Sequential effects and dimensional contrast : some attempts at direct empirical manipulation

Behavioral contrast is the multiple schedule interaction that has received the most attention over the past two decades, but an interaction equally problematic for traditional learning theories is dimensional contrast. This dissertation was concerned with the relationships between sequential effects (specifically local contrast and local dimensional effects) and dimensional contrast.

In Experiment I, the essential features of Catania and Gill's (1964) original demonstration of dimnensional contrast were replicated. Pigeons received variable-interval reinforcement for keypecking during randomized presentations of stimuli in the lower half of a lamp column continuum. Stimuli in the upper half of the continuum were presented in extinction. During the second condition these stimuli were paired with a leaner reinforcement schedule than were stimuli in the lower half.

In Experiment II, pigeons were exposed to stimuli along a lamp colum continuum presented in extinction that were moved in an effort to induce positive dimensional contrast in nearby stimuli paired with variable-interval reinforcement.

In Experiment III, pigeons were exposed to variable-interval reinforcement in seven line-tilt orientations. Initially, only certain stimuli preceded other stimuli in an attempt to influence overall gradient form during the formation of the discrimination. Subsequent conditions used stimulus sequences in which each stimulus preceded all other stimuli.

Dimensional contrast was found not to depend on local contrast in Experiment I, but perhaps on local dimensional effects. Border stimuli in Experiments I and II exerted powerful local dimensional effects and may be involved in producing or modulating dimensional contrast. In Experiment III, equivocal results were obtained.

A review of multiple schedule interactions was presented and the applicability of four models of discrimination (Pavlov, Spence, Blough, and the neural unit) to dimensional contrast was discussed. The neural unit model was deemed most adequate to deal with dimensional contrast in general and data from the aforementioned experiments in particular.