Increased resistance to extinction as a function of double and single alternation and of subsequent continuous reinforcement.

A stimulus-sampling model of the partial reinforcement effect.

“…If the reinforcement schedule is a single-alternation schedule and is continued for a large number of trials, a greatly reduced partial reinforcement effect, if any, is found. Bloom and Capaldi (1961), Campbell et al (1970), Koteskey and Hendrix (1971), and Tyler, Wortz, and Bitterman (1953) found a reduced partial reinforcement effect in singlealternation groups given a large number of trials as compared to randomly reinforced or double-alternation groups. Also, the investigators in several of these articles noted the following pattern in speed of response: (a) early in acquisition the subjects ran faster following a reinforced trial than following a nonreinforced trial; (b) at intermediate stages of acquisition this difference disappeared; and (c) later in acquisition the opposite was true-they ran faster following a nonreinforced trial.…”

“…Double alternation. Bloom and Capaldi (1961), Habu and Ono (1969), and Koteskey and Hendrix (1971) found increased resistance to extinction in their double-alternation groups. This would be explained by the model because at the end of acquisition with a very large number of trials, nearly all of the N 2 elements would be conditioned (since they were never followed by a nonreinforcement) along with many of the NO and Ni elements.…”

“…Support is given for this interpretation in that Theios and McGinnis (1967) pointed out that the Sutherland et al (1965) groups differed in response speed at the end of acquisition and their results may be due to this factor. Campbell, Knouse, and Wroten (1970) and Koteskey and Hendrix (1971) also found that continuous following partial reinforcement resulted in increased resistance to extinction. This is because even though the Ni, N 2 , N 3 , • • •, elements are not affected, more and more No are becoming conditioned during the continuous reinforcement.…”

“…Stimulus-Sampling Axiom 3 has been made more specific than in Koteskey and Hendrix (1971) as shown in Figure 1; rather than arbitrarily assigning values to Ni, N 2l N 3 , • • •, the size of each of these sets of stimulus elements is now assumed to be a decreasing function so that as i -* °°, the number of elements in N,-approaches zero. That is, as Capaldi (1966) points out, less stimulus modification occurs on each successive trial as length of nonreinforced run of trials (N-length) increases.…”

“…If the predictions are wrong, the model can often be modified and new predictions made to see if the modification has enabled the theory to account for the data. Koteskey and Hendrix (1971) proposed such a stimulus-sampling model of Capaldi's (1966) sequential interpretation of the partial reinforcement effect. Although the model was briefly presented there with an experiment to test one of its predictions, space limitations restricted discussion of the model at any length.…”