Printer Friendly

Response-reinforcer contiguity and human performance on simple time-based reinforcement schedules.

Often a schedule of reinforcement is narrowly conceived of as an experimental device that arranges a relationship between the occurrence of a response and the occurrence of a reinforcer (Davey, 1987; Mackintosh, 1983; Schwartz, 1984). This rather simplistic view, however, directs attention away from the complex, interactive behavioral relations that occur on even the most basic of reinforcement schedules. Consider, for example, the fixed-interval (FI) schedule. Keenan and Leslie (1984, 1986) have shown that four major variables acting in concert are responsible for the coordinated maintenance of nonhuman behavior on this schedule. The variables they noted are (a) the fixed reinforcer-reinforcer interval or fixed-time (FT) component, (b) the single response contingency or fixed-ratio 1 (FR 1), (c) the time from one reinforcer presentation to the location of the next response dependency, and (d) response-reinforcer contiguity. The dissection of the FI schedule into these four components now opens to the experimentalist a research strategy involving the employment of other schedule contexts exclusively created from these same four component features (e.g., Keenan & Watt, 1990). Although this approach to the study of reinforcement schedules was developed with nonhuman subjects, this research strategy might also be of benefit to human operant researchers. The current research outlines how this approach may be adopted with human adult subjects. When adult humans are provided with minimal instructions (e.g., "You have to press this button to win points.") before exposure to the FI schedule, two performances usually emerge. Some subjects emit a constant high-rate pattern of responding throughout the interreinforcer interval and report a ratio-based verbal formulation of the contingencies at the end of the experimental session (e.g., Leander, Lippman, & Meyer, 1968). Other subjects demonstrate a low-rate pattern of responding consisting of one or two responses at the end of the interreinforcer interval and report an interval-based verbal formulation of the contingencies (e.g., Lippman & Meyer, 1967; Weiner, 1969). In contrast, when humans are provided with ratio-based instructions (e.g., "You have to press this button a random number of times to get a point."), subjects produce predominantly high-rate patterns and ratio-based verbal reports (Lippman & Meyer, 1967). Despite these findings, however, we still do not fully understand how the FI schedule can give rise to, and maintain, both high-rate and low-rate performances with human subjects (cf. Barnes, 1989; Davey, 1988; Wearden, 1988). The general approach to schedule research outlined previously considered human FI performance (both verbal and nonverbal) as determined, in part, by the behavioral dynamics emerging from the four components of the FI schedule. Given this view, it may be useful to explore human schedule performance through the systematic manipulation of these component features. The two experiments reported here initiated this research strategy. The experiments employed four different schedules constructed from three combinations of the FT and FR 1 components. A schematic representation of the four schedules is presented in Figure 1 (examination of this figure is essential for an appreciation of the complex nature of the component features that define these schedules). On the FI schedule a single response made after the FT component has expired produces an immediate reinforcer; responses made during the FT component have no programmed consequences (thus every reinforcer is contiguous with a response). In contrast, a "conjunctive FT FR 1" schedule imposes no temporal restrictions regarding where, within the FT component, the FR 1 component may be executed. If a response occurs within the FT component, then the reinforcer is delivered when the FT component terminates (thus response-reinforcer contiguity is no longer guaranteed); if the FT component expires without a response having been made, the reinforcer is produced immediately by the next response. During Experiment 1 subjects were exposed to these two schedules of reinforcement.

A third combination of FT and FR 1 components is the "recycling conjunctive FT FR 1." Here a reinforcer is presented at the termination if of the FT component if, and only if, a response has occurred within that component. Where the FT component terminates without a response having been made, reinforcer delivery is omitted and the next FT component begins immediately. A modified version of this schedule incorporates a "zero-delay window" which is inserted at the terminal end of the FT component. Responding during this period produces the reinforcer immediately and terminates the FT component, and thus there is greater opportunity for response- reinforcer contiguity. Note, however, that if the response occurs outside the zero-delay window, reinforcement is presented at the end of the FT component, and thus response-reinforcer delays may still occur on this schedule. In Experiment 2 subjects were exposed to the normal and modified recycling conjunctive schedules. Because the reliability of retrospective verbal reports often decreases as a function of time and number of manipulations of the independent variable (see Ericsson & Simon, 1984), subjects in both experiments were exposed to a single session, with only one transition across schedules during that session (pilot studies showed that when human performance changed across schedules the change often occurred very rapidly and remained stable thereafter). Immediately at the end of the session subjects were given a questionnaire asking them to report on their performance. The order in which the schedules were introduced was counterbalanced across subjects. Additionally, for both experiments one group of subjects received minimal instructions, and a second group received ratio- based instructions. In this way it was possible to examine some of the interactive effects that may occur between verbal instructions and the structure of reinforcement schedules.

EXPERIMENT 1

Method

Subjects

Forty non-psychology undergraduate students, 24 female and 16 male, enrolled at the University of Ulster at Coleraine, served as subjects. Their ages ranged from 18 to 32 years (median = 19). All subjects were recruited through faculty notice-board advertisements and personal contacts. Subjects could earn a maximum of [pounds] 1.50 ($3.00 approx) during their participation in the experiment.

Apparatus and Materials

Subjects sat at a table in a small experimental room facing a computer with a 30.5-cm screen which displayed green characters on a black background. All experimental conditions and data collection were controlled by the computer. No watches or clocks were allowed in the experimental room. Procedure

Subjects were randomly assigned to one of two conditions (i.e., 20 subjects to each condition). One group received minimal instructions and the other group received ratio instructions. Instructions were presented on the monitor screen before the session commenced. Minimal instructions were as follows: In order to obtain points (each point is worth 2 pence) from the computer all you have to do is press the space-bar at the bottom of the keyboard. Press the letter 'S when you are ready to start. Do not leave until the computer tells you the session is over.

Ratio instructions were identical to the minimal instructions except that the end of the first sentence read as follows: "a random number of times." If subjects asked any questions they were referred to these instructions. The experimenter left the room before the subject pressed the "S" key, which started the session.

Each subject was exposed to one session. Ten subjects from the minimally instructed group and 10 subjects from the ratio- instructed group were exposed to 25 trials on a FI 10-s schedule of reinforcement followed immediately by 50 trials on a conjunctive FT 10-s FR 1 schedule. The remaining 10 subjects from each group were exposed to 25 trials on the conjunctive followed immediately by 50 trials on the FI schedule. The transition across schedules was unsignaled.

During exposure to the FI schedule the first response (i.e., press on the space-bar) after the 10-s interval had elapsed produced a 1- s tone from the computer, and the following message appeared on the monitor screen: "You have earned a point." After 2 s the screen cleared and the next interval began immediately. On the conjunctive schedule one or more responses at any time during the FT10-s interval produced the message and a single 1-s tone at the end of the interval, and when the screen cleared the next interval began immediately. If a subject did not respond during the FT interval, the first response after the interval had elapsed produced the tone and the message immediately, and when the screen cleared the next interval commenced. Table 1 presents the obtained rates of reinforcement on each schedule for each subject with either minimal instructions or ratio instructions.

[TABULAR DATA OMITTED]

When a session terminated (after a total of 75 trials) the computer gave a random sound sequence and the following message appeared on the monitor screen: "The session is over. Please leave and report to the experimenter." Each subject was required to complete, in writing, the following questionnaire before leaving the laboratory:

1. What did you think the experiment was about?

2. Did you always employ the same strategy for obtaining points?

3. You answered YES describe the strategy you employed and why, but if you answered NO describe the strategies you employed and explain why you changed strategies during the session.

4. Describe, as best you can, the rules you think the computer was using to deliver points to you.

Subjects were not asked any questions directly by the experimenter, but if any comments were volunteered by the subjects they were recorded on their questionnaire sheet after the subject had left.

Results

Figures 2 to 5 show the results from the entire session for each subject. The upper panel for each subject shows the number of responses emitted on each trial, and the lower panel shows the postreinforcement pause (PRP) on each trial. As outlined earlier the FT 10-s FR 1 conjunctive schedule does not guarantee response- reinforcer contiguity, and thus delays between a given response and the delivery of a point may occur on this schedule. The first response-reinforcer delay of 1 s or more that occurred on the conjunctive schedule is marked by a downward pointing arrow (for convenience these, and all response-reinforcer delays equal to or greater than 1 s, will be referred to as 1-s delays). The three numbers located in the upper panel for each subject record the following events on the conjunctive schedule after this first 1-s delay: The first figure indicates the number of 1-s delays, the second figure shows the number of PRPs that extended beyond the FT 10-s interval (i.e., programmed response-reinforcer contiguities), and the third figure records the number of accidental response- reinforcer contiguities (i.e., one or more responses occurring in the tenth second of the FT interval). Note that for Subjects 33 and 40 no delays occurred on the conjunctive, and thus the three numbers read 0/0/0. No consistent relationships were obtained between these values and subjects' schedule performance. Nevertheless, this information is included in order to provide a more detailed record of the obtained response-reinforcer interactions for each subject on the conjunctive schedule.

Minimal Instructions

FI to conjunctive. Figure 2 shows data for Subjects 1 to 10. Five subjects (3, 5, 6, 9, and 10) showed high response rates (i.e., five or more responses per interval) and short PRP durations (less than 5 s) on FI 10-s followed by a decrease in response rates and an increase in PRPs during exposure to the conjunctive FT 10-s FR 1 schedule. Although S7 also showed a similar change in response rate, PRPs were relatively unaffected. These changes in schedule performance always occurred within one to four trials after the first 1-s delay. All these subjects reported that they assumed a ratio requirement at the beginning of the session, but realized that a time restriction was involved after the first response- reinforcer delay occurred (e.g., "At first I thought you had to keep pressing, but then I got a point when I wasn't pressing, and that made me think that it must have something to do with time."). It should be noted in the description of the results that the first 1-s delay was not necessarily the first delay that a subject actually experienced (i.e., subjects may have noticed delays of less than 1 s). By way of recognizing this possibility a subject's report which refers to a delay will be described as a response- reinforcer delay rather than a 1-s delay. Three subjects (2, 4, and 8) emitted low response rates and extended PRPs across the FI and conjunctive schedules (but S2 showed a dramatic reduction in PRP durations toward the end of the session). These subjects reported that points were delivered only after a certain period of time throughout the whole session. Subject S2, however, also reported the change in the location of the single response contingency, and the change in the reinforcer to response dependency interval, across schedules; "at first you had to wait and press, but then you could press and wait."

Subject S1 showed a gradual increase in response rate and inconsistent PRPs across trials on the FI and, initially, on the conjunctive schedule. However, rates reduced quite abruptly and PRPs increased after the first 1-s delay. This subject reported; "To begin with I wasn't sure whether you had to wait or keep pressing for points, but then I got a point before I pressed which meant that some sort of time interval was operating."

Response-reinforcer interactions across subjects after the first 1- s delay were as follows. The number of 1-s delays ranged between 8 and 47, the number of programmed contiguities ranged between 1 and 17, and the number of accidental contiguities ranged between 0 and 17.

Conjunctive to FI. Figure 3 shows results for Subjects 11 to 20. All subjects emitted predominantly low response rates on the conjunctive schedule (the first 1-s delay always occurred within the first six trials). Seven subjects (11, 13, 15, 16, 18, 19, and 20) continued to show low response rates across the FI. The three remaining-subjects (12, 14, and 17) showed an increase in responding after transferring to the FI schedule, but two of these (14 and 17) showed an abrupt return to a low-rate performance towards the end of the session. All subjects reported that the occurrence of a response-reinforcer delay indicated some form of temporal limitation on point delivery. Subject S12 also suggested that there may have been a ratio requirement (i.e., "most of the time you got points every fifteen seconds or so, but sometimes you got them more quickly if you kept pressing, I think.").

The PRPs were variable across subjects. Five subjects (11, 14, 15, 17, and 20) showed mainly short PRPs on the conjunctive with an abrupt shift to extended PRPs on the FI; this shift occurred toward the end of the session for S14 and S17. The verbal reports obtained from these five subjects always referred to the change in the location of the single response contingency, and to the change in the reinforcer to response dependency interval, across schedules. In the words of S14; "at first I pressed after I got a point and waited, but later this didn't seem to work. I started to press really quickly, but after a bit I worked out you had to wait about twelve seconds and then press." For Subjects 16 and 19 there was a more gradual increase in PRP durations across the whole session on both schedules. Verbal reports were congruent with these performances. In the words of S16; "to begin with I kept pressing just after I got a point, but then I started to count to work out how long you had to wait between points." The PRPs were extended for S13 and S18 throughout most of the session. Again, verbal reports reflected these durations (e.g., "I soon learned that you had to wait after a point before pressing."). The PRPs were variable for S12, and this subject reported; "I know it had something to do with time intervals, but I couldn't work out exactly how long you had to wait before you could get a point."

Response-reinforcer interactions after the first 1-s delay were as follows. The number of 1-s delays ranged between 4 and 24, the number of programmed contiguities ranged between 0 and 12, and the number of accidental contiguities ranged between 0 and 8.

Ratio Instructions

FI to conjunctive. Figure 4 shows data for Subjects 21 to 30. Eight subjects (all but S23 and S25) showed high response rates and short PRP durations on FI 10-s followed by a decrease in response rates and an increase in PRPs during exposure to the conjunctive FT 10-s FR 1 schedule. Although S23 also showed a similar change in response rate, PRPs were relatively unaffected. For five of the subjects (23, 24, 26, 27, and 29) these changes in schedule performance occurred quite abruptly after the first 1-s delay, and four of them (21, 22, 28, and 30) showed this shift more gradually following this first delay. All nine of these subjects reported that they initially believed the ratio instructions, but realized that a time restriction was involved after the first response- reinforcer delay occurred (e.g., "I believed the computer at first and just kept pressing, but then I got a point when I was resting my hand. This made me think that you could only get a point after a certain amount of time, and not before.").

Subject 25 emitted a low response rate and extended PRPs across FI and conjunctive schedules. A 1-s delay occurred on the 50th trial, but had no effect on the subject's subsequent performance. This subject reported; "The first time I pressed the space-bar I got a point, so I knew that you didn't have to keep pressing to get points."

Response-reinforcer interactions across subjects after the first 1- s delay were as follows. The number of 1-s delays ranged between 0 and 41, the number of programmed contiguities ranged between 0 and 25, and the number of accidental contiguities ranged between 0 and 11.

Conjunctive to FI. Figure 5 shows results for Subjects 31 to 40. Six subjects (31, 32, 34, 35, 36, and 39) showed high response rates during initial exposure to the conjunctive, but rates decreased within a maximum of seven trials following the first 1-s delay. For Subjects 37 and 38 response rates were low across all trials on the conjunctive schedule (the first 1-s delay occurred on the first trial for S38, and on the second trial for S37). All eight of these subjects continued to emit predominantly low response rates during exposure to the FI schedule, although there was a slight increase for S36. These subjects reported that the occurrence of a response-reinforcer delay indicated that the experimental instructions were incorrect, and that there was some form of temporal limitation on point delivery. The PRPs were variable across these eight subjects. For three subjects (32, 35, and 39) there was an increase in PRPs during exposure to the conjunctive schedule. This early shift in PRPs was reflected in the subjects' verbal reports. In the words of S32; "I soon worked out that the instructions were wrong and you had to wait about fifteen seconds before you could get a point."). The PRPs for five subjects (31, 34, 36, 37, and 38) remained short on the conjunctive but increased on the FI schedule. These subjects' verbal reports referred directly to the change in the location of the single response contingency, and to the change in the reinforcer to response dependency interval, across schedules; in the words of S37; "For a while, I pressed as soon as I got a point and waited. Later the computer stopped giving points this way; you had to wait about sixteen seconds before pressing for a point."

Subjects 33 and 40 emitted high response rates and short PRPs throughout the entire session. A 1-s delay did not occur for either of these subjects on the conjunctive schedule. Both subjects reported a ratio-based verbal formulation of the contingencies (e.g., "you just had to keep going to get points."). Response- reinforcer interaction after the first 1-s delay were as follows. The number of 1-s delays ranged between 9 and 22, the number of programmed contiguities ranged between 0 and 6, and the number of accidental contiguities ranged between 0 and 5. Note that these figures do not include results for Subjects 33 and 40, because neither of them encountered a 1-s delay.

Discussion

Previous research has shown that minimally instructed subjects tend to produce either a high-rate or a low-rate performance on the FI, whereas most ratio-instructed subjects produce high rates on this schedule (e.g., Weiner, 1969; Lippman & Meyer, 1967). In the present experiment, similar performances were demonstrated by those subjects who were first exposed to the FI. The conjunctive schedule, however, produced consistent low-rate performances for most subjects, even when they were given ratio-based verbal instructions. Furthermore, those subjects who were first exposed to the conjunctive schedule tended to continue with a low-rate performance on the FI. So although the conjunctive and FI schedules are structured from the same four components, the differing behavioral dynamics inherent in their structural diversity plays an important role in determining (a) current schedule performance, (b) performance on subsequent schedules, and (c) the effects of prior verbal instructions on schedule performance.

These findings have important implications for the analysis of human behavior on the FI schedule. Consider the general change in performance in the FI to conjunctive condition. For most subjects (both minimal and ratio instructed), performance often changed from high rate to low rate following a response-reinforcer interaction specific to the conjunctive schedule (i.e., a response-reinforcer delay). Although tentative, this finding suggests that the high- rate performance observed for some subjects, when exposed to the FI schedule first, may be partly generated by the consistent response- reinforcer contiguities that occur on this schedule. This conclusion is bolstered by evidence that increasing response- reinforcer contiguity increases response rates, and also increases the extent to which subjects report that the response actually caused the reinforcer to occur (e.g., Shanks & Dickinson, 1988; Wasserman & Neunaber, 1986).

It should be noted, however, that variation in the frequency of response-reinforcer contiguity was not the only change in response- reinforcer relations that could occur across the FI and conjunctive schedules. The structural differences in these schedules (e.g., the shift in the location of the single response contingency) meant that a low-rate, short PRP, performance produced on the conjunctive schedule would lead to a reduction in the rate of reinforcement on the FI schedule. Indeed, this effect appeared to play a role in the behavioral changes observed for some subjects. For example, after transferring from the conjunctive to the FI, Subject 14 initially increased response rate and reported that he "started to press really quickly" because pressing and waiting "didn't seem to work." In view of this, it is not possible to say whether response- reinforcer contiguity, on its own, played a role in determining the high-rate performance shown by some subjects. In order to examine more closely the effects of response-reinforcer contiguity per se, Experiment 2 employed two recycling conjunctive FT FR 1 schedules of reinforcement (i.e., a normal version and a modified version). The structural features of these schedules was identical except for the insertion of a "zero-delay window" in the terminal 2 s of the FT interval of the modified version. In contrast to the normal recycling conjunctive, the modified version provides greater opportunity for response-reinforcer contiguity (note, however, that the occurrence of response-reinforcer delays are possible on the modified schedule). Employment of these two schedules thus allowed for an examination of the specific effects of response-reinforcer contiguity on schedule performance, and the interactive effects of contiguity with prior verbal instructions.

To understand fully the rationale for Experiment 2 consider, again, the general findings from Experiment 1. When 20 subjects were exposed to the FI schedule first, 15 of these subjects produced predominantly high rates and then shifted to mainly low rates on the conjunctive. However, when subjects were exposed to the conjunctive first, 17 out of 20 subjects tended to produce low rates, and these 17 subjects also produced the low-rate performance on the FI schedule. If response-reinforcer contiguity helped determine the high-rate performance observed when most subjects were exposed to the FI schedule first, then the following results should obtain. In a modified to normal condition, the increased opportunity for contiguity on the modified recycling conjunctive should produce high rates for most subjects, providing that no response-reinforcer delays occur. Subsequent exposure to the normal recycling conjunctive, with its decreased opportunity for contiguity, should then bring about a shift to a low-rate performance. In a normal to modified condition, subjects should produce predominantly low rates on the normal schedule, and these low rates should persist on the modified schedule. Experiment 2 tested these predictions.

EXPERIMENT 2

Method

Subjects

Twenty non-psychology undergraduate students, 11 female and 9 male, enrolled at the University of Ulster at Coleraine, served as subjects. Their ages ranged from 19 to 26 years (median = 21). All subjects were recruited through faculty notice-board advertisements and personal contacts. Subjects could earn a maximum of [pounds] 1.50 during their participation in the experiment. Apparatus and Materials

Apparatus and materials were identical to those employed during the previous experiment.

Procedure

All procedural aspects were identical to Experiment 1 with the following exceptions. A normal recycling conjunctive FT FR 1 schedule and a modified recycling conjunctive FT FR 1 schedule were used during Experiment 2. On the normal recycling conjunctive schedule a response at any time during the FT interval produced the tone and the message at the end of the interval, and when the screen cleared the next interval began immediately. Failure to respond during the FT interval meant that the FT interval ended without a tone or point delivery and the next FT interval began immediately. The same contingencies operated on the modified version of the recycling conjunctive schedule, except that a response anywhere within the terminal 2 s of the FT interval produced the point immediately.

Five subjects from the minimally instructed group and five subjects from the ratio-instructed group were exposed to 25 trials on a modified recycling conjunctive FT 10-s FR 1 schedule of reinforcement followed immediately by 50 trials on a normal recycling conjunctive FT x-s FR 1 schedule. The inclusion of a 2-s zero-delay window on the modified schedule meant that subjects could obtain a point before the end of the FT 10-s interval. In order to keep the obtained FT interval constant across schedules the obtained mean FT interval on the modified conjunctive schedule was calculated across the first 25 trials, and this value was then employed as the FT value on the normal recycling conjunctive.

The remaining 10 subjects (five from the minimally instructed group and five from the ratio-instructed group) were exposed to 25 trials on the normal recycling conjunctive FT 10-s FR 1 schedule followed immediately by 50 trials on the modified recycling conjunctive FT 10-s FR 1 schedule. There was no obvious method for equating the obtained FT intervals across schedules when moving from normal to modified recycling conjunctives. Nevertheless, interpretive problems should be minimal, in that the primary focus here is on the response-generating effects of contiguity for subjects exposed to the modified schedule first. In other words, because those subjects exposed to the normal conjunctive first are expected to produce the low-rate performance across both schedules (i.e., no change in performance), any minor differences in the obtained FT interval across schedules would be unimportant (in fact, the difference was never greater than 0.38 s). Table 2 shows the obtained rates of reinforcement on each schedule, and the obtained FT intervals on the modified schedules, for each subject with either minimal instructions or ratio instructions.
Table 2
Experiment 2: Rates of Reinforcement and Obtained FT Intervals for
Each Subject on Each Experimental Condition
 Minimal Instructions
 Modified right arrow Normal
 (25 trials) (50 trials)
 Reinforce- Reinforce- Obtain FT
Subject ments/Min. ments/Min. Interval
S41 6.43 6.43 8.33
S42 5.19 5.49 9.72
S43 5.24 5.44 9.63
S44 5.25 5.45 9.60
S45 6.36 5.55 8.43
 Modified right arrow Normal
 (25 trials) (50 trials)
 Reinforce- Reinforce- Obtain FT
Subject ments/Min. ments/Min. Interval
S46 5.05 5.44 9.62
S47 5.46 5.37 9.96
S48 5.46 5.02 9.65
S49 5.25 4.69 9.91
S50 5.05 4.36 9.69
 Ratio Instructions
 Modified right arrow Normal
 (25 trials) (50 trials)
 Reinforce- Reinforce- Obtain FT
Subject ments/Min. ments/Min. Interval
S51 6.43 5.85 8.33
S52 6.41 5.85 8.37
S53 6.28 5.60 8.55
S54 6.38 5.56 8.42
S55 4.98 5.19 9.72
 Modified right arrow Normal
 (25 trials) (50 trials)
 Reinforce- Reinforce- Obtain FT
Subject ments/Min. ments/Min. Interval
S56 5.05 5.30 9.90
S57 5.46 4.94 9.81
S58 4.86 4.42 9.81
S59 4.03 5.42 9.87
S60 5.25 5.05 10.00


Results

Figures 6 and 7 show the results from the entire session for each subject. The upper panel for each subject shows the number of responses emitted on each trial, and the lower panel shows the PRP on each trial. Where a response-reinforcer delay of 1 s, or more, first occurred during a session this event is marked by a downward pointing arrow. The two figures located in the section of the upper panels which show data for the modified recycling conjunctive refer to obtained response-reinforcer interactions on this schedule after the first 1-s delay. Where such a delay does not occur the figures simply record the response-reinforcer interactions across a subject's entire exposure to the schedule. The figure appearing on the left records the total number of response-reinforcer delays greater than, or equal to, 1 s (i.e., 1-s delays). The right-hand figure refers to the total number of trials in which a response occurred within the 2-s zero-delay window (i.e., programmed response-reinforcer contiguities). The two figures located in the section of the upper panels which show data for the normal recycling conjunctive, refer to obtained response-reinforcer interactions on this schedule after the indicated first 1-s delay. The left-hand figure records the number of response-reinforcer delays greater than, or equal to, 1 s. The right-hand figure records the number of accidental response-reinforcer contiguities (i.e., one or more responses in the final second of the FT interval). No consistent relationships were obtained between these measures and subjects' schedule performance.

Minimal Instructions

Modified to normal Figure 6 (upper panels) shows data for Subjects 41 to 45. Two subjects (41 and 45) showed high response rates and short PRP durations on the modified recycling conjunctive schedule, followed by a decrease in response rates and an increase in PRPs (which were variable) during exposure to the normal recycling conjunctive schedule. These changes in schedule performance occurred gradually for S41, and quite abruptly for S45, after the first 1-s delay. These subjects reported that they assumed a ratio requirement at the beginning of the session, but realized that a time restriction was involved after the first response-reinforcer delay occurred (e.g., "To begin with I thought you had to press the space-bar continuously. But then a point came up on the screen when I was stretching; this meant that you had to press and wait for the points, which is what I did."). The three remaining subjects (42, 43, and 44) emitted low response rates and variable PRPs across the modified and normal recycling conjunctive schedules (the first 1-s delay always occurred within the first three trials). These subjects reported that the occurrence of a response-reinforcer delay emphasized the temporal nature of the contingencies (e.g., I got a point when I didn't press. This told me that you had to wait for points."). Response-reinforcer interactions across subjects on the modified recycling conjunctive, after the first 1-s delay during the session, were as follows. Delays ranged between 14 and 17, and programmed contiguities ranged between 5 and 7. Note that these figures do not include results for Subjects 41 and 45, because neither of them encountered a 1-s delay on the modified schedule (i.e., 0 delays and 25 programmed contiguities). Response- reinforcer interactions on the normal recycling conjunctive, after the first 1-s delay during the session, were as follows. Delays ranged between 31 and 48, and accidental contiguities ranged between 0 and 8.

Normal to modified. Figure 6 (lower panels) shows results for Subjects 46 to 50. All subjects emitted predominantly low response rates and variable PRPs across both normal and modified recycling conjunctive schedules (the first 1-s delay always occurred within the first two trials). All subjects reported that the occurrence of a response-reinforcer delay indicated some form of temporal limitation on point delivery (e.g., "When the computer gave me a point when I wasn't pressing, I knew it must have something to do with time.").

Response-reinforcer interactions across subjects on the normal recycling conjunctive, after the first 1-s delay during the session, were as follows. Delays ranged between 14 and 24, and accidental contiguities ranged between 0 and 7. Response-reinforcer interactions on the modified recycling conjunctive, after the first 1-s delay during the session, were as follows. Delays ranged between 23 and 48, and programmed contiguities ranged between 1 and 15. Ratio Instruction

Modified to normal. Figure 7 (upper panels) shows data for Subjects 51 to 55. Four of the five subjects (51 to 54) showed high response rates and short PRP durations on the modified recycling conjunctive schedule, followed by a decrease in response rates and an increase in PRPs (which were variable) during exposure to the normal recycling conjunctive. For one subject (53) the shift in performance occurred gradually following the first 1-s delay, but for the other three (51, 52, and 54) there was an abrupt shift in performance following this delay (i.e., within a maximum of two trials). All four of these subjects reported that they initially believed the ratio instructions, but realized that there was a temporal restriction on point delivery when the first response- reinforcer delay occurred (e.g., "At first, I thought the instructions to press a random number of times were correct. But then a point popped onto the screen when I wasn't pressing the space-bar, and it became obvious that you could wait for points.").

Subject 55 emitted a low response rate and variable PRPs across the modified and normal recycling conjunctive schedules (the first 1-s delay occurred on the first trial). This subject reported; "The instructions were wrong. The first time I got a point it came when I wasn't pressing the space-bar; you just had to press once or twice and wait for about eleven or twelve seconds." Response- reinforcer interactions on the modified recycling conjunctive were as follows. Subjects 51 to 54 did not encounter any 1-s delays during their exposure to the recycling conjunctive, and thus their response-reinforcer interactions were constant (i.e., 0 delays and 25 programmed contiguities). The interactions after the first 1-s delay for S55 may be obtained from Figure 7. Response-reinforcer interactions on the normal recycling conjunctive, after the first 1-s delay during the session, were as follows. Delays ranged between 21 and 42, and accidental contiguities ranged between 2 and 6.

Normal to modified. Figure 7 (lower panels) shows results for Subjects 56 to 60. For three subjects (56, 59, and 60) response rates were low and PRPs were variable across most trials on the normal recycling conjunctive schedule (the first 1-s delay occurred on the second trial for each subject). Subjects 57 and 58 showed high response rates and short PRPs during initial exposure to the normal recycling conjunctive, but rates decreased abruptly and PRPs increased and became variable following the first 1-s delay. All five of these subjects continued to emit predominantly low response rates and variable PRPs during exposure to the modified recycling conjunctive. All five subjects reported that the occurrence of a response-reinforcer delay indicated that the experimental instructions were incorrect, and that there was some form of temporal limitation on point delivery (e.g., "When I got a point without pressing the space-bar I knew the computer was wrong, and that you had to wait between points.").

Response-reinforcer interactions across subjects on the normal recycling conjunctive, after the first 1-s delay, were as follows. Delays ranged between 6 and 19, and accidental contiguities ranged between 1 and 5. Response-reinforcer interactions on the modified recycling conjunctive, after the first 1-s delay, were as follows. Delays ranged between 31 and 46, and programmed contiguities ranged between 0 and 8.

Discussion

The main purpose of Experiment 2 was to examine the possibility that response-reinforcer contiguity may help determine a high-rate performance when subjects are exposed to the modified recycling conjunctive schedule first. The current findings provide positive evidence for this suggestion in the following five ways: (a) 6 out of 10 subjects, exposed to the modified schedule first, produced high rates on this schedule with response-reinforcer contiguity occurring on every trial; (b) these 6 subjects subsequently emitted low-response rates following the first 1-s delay on the normal recycling conjunctives; (c) the remaining 4 subjects who produced low rates did so only after the occurrence of a 1-s delay on the modified schedule; (d) all 10 subjects exposed to the normal schedule first (1-s delays occurred for all subjects on this schedule), produced low rates across both reinforcement schedules; and (e) all subjects reported that the occurrence of response- reinforcer delays alerted them to the temporal nature of the contingencies. In summary, the increased opportunity for contiguity on the recycling conjunctive may initially prevent behavioral control by the FT and single response contingency components. Once a response-reinforcer delay occurs, however, contiguity ceases to function in this preventative manner. One criticism of the modified recycling conjunctive schedule is that it allows for slight variability, across trials, in the rate of reinforcement (i.e., a subject could obtain a point anywhere between 8-10 s on every trial). It could be argued, therefore, that those subjects who emitted high rates on the modified schedule were "aware" of the temporal nature of the contingencies, but continued to produce high rates simply in order to maximize the rate of reinforcement. There are three reasons, however, why this seems unlikely: (a) With no timepiece available it would be difficult for subjects to discriminate the temporal variability across trials; (b) subjects continued with the high-rate performance on the normal schedule until the occurrence of the first 1-s delay; and (c) none of the subjects, including those who produced low rates on the modified schedule, made verbal reports which referred to the presence of the zero-delay window.

Results suggest an interesting interaction between the prior verbal instructions and the effects of contiguity. Specifically, it was found that in the modified to normal condition four out of five ratio-instructed subjects produced the high-rate performance across the modified schedule, whereas only two out of five minimally instructed subjects produced this performance. Although tentative, this finding indicates that the ratio-based verbal instructions may have increased the response generating effects of contiguity. As yet, however the exact nature of this interaction remains unclear.

GENERAL DISCUSSION

The findings reported here indicate that human behavior on a simple time-based schedule of reinforcement is, to some extent, a function of the dynamics arising from the component features or structure of the schedule. In Experiment 1 most subjects, who were exposed schedule first, produced high rates on this schedule. With the introduction of the conjunctive schedule, however, these subjects shifted to low rates following the first occurrence of a 1-s delay between a response and point delivery. Verbal reports suggested that these subjects' verbal formulations of the contingencies were ratio based before the first 1-s delay, but became interval based following this delay. These findings indicated a response generating function for response-reinforcer contiguity on the FI schedule. It was noted, though, that because the structure of the FI and conjunctive schedules differed in a number of ways it was not possible to determine whether contiguity, on its own, played a crucial role in producing the high-rate performance. To examine the response-generating function of contiguity more closely, two recycling conjunctive schedules were employed in Experiment 2. The modified version of this schedule provided greater opportunity for contiguity than the normal version, but the other elements of the schedules remained relatively constant. Again, some subjects, and particularly those who received ratio-based instructions, produced high rates before the first 1-s delay, and predominantly low rates after this delay. Verbal reports suggested that verbal formulations of the contingencies changed from ratio to interval in nature following this delay. These findings provide further support for the idea that response-reinforcer contiguity may help to produce the high rates and ratio-based verbal formulations observed with some subjects on the FI schedule. At present, it remains unclear why contiguity has this response-generating effect with only some human subjects, and why this effect consistently fails to occur when subjects have been previously exposed to response-reinforcer delays. To answer these questions it may be necessary to consider aspects of the context other than those provided by the structural components of the reinforcement schedule (see Barnes & Keenan, 1989, 1993a, 1993b).

One criticism of the present study could be that subjects were exposed to two schedules in only one brief session, and thus we cannot consider behavior on either schedule as approaching steady- state performance. Although future research should attempt to replicate the current findings using multiple sessions, it is important that a relationship between response-reinforcer contiguity and human schedule performance was identified using the current single-session methodology. The very fact that this behavioral effect emerged in only one brief session emphasizes the critical role played by contiguity in controlling human behavior on reinforcement schedules.

Another criticism of this research could be its reliance on retrospective verbal reports as a method for analyzing the effects of the schedule components on subjects' verbal behavior, and the subsequent effects of this verbal behavior on nonverbal schedule performance. Basically, the retrospective verbal report presents an interpretative problem, mainly because it is not possible to determine whether the reports are veridical with the covert verbalizations that actually occurred during the session (see Shimoff, 1986). In order to circumvent this problem perhaps future researchers should consider employing other methods for collecting subjects' covert verbalizations during exposure to schedules (e.g., the "think aloud" technique; see Hayes, 1986). Indeed, the relatively powerful schedule control seen here indicates that the present procedures should provide an ideal context for exploring alternative methods for recording the interaction between verbal and nonverbal behavior on schedules of reinforcement. One of the more common findings reported in the human operant literature is an apparent "insensitivity" to the contingencies specified by reinforcement schedules (see Lowe, 1979; Wearden, 1988; but see Chase & Danforth, 1991, for a review of recent research that has demonstrated schedule sensitivity in human subjects). Indeed, the human high-rate FI performance has often been highlighted as an example of this insensitivity. It is interesting then that the current findings, and especially those obtained on the modified recycling conjunctive, suggest that this high-rate performance is not insensitive to the FI schedule per se. Rather, this performance demonstrates sensitivity to a particular feature of the FI schedule, namely response-reinforcer contiguity. Furthermore, the performance and verbal reports produced by some subjects (e.g., S14) indicate that other components of the FI may also play a role in generating high rates. Overall these findings suggest that the idea of human "insensitivity" to reinforcement schedules may have been premature, insofar as our understanding of contingencies is not yet sophisticated enough to fully appreciate the behavioral dynamics that emerge on schedules of reinforcement (cf. Lee, 1988, Chapter 7).

Another common finding reported in the human operant literature is the extent to which prior verbal instructions may override the reinforcement contingencies of the schedule (e.g., Lippman & Meyer, 1967). Indeed, some researchers have reported that subjects will sometimes follow instructions even though doing so produces a reduction in the rate of reinforcement (e.g., Hayes, Brownstein, Zettle, Rosenfarb, & Korn, 1986). In the present study the ratio- based instructions appeared to override the FI and modified recycling conjunctive schedules for most subjects. It is important to understand, though, that the view of simple time-based schedules proposed here suggests that instructions did not simply override the schedule. Rather, the ratio-based instructions appeared to facilitate the response-generating function of contiguity. In other words, it appears that in some cases the verbal instructions might interact with the schedule and affect the extent to which specific component features control behavior.

Other researchers have found that the effects of inaccurate instructions may break down in some cases when subjects contact the reinforcement contingencies (e.g., Galizio, 1979). A similar effect was observed here for those ratio-instructed subjects who produced low rates. Nevertheless, there was one important difference. In those studies where the effects of instructions broke down, subjects first incurred quite large reductions in the rate of reinforcement for following the instructions (see Catania, Shimoff, & Matthews; 1989; Chase & Danforth, 1991). In contrast, results from the present study showed that a reduction in the opportunity for response-reinforcer contiguity, with no reduction in the rate of reinforcement, appeared to attenuate the effects of ratio-based instructions (cf. Centti, 1991). Clearly, the co-determining effects of ratio-based instructions and response-reinforcer contiguity require systematic analysis. Without a clear understanding of these effects it is likely that researchers will fail to identify the exact manner in which verbal instructions control, or fail to control, human schedule performance.

At the beginning of the introduction it was argued that even a simple time-based schedule, such as the FI, represents an intricate and dynamic environmental context. This molecular analysis of reinforcement schedules, however, complicates our understanding as to why schedules have their particular effects. Indeed, despite the wealth of research that followed the early work of Ferster and Skinner (1957), we are still some distance from understanding the principles operating on any of the four basic schedules of reinforcement. Zeiler (1984) went so far as to say:

The attempt to simplify by finding what component independent variables are brought into play by a schedule specification cannot succeed because of the complexity of the interactions, and also because many of the controlling variables arise indirectly through the interplay of ongoing behavior and the contingencies.

He went on to argue that it might even be more appropriate to view schedules as "irreducible independent variables" and that the role of theory would be to integrate their effects at a higher level. However, the procedures and findings reported here may help to redress Zeiler's concerns. Specifically, it has been shown that an improved understanding of the four structural components of the FI schedule can help illuminate some of the molecular behavioral relations that occur when humans are exposed to this schedule.

References

BARNES, D. (1989). Behavior-behavior analysis, human schedule performance, and radical behaviorism. The Psychological Record, 39, 339-350. BARNES, D., & KEENAN, M. (1989). Instructed human fixed- interval performance: The effects of the experimental setting. The Psychological Record, 39, 351-364.

BARNES, D., & KEENAN, M. (1993a). A transfer of functions through derived arbitrary and non-arbitrary stimulus relations. Journal of the Experimental Analysis of Behavior, 59, 61-81.

BARNES, D., & KEENAN, M. (1993b). Concurrent activities and instructed human fixed-interval performance. Journal of the Experimental Analysis of Behavior, 59, 501-520.

CATANIA, A. C., SHIMOFF, E., & MATTHEWS, B. A. (1989). An experimental analysis of rule-governed behavior. In S. C. Hayes (Ed.), Rule-governed behavior: Cognition, contingencies, and instructional control (pp. 119-150). New York: Plenum.

CERUTTI, D. T. (1991). Discriminative versus reinforcing properties of schedules as determinants of schedule-insensitivity in humans. The Psychological Record, 41, 51-67.

CHASE, P. N., & DANFORTH, J. S. (1991). The role of rules in concept learning. In L. J. Hayes & P. N. Chase (Eds.), Dialogues on verbal behavior (pp. 205-225). Reno, NV: Context Press.

DAVEY, G. C. L. (1987). Animal learning and conditioning. London: Macmillan Education.

DAVEY, G. C. L. (1988). Trends in human operant theory. In G. C. L. Davey & C. Cullen (Eds.), Human operant conditioning and behavior modification (pp 1-14). Chichester, England: John Wiley & Sons.

ERICSSON, K. A., & SIMON, H. A. (1984). Protocol analysis: Verbal reports as data. Cambridge, MA: MIT Press.

FERSTER, G. B., & SKINNER, B. F. (1957). Schedules of reinforcement. New York: Appleton-Century-Crofts.

GALIZIO, M. (1979). Contingency-shaped and rule governed behavior: Instructional control of human loss avoidance. Journal of the Experimental Analysis of Behavior, 31, 53-70.

HAYES, S. C. (1986). The case of the silent dog - Verbal reports and the analysis of rules: A review of Ericsson and Simon's Protocol analysis: Verbal reports as data. Journal of the Experimental Analysis of Behavior, 45, 351-363.

HAYES, S. C., BROWNSTEIN, A. J., ZETTLE, R. D., ROSENFARB, I., & KORN, Z. (1986). Rule-governed behavior and sensitivity to changing consequences of responding. Journal of the Experimental Analysis of Behavior, 45, 237-256. KEENAN, M., & LESLIE, J. C. (1984). Separating response dependency and response-reinforcer contiguity within a recycling conjunctive schedule. Journal of the Experimental Analysis of Behavior, 41, 203-210. KEENAN, M., & LESLIE, J. C. (1986). Varying response-reinforcer contiguity in a recycling conjunctive schedule. Journal of the Experimental Analysis of Behavior, 45, 317-332.

KEENAN, M., & WATT, A. A. (1990). Concurrent behavior and response- reinforcer contiguity. The Psychological Record, 40, 127-138.

LEANDER, J. D., LIPPMAN, L. G., & MEYER, M. E. (1968). Fixed- interval performance as related to subjects' verbalizations of the reinforcement contingency. The Psychological Record, 18, 469-474.

LEE, V. L. (1988). Beyond behaviorism. Hillsdale, NJ: Lawrence Erlbaum Associates.

LIPPMAN, G., & MEYER, M. E. (1967). Fixed-interval performance as related to instructions and to subjects' verbalizations of the contingency. Psychonomic Science, 8, 135-136.

LOWE, C. F. (1979). Determinants of human operant behavior. In M. D. Zeiler & P. Harzem (Eds.), Advances in analysis of behavior: Vol 1. Reinforcement and the organization of behavior (pp. 159-192). Chichester, England: John Wiley & Sons.

MACKINTOSH, N. J. (1983). Conditioning and associated learning. Oxford. England: Clarendon Press.

SCHWARTZ, B. (1984). Psychology of learning and behavior (2nd ed.). New York: W. W. Norton & Company.

SHANKS, D. R., & DICKINSON, A. (1988). Associative accounts of causality judgement. In G. M. Bower (Ed.), The psychology of learning and motivation (pp. 229-261). San Diego: Academic Press.

SHIMOFF, E. (1986). Post-session verbal reports and the experimental analysis of behavior. The Analysis of Verbal Behavior, 4, 19-22.

WASSERMAN, E. A., & NEUNABER, D. J. (1986). College students' responding to and rating of contingency relations: The role of temporal contiguity. Journal of the Experimental Analysis of Behavior, 46, 15-35.

WEARDEN, J. H. (1988). Some neglected problems in the analysis of human operant behavior. In G. C. L. Davey & C. Cullen (Eds.), Human operant conditioning and behavior modification (pp. 197-224). Chichester, England: John Wiley and Sons.

WEINER, H. (1969). Human behavioral persistence. The Psychological Record, 20, 445-456.

ZEILER, M. D. (1984). The sleeping giant: Reinforcement schedules. Journal of the Experimental Analysis of Behavior, 42, 485-493.
COPYRIGHT 1994 The Psychological Record
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1994 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Barnes, Dermot; Keenan, Michael
Publication:The Psychological Record
Date:Jan 1, 1994
Words:8512
Previous Article:Schedule-induced behavior occurs in humans: a reply to Overskeid.
Next Article:Stimulus equivalence and relational frame theory.
Topics:


Related Articles
Differential reinforcement of other behavior and response suppression: the effects of the response-reinforcement interval.
The effects of terminal-link stimulus arrangements on preference in concurrent chains.
A matching law analysis of the effect of amphetamine on responding reinforced by the opportunity to run.
Variable-interval schedule control following response acquisition with delayed reinforcement.
Periodic response-reinforcer contiguity: temporal control but not as we know it!
AN OPERANT BLOCKING INTERPRETATION OF INSTRUCTED INSENSITIVITY TO SCHEDULE CONTINGENCIES.
DIFFERENTIAL OUTCOMES EFFECT ON INSTRUMENTAL SERIAL FEATURE-AMBIGUOUS DISCRIMINATION IN RATS.
CONTINGENT INCENTIVE VALUE IN HUMAN OPERANT PERFORMANCE.
DELIVERING DIFFERENT REINFORCERS IN EACH HALF OF THE SESSION: EFFECT OF REINFORCEMENT RATE.
Within-session decreases in operant responding as a function of pre-session feedings.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters