Stimulus Characteristics and Reinforcement in Probability Learning

Abstract

This study attempted to examine the influence of variations in stimulus characteristics on choice behavior in a binary-choice probability learning task. Earlier work by Snelbecker and Schwaab (3) suggested that stimulus characteristics may influence response choice and latency on a two-choice operant learning cask. In more recent work (1) some Ss who had to choose between two pictures of racing cars on a probability learning task stated that they had made their choices on the basis of which car was designed better, which was driven on a slight downhill grade, which had the higher idenrification number, etc. Other stimuli produced other attempts at deduction. It seemed, then, that at least some Ss were reacting to the task as a cognitive puzzle to which there was a correct answer rather than as a probability learning task in which there was no perfect solution available. This suggested that characteristics of the stimuli might influence probability learning independent of the schedule of reinforcement for choices. Such a stimulus would have an effect early in the program afh~ch would dissipate with many trials. Stimuli congruent with the reinforcement conrlngcncles would be expected to facilitate learning and reduce response latency. The opposlre was expected for incongruent stimuli. Further, it was predicted that increasing the strength of the reinforcer (in this case by adding points to a bell-ring and a light-flash) would tend to weaken any stimulus effects. A 75% to 25% pcobability learning procedure similar to that pioneered by Siegel (2) was used, with each S being asked to respond to the identical probability learning task differing only in the type of information conveyed by one of five pairs of discriminant stimuli. Ss were also divided on the basis of whether they did or did not receive points on a counter as positive reinforcement.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 0 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.