Children who emit self-injurious behavior are rare but the counterintuitive nature of self-abuse and its obvious undesirability have generated much theoretical and applied interest. Self-abuse in nonhumans is regularly induced by chronic exposure to sympathomimetic agents, and commonly occurs in children suffering from physiological disorders such as the Lesch-Nyhan and de Lange syndromes, but most cases of human self-mutilation are not associated with specific physiological antecedents. Learning theorists speculate that much self-abuse is maintained as an operant response that serves to manipulate the child’s environment, although the behavior initially appears as an unlearned response to certain conditions. Many kinds of treatments have been used in attempts to control self-abuse, including physical restraints, drug administration, and various procedures involving reinforcement and punishment operations. The present paper critically discusses procedures used to manage self-abuse with respect to demonstrated efficacy, ease of implementation, and limitations. Direct comparisons between treatments are problematic, in part because literature in this area consists largely of single-case studies. However, punishment is surely one of the most effective techniques yet used to control self-abuse, yet legal and ethical considerations limit the use of this procedure to particularly severe and intractable cases, making selection of an appropriate initial treatment a difficult choice at best.

The effects of tripelennamine (3, 6, 12, 18, and 24 mg/kg) and pentazocine (5, 10, 20, 30, and 40 mg/kg), given alone and in selected combinations, were determined in rats performing under a fixed-ratio 30 schedule of food delivery. Each drug alone produced generally dosedependent decreases in response rates. Combinations typically produced effects identical in direction to, and occasionally greater in magnitude than, those predicted by a simple additive model.

The effects of clozapine and several other neuroleptic drugs were examined in rats responding under fixed-consecutive-number (FCN) schedules with minimum response requirements of 4 and 8. Under these schedules, rats were trained to respond either 8 or more times or 4 or more times on one lever, and then respond once on a second lever. In one component of these schedules, an external discriminative stimulus was presented following the completion of the response requirement on the first lever, whereas no stimulus change was programmed under the other. Under the FCN 8 schedule without the external discriminative stimulus, clozapine produced large dose-dependent decreases in accuracy (percent of reinforced response runs), whereas molindone produced small decreases in accuracy. Neither clozapine or molindone, however, altered accuracy under the FCN 4 without the external discriminative stimulus. Under these same schedules, loxapine, chlorpromazine, haloperidol and thioridazine produced small increases in accuracy at intermediate doses without affecting accuracy at the low and high doses. None of the neuroleptics evaluated produced accuracy-altering effects under the FCN schedules with the external discriminative stimulus. In general, all of these drugs decreased response rates in a dose-dependent fashion. The order of potency for the rate-decreasing effects of these drugs was loxapine greater than haloperidol greater than molindone greater than clozapine = chlorpromazine greater than thioridazine. Thus, the effects of clozapine on accuracy under the FCN schedules without the external discriminative stimulus differed qualitatively from those of other neuroleptic agents.

Under the fixed-consecutive-number schedule (FCN), pigeons were reinforced for responding eight or more times on one response key (work key), and then responding once on a second response key. In one component of this schedule, an external stimulus signalled the completion of the response requirement on the work key (FCN 8-SD), whereas no stimulus change was programmed under the other (FCN 8). Across a range of doses, the mu opioid agonist morphine, the kappa opioid agonist U50,488 and the opioid antagonist naloxone had no consistent effect on accuracy under either FCN schedule. Naloxone and U50,488 produced a general flattening of the conditional probability functions by decreasing the conditional probability of response runs exceeding the minimum response requirement of eight consecutive responses on the work key. The sigma agonists phencyclidine and (+)N-allylnormetazocine and the nonopioid analgesics clonidine and l-nantradol produced large decreases in accuracy under the FCN 8 and small decreases under the FCN 8-SD. With the exception of (+)N-allylnormetazocine, these drugs consistently increased the conditional probability of responses runs shorter than the minimum response requirement on the work key. These findings indicate that the accuracy-altering effects of some opioid and nonopioid analgesics depend in part on the type of discrimination task.

The effects of several opioid agonists and antagonists were examined in pigeons performing under a delayed matching-to-sample procedure. The mu agonists morphine and l-methadone, the kappa agonists U 50,488 and ethylketocyclazocine, and the opioid antagonist naloxone had no effect on the accuracy of responding. These drugs were, however, behaviorally active as evidenced by the dose-dependent decreases in rates of responding associated with their administration. In contrast, the sigma agonists (+) N-allylnormetazocine and phencyclidine decreased the accuracy of responding in a dose-dependent fashion. The relative magnitude of these drug-induced decreases in accuracy were similar across the no delay (0-s), short (2-s), and long (8-s) delay intervals. For these drugs, accuracy-decreasing effects were obtained only at doses that reduced rates of responding. The results of the present investigation parallel those reported in pigeons responding under drug discrimination tasks, in which the discriminative stimulus properties produced by the mu and kappa agonists are similar to each other but distinguishable from those produced by the sigma agonists.

Pigeons were trained to discriminate a dose of either 4.2 mg/kg of U50,488 or 1.0 mg/kg of morphine from water using a two-key drug discrimination procedure. In U50,488-trained pigeons, the kappa agonist bremazocine occasioned drug-appropriate responding during substitution tests, whereas ethylketocyclazocine and ketocyclazocine occasioned intermediate levels of drug-appropriate responding up to and including doses that markedly suppressed response rates. The mu agonists morphine, l-methadone and fentanyl produced responding predominantly on the water-appropriate key. In morphine-trained pigeons, l-methadone, fentanyl, ethylketocyclazocine and ketocyclazocine, but not U50,488 and bremazocine, occasioned drug-appropriate responding. Nonopioid compounds, such as d-amphetamine, pentobarbital, phencyclidine and (+)-SKF 10,047 produced responding predominantly on the water-appropriate key in both U50,488- and morphine-trained pigeons. During tests of antagonism, a 0.1 and 1.0 mg/kg dose of naloxone antagonized completely the discriminative stimulus properties of the training dose of U50,488 and morphine, respectively. In addition, morphine displayed a substantially longer duration of action than U50,488, in that intermediate levels of drug-appropriate responding were evident as long as 4 hr after the administration of morphine and only 1 hr after the administration of U50,488. Over a period of approximately 8 months, the dose-effect curves for the discriminative stimulus properties of both drugs were unchanged. The present findings illustrate further the unique behavioral response of pigeons to the discriminative stimulus properties of the kappa agonists, and establishes that pigeons can discriminate between mu and some kappa agonists.

Pigeons were exposed to schedules of food delivery that consisted of two sequential fixed ratios. When alternative sequences provided two food deliveries per 50 responses, the schedule with the shorter initial fixed-ratio value was consistently preferred. Progressively reducing from 1.0 to .25 the probability of food delivery following completion of the second fixed ratio of the sequence with the shorter initial fixed ratio did not reduce preference for this sequence. Moreover, the sequence with the shorter initial fixed ratio also was preferred when the probability of food delivery following completion of the initial ratio in that sequence was progressively reduced from 1.0 to .5, although preference shifted to the alternative when the probability was reduced to 0. These findings suggest that the length of the initial fixed ratio was a primary determinant of choice. Subsequent manipulations demonstrated, however, that when the initial fixed ratios of the two alternatives were equal, changes in the ratio value and probability of food delivery following completion of the second fixed ratio lawfully affected choice.

The effects of several opioid agonists and the opioid antagonist naloxone were examined in rats responding under a fixed-consecutive-number (FCN) schedule. Under this schedule, a reinforced response run consisted of responding eight or more times on one response lever, and then responding once on a second response lever. In one component of this schedule, an external discriminative stimulus signalled the completion of the response requirement on the first lever, whereas no stimulus change was programmed in the other. Morphine, 1-methadone, U50488, ketocyclazocine, phencyclidine, and (±)N-allylnormetazocine decreased the percent of reinforced response runs (accuracy) under the FCN schedule without the external discriminative stimulus, but had no effect under the FCN schedule with the external discriminative stimulus. Naloxone and bremazocine, in contrast, had no effect on the accuracy of the discrimination under either FCN schedule. With the exception of bremazocine and U50488, which increased rates of responding at low doses, all drugs produced comparable decreases in rates of responding under both FCN schedules. During tests of antagonism, a 0.1 mg/kg dose of naloxone reversed completely the accuracy-decreasing effects produced by U50488 and morphine. The rate-decreasing effects of morphine and U50488 were reversed completely by a 0.01 and 1.0 mg/kg dose of naloxone, respectively. These results suggest that the addition of an external discriminative stimulus can modulate the disruptive effects the opioids, and that mu, sigma and some kappa agonists produce similar effects when evaluated under the FCN schedules.

The present study examined the effects of phenobarbital (5, 10, 20, and 40 mg/kg), phenytoin (2.5, 5, 7.5, and 15 mg/kg), and valproic acid (40, 60, 80, and 120 mg/kg), and those of phenobarbital (10 and 20 mg/kg) in combination with phenytoin (2.5, 5, and 7.5 mg/kg) or valproic acid (40, 60, and 80 mg/kg), on the delayed-matching-to-sample performance of pigeons. In general, high doses of each individual drug reduced accuracy. Drug combinations also reduced accuracy relative to control values. Reductions in accuracy produced by drug combinations were very similar in magnitude to those predicted by a response-addition model of drug interaction.

The effects of the anticonvulsant drugs clonazepam and ethosuximide were examined in pigeons performing under a fixed-consecutive-number schedule with and without an added external discriminative stimulus. Under these schedules, food was delivered whenever subjects responded between and 8 and 12 times on one response key (work key), and then responded once on a second response key (reinforcement key). For one group, an external discriminative stimulus signalled completion of the response requirement on the work key, while no stimulus change was programmed for the other group. Clonazepam (0.06–0.75 mg/kg) produced dose-dependent decreases in percentage of reinforced runs and rate of responding for both groups. The magnitude of the accuracy-decreasing effect was generally greater in the group without the external discriminative stimulus. For this group, the higher doses of clonazepam produced pronounced increases in switching to the reinforcement key before completing the minimum requirement of eight consecutive responses on the work key. No consistent patterns of errors were evident for the subjects with the added external discriminative stimulus. Although ethosuximide (20–160 mg/kg) produced dose-dependent decreases in rate of responding, it had little effect on the percentage of reinforced runs or the run length distributions. These findings are consistent with previous reports indicating that clonazepam, but not ethosuximide, substantially disrupts performance under operant tasks requiring conditional discriminations. These data also suggest that the addition of an external discrimination stimulus attenuates the disruptive behavioral effects of clonazepam.

The effects of clonazepam (0.25–4.0 mg/kg) and phenobarbital (10–80 mg/kg) were examined in pigeons responding under a multiple fixed-ratio 50 fixed-interval 90-sec schedule of food delivery. When given acutely 30 min prior to testing, the lowest dose of each drug generally increased responding under both components of the multiple schedule, whereas high doses decreased responding. Time-course determinations revealed maximum rate-reducing effects when phenobarbital was given 30 min prior to testing; 15- and 30-min presession intervals were similarly effective with clonazepam. When a 120-min presession interval was used, responding generally exceeded control rates for clonazepam, but not for phenobarbital.

The behavioral effects of phenytoin, phenobarbital, clonazepam, valproic acid, and ethosuximide were evaluated in food-deprived performing under automaintenance and negative automaintenance procedures. Under the automaintenance procedure, brief periods of key illumination were followed by food delivery without regard to the subject's behavior. In most instances, when drugs were not given this procedure engendered high rates of keypecking during almost all key illuminations (trials). Acute administrations of phenytoin (10–20 mg/kg), valproic acid (40–120 mg/kg) produced generally dose-dependent decreases in percent trials with a response and rate of responding. Although phenobarbital (30–60 mg/kg) and clonazepam (2.5–7.5 mg/kg) produced little obvious effect on percent trials with a response, these drugs generally increased rate of responding. Under the negative automaintenance procedure, food delivery followed only key illuminations during which keypecking did not occur. Keypecking occurred at a low rate under this procedure, with no responses occurring during the majority of key illuminations. Thus, this procedure appeared to involve responding elicited by respondent conditioning but suppressed by the response-dependent omission of food. Across the same dose ranges evaluated under the automaintenance procedure, clonazepam and phenobarbital generally increased percent trials with a response and rate of responding in dose-dependent fashion. Phenytoin similarly increased percent trials with a response but had little consistent effect on rate of responding. Ethosuximide and valproic acid failed to affect responding under this procedure.

The effects of the anticonvulsant drugs valproic acid, phenytoin, phenobarbital and diazepam were examined in pigeons performing under a fixed-consecutive-number schedule with and without an added external discriminative stimulus. Under these schedules, a reinforced response run consisted of responding between eight and 12 times on one response key (work key) and then responding once on a second response key (reinforced key). For one group of pigeons, an external discriminative stimulus signaled completion of the response requirement on the work key, whereas no stimulus change was programmed for the other group. Phenobarbital (5-60 mg/kg) and diazepam (0.5-6 mg/kg) produced large decreases in reinforced response runs (accuracy) and rates of responding. The magnitude of these accuracy- and rate-decreasing effects was larger in the group without the external discriminative stimulus. Under both schedules, these drugs produced pronounced increases in the probability of switching to the reinforcement key before completion of the minimal response requirement on the work key. Valproic acid (20-160 mg/kg) and phenytoin (1.25-10 mg/kg) also decreased both accuracy and rates of responding. Although the accuracy-decreasing effects of these drugs were relatively small in magnitude, they were consistently larger in the group without the external discriminative stimulus. These data suggest that the addition of an external discriminative stimulus attenuates the disruptive behavioral effects of anticonvulsant drugs.

The acute and chronic effects of the antiepileptic drugs clonazepam (0.06, 0.13, and 0.25 mg/kg), phenytoin (2.5, 5, 7.5 mg/kg), ethosuximide (40, 80, and 120 mg/kg), and valproic acid (40, 80, and 120 mg/kg) were evaluated in pigeons responding under a repeated acquisition procedure. At certain doses, acute administrations of all drugs impaired learning (i.e., increased errors) and reduced rate of responding. Appreciable tolerance developed to these effects with chronic exposure, although the physiological mechanism responsible for this outcome is unknown.

The effects of phenobarbital, clonazepam, valproic acid, ethosuximide, and phenytoin were examined in pigeons performing under a delayed matching-to-sample procedure. Clonazepam, valproic acid, ethosuximide, and phenytoin typically reduced the rate of responding to the sample stimulus, whereas phenobarbital usually increased response rates at high doses. Phenobarbital, clonazepam, and valproic acid produced generally dose-dependent decreases in accuracy; ethosuximide and phenytoin failed to do so. These results suggest that there are qualitative as well as quantitative differences in the effects of anticonvulsant drugs under the delayed matching-to-sample procedure.

The present study examined the effects of phenytoin (20, 30, 40, and 50 mg/kg), phenobarbital (10, 20, 30, and 40 mg/kg), and valproic acid (80, 120, 160, and 240 mg/kg), and those of phenobarbital (10 and 30 mg/kg) combined with phenytoin (20, 30, and 40 mg/kg) or valproic acid (80, 120, and 160 mg/kg), on the lever pressing of rats maintained under fixed-ratio and interresponse-time-greater-than-t schedules of food delivery. High doses of each individual drug significantly decreased mean group response (and reinforcement) rate under the fixed-ratio schedule. No dose of an individual agent significantly affected mean group response rate under the interresponse-time-greater-than-t schedule, although high doses of phenobarbital and valproic acid significantly reduced the mean group reinforcement rate under this schedule. When given in combination, phenobarbital and phenytoin and phenobarbital and valproic acid significantly reduced response (and reinforcement) rate under the fixed-ratio schedule and reinforcement rate under the interresponse-time- greater-than-t schedule. These reductions did not significantly differ in magnitude from those predicted by an additive model of drug interaction.

Pigeons were exposed to autoshaping procedures under which 50% of red key illuminations were followed by 9-sec food deliveries, and 50% of blue key illuminations were followed by 3-sec food deliveries. When all key illuminations were 6 sec, pigeons preferred the red stimulus. Subsequent manipulations demonstrated that preference could be shifted to the blue stimulus by either increasing the duration of the red stimulus or imposing a delay interval between the offset of that stimulus and food delivery. A final experiment demonstrated that, in two of three subjects, preference for key illuminations associated with longer, but delayed, food deliveries generally increased as the duration of all key illuminations was lengthened. These results, obtained under conditions where keypecking had no programmed consequences, are similar to those previously observed under procedures involving a positive response-food dependency.