Operant methodology in the study of learning

A series of experiments is described in which operant methodology is used to study the effects of drugs on "learning. " Emphasis is placed on the technique of repeated acquisition as a behavioral baseline for studying this type of transition state. In this technique, each subject is required to learn a new discrimina-tion each session. Multiple-schedule procedures are also described in which acquisition is compared to a "performance " task, where the discrimination is the same each session. The learning baseline is more sensitive to the disruptive effects of a variety of drugs (e.g., cocaine, d-amphetamine, haloperidol) than is the performance baseline. This general finding obtains across procedural variations and species (pigeons and monkeys). The potential usefulness of these procedures for studying both acute and chronic be-havioral toxicity is discussed. Learning has traditionally been defined as the ac-quisition of new behavior. Within an operant framework, however, the word learning implies transitional behavior that is progressing towards a steady state (1). Such transition states may reflect a variety of circumstances and behaviors. For example, a transition state may simply describe the initial acquisition of a lever-press response. Another type of transition state is the change in steady-state behavior that occurs when a single schedule of reinforcement is changed, e.g., from fixed ratio to fixed interval (2). A transition state may also describe the acquisition of more complex sequences of behavior, which is the focus of the present paper. The development of an operant technique for the study of variables affecting transition states began with the work of Boren (3). This technique, termed repeated acquisition of behavioral chains, requires a subject to respond in a predetermined sequence on some number of operanda with a reinforcer de-livered at the end of the sequence. Each session the subject is required to learn a different sequence of responses. Over time, both the pattern of acquisi-tion and the number of errors reach a steady state from session to session. This steady state of transi