Generality of persistence: The role of stimulus and response factors in persistence to punishment

Previous research has shown that intermittent punishment of a response increases its persistence to continuous punishment and that intermittent punishment training in one situation produces persistence to continuous punishment in other situations. Experiment 1 showed that as long as the instrumental response and punishing stimulus were held constant from intermittent to continuous punishment marked differences between these situations had no decremental effect on persistence. Experiment 2 showed that intermittent punishment training of one response resulted in substantial persistence to continuous punishment of a different and apparently incompatible response and that such response change had no more than a marginal effect on persistence. The results were seen as requiring some revision to the traditional conditioning-model interpretation of persistence to punishment.     

Fixed-ratio punishment

Responses were maintained by a variable-interval schedule of food reinforcement. At the same time, punishment was delivered following every nth response (fixed-ratio punishment). The introduction of fixed-ratio punishment produced an initial phase during which the emission of responses was positively accelerated between punishments. Eventually, the degree of positive acceleration was reduced and a uniform but reduced rate of responding emerged. Large changes in the over-all level of responding were produced by the intensity of punishment, the value of the punishment ratio, and the level of food deprivation. The uniformity of response rate between punishments was invariant in spite of these changes in over-all rate and contrary to some plausible a priori theoretical considerations. Fixed-ratio punishment also produced phenomena previously observed under continuous punishment: warm-up effect and a compensatory increase. This type of intermittent punishment produced less rapid and less complete suppression than did continuous punishment.     

Transfer of approach responding between punishment and frustrative nonreward sustained through continuous reinforcement

A block of continuously reinforced nonpunished trials was interpolated between acquisition of a runway response with either partial reinforcement or intermittent punishment and subsequent tests for resistance of that response to the suppressive effects of either extinction or continuous punishment. As previous investigations have shown, both the partial reinforcement effect (PRE) and the intermittent punishment effect (IPE) were sustained through the block of continuously reinforced nonpunished trials. Furthermore, the increased resistance to extinction following intermittent punishment and the increased resistance to punishment following partial reinforcement were also sustained through the interpolated continuous reinforcement. These results support a hypothesized similarity of punishment and frustrative nonreward and were interpreted in an extension of Amsel's conditioning model theory of the role of nonreinforcement in the PRE.     

Transfer of approach responding between punishment, frustrative nonreward, and the combination of punishment and nonreward

Rats trained to make an approach response with either partial reward, intermittent punishment, or a combination of partial reward and intermittent punishment, were tested for persistence to extinction, punishment with reward, or punishment during extinction. Partial reward, alone or with punishment, produced greates resistance to extinction, while intermittent punishment, alone or with partial reward, produced greatest persistence to punishment with reward. Transfer of persistence from partial reward to punishment with reward and intermittent punishment to extinction was also demonstrated. However, partial reward alone did not increase persistence to punishment during extinction, whereas intermittent punishment and partial reward combined with intermittent punishment did increase such persistence. These results were interpreted in Amsel's (1958, 1962) conditioning-model theory by extending the hypothesized similarity of frustrative nonreward and punishment.     

Fixed-interval punishment

The average rate of bar-pressing maintained by a variable-interval schedule of milk reinforcement in 33 rats was found to be a decreasing function of intensity of concurrent punishment and, over a wide range of shock intensities, was inversely related to punishment frequency. Cumulative records were, however, negatively accelerated during 30-min punishment sessions with complete suppression occurring earlier and earlier (after fewer and fewer shocks) as intensity increased. In addition, acceleration was often observed between successive fixed-interval shock presentations and, at low and moderate intensities, bursts of responding occurred after each shock. The time to recover between punishment sessions (post-punishment recovery) was an increasing monotonic function of punishment intensity.     

Motivational aspects of escape from punishment

Punishment and escape were studied simultaneously by allowing a subject to escape from a stimulus situation in which responses were punished, into a stimulus situation in which responses were not punished. The frequency of the punished responses was found to be an inverse function of the intensity of punishment, whereas the frequency of the escape response was a direct function of the intensity of punishment. Both of these functions were obtained under three different schedules of food reinforcement. The strength of the escape behavior was evidenced by (1) the emergence of the escape response even when the frequency of food reinforcement decreased as a consequence of the escape response, (2) the maintenance of the escape response by fixed-interval and fixed-ratio schedules of escape reinforcement, and (3) the occurrence of escape responses at intensities of punishment that otherwise produced only mild suppression of the punished response when no escape was possible. This last finding indicates that a subject may be driven out of a situation involving punishment even though the punishment is relatively ineffective in suppressing the punished responses when no escape is possible.     

Multiple punishment schedule

Concurrent schedules of punishment and fixed-ratio reinforcement were studied in six white Carneaux pigeons. Two stimuli were alternated on the translucent key which served as the response mechanism. The pigeons were exposed to continuous punishment in one stimulus and first response punishment in the other. It was found that continuous punishment suppressed the fixed-ratio performance more than did punishment of the first response. Typically, a ratio run of responses was completed in spite of the onset of continuous punishment. Responding was then reduced for the remainder of the continuous punishment stimulus, except for those times when responding was initiated near the end of the stimulus. The results indicate that a multiple schedule can be used to compare different schedules of punishment in the same organism.     

An interresponse time analysis of variable-ratio punishment

An interresponse time analysis was used to study the effects of variable-ratio punishment schedules on the temporal pattern of reinforced responding. Twelve pigeons responded on a baseline variable-interval schedule of food reinforcement. A variable-ratio ten schedule of electric shock punishment was then introduced. The shock intensity was systematically increased to the highest intensity at which responding could be maintained. At this intensity, the mean variable-ratio value was increased and then decreased. Variable-ratio punishment resulted in an increased relative frequency of very short unreinforced interresponse times (response bursting). Increased response bursting accounted for instances of response rate facilitation. In addition, shock was followed by interresponse times of decreasing mean length over the first several responses after shock.     

Continuous punishment of free-operant avoidance in the rat

Three groups of albino rats were trained under a free-operant avoidance (Sidman) procedure with equal shock-shock and response-shock intervals. After stable performance was achieved, the animals were concurrently exposed to a brief electric shock after each response. The procedures were as follows: Punishment Schedule I: punishment shock was introduced at an intensity approximately one quarter that of avoidance shock; increments of nearly this same size were made as stable performance was achieved at succeeding punishment shock intensities. Punishment Schedule II: punishment shock was introduced at approximately one-half the intensity of avoidance shock; after stable performance, punishment shock was increased to the same intensity as avoidance shock. Punishment Schedule III: punishment shock was introduced and maintained at the same intensity as avoidance shock. Punishment was continued for all groups until one of two suppression criteria was attained. All animals made fewer responses and received more avoidance shocks as a function of increasing punishment shock. Half of the animals under Punishment Schedule I required punishment shock higher than avoidance shock to meet their assigned suppression criterion. A comparison of all procedures showed that suppression was greater when punishment shock was initially at high intensity.     

A measure of intensity of parental punishment

To test the possibilities that antecedents and consequents of certain levels of intensity of punishment may have far-reaching effects on behavior, a self-report measure of parental intensity of punishment was developed. Consisting of 33 different situations to which parents react, scores from the instrument were found to be reliable over time. Further construct validity was demonstrated with a variety of predicted relationships between punishment intensity and parent and child characteristics. Five samples of subjects were used with over 400 families, high intensity of punishment scores were found to be related to mothers' lack of warmth and to her intrusiveness when playing with her child; children's maladjusted behavior; aggression and impulsivity in boys; withdrawal and inhibition in girls, and uncooperativeness. Inconsistency of parental punishment also was associated with maladjustment. Results were discussed in terms of Rotter's social learning theory.     

The effects of punishment intensity on squirrel monkeys

Responses of squirrel monkeys were maintained by a variable-interval schedule of food reinforcement. Concurrently, punishment consisting of a brief electric shock followed each response. As has been found for pigeons and rats, punishment did not produce extreme, all-or-none reactions. By gradually increasing the punishment intensity it was possible to produce response rates intermediate to no suppression and complete suppression. Similarly, the moment-to-moment response rate was free of extreme fluctuations. A “warm-up” effect occurred in which the punished responses were especially suppressed during the initial part of a session. The pre-punishment performance was negatively accelerated within a session, and punishment reduced the degree of negative acceleration. When punishment was discontinued, responding recovered immediately except when suppression had been complete or prolonged. When the punishment intensity was decreased gradually, more suppression resulted at a given intensity than when intensity was increased gradually. This suggests a “behavioral inertia” effect wherein behavior at a new punishment intensity is biased toward the behavior at the previous value. A corollary generalization is that the larger the change in intensity, the less the behavior at the new value will be biased toward the behavior at the previous value.     

Variable-interval punishment during variable-interval reinforcement

Variable-interval punishment was superimposed on a variable-interval food reinforcement baseline for three groups of rats. The value for the variable-interval punishment schedule was the same as that for the variable-interval food reinforcement schedule, although the two schedules were programmed independently. The three groups were Variable Intervals 0.5, 1, and 3 min. Little or no suppression occurred in the three groups at mild (0.2 and 0.4 ma) intensities of punishment, but at 0.6 ma, complete suppression occurred almost uniformly. During continued punishment, there was no consistent recovery toward the pre-punishment baseline at suppressing intensities of punishment. After punishment was discontinued, recovery from suppression was more rapid the lower the punishment intensity, and the lower the value of the variable-interval schedules of reinforcement and punishment.     

Some effects of punishment shock intensity upon discriminative responding

Three pigeons received visual discrimination training under both multiple variable-ratio extinction and variable-interval extinction schedules. All birds developed nearly perfect discrimination. When punishment for every tenth response during food reinforcement was presented, responding decreased as shock intensity increased. At the same time, responding during extinction, which was not punished, increased at intermediate punishment intensities, but returned to low levels under severe punishment. A second procedure, in which punishment and no-punishment sessions alternated unsystematically, was employed with two of the birds. The results under this procedure essentially replicated the data obtained as punishment shock intensity increased gradually.     

Discriminated punishment: avoidable and unavoidable shock

Warning stimuli for two punishment conditions were alternated with periods of appetitive responding by rats. In either warning stimulus, the first response produced a brief shock, terminated the stimulus, and started an interval during which the baseline appetitive schedule was in effect. Not responding resulted in stimuli of random duration, which terminated with a shock under one condition and without a shock under the other. Each subject was exposed to several shock intensities, with trials for the two conditions programmed during alternate portions of the session. In general, response frequency in the warning signal for either condition decreased with increasing intensity; however, at a given intensity, responding was more frequent in the stimulus invariably terminating with shock than in the stimulus terminating without shock when no response was made. The frequency difference was greatest at intensities intermediate between those producing minimal and maximal suppression.     

Chlordiazepoxide and resistance to punishment

Rats were given intermittent electric foot-shock during food-rewarded alley training. In the test phase, food and shock were given on every trial. These animals persisted in running down the alley in the test phase compared to those without prior shock experience. The effects of chlordiazepoxide (CDP) on this learned resistance to punishment were examined using a long and short interval between trials. It was found that CDP abolished the effect at a long inter-trial interval, but left it unaltered if the interval was short. The results match those found previously with an analogous effect using non-reward. It is suggested that the effects of punishment and non-reward may be mediated by a common process, and that the benzodiazepines may act on this process.     

EFFECTS OF INTERMITTENT PUNISHMENT ON SELF-INJURIOUS BEHAVIOR: AN EVALUATION OF SCHEDULE THINNING

Although the use of punishment often raises ethical issues, such procedures may be needed when the reinforcers that maintain behavior cannot be identified or controlled, or when competing reinforcers cannot be found. Results of several studies on the effects of intermittent schedules of punishment suggest that therapists must use fairly rich schedules of punishment to suppress problem behavior. However, residential caretakers, teachers, and parents often have difficulty implementing programs that require constant monitoring of the client's behavior. In this study, we examined the feasibility of gradually thinning the delivery of punishment from a continuous schedule to an intermittent schedule during the course of treatment for self-injurious behavior (SIB). Results of functional analyses for 5 individuals who had been diagnosed with profound mental retardation indicated that their SIB was not maintained by social consequences. Treatment with continuous schedules of time-out (for 1 participant) or contingent restraint (for the other 4 participants) produced substantial reductions in SIB. When they were exposed to intermittent schedules of punishment (fixed-interval [FI] 120 s or FI 300 s), SIB for all but 1 of the participants increased to levels similar to those observed during baseline. For these 4 participants, the schedule of punishment was gradually thinned from continuous to FI 120 s or FI 300 s. For 2 participants, SIB remained low across the schedule changes, demonstrating the utility of thinning from continuous to intermittent schedules of punishment. Results for the other 2 participants showed that intermittent punishment was ineffective, despite repeated attempts to thin the schedule.     

The effect of punishment shock intensity upon responding under multiple schedules

In the first of two experiments, responses of two pigeons were maintained by multiple variable-interval, variable-ratio schedules of food reinforcement. Concurrent punishment was introduced, which consisted of a brief electric shock after each tenth response. The initial punishment intensities had no lasting effect upon responding. Then, as shock intensity increased, variable-ratio response rates were suppressed more quickly than variable-interval response rates. When shock intensity decreased, variable-interval responding recovered more quickly, but the rates under both schedules eventually returned to their pre-punishment levels. In the second experiment, the following conditions were studied in three additional pigeons: (1) With each shock intensity in effect for a number of sessions, punishment shock intensity was gradually increased and decreased and responding was maintained by multiple variable-ratio, fixed-ratio schedules of food reinforcement; (2) Changes in punishment shock intensity as described above with responding maintained by either a variable-ratio or a fixed-ratio schedule, which were presented on alternate days; (3) Session-to-session changes in shock intensity with responding maintained by multiple variable-ratio, fixed-ratio schedules. Responding under the two schedules was suppressed to approximately the same extent by a particular shock intensity. Also, post-reinforcement pauses under the fixed-ratio schedule increased as response suppression increased.     

Timeout as a punishing stimulus in continuous and intermittent schedules

The effectiveness of a brief period of isolation (timeout) in the control of disruptive behavior emitted by a retarded child in a preschool classroom setting was examined. Timeout was shown to be an effective punishing stimulus, and its control of the child's disruptive behavior was investigated under four schedules of intermittent timeout. The results suggest that as a larger percentage of responses were punished, a greater decrease in the frequency of that response occurred. This inverse relationship between the percentage of responses punished and the frequency of the response did not appear to be linear, but rather a non-linear function. This function suggests that some schedules of intermittent punishment may be as effective as continuous punishment, at least in the case of the continued suppression of a response that has already been reduced to a low frequency.     

Reinforcement and punishment effects in concurrent schedules: A test of two models

The joint effects of punishment and reinforcement on the pigeon's key-peck response were examined in three choice experiments conducted to compare predictions of Farley and Fantino's (1978) subtractive model with those made by Deluty's (1976) and Deluty and Church's (1978) model of punishment. In Experiment 1, the addition of equal punishment schedules to both alternatives of a concurrent reinforcement schedule enhanced the preference exhibited for the more frequent reinforcement alternative. Experiment 2 demonstrated decreases in the absolute response rate for each member of a concurrent reinforcement schedule when increasing frequencies of punishment were added to each alternative. Experiment 3 found that preference for the denser of two reinforcement schedules diminished when the absolute frequencies of reinforcement were increased by a constant factor and conditions of punishment for both alternatives were held constant. Diminished preferences were obtained regardless of whether the frequency of punishment associated with the denser reinforcement schedule was greater or less than that associated with the lean reinforcement alternative. The results from all three experiments uniquely supported Farley and Fantino's (1978) subtractive model of punishment and reinforcement.     

The partial punishment effect following minimal acquisition training: Sodium amobarbital and the stimulus properties of early punished trials

In a runway investigation, six groups of rats received limited runway training such that partial punishment, partial reinforcement, or continuous reinforcement was accompanied by sodium amobarbital or saline. Following an interpolated phase of continuous reinforcement without injections, all groups were given punished extinction. The entire experiment was conducted under widely spaced conditions (ITI = 24 hr). It was found that partial punishment increased resistance to punished extinction relative to partially and continuously reinforced controls when acquisition was given under saline. When partial punishment training was accompanied by amobarbital this effect was eliminated. The drug was observed to have no effect on the punished extinction performance of the partial reinforcement and continuous groups, respectively. Moreover, the partial reinforcement effect (PRE) did not generalize to punished extinction. These data provide information concerning the difference between the stimuli associated with the early trials of punishment and nonreward and indicate that the former but not the latter contain emotional elements.