Sometimes-competing retrieval (SOCR): A formalization of the comparator hypothesis.

Abstract

Cue competition is one of the most studied phenomena in associative learning. However, a theoretical disagreement has long stood over whether it reflects a learning or performance deficit. The comparator hypothesis, a model of expression of Pavlovian associations, posits that learning is not subject to competition but that performance reflects a complex interaction of encoded associative strengths. That is, subjects respond to a cue to the degree that it signals a change in the likelihood or magnitude of reinforcement relative to that in the cue's absence. Initially, this performance-focused view was supported by studies showing that posttraining revaluation of a competing cue often influences responding to the target cue. However, recently developed learning-focused accounts of retrospective revaluation have revitalized the debate concerning cue competition. Further complicating the picture are phenomena of cue facilitation, which have been addressed less frequently than cue competition by formal models of conditioning of either class. The authors present a formalization and extension of the comparator hypothesis, which results in sharpened differentiation between it and the new learning-focused models.

Conditioned inhibition and reinforcement rate.

“…The results reported by Hallam et al (1990) are uniquely consistent with the predictions of the Comparator Hypothesis (Stout & Miller, 2007), and therefore that model is contradicted by our finding that extinction of the training CS+ increases the inhibitory strength of the CS-. In the Comparator Hypothesis, the mechanism by which a CSinfluences responding is very different from that of other associative models, in that the CS-itself has no inhibitory strength (and may even have some excitatory strength).…”

Enhancement of responding to A after A+/AX+ training: Challenges for a comparator theory of learning.

“…In the case of our experiments, higher order comparator processes become important only if the context, C, is assumed to enter into associations with the US and with the other experimental stimuli. Stout and Miller (2007) have assumed that associations are unidirectional, which will lead to a total of 12 associations that need to be considered: A-C, A-X, A-US, B-C, B-US, C-A, C-B, C-X, C-US, X-A, X-C, and X-US. To appreciate how these associations affect first-order comparator processes concerning the target association A-US, consider rows 1a to 1e of Table 3.…”

“…When two first-order comparator stimuli could influence the response to the target CS, or when two second-order comparator stimuli could influence the impact of a first-order comparator stimulus (rows 1a to 1e of Table 3), then the influence of the two comparator stimuli was determined by the sum of their individual effects (providing the sum never exceeded 1). These principles and parameter values were taken from the sometimes-competing retrieval model described by Stout and Miller (2007).…”

“…For this simulation, there will be an influence of nonreinforcement because on Aϩ trials the association between A and X will be weakened according to Equation 2. In keeping with assumptions made by Stout and Miller (2007), the value of the learning rate parameter for nonreinforcement (k1) was set to 0.18 on these trials. Despite this increase in the complexity of the design, our simulation of the comparator hypothesis reveals that the strength of conditioned responding to A should still fall, and it should be noted that other simulations we have conducted reveal that this prediction remains true even when the value of k1 is set to the maximum value of 1.…”

A comparator view of Pavlovian and differential inhibition.

“…Critical here is the relative magnitude of these two effects. However, for the last few years the data from all experiments conducted in our laboratory to test the extended comparator hypothesis have been evaluated with respect to a single consistent set of parameters that are fully described in Stout and Miller's (2006) mathematical implementation of this model. With these parameters, even with massed trials the context does not become a sufficiently strong first-order comparator to compensate for the degradation of A's role as a first-order comparator stimulus in the case of Pavlovian conditioned inhibition.…”