FOR BETTER OR WORSE: EFFECT OF UPCOMING REINFORCER TYPE ON RATS' LEVER PRESSING FOR LOW-CONCENTRATION SUCROSE REINFORCERS.

Recent research has demonstrated that food-pellet reinforcement upcoming in the second half of an experimental session can increase operant response rates for low-concentration liquid-sucrose reinforcers in the first half. The present study investigated whether an upcoming reinforcer other than food pellets could produce a similar increase (Experiment 1) and whether a decrease in responding would be observed by worsening the upcoming conditions of reinforcement (Experiment 2). Experiment 1 demonstrated that rats' rates of lever pressing for 1 or 5% liquid-sucrose reinforcers delivered by a random-interval 60-s schedule during the first half of a 50-mm session were similarly increased by the delivery of food or sucrose pellets in the second half. Experiment 2 demonstrated that rates of lever pressing for 5% liquid-sucrose reinforcers in the first half of the session were increased and decreased by upcoming food-pellet and 1% liquid-sucrose reinforcement, respectively. The results from both experiments represen t induction (i.e., a direct relationship between rate of responding and the upcoming conditions of reinforcement). They suggest that the induction may be both general (i.e., not limited to specific upcoming reinforcers) and symmetrical (i.e., manifesting in either an increase or a decrease in responding), the latter of which provides a potential link to the study of behavioral contrast.

Behavioral contrast (Reynolds, 1961) is said to occur when a subject's rate of responding in one, unchanged, component of a multiple schedule of reinforcement varies inversely with changes in the conditions of reinforcement in another component (McSweeney & Norman, 1979). Behavioral contrast procedures typically employ pigeons to peck keys for reinforcers delivered at high rates on multiple schedules that have short components which deliver the same reinforcer and that alternate many times during the experimental session (see McSweeney & Weatherly, 1998, for a review). When these procedures are employed, contrast reliably results.

However, if one deviates from these procedural details, then the opposite of contrast (i.e., induction) may occur. Weatherly, Stout, McMurry, Rue, and Melville (1999, Experiment 2), for instance, had rats lever press on a multiple schedule with long components (i.e., 30 min) that delivered different reinforcers (5% liquid sucrose or a food pellet). Specifically, subjects responded in 60-min sessions in which 5% liquid-sucrose reinforcers were delivered during the first 30 min of the session and 45-mg food-pellet reinforcers were delivered during the second 30 min. Across conditions, rate of food-pellet reinforcement was varied from a variable-interval (VI) 15-s to a VI 240-s schedule in the second half of the session while the rate of sucrose reinforcement in the first half was held constant at a VI 60-s schedule. Results showed that rates of responding for sucrose in the first half of the session increased (induction [1]), rather than decreased (contrast), as the rate of food-pellet reinforcement increased. In other words, the rats' rates of responding for the liquid sucrose varied directly with the upcoming conditions of reinforcement.

Weatherly, Stout, Rue, and Melville (2000) replicated this induction. They had rats respond in 50-min sessions in which liquid-sucrose reinforcers were available during the first 25 min of the session and, across conditions, either the same sucrose reinforcers or food pellets were available during the second 25 min. Higher rates of responding were observed in the first half of the session when food pellets, rather than sucrose reinforcers, were delivered in the second half. The results also showed that the size of induction varied inversely with the concentration of sucrose. Upcoming food-pellet reinforcement had a large effect on responding for 1% sucrose (which maintains very low rates of responding). It had little to no effect on responding for 25% sucrose (which maintains moderate to high rates of responding).

Weatherly, Davis, and Melville (2000) attempted, across three experiments, to pinpoint procedural factors contributing to this induction. They investigated whether the presence or absence of a discriminative stimulus for the second half of the session (Experiment 1) or the predictability of the timing of the switch from liquid-sucrose to food-pellet reinforcement (Experiment 2) would alter the induction. Neither did. However, their Experiment 3 demonstrated that the certainty of the upcoming food-pellet reinforcement was influential. Across conditions, the probability that food-pellet reinforcement would be delivered in the second half of the session was varied from 0.0 to 1.0. Induction was observed for responding for sucrose in the first half of the session when the probability of food reinforcement in the second half of the session was 1.0 (compared to when it was 0.0). Altering the probability of upcoming food-pellet reinforcement (0.25, 0.50, & 0.75) altered the induction. Increases in the rate of respon ding across the first half of the session, measured by the slope, were directly related to the probability of upcoming food-pellet reinforcement. Thus, the results of Weatherly, Davis, et al. (2000) suggested that the probability, and not discriminability or timing, of upcoming differential reinforcement contributes to the induction.

The above studies demonstrate that upcoming food-pellet reinforcement can increase rate of responding for low-concentration sucrose reinforcers. However, their results are limited. Each used upcoming food-pellet reinforcement to produce induction. Moreover, each produced positive induction (i.e., response rates were increased by improving the upcoming conditions of reinforcement). Thus, they can not speak to whether an upcoming reinforcer other than a 45-mg food pellet can produce induction or whether induction may also manifest in a decrease in responding when the upcoming conditions of reinforcement are worse than the current conditions (i.e., negative induction).

The present study was designed to make these determinations. Making them was deemed important for several reasons. First, determining whether different upcoming reinforcers can produce positive or negative induction will help determine the generality of the effect. If only positive induction is observed and is only produced in rats when food-pellet reinforcement is upcoming, then the effect is very limited. On balance, if both positive and negative induction can be produced by different upcoming reinforcers, then the scope of the effect is potentially very broad. Second, by determining whether different upcoming reinforcers can produce positive or negative induction, a theoretical link between the present effect and that of behavioral contrast may be established. More specifically, it has been argued that behavioral contrast is a symmetrical effect (i.e., alterations of the same factor, such as rate of reinforcement, can produce either positive or negative behavioral contrast; see McSweeney & Weatherly, 1998) . If the present effect is related to behavioral contrast (e.g., the inverse of the conditions that produce behavioral contrast), then one would expect that the present induction would also be symmetrical.

Two experiments were conducted. Experiment 1 attempted to determine whether an upcoming reinforcer other than a Noyes food pellet could produce induction. Experiment 2 attempted to determine whether positive or negative induction would be observed when the upcoming conditions of reinforcement were better or worse, respectively, than the current conditions.

Experiment 1

Previous studies (e.g., Weatherly, Davis, et al., 2000; Weatherly, Stout, et al., 2000) have demonstrated that rats responding for low-concentration liquid-sucrose reinforcers (e.g., 1% or 5% sucrose) will display increased response rates when food-pellet (P. J. Noyes, Formula A/l), rather than the same liquid-sucrose, reinforcement will be available in the second half of the session. The present experiment investigated whether upcoming sucrose-pellet (P. J. Noyes, Formula F) reinforcement would produce a similar effect. A sucrose pellet was chosen for several reasons. For one, this type of pellet varies from the previously used food pellet in a number of ways, including color, taste, nutritional value, and caloric content. Secondly, sucrose pellets also maintain higher operant rates of responding than low-concentration liquid-sucrose reinforcers (unpublished pilot data from our laboratory). Thus, it was potentially possible to compare any induction produced by upcoming sucrose pellets to that produced by upc oming food pellets.

Method

Subjects. The subjects were 8 experimentally naive male Sprague-Dawley rats obtained from Charles River Laboratories. Subjects were housed individually, had free access to water (only) in the home cage, and experienced a 14/10 hr light/dark schedule (0600 - 2000). Subjects were maintained at approximately 85% of their free-feeding body weight via postsession feedings when necessary or by daily feedings on days in which sessions were not conducted.

Apparatus. The apparatus was an experimental chamber for rats (MED Associates, ENV-008) that measured 23.5 by 30.5 by 21 cm. A 5cm response lever was located 1.5 cm from the left edge of the front panel and 8 cm above the grid floor. The lever extended 2 cm into the chamber. A force of approximately 0.25 N was required to depress the lever. An identical lever was symmetrically located on the right side of the front panel. It was not used in the present experiment. A 2.5-cm diameter stimulus light was located 6.5 cm above each lever. A 5- by 5-cm aperture, which could allow access to reinforcement, was centered on the front panel, 5 cm above the floor. A liquid-drop dispenser (MED Associates, ENV-201A) and a pellet dispenser (MED Associates, ENV203) were located behind the front panel. Both could deliver reinforcers into a 2.5-cm diameter cup recessed within the aperture. A houselight was centered on the back wall of the chamber, 2 cm below the ceiling. The experimental chamber was housed in a sound-attenuatin g chamber, with a ventilating fan masking noise from the outside. Experimental events were programmed and data were recorded by an IBM-compatible computer that was running MED-State software and connected to a MED-Associates interface. The computer and experimental chamber were located in the same room.

Procedure. Subjects were trained to press the left lever using shaping by successive approximations. Once each subject had pressed the lever more than 100 times, it was placed on the experimental procedure.

Subjects responded in sessions that were 50 min in length. Pressing the left lever was reinforced on a random-interval (RI) 60-s schedule throughout the session. Reinforcers were programmed at a probability of 0.0167 every 1 s as long as a reinforcer was not already scheduled for delivery. That is, a scheduled reinforcer needed to be collected before the next interreinforcer interval began. Neither the session timer nor the interreinforcer interval advanced during the delivery of a reinforcer. The houselight and the light above the left lever were illuminated throughout the session.

Three types of conditions were conducted. In one (Suc-Suc), 0.2 ml of liquid sucrose served as the reinforcer in both halves of the session. In another (Suc-FoodPel), liquid-sucrose reinforcers were delivered in the first 25 min of the session and a 45-mg Noyes food pellet (Formula A/I) served as the reinforcer in the second 25 min. In the third condition (SucSucPel), liquid-sucrose reinforcers were delivered in the first half of the session and a 45-mg Noyes sucrose pellet (Formula F) served as the reinforcer in the second half. Each condition was conducted for a total of 20 sessions with sessions conducted daily, 5 to 7 days per week.

The subjects were divided into two groups. The first group responded for 1% liquid sucrose (v/v mixed with tap water). The second group responded for 5% liquid sucrose. One pair of subjects in each group received the following order of conditions: Suc-SucPel, Suc-FoodPel, and Suc-Suc. The other pair of subjects received the reverse order of conditions. Once each group had completed the three conditions, the sucrose concentration for which they responded was reversed. Each pair of subjects then responded in the same three conditions, but in the reverse order as they experienced them when responding for the previous sucrose concentration.

Results and Discussion

Figure 1 presents the results observed during conditions in which subjects responded for 1% (top graph) or 5% sucrose (bottom graph). It presents rate of responding (in responses/min) across successive 5-min intervals of the session during the Suc-Suc (closed squares), Suc-SucPel (open squares), and Suc-FoodPel (closed circles) conditions. Each function represents the mean for all subjects responding during the final five sessions of a particular condition. The error bars represent the standard error of the mean for responding during that particular 5-min interval.

The data in Figure 1 leave several visual impressions. First, induction was apparent in the first half of the session. For both concentrations of sucrose, response rates in the first half of the session were higher when either sucrose pellets or food pellets served as the reinforcer in the second half than when the same liquid sucrose served. Second, the presence of induction did not vary noticeably between pellet types. Third, both types of pellets maintained higher rates of responding in the second half of the session than did either liquid-sucrose concentration.

Statistical analyses confirmed these impressions. Responding from the first half of the session was analyzed by conducting a three-way (Sucrose concentration x Second-half reinforcer x 5-min interval) repeated measures analysis of variance (ANOVA) on the rates of responding of individual subjects. Results of this analysis showed significant main effects of sucrose concentration, F(1, 7) = 12.68, p [less than] .01, and second-half reinforcer, F(2, 14) = 8.57, p [less than] .01, indicating that subjects responded significantly higher for the 5% sucrose than for the 1% sucrose and that rates of responding in the first half of the session varied with the type of reinforcer delivered in the second half, respectively. The only other effect to reach significance was the interaction between second-half reinforcer and 5-min interval, F(8, 56) = 3.24, p [less than] .01, indicating that responding changed across the first 25 min of the session differently when different reinforcers were delivered in the second half of t he session. Results from this analysis, and all that follow, were considered significant at p [less than] .05.

Although the significant main effect of second-half reinforcer suggests that induction was observed for responding in the first half of the session, it does not necessarily mean that the upcoming sucrose pellets produced a similar induction effect as did the upcoming food pellets. In other words, the significant main effect could have been produced by the food pellets alone. To make this determination, a follow-up three-way (Sucrose concentration x Pellet type x 5-min interval) repeated measures ANOVA was conducted using data from the first half of the session of the Suc-SucPel and Suc-FoodPel conditions. The main effect of sucrose concentration was again significant, F(1, 7) = 6.31, p [less than] .05, indicating that 5% sucrose maintained higher response rates than did 1% sucrose. However, the main effect of pellet type was not significant, indicating that responding in the first half of the session did not differ between these conditions. The main effect of 5-min interval was significant, F(4, 28) = 3.95, p [less than] .05, indicating that response rates changed across the first half of the session when pellets were delivered in the second half of the session. No other effects reached significance. These results thus support the conclusion that induction was observed when subjects responded for either 1% or 5% sucrose and that both types of pellets produced a similar induction effect.

Responding in the second half of the session was analyzed by conducting a three-way (Sucrose concentration x Second-half reinforcer x 5-min interval) repeated measures ANOVA on responding from all three conditions. In this analysis, the main effects of second-half reinforcer, F(2, 14) = 11.16, p [less than] .001, and of 5-min interval, F(4, 28) = 5.22, p [less than] .01, were significant, indicating that the different reinforcers maintained different response rates and that those rates changed across the second half of the session, respectively. The interactions between sucrose concentration and second-half reinforcer, F(2, 14) = 7.54, p [less than] .01, and between second-half reinforcer and 5-min interval, F(8, 56) = 4.67, p [less than] .001, were also significant. These results suggest that the different reinforcers maintained different absolute rates of responding between the 1% and 5% sucrose conditions and that different within-session changes in responding were observed in the second half of the sessio n for the different reinforcers, respectively.

Again, the main effect of second-half reinforcer does not allow for a conclusion on whether the different pellet types maintained different response rates. Therefore, a follow-up three-way (Sucrose concentration x Pellet type x 5-mm interval) repeated measures ANOVA was conducted on responding from the second half of the session of the Suc-SucPel and Suc-FoodPel conditions. Results indicated that the main effect of 5-min interval, F(4, 28) = 5.75, p [less than] .01, was significant, indicating that rates of responding changed across the second half of the session in these conditions. However, no other effects reached significance, indicating that the different types of pellets did not differ in the behavior they maintained in the second half of the session.

The results of Experiment 1 suggest that induction produced by upcoming reinforcement is not limited to the use of food pellets. Similar induction effects were observed when a solid sucrose pellet served as the upcoming reinforcer. Besides increasing the potential generality of the present induction effect, these results may also provide some insight into the factors that control (or do not control) the effect. Intuitively, the present results would seem to run contrary to the idea that a difference in taste between the liquid sucrose and the upcoming reinforcer plays a role in the induction. That is, a switch from liquid sucrose to a solid sucrose pellet produced induction. Such a conclusion cannot be unequivocally accepted, however, because subjects' behavior was similarly supported by both the sucrose and the food pellets. Thus, it is not possible to conclude that subjects discriminated between the two.

However, because the two types of pellets did differ in a number of ways (e.g., taste, color, caloric content, etc.), it seems more likely that they maintained similar rates of behavior because they had a similar "value" than because subjects did not discriminate between them. Such a conclusion implies that the induction effect may then be influenced by the "value" of the upcoming reinforcer. If such were the case, then one might expect that an upcoming reinforcer with a lower "value" than the present one might produce a negative induction effect (i.e., a decrease in response rate). Experiment 2 explored this idea.

Experiment 2

Experiment 2 was designed in an attempt to produce negative induction. As noted above, all previous studies on the present effect, including Experiment 1, have produced a positive induction effect by bettering the upcoming conditions of reinforcement. Negative induction was attempted by worsening of the upcoming conditions of reinforcement by delivering a reinforcer in the second half of the session that maintained a lower rate of responding than the one delivered in the first.

Method

Subjects and apparatus. The subjects were 4 experimentally naive male Sprague-Dawley rats. They were obtained, housed, and maintained as were those in Experiment 1.

The apparatus was a chamber for rats (Med Associates, ENV-007) measuring 28 by 22 by 28 cm. It was a two-lever chamber, nearly identical to that used in Experiment 1. One difference was that two types of liquid reinforcers could be delivered to the recessed cup via two 22-gauge pipettes, each connected to a separate syringe pump (Med Associates, PHM-100) located outside of the sound-attenuating chamber. A second difference was that a nose-poke apparatus, which was not used, was centered on the back wall. The nose-poke opening was 2 cm in diameter, with the bottom of the opening being located 1.5 cm above the floor.

Procedure. Subjects were trained to press the left lever using shaping by successive approximations. Once each subject had pressed the lever more than 100 times, it was placed on the experimental procedure.

The procedure was modeled after that used by Weatherly, Stout, et al. (2000). Subjects responded in sessions that were 50 min in length. Pressing the left lever was reinforced on a RI 60-s schedule throughout the session, with reinforcers programmed identically as in Experiment 1. The houselight and the light above the left lever were illuminated throughout the session.

Two sets of conditions, constant and random, were conducted. In all conditions, the reinforcer in the first half (25 min) of the session was 0.2 ml of a 5% liquid-sucrose solution. In the constant conditions, the type of reinforcer delivered in the second half of the session (25 min) was held constant across 20 consecutive sessions. Three such conditions were conducted. Type of second-half reinforcer was either 0.2 ml of 5% sucrose (5% - 5%), a 45-mg Noyes food pellet (5% - FP), or 0.2 ml of 1% sucrose (5% - 1%). One-percent sucrose was chosen because it maintains significantly lower response rates than does 5% sucrose (e.g., see results of Experiment 1).

In random conditions, the type of reinforcer delivered in the second half of the session was determined randomly prior to each session. This condition was conducted until each of the three possible reinforcers had been delivered in the second half of the session a total of 20 times. Because the procedure was truly random, some reinforcers were presented in the second half of the session more than 20 times before another reinforcer had been presented 20 times. Sessions subsequent to the 20th session for a particular reinforcer were not analyzed.

Two subjects experienced the constant conditions prior to the random conditions. The other pair of subjects received the reverse order. Within each pair of subjects, 1 subject experienced the constant conditions in the following order: 5% - FP, 5% - 1%, and 5% - 5%. The other subject received the reverse order. Sessions were conducted daily, 5 to 7 days per week.

Results and Discussion

Figure 2 presents the rates of responding for 5% sucrose across successive 5-min intervals in the first half of the session (Intervals 1 through 5) for each condition. It was constructed similarly to Figure 1. Each function represents the mean for all subjects when 1% sucrose (open circles), 5% sucrose (closed squares), or a food pellet (open triangles) served as the reinforcer in the second half of the session. The top graph presents the results from the constant conditions. The middle graph presents first-half responding from the random conditions in which the different reinforcers were delivered in the second half of the session. The bottom graph presents the same data as the middle graph but recalculated in terms of the type of reinforcer that was presented in the second half of the session in the previous session (see explanation that follows).

The top graph of Figure 2 demonstrates that responding in the first half of the session during the constant conditions varied as a function of upcoming reinforcer. Rate of responding for 5% sucrose was lowest when 1% sucrose served as the reinforcer during the second half of the session. Response rates were highest and increased most steeply during the first half of the session, when food pellets were delivered in the second half of the session.

Statistical analyses supported these impressions. A two-way (Second-half reinforcer by 5-min interval) repeated measures ANOVA, conducted on response rates for individual subjects in the constant conditions, produced a significant main effect of second-half reinforcer, F(2, 6) = 8.62, p [less than] .05; main effect of 5-min interval, F(4, 12) = 19.46, p [less than] .001; and interaction term, F(8, 24) = 3.73, p [less than] .01. These results indicate that average rate of responding in the first half of the session differed when different reinforcers were delivered in the second half of the session, that rates of responding changed across the first half of the session, and that the changes were different when different reinforcers were delivered in the second half of the session, respectively.

To better elucidate the effects in the top graph of Figure 2, follow-up two-way (Second-half reinforcer by 5-min interval) repeated measures ANOVAs were conducted on each possible pair of constant conditions. Of particular interest were the main effects of second-half reinforcer, which indicate whether the upcoming reinforcer produced different rates of responding in the first half of the session, and the interaction terms, which indicate whether the upcoming reinforcer produced different changes in responding during the first half. Comparison of the 5% - 1% and 5% - 5% conditions produced a main effect of second-half reinforcer that approached significance, F(1, 3) = 10.01, p [less than] .051, and a nonsignificant interaction. Comparison of the 5% - 5% and 5% - FP conditions produced a nonsignificant main effect of second-half reinforcer but a significant interaction term, F(4, 12) = 3.68, p [less than] .05. Finally, comparison of the 5% 1% and 5% - FP conditions produced a significant main effect of second- half reinforcer, F(1, 3) = 37.13, p [less than] .01, and interaction term, F(4, 12) = 5.35, p [less than] .01. In general, these results suggest that the negative induction effect is best described as a change in the overall rate of responding. The positive induction effect, however, is best described as change in the pattern of responding across the first half of the session.

The middle graph of Figure 2 shows that the differences in the top graph were the result of consistent experience with those reinforcers occurring in the second half of the session. When subjects received the same total amount of experience with the same second-half reinforcers, but on a random basis, responding during the first half of the session did not vary with type of upcoming reinforcer. This impression was tested by conducting a two-way (Second-half reinforcer by 5-min interval) repeated measures ANOVA on response rates for individual subjects in the first half of the session for the random conditions in which a particular second-half reinforcer was delivered during the present session. The analysis produced a significant main effect of 5-min interval, F(4, 12) = 7.82, p [less than] .01, indicating that response rates changed across the first half of the session. However, neither the main effect of second-half reinforcer nor the interaction term was significant, indicating that neither the average rat e of responding nor the pattern of responding differed in the first half of the session when different reinforcers were randomly delivered in the second half.

Because the type of upcoming reinforcer in the random conditions was unpredictable, it would seem unreasonable to expect that it would alter responding during the first half of the session. It would seem more reasonable to suspect that the reinforcer delivered in the second half of the previous session may have influenced responding in the first half of the subsequent session. The bottom graph presents first-half responding from the random conditions based upon the previous session's second-half reinforcer. An identical two-way ANOVA was conducted on these data, which produced the identical results. The main effect of 5-min interval, F(4, 12) = 6.99, p [less than] .01, was significant, but the main effect of second-half reinforcer and interaction term were not, indicating that the previous session's second-half reinforcer did not control subjects' responding in the first half of the next session.

Figure 3 presents results from the second half of the session. As expected, the different reinforcers delivered in the second half of the session maintained different rates of responding. It presents the rates of responding that were observed across successive 5-mm intervals. (Intervals 6 through 10) of the second half of the session in the constant (top graph) and random (bottom graph) conditions when subjects responded for 1% sucrose (open circles), 5% sucrose (closed squares), or food pellets (open triangles). Figure 3 was constructed similarly to Figure 1.

The data in Figure 3 were analyzed by conducting a three-way (Condition x Second-half reinforcer x 5-min interval) repeated measures ANOVA on the response rates of individual subjects in the constant and random conditions. This analysis produced a significant main effect of second-half reinforcer, F(2, 6) = 5.67, p [less than] .05, indicating that the different reinforcers supported different rates of responding. The interaction between second-half reinforcer and 5-mm interval was also significant, F(8, 24) = 2.56, p [less than] .05, indicating that different changes in response rates were observed in the second half of the session for the different reinforcers. No other main effects or interactions reached significance.

General Discussion

A number of recent studies (e.g., Weatherly, Davis, et al., 2000; Weatherly, Stout, et al., 2000) have reported that food-pellet reinforcement upcoming in the second half of the session produces an increase in liquid-sucrose-reinforced response rates in the first half (i.e., induction). The results of the present study replicated and extended this finding. Experiment 1 demonstrated that an upcoming reinforcer other than a food pellet can produce an increase in sucrose-reinforced responding. Experiment 2 demonstrated that a decrease in sucrose-reinforced responding may be observed when the reinforcer delivered in the second half of the session maintains a lower rate of responding than that delivered in the first half. The present results thus increase the generality of the induction by showing that the effect is not limited to just one type of upcoming reinforcer (i.e., food pellets). Moreover, by showing that the induction may manifest in either an increase or decrease in response rate, the results suggest th at the present induction effect shares the quality of symmetry with its seemingly opposite effect, behavioral contrast.

Although the present study appears to broaden the generality of the induction effect, it could potentially be argued that the present results do not represent induction at all. Rather, by delivering reinforcers of different "values" in the second half of the session, the value of the entire session might have been altered. Thus, response rates in the first half of the session may have differed across conditions because subjects were treating sessions as single units with differing values rather than displaying induction. Although the present results cannot rule out this explanation, several pieces of evidence would seem to run counter to it. For one, although the present study used a mixed schedule of reinforcement, the present effect has been observed using multiple schedules of reinforcement, which provide overt discriminative stimuli to allow subjects to discriminate the first and second halves of the session (Weatherly et al., 1999 Weatherly, Davis, et al., 2000). Secondly, the behavior of subjects in the present study suggests that they differentiated between the reinforcers delivered in the first and second halves of the session. As can be seen in Figure 1, and by comparing Figures 2 and 3, when a change in reinforcer type occurred at the midpoint of the session, rates of responding changed as well. Both of these pieces of evidence suggest that subjects were not treating the session as a single unit.

That is not to say, however, that the present effect was not a product of a change in the "value" of the reinforcer delivered in the first half of the session. Such a change in the value of the liquid-sucrose reinforcer may have occurred if Pavlovian conditioning played a role in the current effect. One possible role is that the liquid-sucrose reinforcers and/or the stimuli-that were present during their delivery served as conditioned stimuli for the upcoming differential reinforcement. For example, when these stimuli predicted upcoming food-pellet reinforcement, they might have become conditioned exciters and thus facilitated responding by producing arousal (e.g., Killeen, 1995). Likewise, when they predicted upcoming 1% sucrose reinforcement, they might have become conditioned inhibitors and thus suppressed responding, possibly by causing withdrawal from the operandum (e.g., Wasserman, Franklin, & Hearst, 1974).

This explanation appears consistent with the finding that the effect size of the induction varies inversely with sucrose concentration (Weatherly, Davis, et al., 2000; Weatherly, Stout, et al., 2000). One might expect high concentrations of sucrose to serve as strong reinforcers (i.e., unconditioned stimuli) and therefore be less likely to serve as conditioned stimuli (thus eliminating the induction). On balance, low concentrations of sucrose may serve as weak reinforcers and thus may have enhanced potential to serve as conditioned stimuli. The idea that Pavlovian conditioning plays a role in multiple-schedule interactions is not new (e.g., Gamzu & Schwartz, 1973). It should thus be possible to use existing procedures designed to measure the influence of Pavlovian conditioning in such situations (e.g., Williams, 1991) to determine its role in the present effect.

One explanation for the present induction effect forwarded by Weatherly, Davis, et al. (2000) and Weatherly, Stout, et al. (2000) seems to be challenged by the results of Experiment 2. That explanation was that the induction produced by upcoming food-pellet reinforcement was the result of anticipatory responses for food pellets being added to responses for the presently available liquid-sucrose reinforcers. In other words, subjects may have started to respond for the food pellets prior to their availability. Thus, these responses, added to those for the currently available sucrose, may have produced a rate of responding higher than that observed when food-pellet reinforcement was not upcoming. Although the same explanation could potentially explain the results of Experiment 1 and those from Experiment 2 when food-pellet reinforcement was upcoming, it seems to fail to explain the decrease in responding observed in Experiment 2 when 1% sucrose reinforcement was upcoming. That is, it seems unlikely that upcoming 1% sucrose reinforcement subtracted from responding for presently available 5% sucrose reinforcement.

Although the results of Experiment 2 suggest that response rates in the first half of the session can be increased or decreased depending on the type of reinforcer in the second half of the session, the notion that these effects are symmetrical remains open to question. The increase in responding in the first half of the session when food-pellet, rather than the same 5% sucrose, reinforcement occurred in the second half can be best described as a difference in response pattern. However, the effect of upcoming 1% sucrose reinforcement was one of response rate. When these results are coupled with those of Weatherly et al. (1999, Experiment 4), who found that response rates for food pellets in the first half of the session were insensitive to changes in the rate of 5% liquid-sucrose reinforcement in the second half, it appears that further research is required before one can conclude whether or not the induction is indeed symmetrical.

Finally, it should be noted that the induction observed in the present study was observed under conditions that maintained low baseline rates of responding. This observance may suggest that the induction effect is peculiar to situations that support low response rates. In fact, finding that little or no induction is observed when rats respond for 25% liquid-sucrose reinforcers (Weatherly, Davis, et al., 2000; Weatherly, Stout, et al., 2000), which maintain relatively high rates of responding, is consistent with such a suggestion. This potential limitation in generality, however, does not necessarily diminish the potential importance of the induction. That is although the induction may be limited to situations in which there is little or weak reinforcement, it could be argued that these are exactly the situations that most interest psychologists and behavior analysts.

References

(1.) This result can be considered induction for two reasons. First, increasing the upcoming rate of food-pellet reinforcement represented an increase in the conditions of reinforcement. Second, food-pellet reinforcement, in itself, represented an increase in the conditions of reinforcement compared to low-concentration (i.e., 1% or 5%) liquid-sucrose reinforcement. This latter claim can be supported by several pieces of evidence. Food pellets maintain higher rates of operant responding at the same rate of reinforcement than do low-concentration sucrose reinforcers (e.g., see Weatherly et al., 1999). On a concurrent schedule of reinforcement in which both food-pellet and liquid-sucrose reinforcers are available at an equal rate, rats proportion more responses for food pellets than for sucrose (unpublished research from our laboratory). Finally, negative successive contrast (see Flaherty, 1996) is observed when subjects are switched from food-pellet to liquid-sucrose reinforcement (Weatherly & Moulton, 2001).

FLAHERTY, C. F. (1996). Incentive relativity. New York: Cambridge University Press.

GAMZU, E., & SCHWARTZ, B. (1973). The maintenance of key pecking by stimulus-contingent and response-independent food presentation. Journal of the Experimental Analysis of Behavior, 5, 65-72.

KILLEEN, P. R. (1995). Economics, ecologics, and mechanics: The dynamics of responding under conditions of varying motivation. Journal of the Experimental Analysis of Behavior, 64, 405-431.

MCSWEENEY, F. K., & NORMAN, W. D. (1979). Defining behavioral contrast for multiple schedules. Journal of the Experimental Analysis of Behavior, 32, 457-461.

MCSWEENEY, F. K., & WEATHERLY, J. N. (1998). Habituation to the reinforcer may contribute to multiple-schedule behavioral contrast. Journal of the Experimental Analysis of Behavior, 69,199-221.

REYNOLDS, G. S. (1961). Behavioral contrast. Journal of the Experimental Analysis of Behavior, 4, 57-71.

WASSERMAN, E. A., FRANKLIN, S. R., & HEARST, E. (1974). Pavlovian appetitive contingencies and approach vs. withdrawal to conditioned stimuli in pigeons. Journal of Comparative and Physiological Psychology, 86, 616-627.

WEATHERLY, J. N., DAVIS, C. S., & MELVILLE, C. L. (2000). Induction with upcoming food-pellet reinforcement. Learning and Motivation, 31, 180-199.

WEATHERLY, J. N., & MOULTON, P. L. (2001). The effect of food-pellet reinforcement on rats' rates of lever pressing for 1% sucrose reinforcers across several "contrast" procedures. Learning and Motivation, 32, 193-218.

WEATHERLY, J. N., STOUT J. E., MCMURRY, A. S., RUE, H. C., & MELVILLE, C. L. (1999). Within-session responding when different reinforcers are delivered in each half of the session. Behavioural Processes, 46, 227-243.

WEATHERLY, J. N., STOUT, J. E., RUE, H. C., & MELVILLE, C. L., (2000). The effect of second-half reinforcer type on responding for sucrose in the first half of the session. Behavioural Processes, 49, 43-60.

WILLIAMS, B. A. (1991). Behavioral contrast and reinforcer value. Animal Learning & Behavior, 19, 337-344.

[Graph omitted]

[Graph omitted]

[Graph omitted]