Acquisition of S-R connections: a test of Hull's and Guthrie's theories.

Publisher Website
Google Scholar
Abstract
Hull has postulated that S-R connections are strengthened gradually with repeated reinforcements (N). Guthrie holds that each S-R connection is established in an all-or-none fashion. This expt. was designed to test the validity of these alternative proposals. An eyelid conditioning procedure was used with 32 college males. Stimulus-variability from trial to trial, similarity of on-trial and off-trial stimuli, and other sources of Hull''s various inhibitory potentials were minimized to an unusual extent. The data revealed that when these conditions are controlled, generally neither an individual''s amplitude nor probability of R gradually increase with N. (This is not in accord with expectations based upon Hull''s theory.) Using Guthrie''s theory, for each trial a prediction was made for each individual of whether he would make R. These predictions had an accuracy significant beyond the 5% level for each of 28 subjects. For the group as a whole, the accuracy of predictions based on Guthrie''s theory was significant far beyond the 1% confidence level. An apparent gradual increase in probability of R will result from grouping data, even though each individual in the group jumps from 0% to 100% of R. These and other data of the expt. suggest we have been mistaken in interpreting gradual increases in probability of R or gradual increases in mean amplitude as being due to gradually strengthened S-R connections. When they occur, the gradual increments perhaps are due always to the following factors: (1) Grouping data[long dash]with different individuals forming the S-R connection on different trials, (2) stimulus-variability from one learning occasion to the next or from one test occasion to the nest[long dash]a condition which necessitates forming many S-R connections, with different ones of these S-R connections being formed on different trials; (3) high similarity of on-trial and between-trial stimuli[long dash]with consequent opportunities for unlearning (between trials) of S-R connections formed on the trials. These three conditions (and not gradual strenghening of any S-R connection) may be wholly responsible for the fact that mean amplitude and probability of R often gradually increase with number of reinforced trials.
Keywords

This publication has 1 reference indexed in Scilit: