr to be significantly clustered on any one Int. or ITI. A similar check of H gsr 3Because of the large number of gs and subsequent bulk of data, all calculations were computed and checked on Friden and Monroe calculators with correlations checked, in addition, on a CDC3600 computer. 38 by ITI and Int. was made with analysis of variance (Table 8) and, again, H gsr scores Were not found significantly dif- ferent by ITI, Int., or their interaction. These results supported the random assignment of gs to conditions as not 'biasing the stimulus variables with differential g character- istics.‘ . Another preliminary consideration was the nature of the intererelationship of § characteristics. Both sexes had a mean age of 18.7 years. However, females had H gsrs significantly more relaxed than males as tested by t-test (t = 3.28, df = 118, p < .01). Age (18 years vs. 19-22 years) and H gsr (+ vs. -,0) showed no significant relation- 'ship by chi'square analysis nor did "excited" gs (+ H gsr) differ significantly by age from "relaxed" gs (0 or - H gsr) “as gauged by t-test. (In Experiment II: all non-significant and signifi- cant chi squares can be observed in chi square Table 20; all t-test analyses can be observed in Table 19; all correlational analyses can be observed in Table 16.) Subjective unpleasantness of the US (Sb Un) was found to bear no significant relationship to ITI, or the sex, age, or H gsr of § as analyzed by chi square. As ex- pected, §s at high stimulus intensity (H Int.) significantly more frequently labeled the US as "unpleasant" than gs at the low stimulus intensity (L Int.) as tested by chi square (Chi square = 8.89, df = l, p < .01). At H Int. 53% of the ‘§8 chose "unpleasant" as opposed to 27% of the 83 at L Int. CR Int fre hig 'a'df be the was 39 Figure 7 illustrates the mean per cent occurence of C over all acquisition trials as a function of ITI and R80 Int. Binomial probabilities (Siegel, 1956) indicated that a frequency of occurrence of the CR80 over 15 trials of 41% or higher would be significant at p < .05. Once again, there was a dip of level of conditioning between ITIs of l and 4 minutes. The bottom of the dip was at or below chance level. Analysis of variance (Table 9) showed ITI to significantly affect the level of conditioning (F = 3.54, df = 9,100, p < .005). Int. appears to interact with ITI as in the last experiment with L Int. excelling H Int. at short ITIs and vice versa beyond the 120 sec. ITI. Analysis of variance (Table 9) did not show Int. or its interaction with ITI to be significant over all the ITIs. However, it must be noted that for the four shortest ITIs mean level of conditioning was higher for the L. Int. groups while for the six longer ITIs mean level of conditioning was higher for the H Int. conditions. The cross-over was between 120 and 150 seconds. Tukey (1949) has developed a procedure for testing the sig- nificance of individual comparisons between condition means following an analysis of variance. Winer (1962), labeling it the "honestly significant difference" procedure, gave it laurels as a widely applicable but conservative measure. Edwards (1960a) referred to the procedure more simply as "Tukey's significant gap test." Basically it employs the error mean square of the analysis of variance as a common A.l__ Mean Z trials CR80 occurs 40 .Figure 7.--Mean % occurrence of CR80 at both stimulus intensities over all 15 acquisition trials as a function of III in Experiment II 100 90 f x————x High stimulus intensity ._---. Low stimulus intensity 80 i 70 g,’° —~ ‘-.'—'.' ‘VVn—v_-..... —.~ 60 50 -_- _. . A... o-..“ .~ 40 30 20 10 I L__- - . ,. -- , -._. , s. . 0 3O 60 90 120 150 180 210 240 270 300 ITI in seconds meas vert Int. for midd for tieS lowe Subs IEVe from 41 measure of error variance for t-test comparisons and con- verts this by formula to a minimum difference necessary to all comparisons for a specific level of significance. Table 10 lists selected comparisons and their significance at the 5% level (where applicable) according to Tukey's ‘gap test. Returning now to the question of Int., it is seen from Table 10 that L Int. significantly excelled H Int. at both the 30 and 60 sec. ITIs thus supporting the hypothesis that short ITIs lead to this kind of result. On the other hand, only at 150 and 180 sec. ITIs did H Int. excel L Int. to a significant extent. The evidence for superiority of H Int. was supportive then only at middle range ITI lengths. Turkey's gap test also allows for another look at the effect of ITI length at both intensi- ties. Comparing each ITI to its next highest and next lowest neighbor in time, Table 10 illustrates that the subsequent ITI groupings were quite familiar. At H Int., lexrel of conditioning separated 30, 60 and 90 sec. ITIs from middle range 120, 150, and 180 sec. ITIs, while the latter group was separated from the longer ITIs of 210 to 30 0 sec. This same division was found at L Int. with the additional isolation of the 90 sec. group from both shorter and longer ITIs. Here again are the American, Soviet, and untOuchable ranges. Further analysis of these data-deter- mined ITI groupings (ITI gps.) was made according to the f . o:L-‘Lowing schema: 30 and 60 sec. ITIs were considered as 42 the "short" (5) ITI group; ITIs of 120, 150, and 180 sec. were considered "medium" (m) length ITIs; ITIs of 210, 240, 270, 300 sec. were considered "long" (1) ITIs. The 90 sec. ITI, not clearly a member of the "short" group for both Ints" was not included ill the groupings. Note that the "short," "medium," and "long" ITI groups each correspond to a different area of past American vs. Soviet exploration or lack of exploration. Figure 8 illustrates the general increase in level of conditioning with blocks of acquisition trials observed at most ITIs and intensities. Again, a 5-trial block needed 55% frequency of occurrence of CR8O or better to show con- ditioning significant at p < .05. The overall increase was significant (F = 26.54, df = 2,200, p < .005) by analysis of variance (Table 9) although none of the interactions between trial blocks and ITI or Int. were significant. Figure 9 demonstrates the mean per cent occurrence of CR for all acquisition blocks as a function of ITI and 80 Int. when ITI conditions were pooled into "short," "medium," and "long" groupings. Note the striking similarity between the "V" shapes in this figure and the "V" shapes in Figure l of Experiment I. In Figure 1, each point represents a single ITI falling in the derived ranges of "short," "medium," and "long" as used in Figure 9. Differences between the ITI groupings depicted in Figure 9 were analyzed by t-test. As for Int. differences, only the "short" ITI group showed a significant difference a» ~k.-—.-.1AH . n‘~‘ 43 Figure 8.--Mean Z occurrence of CR80 at both stimulus intensities by S-trial acquisition block for each ITI in Experiment II 100 x___x High stimulus intensity 3 Mean for all III'S 8 .---. Low stimulus intensity 0 O 75 o o__"n_ Mean for both stimulus intensities o 00000 000 N 000 H 00000H00H< N 0H0 N me H 0HmmvanHa N 0H0 N 0H0 H 0H0 .0aH-H0H Ammsv 000 0000 .umm .4 w 0 0H .ucs .m GMHomz .omm ca .uc¢ .m CMchz .cmscawmooll.h.mHnma 111 mm 00. 00. 00. HH. .o.m 00 00.+ H0.1 00. m0.+ m 0NH 1- III III III 111 N 0 N0 H0.+ 00.1 00.1 ~0.+ H1000 00m 0 N0 vo.+ H0.- 0 00.+ m1000 00m 0 NN m0.+ m0.+ H0.+ H0.+ H1000 00m 0 mm H0.+ 0 No.1 m0.+ 01000 00m 0 H0 H0.+ 00.1 No.1 0H.+ H1000 00m 0 N0 m0.+ No.1 H0.- 0 m1000 00m 0 mN 00.+ m0.+ H0.- ~0.+ H1000 0HN 0 0N m0.+ 00.- N0.+ m0.+ $1000 0HN 0 0N m0.H 0 No.1 00.1 H1000 00H 0 m0 H0. m0.+ N0.+ H0.- 01000 00H 0 00H N0.+ m0.+ 00.- 0 H1000 0mH 0 00 HH.+ 00.+ 00.+ NH.+ 01000 0mH 0 N0 00.+ 0 No.1 mo.+ H1000 0NH 0 m0 00.+ 0 m0.+ 00.+ $1000 0NH 0 mm HH.+ 00.1 m0.+ 0H.+ H1000 00 0 0N 0H.+ m0.+ m0.+ N0.+ m-000 00 0 00H N0.+ m0.- ~0.- 0H.+ H1000 00 0 mm N0.+ H0.+ m0.+ 00.+ 31000 00 0 00 00.+ 0 H0.+ mo.+ H1000 0m 0 m0 N0.+ 0 H0.+ 00.+ m-000 0m 2 0 000 000 00 000 00 m N 0H0 N u.30 H 0H0 0.00-EH mchcsoo 0:000 0EHB w m. 000 2H 000 H x .00osHucouul.N 0Hn09 Table 8.--Analysis 112 of variance of H gsr by ITI and Int. in Experiment II Source SS df MS F A: ITI 39.21 9 4.36 0.48 B: Int. 0.00 l 0.00 0.00 A x B 96.75 9 10.75 1.18 error 908.83 100 9.08 Total 1044.79 IS; Table 9.--Ana1ysis of variance of occurrence of CR80 by ITI, Int., and trial blocks in Experiment II Source SS df MS F A: ITI 48.0 9 5.33 3.54* B: Int. 1.0 l 1.00 0.66 A x B 15.5 9 1.72 1.14 error (a) 150.5 100 1.51 C: Trial Blks 58.0 2 29.00 26.54* A x C 31.0 18 1.72 1.58 B x C 0.0 2 0.00 0.00 A x B x C 13.5 18 0.75 0.67 error (b) 218.5 200 1.09 Total 536.0 35; * p < .005 113 Table 10.--Tukey (1949) significant gap test for selected differences in level of conditioning of ITI-Int. conditions in Experiment II Mean % difference* Comparison of % occurrence of CR80 for all blks. 1. ITI (sec.): H Int. 30 vs. 60 + 3.2 60 vs. 90 - 2.3 90 vs. 120 -14.4** 120 vs. 150 . . . . + 7.8 150 vs. 180 . . . . + 5.6 180 vs. 210 . . . . . +10.0** 210 vs. 240 . . . . . - 2.3 240 vs. 270 . . . . . - 1.1 270 vs. 300 . . . . . - 3.3 2. ITI (sec.): L Int. 30 vs. 60 . . . . . + 2.2 60 vs. 90 . . . . . -10.0** 90 vs. 120 . . . . -14.5** 120 vs. 150 . . - 8.8 150 vs. 180 . . . . + 7.8 180 vs. 210 . . . . +14.4** 210 vs. 240 . . . . + 2.2 240 vs. 270 . . . . - 7.8 270 vs. 300 . . . . + 3.4 3. ITI (sec.): H vs. L Int. 30 . . . . . . . +13.3** 60 . . . . . . . -12.2** 90 . . . . . . . + 4.5 120 . . . . . . . + 4.4 150 . . . . . . . -12.2** 180 . . . . . . . -10.0** 210 . . . . . . - 5.6 240 . . . . . . - 1.1 270 . . . . . . . - 7.8 300 C O O O O O O - 1.1 * A + sign indicates increase in per cent occurrence of CR80 from first to second condition in the comparison; a - sign indicates a decrease. ** p < .05 or a critical difference exceeding 114 Table 11.--Ana1ysis of variance of I gsr (all 120 gs) by ITI, Int., and trial blocks in Experiment II Source SS df MS F A: ITI 936 9 104 1.48 A: Int. 271 271 3.86 A x B 1094 122 1.74 error (a) 7020 100 70 C: Trial Blks 2102 1051 17.23* A x C 1549 18 86 1.41 B x C 191 2 96 1.57 A x B x C 1167 18 65 1.07 error (b) 12191 200 61 frotal 26521 359 * p < .005 'Iable 12.--Ana1ysis of variance of I gsr (for 83 gs having a + OE gsr) by trial blocks in Experiment II ¥ ¥ Source SS df MS F frrial blocks 2587 2 1293.5 20.80* error 15302 246 62.2 Total 17889 248 k * p < .005 115 Table 13.--Ana1ysis of variance of OE gsr by ITI and Int. in Experiment II Source SS df MS F A: ITI 1369 9 152 2.22* B: Int. 23 1 23 0.34 A x B 388 9 43 0.63 error 6839 100 Total 8619 119 * p < .05 Table 14.--Ana1ysis of variance of per cent time per §_spent counting by ITI and Int. in Experiment II Source SS df MS F A: ITI 3334 9 370 0.44 1B: Int. 399 1 399 0.47 A x B 10128 9 1125 1.34 error 84210 100 Total 98071 119 Table 15.--Ana1ysis of variance of per cent A. Est. by ITI, Int., and estimate order in Experiment II k ‘ Source SS df MS F A: ITI 23395 9 2599.4 1.26 B: Int. 149 1 149.0 0.07 A x B 18616 9 2068.4 1.00 error (a) 206698 100 2067.0 C: Estimates 4 1 4.0 0.05 A x C 1 9 0.1 0.00 B x C 22 1 22.0 0.30 A x B x C 1403 9 155.9 2.11* error (b) 14300 200 71.5 frotal 264488 239 nu..._‘__‘. 116 Table l6.--Corre1ation analyses in Experiment II Comparison df Correlation CRd (all blks) vs. % A. Est. 118 PPM r = +.028 CRd (all blks) vs. H gsr 118 PPM r = -.300*** CRd (all blks) vs. H gsr 118 Eta r = -.408*** (X.Y) Eta r = -.483*** (Y.X) CRd (all blks) vs. H gsr 118 SRO r = -.309*** CRd (blk 1) vs. I gsr (blk l) 118 PPM r = -.037 CRd (blk 2) vs. I gsr (blk 2) 118 PPM r = -.067 CRd (blk 3) vs. I gsr (blk 3) 118 PPM r = -.120 CRd (all blks) vs. OE gsr 118 PPM r = -.095 I gsr (blk 1) vs. H gsr 118 PPM r = +.164 I gsr (blk 2) vs. H gsr 118 PPM r = +.066 I gsr (blk 3) vs. H gsr 118 PPM r = -.043 I gsr (blk 1) vs. OE gsr 118 PPM r = +.783*** I gsr (blk 2) vs. OE gsr 118 PPM r = +.362*** I gsr (blk 3) vs. OE gsr 118 PPM r = +.312*** I gsr (blk 3) vs. % A. Est 118 PPM r = +.201* OE gsr vs. ITI length 118 PPM r = -.230** OE gsr vs. H gsr 118 PPM r = +.226* OE gsr vs % A. Est. 118 PPM r = +.040 % A. Est vs. H gsr 118 PPM r = +.084 % A. Est. vs % counting per g 118 PPM r = -.024 Final S-S interval vs. final S-R interval 66 PPM r = +.973*** Final S-R interval vs. ITI Est. 66 PPM r = +.790*** Final S-S interval vs. ITI Est. 66 PPM r = +.849*** CR80 (all blks) vs. ,Med. % E.Ant.: Block 3: 115 SRO r = -.338*** All blks: 118 SRO r = -.423*** PPM - Pearson product moment * p < .05 SRO - Spearman rank order ** p < .02 Eta - Correlation ratio *** P < .01 117 Table 17. --Mean per cent time spent per S on each method of synchrony used in Experiment II Method of Synchrony Mean per cent time per S Counting methods a) successive numbers . . . . . . 60.5 b) tapping . . . . . . . . . 15.0 c) counting breaths . . . . . . 5.4 d) repeated mental events . . . . 4.4 e) pulse count . . . . . . . . 3.4 “Feel" . . . . . . . . . . . . 6.1 Guess . . . . . . . . . . . 4.8 Internal clock . . . . . . . . . . 0.2 Sleep . . . . . . . . . . . . 0.1 Listen for cues . . . . . . . . . 0.1 N = 120 100.0 Table 18.--Median CRd and number of Ss by majority method of synchrony in Experiment II Majority method of synchrony N Median CRd U* Counting methods a) successive numbers 69 1.00 76.5 b) tapping 17 1.10 14.5 c) counting breaths 6 1.10 2.0** d) repeated mental events 4 1.10 3.0 e) pulse count 4 1.10 6.0 "Feel" 5 1.00 6.5 No majority method 11 1.00 9.0 Guess 4 0.35 __- 120 * U is the Mann-Whitney U (Siege1,l956) and is based on _the comparison of the _CRds of Ss in each method group with the CRds of the S3 in the Guess group. ** p < .05 Table 19.--S—test analyses 118 Experiment II Comparison df t Age by sex (M vs. F) . . . . 118 0.00 H gsr by sex . . . . . . . . . 118 3.28*** Age by H gsr (+ vs. -,0) . . . . . 118 1.72 CRd by sex . . . . . . . . . . 118 0.69 CRd by age . . . . . . . . . 118 2.39** CRd by H gsr . . . . . . . . 118 3.28*** CR80 of ITI groups by Int.: "Short“ (H vs. L) . . . . 22 2.25* "Medium“ (H vs. L) . . . . 34 0.93 “Long" (H vs. L) . . . . 46 1.07 CRBO Of H Int. by ITI gp.: "Short" vs. "Long" . . . . . 34 0.98 "Short" vs. "Medium" . . . . . 28 1.28 “Long vs. "Medium" . . . . 40 3.06*** CR80 of L Int. by ITI gp.: “Short" vs. "Long" . . . . . 34 2.80*** "Short" vs. "Medium" . . . . . 28 6.15*** "Long“ vs. "Medium" . . . 40 3.42*** I gsr by ITI gp.: "Long" vs. "Short" (Blk 1) . . 70 0.71 "Long" vs. "Short" (Blk 2) . . 70 0.93 "Long" vs. "Short" (Blk 3) . . . 70 0.07 "Long & Short" vs. "Medium" (Blk 1) 106 0.82 "Long & Short" vs. "Medium" (Blk 2) 106 0.04 "Long & Short" vs. "Medium" (Blk 3) 106 2.47** OE gsr by sex . . . . . . . . . 118 4.66*** OE gsr by age . . . . 118 0.61 % time per S counting by H gsr . . 118 1.06 % time per S counting by ITI gp.: "Short" vs. "Long" . . . . 70 1.04 "Short" vs. "Medium" . . . . . 58 0.35 "Medium" vs. "Long" . . . . 82 0.66 % A. Est. by order of est. (lst vs. 2nd) 238 0.06 % A. Est. by ITI gp.: "Short" vs. "Long" . . . . . 70 0.61 "Short" vs. "Medium" . . . . . 58 1.66 "Long“ vs. "Medium" . . . . . 82 1.58 % A. Est. by sex . . . . . . . . 118 1.56 % A. Est. by age . . . . . . . . 118 0.48 % A. Est. by CEDE (Y vs. N) . . . . 99 0.70 % A. Est. by DEE (under vs. over) . . 99 3.97*** % A. Est. by Sb Un (Y vs. N) . . . . 118 0.27 % A. Est. by breath counting method (users vs. non-users) . . . . 118 0.93 * p < .05 ** p < .02 *** p < .01 Table 20.--Chi square analyses 119 Experiment II Comparison df Chi sq. Sex (M vs. F) & ITI (all 10) . . . 9 3.22 Sex & Int. (H vs. L) . . . . . . 1 0.33 Age (18 vs. 19-22) & ITI . . . . . 9 12.74 Age & Int. . . . . . . . . . 1 0.07 H gsr (+ vs. -,0) & ITI . . . . . 9 13.90 H gsr & Int. . . . . . . . . . 1 0.13 Sex & Age . . . . . . . . l 0.78 Sex & H gsr . . . . . . . . . 1 12.70*** Age & H gsr . . . . . . . . l 2.50 Sb Un (Y vs. N & ITI . . . . . . 9 13.05 Sb Un & Int. . . . . . . . . 1 8.89*** Sb Un & Sex . . . . . . . . . l 0.15 Sb Un & Age . . . . . . . . . 1 0.23 Sb Un & H gsr . . . . . . . 1 0.07 Blk 3 I gsr (+ vs. -,0) & Sex . . . 1 1.96 Blk 3 I gsr & Age . . . . . . . l 2.29 OE gsr (+ vs. -,0) & Sex . . . . . 1 13.97*** OE gsr & Age . . . . . . l 0.00 Counting methods (Y vs. N) & Sex . 1 0.26 Counting methods & Age . . . . . l 0.00 Counting methods & H gsr . . . . . 1 8.18*** Counting successive numbers & Sex . . l 0.82 Counting successive numbers & Age . . 1 1.68 Counting successive numbers & H gsr . l 0.00 Tapping & Sex . . . . . . . . 1 1.24 Tapping & Age . . . . . . . . 1 0.00 Tapping & H gsr . . . . . . . 1 0.00 Counting breaths & Sex . . . . . 1 0.00 Counting breaths & Age . . . . . 1 6.38** Counting breaths & H gsr . . . . 1 0.00 Repeated mental events method & Sex . 1 0.00 Repeated mental events method & Age . 1 4.87 Repeated mental events method & H gsr 1 0.04 Counting pulse & Sex . . . . . . 1 1.38 Counting pulse & Age . . . . . . 1 0.00 Counting pulse & H gsr . . . . . 1 0.00 “Feel" method & Sex . . . . . . l 3.49 "Feel" method & Age . . . . . . 1 0.00 "Feel" method & H gsr . . . . . . 1 3.07 No majority method & Sex . . . . . 1 0.43 No majority method & Age . . . . . 1 0.00 *1 *** p < .01 120 Table 20.--Continued. Comparison df Chi sq. No majority method & H gsr . . . l 1.32 Guess method & Sex . . . . . . . 1 0.00 Guess method & Age . . . . . . 1 10.00 Guess method & H gsr . . . . 1 0.19 CEDE (Y VS. N) & ITI . . . . . . 9 12.27 CEDE & Int. . . . . . . . l 0.15 CEDE & Sex . . . . . . l 0.03 CEDE & Age . . . . . . 1 2.80 CEDE & H gsr . . . . . . l 0.03 DEE & ITI . . . . . . 9 7.10 DEE & Int. . . . . . l 0.01 DEE & Sex . . . . . . . . . 1 0.04 DEE & Age . . . . . . . 1 2.90 DEE & H gsr . . . . . . . . . 1 0.68 “Ons" (Y vs. N) & ITI . . . . . 9 3.15 “Ons" & Int. . . . . . . . . 1 2.25 "Ons" & Sex . . . . . . 1 1.10 “Ons" & Age . . . . . . . 1 0.02 "Ons" & H gsr . . . . . . . 1 0.02 “Ons" vs. “Offs" & ITI . . . . 9 4.74 "Ons“ vs. "Offs" & Int. . . . 1 1.24 "Ons" vs. "Offs" & Sex . . . 1 1.12 "Ons“ vs. "Offs" & Age . . . . 1 0.01 "Ons" vs. "Offs" & H gsr . . . . 1 0.00 CRd over 1.00 (Y vs. N) & CED . . 1 0.29 CRd over 1.00 & DEE . . . . . . 1 0.01 CRd over 1.00 & "Ons" (Y vs. N). . l 0.71 CRd over 1.00 & "Ons“ vs. "Offs" . 1 0.11 CRd over 1.00 & Sb Un . . . . . l 0.01 CEDE & DEE . . . . . . . . 1 0.64 CEDE & Blk 3 I gsr . . . . . 1 0.00 CEDE & OE gsr . . . . . 1 2.98 CEDE & Sb Un . . . . . . . 1 0.92 DEE & Blk 3 I gsr . . . . . . 1 1.12 DEE & OE gsr . . . . . . . . . l 0.05 DEE & Sb Un . . . . . . . . . 1 0.00 "Ons" (Y vs. N) & Blk 3 I gsr . . . 1 2.45 "Ons" & OE gsr . . . . . . . . 1 5.57** "Ons" & Sb Un . . . . . . . . 1 0.32 "Ons" vs. "Offs" & Blk 3 I gsr . . . 1 0.91 "Ons" vs. "Offs" & OE gsr . . . . l 3.96* "Ons" vs. “Offs" & Sb Un . . . . . 1 0.00 Blk 3 I gsr & Sb Un . . . . . . l 0.01 OE gsr & Sb Un . . . . . . . . 1 0.47 * p < .05 ** p < .02 121 Table 21.--Eetest analyses of occurrence of CR8 with ITIs pooled by trial block and stage of learning in Experiment III Comparison (i % occur. in paren.) df t Acq. D2 vs. Acq. D1: Blk 1 (60) vs. Blk 1 (50) 22 1.07 Blk 1 (60) vs. Blk 2 (57) 22 0.34 Blk 1 (60) vs. Blk 3 (65) 22 0.56 Blk 2 (68) vs. Blk 1 (50) 22 2.30* Blk 2 (68) vs. Blk 2 (57) 22 1.35 Blk 2 (68) vs. Blk 3 (65) 22 0.46 Blk 3 (82) vs. Blk 1 (50) 22 4.08** Blk 3 (82) vs. Blk 2 (57) 22 2.95** Blk 3 (82) vs. Blk 3 (65) 22 2.31* All Blks (70) vs. All Blks (57) Ext. D2 vs. Acq. D 2: Blk 1 (48) vs. Blk 3 (82) 22 2.91** Blk 1 (48) vs. Blk 2 (68) 22 1.72 Blk 1 (48) vs. Blk 1 (60) 22 0.92 Blk 2 (32) vs. Blk 3 (82) 22 4.46** Blk 2 (32) vs. Blk 2 (68) 22 3.25** Blk 2 (32) vs. Blk 1 (60) 22 2.31* Blk 3 (20) vs. Blk 3 (82) 22 5.92** Blk 3 (20) vs. Blk 2 (68) 22 4.57** Blk 3 (20) vs. Blk 1 (60) 22 3.45** All blks (33) vs. A11 Blks (70) 22 6.10** Ext. D2 vs. Acq. D1: Blk 1 (48) vs. Blk 1 (50) 22 0.14 Blk 1 (48) vs. Blk 2 (57) 22 0.67 Blk 1 (48) vs. Blk 3 (65) 22 1.48 Blk 2 (32) vs. Blk 1 (50) 22 1.62 Blk 2 (32) vs. Blk 2 (57) 22 2.12* Blk 2 (32) vs. Blk 3 (65) 22 3.05** Blk 3 (20) vs. Blk 1 (50) 22 2.83** Blk 3 (20) vs. Blk 2 (57) 22 3.27** Blk 3 (20) vs. Blk 3 (65) 22 4.41** All Blks (33) vs. A11 Blks (57) 22 4.25** * p < 05 ** p < .01 Table 22.--Ana1ysis of variance of occurrence of CR80 in Acq. at H Int. by ITI (60 and 240 sec.) and by Experiment II vs. Experiment III (D1) Source SS df MS F A: ITI 1.4 l 1.40 0.89 B. EXperiment 0.5 1 0.50 0.32 A x B 0.0 1 0.04 0.02 error 31.6 20 1.58 Total 33.5 23 Table 23.--Ana1ysis of variance of occurrence of CRgo in acq. (D1) by ITI and trial blocks in Experiment III Source SS df MS F A: ITI 0.4 1 0.39 0.34 error (a) 11.6 10 1.16 B: Trial blks 3.1 2 1.54 1.60 A x B 1.7 2 0.85 0.89 error (b) 19.2 20 0.96 Total 36.0 35 Table 24.--Ana1ysis of variance of occurrence of CR80 in Acq. (D2) by ITI and trial blocks in Experiment III Source SS df MS F A: ITI 0.0 1 0.00 0.00 error (a) 4.3 10 0.43 B: Trial blks 7.2 2 3.58 2.28 A x B 2.2 2 1.09 0.70 error (b) 31.3 20 1.57 Total 45.0 35 123 Table 25.--Ana1ysis of variance of occurrence of CRgo in Ext. (D2) by ITI and trial blocks in Experiment III Source SS df MS F A: ITI 1.8 l 1.78 0.31 error (a) 57.6 10 5.76 B: Trial Blks 12.2 2 6.09 4.01* A x B 2.1 2 1.02 0.67 error (b) 30.4 20 1.52 Total 104.1 35 * p .05 Table 26.--Ana1ysis of variance of per cent A. Est. by ITI, successive estimate, and successive day in Experiment III Source SS df MS F A: ITI 88 1 88.0 0.10 error (a) 8599 10 859.9 B: Successive Days 653 1 653.0 7.35* C: Successive Estimates 99 1 99.0 1.11 A x B 143 1 143.0 1.61 A x C 1 1 1.0 0.01 B x C 2 1 2.0 0.02 A x B x C 77 1 77.0 0.87 error (b) 2664 30 88.8 Total 12326 47 ‘ .05 124 Table 27.--Ana1ysis of variance of OE gsr by ITI and succes- sive days of Acq. in Experiment III Source 88 df MS F A: ITI 23563 1 23563 2.79 error (a) 84322 10 8432 B: Successive Days 18593 1 18593 9.98* A x B 73 1 73 0.04 error (b) 18631 10 1863 Total 145182 23 * p . .05 Table 28.-~Ana1ysis of variance of I gsr by ITI, trial blocks, and successive days of Acq. in Experiment III Source SS df MS F A: ITI 29 1 29.00 0.50 error (a) 581 10 58.10 B: Successive Days 162 1 162.00 3.12 C: Trial Blk 593 2 296.50 5.71* A x B 12 1 12.00 0.23 A x C 91 2 45.50 0.88 B x C 162 2 81.00 1.56 A x B x C 70 2 35.00 0.67 error (b) 2596 50 51.92 0.67 Total 4296 71 125 mm mm MI mme HHH am ow mH m me mm mm o m me me om o H me .HeH mg m m m m m H H b mme HHH H m m H H o m me H H m o o o m me H N H o o o I H me .m an amnesz ommo mo Wucmnusooo ucH m H6 mm oo\mmu» w z w w w w m an “no gm om.uom “no.+nm mo.+ mo.| HeH mm cmHowz No.+ Ho.H+ 40.- mo.+ mo.- mo.- m an oum "Hmm m an m m HaH an m: H0 z Hmuoe z z z z z E m an "xmm m.mH m.mH HaH ma x n.mH mH mH mH mH mH mH m mm "mm« HomH u 21 mm mm Hm mm mm Hm HeH omm oomuom Amnzv HBH omm oomH Hmuzv HeH omm comm muommumo mums HH pcmEHHmmxm . >H ucmEHHmmxm m Mom mafiucsoo .Hmm mo .Hmm H .mma .momu ..umm .¢ ..uc¢ .m .ommu mo mocmuusooo .cD am .Mmm m .xmm .mmm "HH ucmEHHmmxm ou ommommo mm >H Dameflummxm CH mcofluflpcoo HRH How mumEESm mumoll.m~ mance 126 no om mm HeH an mlcmmz QOH so me OOH OOH om m an H .umm umH "umm .a Hm om mme HHH Hm. om m me om mm m me H mm «m H me HeH an Houum w cMHomz ".u:« .m HEM mva. mme HH< mam moom m me Hmm OOHN m me mew OHNH H me HRH Sn 0mm cmHomz wbm _bmm Hem mmHm HGOH mew mme HHH mmv aHm mum moom momH GmHm m me ‘ onm Hmm qu mmHm «mm OOHN N me mvs mMH mmv mva HmHH 0H~H I H me m an 0mm cMHomz ".ucm .m oxxhloo (o O V .H mmo 0 CNN Nt-lt-i H AomH u zc mm mm Hm mm mm Hm HeH 0mm oomuom Amaze HeH omm oomH Amaze HeH omm comm wuommumo mama HH ucmEHquxm >H ucmEHHmmxm .cmscHucoouu.m~ mHnma 127 Hm mm mm HHH Mg m m Hm om mm mm o Hm I m Hg m um Mom mcwucsoo Ho.+ mm.- mo.H- Hem Ha oum m mH.+ Hm.H- mm.H- mH.H- mm.H- Hm.- m Ha onm "Hmm mo Ho.- mo.- co. m HHm mo.. 00. 40.- m me HH.+ oo. mo.+ H me HBH an cum M 40.- Ho.+ mo.- Hm.- ~¢.+ mo.- m me Ho.+ mo.- No.+ NH.- HH.u mH.+ m me Ho.+ HH.+ mm.- mm.+ Ho.u mo.+ IH HHm I m an oum “umm H omH\mm u H u H H I u H m Ha ammmo "mmo o~H\~m u H u 2 H u u H m Hg "memo HomH u 21 mm mm Hm mm mm Hm HHH omm oomuom Amuzv HHH 0mm oomH Amuzv HHH omm comm Huommumo mama HH usmEHummxm >H ucmEHummxm .mmscHumooun.m~ mHan 128 Table 30.--Samp1e data sheet from all experiments DATA SHEET TEMPORAL CONDITIONING Date 1965 Age Sex ITI Bulb Time Name of S Examiner I. GSR 0.50 min. 2.50 min. II. Acquisition (XX = UR) Seconds before Time GSR Flash # flash "NOW" said Taken GSR 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 III. Estimation Time in seconds S: Over or under-estimated How much? IV. Questions Method Count When used 1. Method of synchrony: 2. Bulb flash pleasant, unpleasant, or neutral on average? 3. Other comments: APPENDIX B 130 Alternate Methods of CR Definition in Extinction in Experiment III CR is defined as any vocal response within the last 10% of the ITI before the periodic flash would have occurred or within the first 10% of the ITI after the periodic flash would have occurred. Mean % Occurrence of CR by Block of 5 Trials II. S Block 1 BloCk 2 Block 3 1 60 4O 60 2 40 0 60 3 100 100 80 4 40 40 20 5 100 100 100 6 0 0 0 7 100 80 100 8 80 100 20 9 100 100 60 10 0 40 20 11 100 80 100 1_2. m. 29. 0 Mean 67 62 52 CR is defined as any vocal response within the last 20% of the ITI before the periodic flash would have occurred 9E within the first 20% of the ITI after the periodic flash would have occurred. the acceptable time range for a CR as compared to the above definition as well as the definition used in Ex- periment III. [(1) \OCDQO‘WIbWNH Mean % Occurrence of CR by Block of 5 Trials Block 1 80 60 100 100 100 0 100 100 100 40 100 199 82 Block 2 100 20 100 100 100 0 100 100 100 100 100 8_g 83 Note that this method doubles Block 3 100 60 100 100 100 0 100 60 100 80 100 0 75 131 Secondary Elicitation and Relaxation During the Acquisition of a Temporally Conditioned Response Elicitation theory (Denny and Adelman, 1955) is basically an S-R contiguity theory of learning. One of the major postulates is that of secondary elicitation which deals with the omission of an unconditional stimulus or consistent elicitor. Secondary elicitation is defined as the elicitation of a new characteristic class of re- sponse which is typically antagonistic to the original response, and which occurs with the omission of the ori- ginal elicitor (US). As used in an aVOIdance learning context, secondary elicitation means that cues discrimi- nated by S as occurring in contiguity with non-shock (a safe chamber) will acquire the prOperty of eliciting re- 1axationa1 response, just as the cues associated with shock come to elicit arousal and escape responses. Extinction is explained by assuming secondary elicitation brings about relaxational response in the shock area when shock is omitted. While secondary elicitation has thus generally been used to explain the extinction of avoidance behavior, the temporal conditioning situation suggests another application. The acquisition and extinction of a temporally con- ditioned response take place in the presence of common external cues. The CS or cue to be discriminated must of necessity be some periodic interoceptive process. Such a discrimination does in fact occur with repeated presentations of the periodic US. As mentioned in the Introduction (p. 19), the photic US of the preceding experiments elicited immediate URs of arousal as gauged by gsr. The omission of the photic flash during the ITI, once S has discriminated that ITI as not associated with the noxious flash, should therefore lead to relaxational response according to secondary elicitation. The secondary elicitation postulate is being applied here to explain differential behavior during the ITI (as a function of perceiving the ITI as a relaxational cue) rather than an explanation of extinction. Reynierse (1964) used the same sort of explanation to account for the effect of a delayed avoidance trial on resistance to extinction (in rats). Interestingly enough, the gsr data support this point of View. As Figure 13 (p. 51) shows, those ITIs associated with significant acquisition of the temporal response showed significant relaxation (as measured by gsr) from the midpoint of one ITI to the next by the third block of trials. Those ITIs not associated with significant acquisition of the tem- poral response showed greater mean arousal from the midpoint of one ITI to the next. Whether this differential relaxation was a result or a cause of the temporal discrimination cannot yet be determined from the data. That the latter is also a possibility is suggested by the significant correlation be- tween pre-treatment relaxation and level of conditioning. "I1111111111111111111