Contingent partial reinforcement and the anticipation of correct alternatives.

Abstract

Estes and Straughan (6) developed an equation for the two-choice noncontingent reinforcement situation.Here S has two alternative responses either one of which may be followed by one of two environmental events. The occurrence of the events is controlled entirely by E and is independent of S's choices. A typical situation is one where there are two levers (alternative responses) with a light over each one (environmental events); S is to predict which light is going to flash next by pressing one of the levers. After S has pressed one of the levers E may then flash either the left or the right light according to a predetermined probability schedule.Contingent reinforcement, however, differs from noncontingent in that the events following the responses are, in part, S-controlled. Typically S has two alternative responses, and one of two environmental events may subsequently occur, but they are not independent of S's responding. In such a situation, there may be two levers and a ground-glass screen behind which are red and white bulbs. The levers are controlled by E so that either one can switch on either light and are manipulated according to a predetermined probability schedule. The S is to predict which lever will give, say, a red light. If E has set 1 The present research was conducted under grant M-814C from the National Institute of Mental Health. 2 The collection and analysis of data was mainly done by Paul J. Woods. Howard Brand and James M. Sakoda are senior staif on the project.

Probability learning: Left-right variables and response latency.

“…In some two-choice probabilitylearning studies, human Ss given contingent reinforcement matched their choice percentages to the reinforcement percentages when the probabilities of positive reinforcement summed to one (Goodnow & Postman, 1955;Neimark, 1956;Sidowski & Gregovich, 1963). In others, 6s tended to overshoot the probabilitymatching percentages (Brand, Sakoda, & Woods, 1957;Bush & Mosteller, 1955, pp. 294-296;Edwards, 1956;Morin, 1955;Wyckoff & Sidowski, 1955).…”

Probability learning in children.

“…At the same time, it has been found that under certain conditions Ss tend to maximize their frequency of choosing the more frequently reinforced stimulus. Among these are the presentation of a gambling task or a game of skill rather than a problem-solving task (7,9), an increased number of training trials (3,6), the presence of a distracting stimulus (18,19), and the lack of information concerning the appropriateness of some responses (2,3,14).…”

The relationship between probability difference, (!p1—!p2), and learning rate in a contingent partial reinforcement situation.

“…Apparatus.-The two-choice contingent reinforcement apparatus used has been fully described elsewhere (Brand et al, 1957). Briefly, it consists of two levers (microswitches) at table height in front of 5 and a vertical panel containing red and white reinforcing lights and a green signal light.…”

“…In contrast, the contingent equation for a two-choice situation takes the following form (Brand, Sakoda, & Woods, 1957):…”