Resistance to extinction as a function of the number of reinforcements.

Abstract

The present paper is an attempt to give a quantitative description of one of the important relationships postulated by conditioned-response theories of learning. An extensive experimental literature, comprehensively reviewed by Hilgard (2), points to the essential similarity between 'reward' learning and 'reinforcement' learning, suggesting the possibility of deriving both types from a single set of postulates. Hull (5, 6) has attempted the theoretical derivation of a large number of the mechanisms of adaptive behavior, notably those of trial-and-error learning, from postulates based on a number of the phenomena found in conditioning. The predictive value of such a theoretical system depends in large part on a knowledge of the quantitative nature of the phenomena in question, particularly reinforcement, extinction, and spontaneous recovery.The strength of a response tendency as some function of the amount of reinforcement given it, is a relationship whose determination is dependent on the method used for testing the strength of the tendency. The most commonly used indicators are: magnitude of response, frequency of appearance of response, latency of response, and the resistance of the response to experimental extinction. The last mentioned measure was suggested by the following statement of Pavlov (7): "... The greater the intensity of the excitatory process, the more intense must be the inhibitory process to overcome it, and therefore the greater number of unreinforced repetitions necessary to bring about complete extinction" (p. 61).1 This investigation is a part of a coordinated research program of the Institute of Human Relations, Yale University. It was carried out under the direction of Professor Clark L. Hull, to whom the writer is greatly indebted for interest and assistance in the problem.

Representational gain in cortical area underlies increase of memory strength

“…Group performance improved across training sessions [F (24,199) = 27.37; P < 0.0001], to an asymptotic level of 76.2% after 24.8 sessions (Fig. 1B).…”

Evidence for distinct serial processes in animals: The multiplicative-factors method

“…Such theories used to be more common; Hull (1943), for example, proposed that habit strength (changed, say, by the schedule of reward) and drive (changed, say, by hours of deprivation) multiplied to produce response strength. His proposal was based on data similar to what we are considering here (Perin, 1942;Williams, 1938), but Hull's proposal differs in two ways from conclusions based on the MFM. For Hull, the multiplicative relationship betweenhabit strength and drive was an assumptionHull assumed something resembling the data that he wanted to explain.…”

RESPONSE DURATION IN OPERANT LEVEL, REGULAR REINFORCEMENT, AND EXTINCTION1

“…Two stainless steel cages, each containing a lever "Sidman (1952) has argued that the Perin (1942) and Williams (1938) data, which used groups of subjects to determine each point, do not allow us to specify, for individual subjects, the form of the function relating resistance to extinction to number of reinforcements. However, their data do indicate that the distribution is monotonic, and this is the only assumption necessary in the present study.…”