Donald A. Overton

State-Dependent Learning Produced by Depressant and Atropine-Like Drugs*

SummaryRats were trained to escape from shock in a T maze. In this situation a variety of drugs were tested and compared with regard to their ability to produce state-dependent, or dissociated, learning. Three major results were obtained. First, several depressant drugs produced a state in which learning was partially dissociated from learning which occurred in the nondrug state. Pentobarbital, phenobarbital, alcohol, urethane, and meprobamate were approximately equivalent in their actions and were interchangeable. Second, atropine produced a drug state in which learning was partially dissociated from learning in the nondrug state and from learning which occurred under the depressant drugs. The effects of atropine which produced this state neither mimicked nor antagonized those effects of the depressant drugs which allowed them to dissociate learning. This result is especially important as it indicates that not all dissociation phenomena are the result of some single unidimensional process (eg., activation level changes produced by the drugs). Third, many drugs which dissociate learning in other experimental situations have weak or nonexistent dissociative effects on escape learning in the T maze. The present technique appears to be useful for directly comparing the dissociative effects of a variety of drugs.

Comparison of the degree of discriminability of various drugs using the T-maze drug discrimination paradigm

This paper reports preclinical data that may predict the amount of state-dependent learning likely to be produced in humans by various psychoactive drugs. In a T-maze, rats were required to turn right when drugged and left when not drugged to escape from electric shock. The number of training sessions required to learn this drug versus no drug discrimination was used as an indicator of the degree of discriminability of the training drug. Using this procedure, the discriminability of more than 100 common psychoactive drugs was determined at one or more doses. Sessions to criterion usually decreased as dosage was increased. Maximum discriminability occurred at the highest usable dose in most cases, and differed considerably for drugs of various types. The results suggest that the majority of psychoactive drugs can be investigated by use of the drug discrimination technique, and that state-dependent learning effects will not accompany clinical use of most psychoactive drugs unless intoxicating doses are used.

Discriminative Control of Behavior by Drug States

Centrally acting drugs can be used in place of discriminative stimuli, and often acquire response control with surprising rapidity when used in this way. Also, without discriminative training, the performance of behaviors learned while an animal is drugged may appear conditional upon the drug state present during response acquisition; the response may fail to transfer, or transfer only partially, into different drug states.

State Dependent Learning and Drug Discriminations

In these words, Combe first described state dependent learning in 1830. During the last 20 years, state dependent learning (SDL) and drug discriminations (DD) have been the subject of several hundred published reports. The intent of this paper is to review our present state of knowledge regarding these two closely related phenomena.

State-Dependent Learning Produced by Alcohol and Its Relevance to Alcoholism

A behavior learned while an animal is drugged sometimes fails to appear during subsequent nondrug test trials, although it appears reliably whenever the drug is readministered. Conversely, if the same behavior is learned when the animal is not drugged, it may then be performed only as long as the animal remains undrugged. For some reason, the ability to perform appears to be conditional upon the drug conditions present during initial acquisition (Overton, 1964). Experimental workers generally refer to this surprising phenomenon as “dissociated” learning or as “state dependent” learning. Clinically trained readers may find it convenient to think of the drug as inducing a sort of temporary “fugue” state separated from the nondrug state by a partial or complete amnesic barrier.

Historical context of state dependent learning and discriminative drug effects.

Drug-induced state dependent learning (SDL), as well as similar effects on memory retrieval exercised by physiological states, have been known since 1830. Before 1950, understanding of this area derived primarily from clinical descriptions of somnambulism, dream recall, fugue states, and cases of multiple personality. After 1950, experimental demonstrations of the properties of SDL and drug discriminations (DDs), along with a series of changes in the DD procedure, have led to the DD paradigm that is currently employed, and which has properties that make it an extremely useful tool for preclinical investigation of a variety of pharmacological and psychological questions. These conceptual and technical developments have resulted in widespread acceptance of the DD paradigm as a preclinical research method. This paper reviews the nineteeth and twentieth century history of clinical observations, concepts, and experiments, that have led to our current status of knowledge about drug discriminations and SDL.

Comparison of ethanol, pentobarbital, and phenobarbital using drug vs. drug discrimination training

Rats learned drug vs. drug (D vs. D) or drug vs. no drug (D vs. N) discriminations in a T-maze shock-escape task with various doses of pentobarbital, phenobarbital, or ethanol. Dose-effect curves were obtained for each drug using D vs. N training. After D vs. N training with any one of these drugs, rats made D choices during substitution tests with the other two drugs, suggesting drug interchangeability. D vs. D training also showed that pentobarbital and phenobarbital were virtually indistinguishable from one another. However, ethanol was readily discriminated from pentobarbital, showing that the two drugs differed. The results show the utility of D vs. D training as a method for studying drug differences that may be too small to detect with substitution tests.

Influence of shaping procedures and schedules of reinforcement on performance in the two-bar drug discrimination task: a methodological report.

Rats were trained to discriminate phenobarbital (50 mg/kg) vs saline in a two-bar, thirst-motivated operant drug discrimination task. Presses on bar 1 were reinforced when the rat was drugged and presses on bar 2 when not drugged. Several shaping procedures and schedules of reinforcement were compared to determine which would allow drug discriminations to be learned most rapidly, and which would result in the higherst asymptotic accuracy of discrimination. The discrimination was learned more rapidly when differential drug conditions were used from the very first day of training than when shaping on both bars was completed before the differential drug conditions were introduced. Variable interval (VI), fixedratio (FR), and differential reinforcement of low rate (DRL) schedules were tested as well as several more complex schedules. Asymptotic accuracy of discrimination was highest when FR, interlocking FR 10/FI90″, or DRL 16″ schedules were used; accuracy was intermediate when a tandem VI 20″/FR 10 schedule was used; accuracy was relatively low when a VI 20″ schedule was employed. When the optimal shaping procedure and schedule of reinforcement were used, highly accurate drug discriminations were learned much more rapidly than has previously been reported. The results are discussed with reference to control of differential responding by contextual and discriminative stimuli.

Problems associated with application of the contingent negative variation to psychiatric research.

As a first step in applying Walters contingent negative variation (CNV) in comparative studies of psychiatric patients, data were obtained concerning the issues of contamination by activity arising in the orbit and of individual variability between different test paradigms. Subjects were 20 college students and 10 psychiatric inpatients. Looking only at the conventional CNV recording leads (vertex-mastoid), a significant CNV was obtained in two experimental paradigms. However, analysis of the spatial distribution of the potential strongly suggested that much of the recorded vertex-mastoid negative shift can be accounted for by activity associated with eye movement. The distribution of CNV closely paralleled the distribution of vertical eye movement potentials. Results in patients and nonpatients were similar. The vertex-mastoid lead placement appears to be a poor one for recording relatively uncontaminated CNV; vertex-temporal derivation appears promising, although a smaller negative shift was recorded there. CNV measurements of the same subjects obtained under two experimental conditions were poorly correlated with each other, suggesting that the CNV responsiveness under any one condition does not characterize the individual. It was concluded that application of CNV recording to psychiatric research presents serious methodological and interpretive difficulties.

Discriminable effects of antimuscarinics: Dose response and substitution test studies

In a shock escape T-maze task, rats were trained to turn right following one drug treatment and left following a second drug treatment. The specific drug and dose conditions were the only discriminative cues available to the animals. The number of training sessions before criterion performance indicated the discriminability of the two training conditions. Drug vs no drug training showed that discriminability was proportional to dosage for low doses, but was constant over a range of higher doses. Such an asymptote of discriminability was observed with scopolamine, atropine, benactyzine and Ditran (JB 329), and was shown not to result from tolerance. High dose vs low dose discriminations involving scopolamine were learned very slowly if both doses were within the asymptotic range; this indicates that similar discriminable effects were produced by high and low doses. To compare various drugs, substitution tests were administered to trained rats. The four antimuscarinic drugs generally substituted for one another but did not mimic and were not mimicked by drugs in other pharmacological classes. Some exceptions to this pattern were noted. The discriminable effects of scopolamine were partially antagonized by physostigmine. The results indicate that the antimuscarinic drugs share discriminable actions probably produced by their anticholinergic actions. The asymptote of action at high doses appears genuine, possibly reflecting receptor saturation.