Printer Friendly

Periodic response-reinforcer contiguity: temporal control but not as we know it!

When significant events in the environment occur on a regular basis, people adapt by producing regularities in their behavior. Some of these events may be separated in time by a year or more (e.g., anniversaries) and others occur on a smaller time scale (e.g., weekends off work, lunch breaks during the day, etc.). In the laboratory, simulations of periodicity in behavior have been studied with both humans and nonhumans. The general approach has been to arrange a contingency between a selected behavior and an environmental event to ensure that the environmental event occurs at regular intervals. The technical term for this arrangement is a 'schedule of reinforcement.' A wide variety of schedules of reinforcement have been studied in the laboratory and each is associated with a distinctive pattern of behavior (Catania, 1998).

One of the most studied laboratory procedures for investigating periodicity in behavior is the fixed-interval (FI) schedule of reinforcement. Reinforcer delivery on this schedule is dependent upon the occurrence of a single response after a fixed period of time has elapsed since the previous reinforcer presentation. Baseline performance on this schedule is typically described as involving a postreinforcement pause (PRP) followed by either an accelerating or a constant response rate up to the next reinforcer delivery (Baron & Leinenweber, 1994; Cumming & Schoenfeld, 1958; Dews, 1970; Ferster & Skinner, 1957; see Hyten & Madden, 1993, for a discussion of problems arising from imprecision in the description of human FI performance). In the analysis of this performance a variety of techniques of have been employed. These can be grouped together according to whether they involved simple parametric investigations of the interreinforcer interval, manipulation of the single response contingency, disruption of responding during the interreinforcer interval by the presentation of other stimuli, or the replacement of occasional reinforcer presentations by other stimuli (for extended discussions of these and other related procedures see Davey, 1987; Keenan, 1986; Lowe & Wearden, 1981; Richelle & Lejeune, 1980; Staddon, 1983; Zeiler, 1977).

The analysis of patterns of behavior on schedules generally has proven difficult because even on the simplest of schedules it is recognized that behavior is multiply determined (Morse & Kelleher, 1997; Zeiler, 1997). Thus, although the formal description of a schedule may reference simply the programmed relation between the behavior and the environmental event, closer inspection shows that other variables operate collectively to produce baseline responding. For example, Keenan and Leslie (1986) (see also Keenan & Toal, 1991) offered a structural analysis of the independent variables that collectively define a FI schedule. They pointed out that there were four variables acting in concert: (a) the time between reinforcer presentations; (b) the single response contingency; (c) response-reinforcer contiguity; and (d) the time from one reinforcer presentation to the location in time of the next response dependency. The inspiration for this work came from the effects observed on another schedule that is similar in makeup to a FI schedule, a recycling conjunctive fixed-time (FT) fixed-ratio (FR) 1 schedule.

A FI schedule can be seen as a tandem FT FR 1 schedule of reinforcement. Thus, once the FT component expires, and only then, a FR 1 contingency comes into operation. A major effect of this particular construction is that it ensures periodic occurrences of response-reinforcer contiguity. A recycling conjunctive FT FR 1 schedule is similar to a FI schedule in that it too has a single response contingency and it also presents reinforcement at regular intervals. However, unlike the FI schedule a single response executed any time during a FT component results in reinforcer delivery at the end of that FT component. Also, if a response fails to occur during a FT component, that component ends without any stimulus event and the next FT component begins immediately. Although periodic reinforcement is assured in this arrangement, periodic response-reinforcer contiguity is not guaranteed. Baseline performance on this schedule is characterized by a variety of response-reinforcer delays, low overall rates of responding, and a response distribution markedly different from that found on a FI schedule (cf. Keenan & Watt, 1990).

Notwithstanding this marked difference in performance between these two schedules, Keenan and Leslie (1986) demonstrated how a modified version of the recycling conjunctive FT FR 1 could be an extremely useful tool for exploring the effects of periodic response-reinforcer contiguity. The response strengthening effects of response-reinforcer contiguity have been well documented (e.g., Thomas, 1981), and from Keenan and Leslie's study it would seem that these effects within the context of regular occurrences of contiguity are a major contributing factor to overall performance on a FI schedule. Using the framework of the recycling schedule, Keenan and Leslie designed a procedure that produced an increased incidence of contiguity but did not alter the single response contingency, and at the same time had minimal effects on the duration of the FT component. They did this by changing the consequences of responding in either the final 2 s or 4 s of the FT component. A response during this terminal window of the FT component would not normally be contiguous with reinforcer presentation. Thus, for example, if a response occurred at 28 s on a 30 s FT component, then, assuming no other response occurred, there would be a delay of 2 s between this response and reinforcer delivery. In a modified version of the procedure, however, a response in a terminal window of a FT component produced a reinforcer immediately and terminated that component.

To recap, throughout all FT components on a normal recycling conjunctive FT FR 1 schedule only one response is required within each component for a reinforcer to be delivered at the end of that component. In the modified version, however, if a response occurred in a terminal window of a FT component, response-reinforcer contiguity occurred. Keenan and Leslie found that performance on this modified schedule was markedly different from that observed on the normal schedule; the PRP was relatively unaffected, the usual pause-respond-pause patterning during the FT component was replaced by Fl-like patterning and there was a three-fold increase in overall response rate for some animals. The occurrence of Fl-like patterning and the occurrence of a relatively large number of responses during the FT component, even though only one response was necessary for reinforcer presentation, are precisely the behavioral characteristics that have intrigued researchers interested in the FI schedule.

Keenan and Watt (1990) highlighted the importance of this finding. This procedure is exciting because it not only gives the experimenter a certain degree of control over contiguity during the interreinforcer interval, but it does so without contaminating the spontaneous regulation on the schedule or without introducing major changes in the response contingency. Furthermore, differing response rates can be produced without introducing significant changes in overall rate of reinforcement. (p. 129)

In other words, two schedules currently exist that differ radically in their formal specifications (i.e., the FI and the modified recycling conjunctive FT FR 1) but which produce similar patterns of behavior. The implications are that key variables shared by each schedule are responsible for the similarity in the patterning of behavior that is observed. A primary candidate is periodic response-reinforcer contiguity.

In this study the versatility of the modified recycling conjunctive FT FR 1 was explored further. The primary objective of the first experiment was to examine the possibility that the procedure could be fine tuned so as to provide control over the percentages of obtained response-reinforcer contiguities. This was done by varying, in a semi-random manner, the probability that the first response in the terminal segment of the FT component would produce reinforcement immediately. Another objective of the experiment arose incidentally from data obtained early in training. When response-reinforcer contiguities were first introduced, an unexpected behavior developed for some of the rats. They stood on the lever immediately after reinforcer presentation and began gnawing at various areas on the plastic walls and on the light fittings. This behavior was so vigorous that special metal panels had to be inserted into the chambers to cover up holes that were developing. This casing was inserted about 8 days after the abnormal responding had been established. It was decided to continue with the rats that engaged in this behavior because it was assumed that 'normal' responding would return after extended exposure to periodic reinforcement. When this proved not to be the case, it was decided to continue the experiment with these rats to see what behavior would develop eventually. The findings obtained add to the existing body literature concerned with the role played by behavioral history in determining schedule performance (cf. Bickel, Higgins, Kirby, & Johnston, 1988; Johnston, Bickel, Higgins, & Morris, 1991; Leinenweber, Nietzel, & Baron, 1996; Tatham, Wanschisen, & Hineline, 1993; Wanchisen, Tatham, & Mooney, 1989).

Experiment 1



Nine experimentally naive 6-months old, male albino Wistar rats were used initially. They were caged in groups of three with water freely available, and they were maintained at approximately 85% of their free-feeding weights by feeding after experimental sessions.


Eight Campden Instruments rat test chambers with plastic walls and light fittings were used. Later on in the study, metal casing had to be fitted to the inside of the chambers because some rats chewed holes in the walls and in the light fittings (see below). A single retractable lever was positioned in the center of the magazine wall. The magazine itself comprised of a recessed tray situated at floor level in the bottom left-hand comer of the wall. The reinforcer used was a 45-mg precision pellet (BioServ. Inc.) which was accompanied by a 3-s illumination of the tray light. Contingencies were controlled by an Apple II+ computer programmed in BASIC.


The sequence of conditions and the numbers of sessions in each condition for each rat are given in Table 1. After initial training on a continuous reinforcement schedule, all rats were transferred to a recycling conjunctive fixed-time (FT) 30 s fixed-ratio (FR) 1 schedule of reinforcement. On this schedule one response was required anywhere inside a 30-s cycle (i.e., the FT component) to produce a reinforcer at the end of that cycle. Failure to respond in any cycle meant that that cycle ended without a reinforcer, or the delivery of any other stimulus, and the next cycle began immediately. Unlike more conventional schedules, response-reinforcer contiguity (hereafter called "zero-delay" (ZD) reinforcement; cf. Keenan & Leslie, 1986) is not explicitly programmed on this schedule. Because the conditions which followed this initial condition were designed to manipulate the probability of obtainable programmed ZDs, this condition is referred to as 'Condition p = 0.'

Each session terminated after either 100 reinforcements or after 90 min, whichever occurred first, and all rats received one session daily from Monday through Saturday. Given that at least one response was required inside a FT component, the value obtained when the total number of reinforced cycles is expressed as a percentage of the total number of cycles that have elapsed in a session was used as a measure of performance efficiency. Behavior was considered stable in this, and in all other conditions, when average performance efficiency over six consecutive sessions was at least 70% and when there were no systematic directional changes in overall response rate. This stability criterion was used because of the intrinsic problems in controlling for the rate enhancing effects of accidental zero-delay reinforcements (see below).
Table 1

Sequence of Conditions, Numbers of Sessions, and the Percentage
Efficiency in Each Condition in Experiment 1

Rat Condition Number of sessions % Efficiency

K1 p = 0 32 82.0 (6.4)
 p = 0.2 22 80.1 (4.9)
 p = 0.4 3
 p = 0.6 18 95.2 (3.3)
 p = 0.8 20 98.5 (0.5)
 p = 1.0 15 97.5 (1.8)
 Extinction 3
 FI 26 s 19

K2 p = 0 38 88.0 (9.4)
 p = 0.2 24 80.1 (11.8)
 p = 0.4 21 85.3 (12.5)
 p = 0.6 16 93.8 (2.6)

K3 p = 0 31 96.6 (1.9)
 p = 0.2 26 99.6 (0.5)
 p = 0.4 4
 p = 0.6 18 97.0 (2.5)
 p = 0.8 17 96.9 (0.9)

K4 p = 0 33 77.3 (6.2)
 p = 0.2 16 85.6 (3.7)
 p = 0.4 3
 p = 0.6 17 97.5 (1.3)
 p = 0.8 17 98.4 (0.8)
 p = 1.0 15 98.3 (0.8)
 Extinction 3
 FI 26 s 24

K6 p = 0 29 90.4 (2.7)
 p = 0.2 16 96.3 (2.3)
 p = 1.0 20 97.1 (1.3)
 p = 0.4 18 98.8 (0.4)
 p = 0.8 19 98.3 (0.4)
 Extinction 3
 FI 26 s 21

K8 p = 0 29 87.4 (6.5)
 p = 0.4 24 99.3 (0.8)
 p = 1.0 18 98,8 (0.4)
 p = 0.2 23 98.6 (0.8)
 p = 0.8 16 98.8 (0.4)
 p = 0.6 18 98.4 (1.3)
 Extinction 3
 FI 26 s 20

K10 p = 0 31 77.3 (2.4)
 p = 0.6 24 96.8 (2.6)
 p = 1.0 18 96.4 (1.9)
 p = 0.2 16 88.8 (5.1)
 p = 0.4 16 87.5 (3.3)
 p = 0.8 15 90.0 (5.4)
 Extinction 3
 FI 26 s 19

K12 p = 0 38 81.4 (8.6)
 p = 0.8 30 74.8 (6.2)
 p = 0.4 14 72.7 (2.8)
 p = 1.0 17 90.8 (3.3)
 p = 0.6 19 92.0 (3.8)
 p = 0.2 15 80.5 (3.0)
 Extinction 3
 FI 26 s 19

K15 p = 0 31 82.2 (7.0)
 p = 1.0 35 98.7 (1.4)
 p = 0.8 21 97.9 (0.8)
 p = 0.2 17 96.8 (1.9)
 p = 0.6 16 95.5 (1.8)
 Extinction 3
 FI 26 s 24

Note. Figures in parentheses are the standard deviations of the
session means for the percentage efficiency.

In the next series of conditions the basic recycling conjunctive schedule was modified so as to introduce a degree of control over the frequency of obtained ZDs. Again only one response was required anywhere inside a FT component for reinforcement to be delivered at the end of that component. This time, however, if a response occurred in the terminal 4 s of a FT component, the reinforcer was delivered immediately, depending upon the programmed probability of its occurrence. For example, if the probability of obtaining a ZD for the first response in the terminal 4 s was set at p = 1.0, then this response produced a reinforcer immediately and terminated the cycle. Thus, only one reinforcer was presented at the end of a cycle. If, however, the probability of obtaining a ZD was set at p = 0.4, then the first response in the terminal 4 s was only effective in producing a ZD on a maximum of 40 out of 100 occasions. Where a ZD was not produced by the first response in the terminal 4 s of a cycle, further responses were also ineffective in producing a ZD. Note, however, that a reinforcer was always delivered at the end of a FT component if the basic FR 1 requirement had been satisfied anywhere during it. This aspect of the schedule arrangement creates the possibility that accidental ZDs may occur. In summary, each cycle ended with a reinforcer if at least one response had occurred anywhere during it. Additionally, the probability of the first response in the terminal 4 s of a cycle producing a ZD was varied across conditions.

For Rats K1, K2, K3, and K4 the probability of programmed ZDs was increased across conditions. However, because Rats K2 and K3 never actually obtained more than 20% of scheduled programmed ZDs in any condition, they were eventually eliminated from the study. For Rats K6, K8, K10, K12, and K15, the probability of programmed ZDs was varied in a semirandom manner across conditions.

After the final session on the recycling conjunctive schedule, all rats were exposed to three sessions of extinction. During extinction the same recycling procedure was in effect, only this time food pellets were removed from the dispenser. Finally, all rats were exposed to a fixed-interval (FI) 26 s schedule of reinforcement. This value of the FI schedule was chosen because it matched the time on the recycling conjunctive schedule after which ZDs were available. Each session terminated after 100 reinforcers had been delivered.


The results presented here for each rat are means calculated over the last five sessions in each condition. All rats except K1, K10, and K12 engaged in the unexpected gnawing behavior.

Response Distributions

Figures 1 and 2 show the response distribution in successive seconds after food for each rat (except K2 & K3) in each condition. In Condition p = 0, responding was characterized by few responses throughout the interval for all rats. For K4, K6, K8, and K12, a pause-respond-pause pattern predominated, while for the other rats responding was relatively constant after about 10-15 s into the interval. The final condition for each rat was the FI schedule. Across rats a number of patterns emerged. K1, K10, and K12 (the rats who did not engage in gnawing) produced typical FI patterning in that after extended pausing responding accelerated up to the end of the interreinforcer interval. For the other rats, some unusual patterns were maintained. Responding for K4 and K8 began very early after reinforcer delivery and accelerated rapidly throughout the remainder of the interval. K6 and K15 produced bimodal response patterns with a peak in responding at about 5 seconds after reinforcer delivery and another peak towards the end of the interval.

Between the first and last conditions the patterning across rats mirrored the differences in responding shown on the FI schedule. A consistent finding across rats, with the exception of K12, was that once responding had increased in the terminal segment of the interreinforcer interval in any one condition it tended to remain elevated to the same extent. For K12, the increase in the rate of responding during the terminal segment of the interreinforcer interval in Condition p = 0.8 was diminished substantially in the subsequent condition (Condition p = 0.4).

Response Rate and ZDs

Figure 3 shows response rate and the total numbers of obtained ZDs, programmed and accidental, for each rat in each condition. (An accidental ZD was defined as a delay between a response and a reinforcer which was less than 1 s.) In the first condition (Condition p = 0) the maximum overall response rate of just over 10 responses per minute was produced by K15. The maximum number of accidental ZDs obtained by any rat in this condition was about 20 out of a possible 100, and this was by K15.

Across animals there were variations in the extent to which response rate and numbers of programmed ZDs were related. For K1 and K4 there was some correspondence between the increasing probability of programmed ZDs across conditions and the total numbers of ZDs obtained. Up to Condition p = 1.0, increases in the percentage of obtained ZDs were mirrored by increases in overall response rate such that the highest rates on the recycling conjunctive schedule were recorded in this condition. Thereafter, however, the effect of the FI schedule on overall response rate differed for these two rats. For K1, response rate decreased quite substantially, and for K4 it continued to rise slightly.

The remaining rats (K6, K8, K10, K12, & K15) had the probability of programmed ZDs varied in a semirandom manner across conditions. Two main effects emerged. Firstly, once the total numbers of ZDs for K6 and K8 had reached a maximum, subsequent decreases in the probability of programmed ZDs had little effect on their frequencies; there was, though, a high percentage of accidental ZDs. Overall response rates for these two rats tended to increase slightly across conditions.

The second type of effect which emerged across conditions was seen for K10, K12, and K15. In the main, relatively large increases or decreases in the numbers of ZDs produced corresponding changes in the overall response rates. It was noted again that decreases in the probability of programmed ZDs did not prevent substantial numbers of accidental ZDs from occurring.

Although K2 and K3 were eliminated eventually from the study, some of their results are included [ILLUSTRATION FOR FIGURE 3 OMITTED]. The reason for this is to demonstrate some interesting difficulties encountered in gaining control over the production of ZDs by the procedures used here. Both rats developed a very pronounced pause-respond-pause pattern with almost 90% of responses occurring in the middle 10 s of the interval after food. Consequently, as the probability of programmed ZDs was increased across conditions for these rats, there was not a corresponding increase in the total numbers of obtained ZDs. In fact, for K2 there were more ZDs in Condition p = 0.2 than in Condition p = 0.6.


The median postreinforcement pause (PRP) durations are shown for all rats (except K2 & K3) in Figure 4. Pauses ranged from about 7-17 s in this first condition. Thereafter the effects on pausing varied across rats. For Rats K1, K10, and K12 the largest PRPs occurred in the final condition (the FI schedule). For the other animals there was a general tendency for PRPs to decrease across conditions so that minimal pausing occurred on the FI schedule.


Previous studies with Sprague Dawley rats have shown that the recycling conjunctive FT x s FR 1 schedule produces low overall response rates. Also, responding during the interreinforcer interval is characterized by either a pause-respond-pause pattern, or else a pause followed by a constant low response rate throughout the rest of the interval. Similar findings were reported here across all rats in Condition p = 0. During this condition, the incidence of obtained, accidental ZDs was very low. The introduction of programmed ZDs produced a number of effects. When K2 and K3 are excluded (because the numbers of obtained ZDs did not correspond with the opportunity for obtaining them), the main effect for other rats was an increase in overall rate of responding (cf. Keenan & Leslie, 1986). For some rats (K4, K6, & K8), once response rate had increased, it was maintained at this level even though the probability of programmed ZDs was varied across conditions. This can be accounted for by noting that the numbers of accidental ZDs kept the overall total of ZDs relatively high. Other rats (K10, K12, & K15) showed evidence of differential control over response rates by programmed ZDs. That is, when variations in the probability of programmed ZDs resulted in variations in the overall numbers of obtained ZDs overall response rates varied accordingly.

One of the most intriguing aspects of the findings concerns the effects on response patterning. Usually schedules of reinforcement are noted for the consistency in response patterning that they produce across nonhumans. This was not the case here. Instead a variety of performances was obtained across rats on the final FI schedule. That is, they exhibited predictable, stable patterns of responding as a function of their history of responding on the modified recycling conjunctive schedule (cf. Johnston et al., 1991). Of these, only K1, K10, and K12 produced 'typical' FI patterning. The performances of these rats prior to the FI condition were reminiscent of findings reported previously by Keenan and Leslie (1986). That is, they resembled FI patterning. Interestingly, these rats did not engage in the unusual gnawing behavior noted earlier.

For the other rats who did engage in this behavior there were some highly unusual patterns produced on the final FI schedule. K4 and K8 increased responding across the interreinforcer interval, and they had unusually small PRPs. Similar PRP durations occurred for K6 and K15, but patterning was bimodal. These unusual response patterns are all the more remarkable when you bear in mind that each rat had over 100 sessions of periodic food presentation prior to being exposed to the FI schedule. In other words, even though there was evidence for periodicity in behavior, the patterning normally associated with temporal control on FI schedules did not emerge (Richelle & Lejeune, 1980). Instead, unintentional history effects occurred. These findings were fortuitous and they are the first recorded instances of such behavior that is uncontaminated by different response requirements and different interreinforcer intervals prior to exposure on a FI schedule (Bickel et al., 1988; Johnston et al., 1991; Leinenweber et al., 1996; Tatham et al., 1993; Wanchisen et al., 1989). An account of the unusual patterns obtained for these animals is deferred until the General Discussion.

In general, these findings have been relatively successful in regards to the first objective of the study. They demonstrate that it is possible to attain a fair degree of control over the numbers of obtained response-reinforcer contiguities on a temporally based schedule that delivers periodic food reinforcement. Precise control, however, was not possible because the rate-enhancing effects of ZDs, whether accidental or programmed, ensured that the overall numbers of obtained ZDs was often higher than programmed. Following on from this, though, the results show how both response rate and patterning are affected by the incidence of ZDs. This is an important finding in so far as all previous research with the FI schedule has confounded the contribution played by the temporal distribution of response-reinforcer contiguities with that played by the temporal distribution of response dependencies (Keenan, 1982).

Results from two animals (K2 & K3) show also some of the difficulties in working with this schedule. If responding does not occur in the terminal window of the FT component ZDs do not occur and responding does not increase in rate. This was the case for these animals despite the fact that they were exposed to about 100 sessions each of periodic reinforcer presentations. The response patterning produced by these animals has an important bearing on theoretical accounts of FI performance and this is discussed later in the General Discussion.

In the next experiment, the objective was simply to see how similar were the performances of both the modified recycling conjunctive with maximum opportunity to obtain ZDs (i.e., Condition p = 1.0) and the FI schedule. The experiment had commenced at the same time as Experiment 1 and thus none of the findings reported above had yet been obtained. It had been assumed on the basis of previous work with the recycling conjunctive that it would be a relatively straight forward task. However, similar disruptions in patterning occurred for the same reasons (i.e., excessive gnawing of the plastic casing of the chamber). As before, rats that engaged in this behavior were retained for the study because it was assumed (incorrectly as it turned out) that the disruption would abate after continued exposure to periodic reinforcement.

Experiment 2



Six experimentally naive 6-months old, male albino Wistar rats were used initially. They were caged in pairs with water freely available, and they were maintained at approximately 85% of their free-feeding weights by feeding after experimental sessions.


The apparatus described previously was also used here.


The sequence of conditions and the numbers of sessions in each condition for each rat are given in Table 2. After initial training on a continuous reinforcement procedure, two rats (K26 & K28) were transferred to a recycling conjunctive FT 30 s FR 1 schedule of reinforcement (Condition p = 0; see Experiment 1). The other rats (K32, K33, K34, & K35) were transferred to the modified version of this schedule with programmed ZDs (Condition p = 1.0; see Experiment 1). Throughout all conditions a session terminated after 100 reinforcements, or after 90 min, whichever occurred first, and all rats received one session daily from Monday through to Saturday. The stability criteria used in Experiment 1 were also used here.

Using an ABA reversal design, Rats K26 and K28 were transferred to Condition p = 1.0 before returning to Condition p = 0. Following the last session in Condition p = 0, the food pellets were removed from the dispenser and the rats were exposed to three sessions of extinction with this procedure. Thereafter they were transferred to a FI 26 s schedule of reinforcement. Finally, three sessions in extinction with the FI procedure (i.e., using an empty pellet dispenser) preceded a return to Condition p = 1.0

For the other rats an ABA reversal design was used whereby they were transferred from Condition p = 1.0 to a FI 26 s schedule and then back to Condition p = 1.0. Three sessions of extinction with the procedure last used separated the transfer across conditions for each rat.
Table 2

Sequence of Conditions, Numbers of Sessions, and Percentage
Efficiency in Each Condition in Experiment 2

Rat Condition Number of sessions % Efficiency

K26 p = 0 25 93.4 (4.5)
 p = 1.0 30 100.0 (0.0)
 p = 0 20 98.1 (1.0)
 Extinction 3
 FI 26 s 18
 Extinction 3
 p = 1.0 20 97.9 (0.8)

K28 p = 0 26 98.1 (2.4)
 p = 1.0 29 100.0 (0.0)
 p = 0 34 98.4 (0.8)
 Extinction 3
 FI 26 s 22
 Extinction 3
 p = 1.0 15 98.8 (0.4)

K32 p = 1.0 28 99.0 (0.0)
 Extinction 3
 FI 26 s 23
 Extinction 3
 p = 1.0 19 99.0 (0.0)

K33 p = 1.0 30 98.3 (1.7)
 Extinction 3
 FI 26 s 19
 Extinction 3
 p = 1.0 20 96.8 (2.1)

K34 p = 1.0 27 98.8 (0.4)
 Extinction 3
 FI 26 s 25
 Extinction 3
 p = 1.0 20 99.9 (0.0)

K35 p = 1.0 22 98.6 (0.8)
 Extinction 3
 FI 26 s 26
 Extinction 3
 p = 1.0 20 99.9 (0.0)

Note. Figures in parentheses are the standard deviations of the
session means for the percentage efficiency.


The results presented here for each rat are means calculated over the last five sessions in each condition. All rats except K33 engaged in unexpected gnawing behavior. There were over 90 ZDs for all rats in Condition p = 1.0.

Response Distributions

Figure 5 shows the response distribution in successive seconds after food for each rat in each condition. In Condition p = 0 for K26 and K28, patterning was very similar in that responding started off at a low level within the first few seconds of the interval and it accelerated slowly across the interval before leveling off in the final 10 seconds. When programmed ZDs were introduced in Condition p = 1.0 changes in response patterning were comparable for both rats. That is, there was an abrupt increase in responding in the first 5 s after food followed by an acceleration in responding throughout the remainder of the interval. When Condition p = 0 was reintroduced the abrupt increase in responding seen at the start of the interval remained, but it was followed by a lower and fairly constant rate thereafter.

For K32 and K35 the first exposure to Condition p = 1.0 produced patterns similar to those seen with K26 and K28. Patterning for the other two rats was different. K34 responded at a fairly constant rate across the interval, whereas patterning for K33 looked like that normally found on a FI schedule. When all of the rats were exposed to the FI schedule one main effect emerged. That is, the major characteristics of response patterning in the previous condition for each rat were replicated. Patterning for four rats (K26, K28, K32, & K34) appeared bimodal with a slight decrease in responding up to the first 10 s of the interval (20 s for K34) followed by acceleration thereafter.

In the final condition (Condition p = 1.0), the performance observed during the FI schedule for each rat was replicated, with slight increases in overall responding across the interval for K34 and K35.

This research was conducted when Michael Keenan was holder of a European Exchange Fellowship to the University of Cologne in Germany. I thank Prof. W. F. Angermeir for the facilities and support during my time there. Reprints may be obtained from Michael Keenan, School of Behavioural and Communication Sciences, University of Ulster, at Coleraine, Cromore Road, Coleraine, County Londonderry, N. Ireland, BT 52 1SA. E-mail:

Response Rate

Overall response rates for each animal in each condition are shown in Figure 6. For K26 and K28 the first transition to Condition p = 1.0 produced a three-fold increase in response rate. When K26 was returned to Condition p = 0 response rate decreased markedly; this did not happen for K28. In subsequent conditions, response rates were similar in each condition for these two rats. Across the other rats different effects emerged. For K33 and K35 response rate was comparable throughout all conditions. Rate decreased across conditions for K32. Rat K34 produced a staggering rate of about 60 responses per minute in the first condition. This dropped to just under 40 responses per minute in the second condition and increased slightly again in the final condition.


PRPs for each animal in each condition are given in Figure 7. Rats K26 and K28 produced pauses of about 17-18 s in the first condition. Thereafter there were substantial reductions in pause duration so that by the final three conditions they were pausing for less than 5 s after each reinforcer delivery. Pausing for K32, K34, and K35 was minimal also, but this was more extreme for K34. Rat K33 consistently produced the longest pauses in all conditions, between 17 and 20 s.


The primary objective of Experiment 2 was to compare performances produced by two schedules that each produce periodic response-reinforcer contiguity (ZDs). These schedules were the modified recycling conjunctive FT 30 s FR 1 (Keenan & Leslie, 1986) that provided maximum opportunity to obtain ZDs (Condition p = 1) and a FI 26 s schedule of food reinforcement. Of the six rats that were used, only one rat (K33) produced results that were not complicated by an unexpected history of gnawing. The rats who gnawed all had unusually elevated levels of responding in the first 10 seconds immediately after reinforcer delivery. For K26 and K28, who were first exposed to the normal recycling conjunctive, elevations in responding that occurred during this period on the modified schedule were not reversed when the normal schedule was reintroduced; such findings occurred also in Experiment 1. In addition, exposure to the FI contingencies did not produce substantial changes in patterning across animals. Rather, patterning on the FI schedule was determined by patterning on the immediately preceding schedule. Across all of these rats it might have been expected that continued exposure to periodic reinforcer presentation per se would have resulted in extended pausing after reinforcer delivery, as is normally the case (Richelle & Lejeune, 1980). Instead, the unusual patterns of responding on both the modified recycling conjunctive schedule and on the FI schedule remained resistant to change.

For K33, performance on the modified recycling conjunctive schedule and on the FI schedule was undisturbed by gnawing. FI-like patterning occurred on the modified schedule and across conditions performances were virtually indistinguishable. This finding, in conjunction with results in Experiment 1 by K1, K10, and K12 lends support for the view that periodic response-reinforcer contiguity on a FI schedule is a major determinant of response patterning (see also Keenan & Toal, 1991; Keenan & Watt, 1990).

The ability of the modified recycling conjunctive schedule to produce FI-like patterning holds much promise for other research. Future studies could, for example, map in more detail than presented here the similarities and differences in responding on both schedules (see Richelle & Lejune, 1980, for extended discussions on molecular and molar levels of analysis of FI performance). This information might help to pinpoint which aspects of the dynamics on these schedules is responsible for the different performances. This focus on dynamics would necessitate a systems(1)-based language (Keenan & Toal, 1991) that might open the way for classifying schedules in a way that transcends the traditional methods used at present. Keenan and Watt (1990) touched on this issue when they discussed the differences between four schedules, the FI (i.e., tandem FT FR 1), the conjunctive FT FR 1, the recycling conjunctive FT FR 1, and the FT schedule. Although three of these schedules are constructed from the same elements, it is the differing dynamics inherent in their different structures that crucially determines baseline performance. This point was emphasized when a comparison was made between the FT schedule and the other schedules:

In view of the fact that a FI schedule represents but one of a number of systems comprising a response contingency and periodic food presentation, ... it is incumbent upon us to use other related systems in order to clarify the manner in which this particular combination is distinctive. (pp. 128-129)

Other studies of the modified recycling conjunctive might be able to determine also whether or not there is a threshold of exposure to response-reinforcer contiguity that leads to rapid changes in patterning and rate of responding. One thing that was not clear from the results obtained here was the minimum relative frequency of contiguity needed to produce FI-like patterning. Any future studies that address this issue must deal with the relative effects of accidental and programmed contiguities without contaminating the single response contingency on the schedule.

General Discussion

Two general conclusions emerged from these experiments. Firstly, FI-like patterning can occur on a modified recycling conjunctive schedule that allows for a relatively high frequency of periodic response-reinforcer contiguity. Secondly, the pattern of responding obtained on this schedule can determine subsequent performance on a FI schedule.

If we look at the first conclusion we can ask the following questions. How are such marked differences in the formal specifications of the FI and modified recycling conjunctive schedules to be reconciled with the similarity of performance they can produce? Do they produce similar performances by different means? The answer to these questions must await the results of future studies that address the nature of the processes which occur during the acquisition phase of baseline responding. A focus on acquisition would be in keeping with the suggested need to map the behavioral dynamics on these schedules. Another argument for looking at acquisition more closely comes from the unusual patterns of responding which occurred during the interreinforcer interval for many of the animals. Processes occurring during acquisition produced these patterns. Once established they were maintained within the context of periodic response-reinforcer contiguity during baseline conditions. A related argument is that periodic food presentation per se does not result inevitably in extended pausing and bunching of responding in the terminal segment of the interreinforcer interval, a finding that is common with extended exposure on a FI schedule (Richelle & Lejune, 1980). Results from K1 and K2 bear this out.

Although the impact of history effects on FI patterning has been reported elsewhere (e.g., Baron & Leinenweber, 1995) none of these studies provided histories using regular food presentation in conjunction with a single response contingency. It seems that the history effects reported here, although unintentional and interesting in their own right, are fortuitous in so far as they help draw attention to an important and useful distinction between the FI schedule and the modified recycling conjunctive schedule. Although these schedules are similar in the components which comprise their formal specifications, they differ markedly in the dynamics that they each support. Behavior on the modified recycling conjunctive schedule reflects the combined effects of accidental contiguities (between any behavior and reinforcer delivery) and programmed contiguities (between the selected operant and reinforcer delivery) within the context of a system that does not arrange contiguous reinforcement for extended pausing (as would be the case on a FI schedule) and which arranges reinforcement (albeit delayed) for responding during the FT component.

Although these schedules differ in their dynamics because of the way the response contingency is designed, they share the effects of an important variable, the FT component. Numerous studies have shown that on its own a FT schedule of reinforcer delivery controls distinctive temporal distributions of behavior. It has been shown repeatedly that various categories of behavior become organized into a sequential pattern during the interreinforcer interval (Anderson & Shettleworth, 1977; Staddon, 1977). Of particular interest here are those kinds of behaviors (interim behaviors) which occur immediately after food delivery. The gnawing that was observed in this study was probably an instance of this behavior. The main support for this argument comes from the temporal location of the behavior. It reliably occurred only in the first 10-15 seconds of the interval (cf. Keenan & Watt, 1990). Unusually, though, it seems that this behavior became incorporated into the behavior that was selected initially as the operant. This possibility is supported by the finding of variations in the rate of lever depression across conditions during the period immediately after reinforcer delivery.

Because these animals either stood on the lever or chewed it, there is no way to distinguish operant pressing of the lever from depression of the lever that arose incidentally from other schedule-induced gnawing. Hence the unusual patterns that were recorded. Future studies could explore the possibility that interim activities can interact with an operant in the manner suggested by first inducing interim activities on a FT schedule that has a chewable, but inoperative operandum. The FT schedule might be changed then to a modified recycling conjunctive schedule by attaching the response contingency to the operandum. Such a study might look also at the possibility that performances produced here are in some way related to the strain of rat used; previous studies with the recycling conjunctive schedule have Sprague Dawleys whereas albino Wistar rats were used here.

To conclude, the mixture of patterns observed across animals in both experiments also has important implications for how we view the notion of 'temporal control.' To put it briefly, discussion of temporal control often arises in response to the baseline patterning that appears on a schedule where reinforcers are presented at regular intervals of time. Historically the patterning frequently discussed is that found on FI schedules. There are problems, however, with the way in which the notion of temporal control is sometimes used both to describe and explain performance (Leslie, 1996). At the descriptive level, the patterning on a FI schedule is said to reflect temporal control by virtue of its synchronization with the periodicity of the reinforcer presentation. On balance, the explanation for the pattern is sometimes said to reflect an animal's 'ability to discriminate the passage of time.' Blackman (1983) discussed how the seductive nature of temporal patterning in behavior can result in the cognitive metaphor of an internal clock:

A particularly interesting paper in the field of cognitive learning theory is that of Church (1978). The paper reports the results of an extensive series of experiments which reveal how rats' lever-pressing behavior can become functionally related to the passage of time, that is how temporal patternings develop in operant behavior as a result of certain schedules of reinforcement. Behavior analysts would seek ways of capturing and describing (a) the temporal pattern of behavior, (b) the temporal patterning of environmental events, and (c) the functional relationships between these two measures. Church himself (p. 282) recognizes that his experiments 'provide ample evidence that there is a relationship between time and behavior'. However the evolution of Church's attempts to explain [emphasis in original] this relationship is illuminating, ....

Church's data describing functional relationships between behavior and environmental events takes second place to a clock that runs, stops, or runs at different speeds, but which cannot be seen by the experimenter. The clock apparently has to be consulted by a positive decision on the part of the rat, who then decides how to behave as a result of his reading. The behavior analysts' position can be succinctly conveyed by the suggestion that it would be more profitable to conceptualize the rat ass [emphasis in original] the clock. Rather than reflecting the operation of [emphasis in original] a clock, the rat's behavior would now be said to have [emphasis in original] the properties of a clock in certain environmental conditions. The functioning of this clock (i.e., this behavior) may differ in different environmental conditions, and it is the task of experimental analysis to identify these conditions and their effects on the clock (behavior). (pp. 45-46)

Imagine for a moment that the findings reported here were the first of their kind. That is, imagine that the data presented here represented the first findings using schedules with FT components. I suspect that the variety of patterns shown here would preclude the use of the term temporal control as outlined by Church. Explanations that refer to the operation an internal clock as an independent variable would be forced to invent as many different kinds of clocks, or clock rates as there were patterns of responding. Also, because all of the 'clocks' were controlled by the same temporal distribution of reinforcers, one would have to explain why the rats 'decided to do different things at the same time in the interval.' Another problem with the notion of temporal control as an explanatory term is that it requires decisions to be made for determining the criteria that distinguish 'poor temporal control' from 'accurate temporal control.' In regards to the patterns reported here this would be extremely problematical. A natural science perspective of these response patterns, however, has no need for the term temporal control as traditionally conceived. Because they are all instances of adaptive behavior they can not be graded as examples of either poor or accurate temporal control. They are explained only by reference to the dynamics afforded by the structure of the prevailing contingencies.

1 Keenan and Toal (1991) defined a schedule in the following way: "a schedule is more properly conceived as providing an opportunity for observing the dynamic behavioral system that "crystallizes out" when a biological system is exposed to environmental constraints... At any one instance, the characteristics of the behavioral system are dependent upon the interplay between the "plasticity" or dynamic limitations inherent in the adaptiveness of the biological system, and the dynamics imposed across time by the structure of the prevailing contingencies." (p. 113)


ANDERSON, M. C., & SHETTLEWORTH, S. J. (1977). Behavioral adaptation to fixed-interval and fixed-time food delivery in golden hamsters. Journal of the Experimental Analysis of Behavior, 27, 33-49.

BARON, A., & LEINENWEBER, A. (1994). Molecular and molar analyses of fixed-interval performance. Journal of the Experimental Analysis of Behavior, 61, 11-18.

BARON, A., & LEINENWEBER, A. (1995). Effects of a variable-ratio conditioning history on sensitivity to fixed-interval contingencies in rats. Journal of the Experimental Analysis of Behavior, 63, 97-110.

BICKEL, W. K., HIGGINS, S. T., KIRBY, K., & JOHNSTON, L. M. (1988). An inverse relationship between baseline fixed-interval response rate and the effects of a tandem response requirement. Journal of the Experimental Analysis of Behavior, 50, 211-218.

BLACKMAN, D. E. (1983). On cognitive theories of animal learning: Extrapolation from humans to animals? In G. C. L. Davey (Ed.), Animal models of human behavior. New York: John Wiley & Sons Ltd.

CATANIA, A. C. (1998). Learning (4th ed.). Upper Saddle River, NJ: Prentice-Hall, Inc.

CHURCH, R. M. (1978). The internal clock. In S. H. Hulse, H. Fowler, & W. K. Honig (Eds), Cognitive processes in animal behavior. Hillsdale, NJ: Erlbaum Associates.

CUMMING, W. W., & SCHOENFELD, W. N. (1958). Behavior under extended exposure to a high-value fixed-interval schedule. Journal of the Experimental Analysis of Behavior, 1, 245-263.

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

DEWS, P. B. (1970). The theory of fixed-interval responding. In W. N. Schoenfeld (Ed.), The theory of reinforcement schedules (pp. 43-61). New York: Appleton-Century-Crofts.

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

HYTEN, C., & MADDEN, G. J. (1993). The scallop in human fixed-interval research: A review of problems with data description. The Psychological Record, 43, 471-500.

JOHNSTON, L. M., BICKEL, W. K., HIGGINS, S. T., & MORRIS, E. K. (1991). The effects of schedule history and opportunity for adjunctive responding on behavior during a fixed-interval schedule of reinforcement. Journal of the Experimental Analysis of Behavior, 55, 313-322.

KEENAN, M. (1982). Behavioural organisation in schedules of reinforcement. Unpublished D. Phil. thesis. University of Ulster.

KEENAN, M. (1986). Second-order schedules. The Psychological Record, 36, 407-417.

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., & TOAL, L. (1991). Periodic reinforcement and second-order schedules. The Psychological Record, 41, 87-115.

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

LEINENWEBER, A., NIETZEL, S. M., & BARON, A. (1996). Temporal control by progressive-interval schedules of reinforcement. Journal of the Experimental Analysis of Behavior, 66, 311-326.

LESLIE, J. C. (1996). Principles of behavior analysis. Amsterdam: Harwood Academic Publishers GmbH.

LOWE, C. F., & WEARDEN, J. H. (1981). Weber's law and the fixed-interval postreinforcement pause. Behavior Analysis Letters, 1, 27-37.

MORSE, W. H., & KELLEHER, R. T. (1977). Determinants of reinforcement and punishment. In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of operant behavior (pp. 174-200). Englewood Cliffs, NJ: Prentice-Hall.

RICHELLE, M., & LEJEUNE, H. (1980). Time in animal behaviour. Oxford: Pergamon.

STADDON, J. E. R. (1983). Adaptive behavior and learning. New York: Cambridge University Press.

STADDON, J. E. R. (1977). Schedule-induced behavior. In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of operant behavior (pp. 125-153). Englewood Cliffs, NJ: Prentice-Hall.

TATHAM, T. A., WANCHISEN, B. A., & HINELINE, P. A. (1993). Effects of fixed and variable ratios on human behavioral variability. Journal of the Experimental Analysis of Behavior, 59, 349-359.

THOMAS, G. (1981). Contiguity, reinforcement rate and the law of effect. Quarterly Journal of Experimental Psychology, 33B, 33-43.

WANCHISEN, B. A., TATHAM, T. A., & MOONEY, S. E. (1989). Variable-ratio conditioning history produces high- and low-rate fixed-interval performance in rats. Journal of the Experimental Analysis of Behavior, 52, 167-179.

ZEILER, M. D. (1977). Schedules of reinforcement: The controlling variables. In W. K. Honig & J. E. R. Staddon (Eds.), Handbook of operant behavior (pp. 201-232). Englewood Cliffs, NJ: Prentice-Hall.
COPYRIGHT 1999 The Psychological Record
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Keenan, Michael
Publication:The Psychological Record
Date:Mar 22, 1999
Previous Article:Effects of punishment proportion and condition sequence on contrast and induction with humans.
Next Article:Decrease of precurrent behavior as training increases: effects of task complexity.

Related Articles
Differential reinforcement of other behavior and response suppression: the effects of the response-reinforcement interval.
Response-reinforcer contiguity and human performance on simple time-based reinforcement schedules.
The operant-respondent distinction revisited: toward an understanding of stimulus equivalence.
Rats' response rates for 1% sucrose when food-pellet reinforcement is upcoming: effect of upcoming reinforcement contingency.
Positive induction produced by food-pellet reinforcement: component variations have little effect.
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