Optimization Versus Response‐strength Accounts of Behavior

Abstract

Pigeons were run in both single-key and concurrent-key experiments in which, over most of the range of response rates, an increase in response rate gave rise to a continuous decrease in reinforcement rate. In spite of the fact that a low response rate would have produced a high reinforcement rate, all birds responded at relatively high rates, thus keeping reinforcement rates substantially below the maximum possible. In the concurrent-key experiment, in addition to responding at relatively high rates, the birds' ratios of responses approximately matched the corresponding ratios of obtained reinforcers. The results are inconsistent with most theories of optimal performance, which assume that organisms behave in ways that either maximize reinforcement value or minimize deviations from a free-behavior point. On the other hand, the results are consistent with the assumption that reinforcement strengthens the tendency to respond.

Human sensitivity to reinforcement feedback functions

“…However, there are several problems with the original demonstration in humans that make it very important that this effect is replicated. As noted above, there is very little other unambiguous evidence for sensitivity to temporally extended aspects of the schedules (Schachtman & Reed, 1998;Vaughan & Miller, 1984). Hence, there is little strong evidence for the ability to integrate information across time on such contingencies.…”

“…For example, if response rates immediately prior to reinforcement are low, then strengthening the tendency to emit low local rates would reduce overall response rates. According to this view, optimal performance observed over extended periods is only a by-product of these local processes and does not depend on the ability to integrate information across extended periods (e.g., Vaughan & Miller, 1984).Schedules of reinforcement have often been employed to assess sensitivity to the temporally extended (molar) relationship between response output and reinforcement input, often characterized as the reinforcer feedback function. Demonstrating sensitivity to this molar characteristic of a schedule is important, since it would suggest that integration over time is a critical factor in instrumental performance.…”

“…This means that the molecular or local aspects of the schedule could also account for the lower response rates on VI schedules. Vaughan and Miller (1984) suggest that such confounds between the molecular and molar aspects of the contingencies make the use of simple schedules problematic, and that on more complex schedules, there is very little support for a molar view (see also Reed & Schachtman, 1989).The variable-interval-plus-linear-feedback (VI ) schedule (McDowell & Wixted, 1986) is an important schedule to study in the investigation of molar sensitivity. …”

“…In particular, the extent of this ability addresses a distinction between "molar" and "molecular" theorists (cf. Baum, 1981;Vaughan & Miller, 1984). "Molar theorists" hold that behavior is guided by the integration of information about responses made and reinforcers earned, across temporally extended periods.…”

“…For example, if response rates immediately prior to reinforcement are low, then strengthening the tendency to emit low local rates would reduce overall response rates. According to this view, optimal performance observed over extended periods is only a by-product of these local processes and does not depend on the ability to integrate information across extended periods (e.g., Vaughan & Miller, 1984).…”