Dynamics in the fine structure of schedule-controlled behavior.

The variability in the behavioral equilibrium established by six basic schedules was characterized. The measures were the pause preceding the first response in each interreinforcement interval; the mean rate of responding in each interreinforcement interval; and the relative frequency of each interresponse time. The temporal windows ranged across the 780-session exposure, across a session, and across the interreinforcement interval. A display of individual interresponse times as a function of time in the interreinforcement interval indicated clear recurrent responding at somewhat less than 3 Hz in every bird, even after extended exposure to a schedule and regardless of the contingency. No strong sequential dependencies in the interresponse-time distributions were identified. A simulator, based on a simple recurrent pulser, was presented that produced output similar to the obtained data. An archival data base of the behavior chronically maintained by the simple schedules was also generated.     

ON THE INABILITY OF INTERVAL TIME SAMPLING TO REFLECT FREQUENCY OF OCCURRENCE DATA1

Interval time sampling yields the result, percentage of intervals scored. It is rudimentary to note that this measure per se does not constitute a response dimension. It is a useful behavioral measure, therefore, only to the extent that it accurately reflects the nature and degree of the fundamental dimensions from which it is drawn i.e., frequency and duration. The correspondence between scored intervals and response duration is fairly well understood (Journal of Applied Behavior Analysis, 1975, 8 , 463–469; 1977, 10 , 325–332). This study determined the correspondence between scored intervals and response frequency. Eleven, 30-minute experimental sessions that differed along the variables of frequency of occurrence and time per response (the average length of a response per session) were computer simulated. In the first group of four sessions, the frequencies were 45, 100, 150, and 300; in these sessions, all responses ranged from one to three seconds. In the second group of four sessions, the frequencies were 31, 61, 101, and 152; in these sessions, all responses ranged from three to nine seconds. In the last group of three sessions, the frequencies were 25, 34, and 50; here, all responses ranged from nine to 27 seconds. The response distribution within the above ranges was rectangular, with each whole second represented once. The responses were selected by a random number generator, and on each trial every number in the distribution had an equal probability of occurrence. These provisions produced a linear pattern of responding. The time per response in the three groups of sessions were 2, 6, and 18 seconds. For all sessions, event recordings were made and analyzed. The analysis consisted of using partial interval time sampling to determine the percentage of intervals scored; this total was subdivided into intervals containing (1) single responses, (2) multiple responses, (3) continuing responses, and (4) response initiations or terminations. The analysis was performed when the length of the observation interval was 5, 10, 20, 30, 60, and 120 seconds. An additional session drawn from a study that contained real-life data was subjected to this same analysis. The most significant results were derived by finding the ratio of scored intervals containing single responses to the total intervals scored. If every scored interval contained a single response, this ratio would equal 1.0; if no scored interval contained a single response, the ratio would be 0.0. It can be seen that this ratio is an objective expression of the validity of interval time sampling as a measure of response frequency. Of 66 data points (11 sessions × six observation lengths per session), only five were equal to or greater than 0.80. These five points were all found in just two sessions (f = 45, 100). A validity index of less than 0.50 was observed in 49 of the 66 points. Also, the validity index increased, peaked, and then decreased within sessions as the length of the observation interval was increased. The results from the real-life session were in close agreement with those obtained from the simulations. The importance of these and previous findings lies in the demonstration that changes in scored intervals need not represent true behavior change. The data indicate that there are many combinations of behavioral frequency and duration where interval time sampling cannot produce valid measurement results.     

Accuracy of observational data with whole interval,partial interval,and momentary time-sampling recording techniques

The purpose of the study was to compare the accuracies obtained from whole interval, partial interval, and momentary time-sample recording procedures. Two types of accuracies were defined. Between-methods accuracy is the degree to which an observer using a particular recording method agrees with a standard using a continuous recording method. Within-methods accuracy is the degree to which an observer and a standard agree when both use the same recording method. Fifty-four undergraduate students viewed a videotape of a woman twisting her hair and recorded the occurrence or nonoccurrence of the behavior using one of the three recording methods. Tapes were divided into segments with low, intermediate, and high rates of behavior. Partial and whole interval recording obtained poorer between-methods accuracies than momentary time sampling. Results showed that whole interval and momentary time-sampling procedures yielded higher within-method accuracies than partial interval recording. Thus, momentary time-sampling recording provided the advantages of both greater representativeness and as few or fewer observer errors than whole or partial interval recording procedures. Use of the interval recording methods by researchers must be justifiable within this framework.     

Failure to produce response variability with reinforcement

Two experiments attempted to train pigeons to produce variable response sequences. In the first, naive pigeons were exposed to a procedure requiring four pecks on each of two keys in any order, with a reinforcer delivered only if a given sequence was different from the preceding one. In the second experiment, the same pigeons were exposed to this procedure after having been trained successfully to alternate between two specific response sequences. In neither case did any pigeon produce more than a few different sequences or obtain more than 50% of the possible reinforcers. Stereotyped sequences developed even though stereotypy was not reinforced. It is suggested that reinforcers have both hedonic and informative properties and that the hedonic properties are responsible for sterotyped repetition of reinforced responses, even when stereotypy is negatively related to reinforcer delivery.     

CAN A DECAY PROCESS EXPLAIN THE TIMING OF CONDITIONED RESPONSES?

To explain time-scale invariant distributions of response latencies, it appears to be necessary to postulate scalar noise in the remembered intervals, against which the subjective measure of the currently elapsing interval is compared. At least in some cases, the observed variability cannot be due to variability in the subjective intervals written to memory; it must come from noise (variability) in the reading of a memory. The Staddon and Higa proposal offers no explanation for the observed variability, and it is unclear what noise assumption would yield the observed variability, given their assumption that intervals are timed by a nonlinear decay process. The decay process cannot plausibly be represented by the logarithmic function, because it begins and ends at infinity. The assumption of any form of nonlinear timing is inconsistent with the most important result of the time-left experiment, which is that the changeover time increases linearly with the comparison-standard difference.     

Using extinction to promote response variability in toy play

We report the effects of extinction and positive reinforcement on the number of untrained topographies emitted by children with toys. Baseline showed no appropriate toy play. Participants were then trained individually on one topography for each toy. Previously reinforced topographies of toy play were placed on extinction, resulting in the induction of untrained topographies.     

Effects of fixed and variable ratios on human behavioral variability.

The effect that ratio schedules of reinforcement had upon variability of responding was investigated in college students. Subjects were paid $0.02 contingent upon completion of eight presses, distributed in any combination across two push buttons; 256 different sequences were possible. Sequence emission was reinforced according to fixed- and variable-ratio schedules. Ratio requirements of 1, 2, 4 and 8 were presented in alternate components of a multiple schedule. The variability engendered by variable-ratio schedules was also compared to that engendered by fixed ratios. Variability increased with ratio size, irrespective of whether the schedule requirement was fixed or variable. The data demonstrate the similarity between the determinants of human and nonhuman variability, and they illustrate the role of ratio size in determining variability in operant behavior.     

Effects of response variability on the sensitivity of rule-governed behavior

Two experiments examined the relation between response variability and sensitivity to changes in reinforcement contingencies. In Experiment 1, two groups of college students were provided complete instructions regarding a button-pressing task; the instructions stated “press the button 40 times for each point” (exchangeable for money). Two additional groups received incomplete instructions that omitted the pattern of responding required for reinforcement under the same schedule. Sensitivity was tested in one completely instructed and one incompletely instructed group after responding had met a stability criterion, and for the remaining two groups after a short exposure to the original schedule. The three groups of subjects whose responding was completely instructed or who had met the stability criterion showed little variability at the moment of change in the reinforcement schedule. The responding of these three groups also was insensitive to the contingency change. Incompletely instructed short-exposure responding was more variable at the moment of schedule change and was sensitive to the new contingency in four of six cases. In Experiment 2, completely and incompletely instructed responding first met a stability criterion. This was followed by a test that showed no sensitivity to a contingency change. A strategic instruction was then presented that stated variable responding would work best. Five of 6 subjects showed increased variability after this instruction, and all 6 showed sensitivity to contingency change. The findings are discussed from a selectionist perspective that describes response acquisition as a process of variation, selection, and maintenance. From this perspective, sensitivity to contingency changes is described as a function of variables that produce response variability.     

Risk-sensitive foraging theory and operant psychology.

Hastjarjo, Silberberg, and Hursh (1990) have presented data on the foraging behavior of rats and discussed it in terms of risk-sensitive foraging theory. Because risk-sensitive foraging theory is comprised of several different models, it does not lead to general predictions about when an organism should prefer a foraging option with high variance to a foraging option with low variance. Any comparison of data with the predictions of the theory must be based on an appropriate model. I draw attention to various experiments that are potentially relevant to the results reported by Hastjarjo et al. and show how the time period over which the organism must survive can influence a model's predictions about risk sensitivity.     

Enumeration and simulation methods for 0–1 matrices with given marginals

Data in the form of zero-one matrices where conditioning on the marginals is relevant arise in diverse fields such as social networks and ecology; directed graphs constitute an important special case. An algorithm is given for the complete enumeration of the family of all zero-one matrices with given marginals and with a prespecified set of cells with structural zero entries. Complete enumeration is computationally feasible only for relatively small matrices. Therefore, a more useable Monte Carlo simulation method for the uniform distribution over this family is given, based on unequal probability sampling and ratio estimation. This method is applied to testing reciprocity of choices in social networks.The author wishes to thank Cajo ter Braak and John Birks for pointing out the relevance of this subject for ecology; and also Albert Verbeek and Ivo Molenaar, a referee, the Editor, and the Associate Editor for their comments. An earlier version of this paper was presented at the Stockholm Conference on Random Graphs and Applications (April 25–27, 1989), organized with financial support from the Swedish Council of Research in the Humanities and the Social Sciences.