Article Question review

ORIGINAL ARTICLE

The Effects of Punishment on Resurgence in Laboratory Rats

Katie Kestner & Ryan Redner & Erin E. Watkins & Alan Poling

Published online: 3 December 2014 # Association for Behavior Analysis International 2014

Abstract The resurgence of a previously extinguished re- sponse can be an undesired result when implementing behav- ioral interventions in applied settings. The present study evalu- ated in rats a potential strategy for reducing resurgence, arrang- ing punishment as well as extinction for the initially reinforced response. Ten rats pressed a lever (the B1 response) that was initially established with food reinforcement, then extinguished. For half of the rats, electric shock punishment was also arranged for the B1 response. Subsequently, nose poking (B2) was established with food reinforcement, then extinguished. During extinction of B2, substantial levels of B1 responses, that is, of resurgence, were evident in four of five rats not exposed to punishment of the B2 response. Resurgence was evident in only one of five rats exposed to punishment. These results suggest that arranging concurrent punishment and extinction of inappro- priate responding merits attention as a technique for reducing resurgence of that response in applied settings, although this strategy would be viable only if an innocuous and socially acceptable form of punishment proved effective.

Keywords Resurgence . Punishment . Extinction . Operant behavior . Lever-press response . Nose-poke response . Rats

The Effects of Punishment on Resurgence

Extinction-induced resurgence, commonly termed “resur- gence,” is the reemergence of a previously extinguished

response when an alternative, more recently reinforced, re- sponse is no longer effective (i.e., the alternative response is put on extinction) (Cleland et al. 2001). Resurgence is typi- cally demonstrated using a three-phase experimental design. During the first phase, a response (B1) is established and reinforced (e.g., a left-lever press is reinforced under a variable-interval [VI] 30-s schedule). In phase two the initial response (B1) is extinguished, and an alternative response (B2) is then reinforced (e.g., extinction is arranged for left-lever responses and right-lever responses are reinforced under a VI 30-s schedule. The third phase is a resurgence test in which extinction is arranged for both responses (e.g., for left-lever and right-lever presses). Resurgence is evident if the level of B1 responding increases substantially during the resurgence test relative to its level at the end of the second phase.

When a previously reinforced response is no longer an effective means for gaining reinforcement, it may be advanta- geous to return to responses that were reinforced in the more distant past. As Lieving and Lattal (2003) pointed out, resur- gence sometimes is an adaptive process leading to novel be- havior and contributing to creativity and problem solving. Unfortunately, resurgence of inappropriate behavior is also possible (e.g., Goh and Iwata 1994). For example, consider the differential reinforcement of an alternative behavior (DRA) procedure, which is a common and effective intervention for reducing undesired behavior (Petscher et al. 2009). In this procedure, the environment is manipulated so that the unde- sired behavior is no longer reinforced and a more appropriate alternative response is trained (if needed) and reinforced.

The process of implementing a DRA procedure mirrors the first two phases of resurgence studies. The undesired behavior can be considered B1; it is reinforced prior to the implemen- tation of the DRA intervention. Reinforcement is then ar- ranged for an alternative, appropriate response (B2), and B1 is no longer reinforced. If the intervention is then for some reason terminated (e.g., because of a lapse in treatment

K. Kestner: E. E. Watkins: A. Poling (*) Department of Psychology, Western Michigan University, Kalamazoo, MI, USA e-mail: [email protected]

Present Address: R. Redner University of Vermont, Burlington, VT, USA

Psychol Rec (2015) 65:315–321 DOI 10.1007/s40732-014-0107-y

integrity), there is an opportunity for resurgence of the target behavior (i.e., B1). Such an outcome is evident in a study by Volkert et al. (2009), in which individuals diagnosed with a developmental disability first experienced Functional Communication Training (FCT) to decrease undesired behav- ior. FCT can be viewed as a variant of DRA. During the FCT phase, B1 (problem behavior) no longer produced reinforce- ment; instead, an alternative communication response (B2) resulted in reinforcement. When both B1 and B2 were put on extinction, for two of three participants B1 resurged relative to levels observed in the previous condition. In another analysis, resurgence was examined when B2 was reinforced under a thin schedule, which rarely arranged reinforcers, rather than extinguished. Each of the three participants exhibited resur- gence under the thin reinforcement schedule. This finding is particularly troubling because reinforcement schedules are typically thinned when using DRA and FCT procedures in order to make the intervention more manageable for the behavior change agents (Hanley et al. 2001).

The extent to which treatment effects withstand lapses in treatment integrity and other challenges is arguably an impor- tant consideration in assessing the relative merits of various intervention options (Wacker et al. 2011). As a result of the possible resurgence of undesired behaviors in applied settings, researchers have begun to investigate techniques for minimiz- ing the resurgence of unwanted behavior and increasing the persistence of desired behavior. One proposed strategy to minimize resurgence is to train the appropriate response in a different context from that in which the targeted problem behavior has a history of reinforcement. Mace and colleagues (2010) successfully used this strategy in a basic research study with rats and then in an applied setting with humans. Both rat and human participants were trained on the alternative response (B2) in a different room than the room in which the problem behavior (B1) had pre- viously been reinforced. This procedure prevented resur- gence from occurring.

Another possible strategy for reducing the resurgence of problem behavior is to simultaneously arrange punishment and extinction for such behavior. Hagopian et al. (1998) analyzed 21 inpatient cases of FCT with and without extinc- tion, and with and without punishment. FCT without extinc- tion was rarely effective in decreasing problem behavior, and adding the extinction component was effective for about half of the cases. Unfortunately, when a schedule-fading procedure was implemented (in the form of increasing delay to rein- forcement), the procedure became ineffective for almost half of the cases. Adding a punishment procedure effectively sup- pressed the undesired behavior in all cases in which it was applied. These studies did not, however, examine levels of undesired behavior after removal of the punishment procedure

and/or alternative reinforcement procedure. That is, they did not ascertain if punishment attenuated resurgence.

Although not intended to specifically examine the effects of punishment on resurgence, Rawson and Leitenberg (1973) conducted a relevant study with rats. Their between-subjects experiment used a multiple schedule to examine whether making a reinforced alternative response available increased the response-weakening effects of punishment combined with extinction and of extinction alone. Initially, in Phase 1, each of four groups of rats was exposed to a multiple schedule in which presses of one lever (B1) were reinforced with food under a VI schedule (described in the article as VI 30 min, an apparent error). In Phase 2, in one (the SD) component, B1 continued to be reinforced under the VI schedule. In the other (SΔ) component, for all four groups food was never delivered (i.e., extinction was in effect for B1 responses). For two groups in the SΔ component, B1 was punished by electric shocks arranged under a fixed-ratio 1 schedule. For one of the two groups exposed to extinction alone and one of the two groups exposed to both extinction and punishment, responses on a second lever (B2) during the S

Δ component were reinforced with food. Phase 3 essentially replicated Phase 2. In Phase 3, the B2 lever remained available, but responses on this lever were not reinforced (i.e., they were extinguished).

During Phase 2, the mean rate of B1 responding in the S Δ

component was lowest in the group exposed to punishment and extinction with an alternative (i.e., B2) reinforced re- sponse available. It was progressively higher in the group exposed to punishment and extinction with no reinforced response alternative, in the group exposed to extinction with a reinforced response alternative, and in the group exposed to extinction with no response alternative. Despite the clear behavioral effects of punishment in Phase 2, the presence or absence of punishment in Phase 2 did not significantly affect the mean rate of B1 responding in the S

Δ component during Phase 3. This finding demonstrates that the effects of punishing B1 were not enduring under conditions that in some regards resemble those used to engender resurgence. It is, however, the case that in Phase 3 of the Rawson and Leitenberg (1973) study B1 responses during the S

Δ compo- nent were reinforced during Phase 3. In addition, such re- sponses were reinforced throughout all three phases during the SD component. Neither of these features is characteristic of the procedural arrangements commonly used to study resurgence (Cleland et al. 2001; Lieving and Lattal 2003). It is unclear whether the absence of enduring effects of punishment on B1 responding in Phase 3 reported by Rawson and Leitenberg would be evident under conditions in which B1 was punished during Phase 2, when reinforcement was provided for B2, but not for B1. The present study arranged such conditions.

316 Psychol Rec (2015) 65:315–321

Method

Subjects and Apparatus

The subjects were 10 male Sprague-Dawley rats about seven- months-old when the study began. All rats had previous experi- ence with pressing levers for water reinforcers as subjects in an undergraduate learning course. They were individually housed on a 12-hour light/dark cycle with unlimited access to water. Each rat was maintained at 80 % of its free-feeding weight; postsession feeding was utilized as needed to maintain target weights.

Each of the four operant conditioning chambers used in the study was 29 cm long, 24 cm wide, and 21 cm high. Two retractable levers and a 5 cm×5 cm opening where 45 mg food pellets were dispensed were located on the front wall of each chamber. On the opposite wall there were two circular openings (2.25 cm diameter×2 cm deep) with infrared sensors to record nose-poke responses. A ceiling light provided general lighting; lights in each nose-poke opening and above each lever also were illuminated. All lights were off for 3 s following each reinforcer (food pellet) delivery. Each chamber was enclosed in a sound- attenuating box equipped with a ventilation fan and white noise generator to minimize outside sounds. The floors of the operant chambersconsistedofmetalbarsthroughwhichscrambledelectric shocks could be delivered. MED-PC© (St. Albans, VT) software arranged experimental events and recorded data. Experimental procedures were approved by the Western Michigan University Institutional Animal Care and Use Committee.

Procedure

The 10 rats were randomly assigned to two groups, each comprising five animals. The training procedures described below were used for both groups to first establish the sound of food pellet delivery as a conditioned reinforcer and to train lever pressing. Following training, the subjects in each group were exposed in sequence to three phases. Phases 1 and 3 were identical for the experimental and control groups; Phase 2 differed. Each rat was exposed to one experimental session each weekday, at about the same time every day.

Training Phase 1: Response-Independent Food Delivery The purpose of this phase was to establish the sound of pellet delivery as a reinforcer. During Phase 1 sessions, the levers were retracted and a food pellet was delivered every 30 s until 60 pellets were delivered. Each rat experienced this training proce- dure during a single session in each of two consecutive days.

Training Phase 2: Lever Pressing The purpose of Training Phase 2 was to establish responding on the B1 operanda. The position of the reinforced lever was counterbalanced across

rats. For each rat, responses on the operational lever were initially reinforced under a fixed-ratio (FR) 1 schedule, where each press produced food. The FR value was progressively increased to 2, 5, 10, and 15, with the value increased each time 15 reinforcers were delivered. Sessions were terminated after 60 food deliveries. This training phase continued for three consecutive days and all subjects successfully completed the FR 15 schedule by the end of the third session.

Phase 1: Reinforcement of Lever-Press B1 During Phase 1, lever presses (B1) were reinforced with the delivery of a food pellet under a VI 30-s schedule with a 3-s changeover delay. Individual interval values for each phase were calculated based on the procedure described by Fleshler and Hoffman (1962). There were no programmed consequences for presses on the other lever or for responding on either nose-poke operanda. Each session was terminated after 60 reinforcers were delivered. This phase continued until a minimum of 15 sessions had been completed and stable responding was achieved. The stability criterion for Phase 1 and for Phase 2 (described below) was as described by Lieving and Lattal (2003). With this criterion, the mean response rate for the last six sessions was calculated, and the mean for the first three sessions and the mean for the last three sessions of this six-session sample could not differ more than 10 % from the overall mean. If either of the sub-means differed by more than 10 %, the condition was continued for additional sessions until the stability requirement was met.

Training Phase 3: Nose-Poke Training For Training Phase 3, three sessions were conducted in order to train the nose-poke response (B2), which served as the alternative behavior. Nose- pokes in the left hole were designated as B2 for half the rats, selected at random, and nose pokes in the right hole were designated as B2 for the others. One food pellet was placed inside the circular hole at the beginning of each training session. Initially, nose-pokes were reinforced under an FR 1 schedule. The schedule was gradually increased to an FR 15 in the same manner as in the lever-press training sessions. he levers were retracted during this phase, and there were no programmed consequences for nose-pokes in the inactive hole.

Phase 2: DRA (Control Group) or DRA + Punishment (Ex- perimental Group) During Phase 2, the lever-press response (B1) no longer produced food for rats in either the experimen- tal or the control group. For the experimental group, each B1 response immediately produced a 0.5-s 0.6-mA foot shock. There were no programmed consequences for B1 responses by rats in the control group. Food reinforcers were delivered under a VI 30-sschedule for the left nose-poke responses (B2) for both groups. There was a 3-s changeover delay for responses on B1 and B2. This phase continued for at least 15

Psychol Rec (2015) 65:315–321 317

sessions, until stable responding for B2 and a rate of less than three responses per min on B1 was achieved for three consec- utive sessions. The changeover delay was temporarily in- creased to 10 s for subjects not showing significant decreases in responding on B1, because this pattern suggests that a behavior chain of pressing B1, then emitting an operative nose-poke, was established and adventitiously reinforced (Lieving and Lattal 2003). Subjects remained on the 10-s changeover delay until they responded at a rate lower than three responses per min or for six sessions, whichever oc- curred first. All subjects were then returned to the 3-s change- over delay for a minimum of three sessions before moving on to Phase 3, in order to ensure they met extinction and stability criteria. The changeover delay was extended for subjects 10 and 12.

Phase 3: Resurgence Test During Phase 3, all programmed consequences for B2 were discontinued. Thus, there were no programmed consequences for B1 or B2 or the inactive lever and nose-poke operanda during this phase. This phase lasted for five sessions.

Results

During Phase 1, all of the rats acquired the lever-press (B1) response and met the stability criteria to move on to Phase 2 after 15 sessions. During Phase 2, all subjects acquired the nose-poke (B2) response, and a decreasing trend in B1 responding was seen for all subjects. There were substantial differences in the patterns of responding between the experimental- and control-group rats. Responding on B1 by rats in the experimental group (with the punishment contin- gency on B1) dropped to near-zero levels within the first few sessions of Phase 2, and very little B1 responding occurred in subsequent sessions. Rats in the control group took longer for B1 responding to substantially weaken and engaged in more B1 responding throughout Phase 2.

Figures 1 and 2 show B1 and B2 responding for rats in the control and experimental groups, respectively. Four of the five subjects in the control group showed an elevation of B1 responding in Phase 3 relative to the end of Phase 2, demon- strating a resurgence effect. Rat 11 did not show an apprecia- ble increase in B1 responding. With the exception of Rat 16, all rats in the experimental group showed a lack of resurgence (see Fig. 2). Rat 16 demonstrated an increased rate of responding on B1 during the first session of Phase 3. Examining the rate of B2 responding of rats in the two groups across conditions provides a means of accessing possible nonselective effects of punishment. These data indicate simi- lar patterns of B2 responding for rats in the two groups.

Discussion

In the present study, resurgence was evident in all but one rat in the control group and in only one rat in the experimental group. These results suggest that punishing B1 during Phase 2, when that response was also being extinguished, substantially reduced the probability that the response would resurge in Phase 3.

It is noteworthy that rats in the experimental groups emitted fewer B1 responses in Phase 2 than rats in the control group, which demonstrated that the putative pun- ishment procedure was actually punishing. It is also note- worthy that this effect is consistent with results reported in applied settings. As noted previously, Hagopian et al. (1998) analyzed 21 inpatient cases of FCT with and without extinction, and with and without punishment. FCT with extinction was more effective than FCT in reducing undesired behavior, and FCT with both extinction and punishment was most effective. They did not, howev- er, report whether undesired behavior resurged when FCT and other intervention components (i.e., extinction, punish- ment) were terminated, probably because relevant data were not reported in the articles they reviewed. Although the present data suggest that the lowest levels of resur- gence would occur in individuals exposed to FCT with both extinction and punishment, this outcome is not fore- gone. One rat exposed to punishment in the present study did exhibit evidence of resurgence, although the effect was relatively weak, and one rat not exposed to punishment did not exhibit resurgence. In both humans and other animals, the extent to which responding resurges often varies substantially across individuals exposed to the same experimental conditions (e.g., Doughty and Oken 2008; Lieving and Lattal 2003), and the variables responsible for this difference are not well understood.

The present findings differ from those previously reported by Rawson and Leitenberg (1973), who did not observe a lasting suppression effect of punishment using a multiple schedule discrimination procedure. As noted previously, there are several procedural differences in the studies that might account for their different results. One is that shock intensity was considerably higher in the present study (0.25 m) than in the Rawson and Leitenberg (0.06 mA) study. A second is that B1 responses were reinforced in one component of a multiple schedule in the Rawson and Leitenberg study, but not in the present one. A third is that the topography of the B1 and B2 responses (both lever presses) were the same in that study, but different (lever presses and nose pokes) in ours. A fourth is that Phase 3 of the Rawson and Leitenberg study, when resurgence would be evident, was substantially longer than Phase 3 of the present study. Given that resurgence sometimes is evident only after several sessions (e.g., Concado and Lattal 2013), Phase 3 in the present study may have underestimated

318 Psychol Rec (2015) 65:315–321

resurgence. A fifth procedural difference, and arguably the most important one, is that Rawson and Leitenberg arranged reinforcement for B1 in Phase 3 whereas we, as is the con- vention in studies of resurgence (Cleland et al. 2001; Lieving and Lattal 2003), did not do so. Once B1 produced food in Phase 3 of the Rawson and Leitenberg study, such responses could be strengthened by reinforcement and would not be an appropriate measure of resurgence. Their data are presented in blocks of three sessions, each 21 min in duration, and it is impossible to ascertain the number of B1 responses emitted by

rats in each group prior to reinforcement of B1 in Phase 3, which is necessary to provide an uncontaminated measure of resurgence. Therefore, it is inappropriate to view their findings as directly relevant to the effects of punishment on resurgence. Be that as it may, further research is obviously needed to ascertain the range of conditions under which punishment of B1 responses attenuates their subsequent resurgence and the behavioral mechanisms responsible for this effect.

Nonetheless, the present data suggest that punishing B1 responding is likely to be an effective way to reduce the

R e s p o n s e s /M

in u te

Sessions

Fig. 1 Responses per minute on B1 (filled circles) and B2 (open circles) across all three phases for control subjects

Psychol Rec (2015) 65:315–321 319

likelihood that such responding will subsequently resurge under at least some conditions. This outcome was not due to a general suppressive effect of exposure to punishment, be- cause rates of B2 responding were similar in rats that were and were not exposed to faradic punishment. Faradic punish- ment was examined in the present study for the sake of convenience and shock intensity was relatively high, be- cause we wanted to produce rapid and strong response suppression. Faradic punishment is almost never used with humans, but other forms of punishment, such as mild

verbal reprimands, response cost, and timeout are some- times used with humans and an interesting goal for future research is to ascertain whether they are effective in attenuating resurgence. Some relevant research could be done with nonhuman subjects, but ultimately it is critically important to determine whether ethical and practical pun- ishment procedures reduce the resurgence of clinically significant behaviors in actual applied settings. It is our hope that the data reported here encourage other re- searchers to conduct such studies.

Sessions

R e s p o n s e /M

in u te

Fig. 2 Responses per minute on B1 (filled circles) and B2 (open circles) across all three phases for experimental subjects

320 Psychol Rec (2015) 65:315–321

References

Cleland, B. S., Guerin, B., Foster, T. M., & Temple, W. (2001). Resurgence. The Behavior Analyst, 24, 255–260.

Concado, C. R., & Lattal, K. A. (2013). Response elimination, reinforce- ment rate and resurgence of operant behavior. Behavioural Processes, 100, 91–102.

Doughty, A. H., & Oken, G. (2008). Extinction-inducted resurgence: A selective review. Behavior Analysis Today, 9, 27–33

Leshler, M., & Hoffman, H. S. (1962). A progression for generating variable-interval schedules. Journal of the Experimental Analysis of Behavior, 5, 529–530.

Goh, H. L., & Iwata, B. A. (1994). Behavioral persistence and variability during extinction of self-injury maintained by escape. Journal of Applied Behavior Analysis, 27, 173–174.

Hagopian, L. P., Fisher, W. W., Thibault Sullivan, M., Acquisto, J., & LeBlanc, L. A. (1998). Effectiveness of functional communication training with and without extinction and punishment: a summary of 21 inpatient cases. Journal of Applied Behavior Analysis, 31, 211–235.

Hanley, G. P., Iwata, B. A., & Thompson, R. H. (2001). Reinforcement schedule thinning following treatment with functional communication training. Journal of Applied Behavior Analysis, 34, 17–38.

Lieving, G. A., & Lattal, K. A. (2003). Recency, repeatability, and reinforcer retrenchment: an experimental analysis of resurgence. Journal of Experimental Analysis of Behavior, 60, 217–233.

Mace, C. F., McComas, J. J., Mauro, B. C., Progar, P. R., Taylor, B., Ervin, R., & Zangrillo, A. N. (2010). Differential reinforcement of alternative behavior increases resistance to extinction: clinical demonstration, animal modeling, and clinical test of one solution. Journal of the Experimental Analysis of Behavior, 93, 349–367.

Petscher, E. S., Rey, C., & Bailey, J. A. (2009). A review of empirical support for differential reinforcement of alternative behavior. Research in Developmental Disabilities, 30, 409–425.

Rawson, R. A., & Leitenberg, H. (1973). Reinforced alternative behavior during punishment and extinction with rats. Journal of Comparative and Physiological Psychology, 85, 593–600.

Volkert, V. M., Lerman, D. C., Call, N. A., & Trosclair-Lasserre, N. (2009). An evaluation of resurgence during treatment with function- al communication training. Journal of Applied Behavior Analysis, 42, 145–160.

Wacker, D. P., Harding, J. W., Berg, W. K., Lee, J. F., Schieltz, K. M., Padilla, Y. C., Nevin, J. A., & Shahan, T. A. (2011). An evaluation of persistence of treatment effects during long-term treatment of de- structive behavior. Journal of the Experimental Analysis of Behavior, 96, 261–282.

Psychol Rec (2015) 65:315–321 321

Copyright of Psychological Record is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.