An analysis of conditioned reinforcement using a new technique

Abstract

Four experiments which investigated various parameters of conditioned reinforcement with pigeons were reported. In Experiment I, variations of a multiple schedule, in which a key light followed by a tone was intermittently presented (autoshaping sequence) in one component, and the tone was paired in another component with food, were used. The purpose of this experiment was to determine the necessary conditions for the tone to become a conditioned reinforcer. These procedures included presenting only the light-tone pairing component, presenting only food, random presentations of tone and food, and finally pairings of tone and food in the other component of the multiple schedule. The last procedure resulted in one bird learning to key-peck, while the other birds pecked irregularly. These last two birds key-pecked regularly when the key light was constantly on and each peck was reinforced with a tone presentation. It was concluded that the normal autoshaping was prevented by habituation to the stimuli during the earlier procedures. The tone was, however, a conditioned reinforcer, since there was an increase in the level of key-pecking in the last procedure, in which the tone and food were paired. In Experiments II and III the effects of the schedule of conditioned reinforcement and the rate of conditioned reinforcement presentation were investigated. The strength of the conditioned reinforcer was maintained by pairing the tone and food in one component of a multiple schedule, (DRO 5" in effect to prevent pecking in this component) while the subjects received conditioned reinforcement for key-pecking in the other component. There were few consistent effects between birds in either of these procedures, although there was some indication that rate of responding was directly related to rate of conditioned reinforcement. Experiment IV manipulated the pairing of two tones with food in two components of a multiple schedule while responding on two concurrent VI 1' schedules was reinforced with the tones in a third component. Response rates for the tone, followed 100 per cent of the time with food, increased as the percentage of food pairings of the other tone decreased from 100 per cent. The similarity of this effect to behavioral contrast was noted and discussed. The results of the multiple schedule technique should be more general than other methods of maintaining behavior with a conditioned reinforcer. As in much of human behavior the subjects never respond for the primary reinforcer and the conditioned and primary reinforcers are paired at irregular intervals. Experimentation using this technique led to the view that with some experience at a task, conditioned reinforcement becomes less important in the maintenance of behavior as other stimuli (such as the discriminative stimuli) assume partial control of the behavior.