We argue that the propositional and link-based approaches to human contingency learning represent different levels of analysis because propositional reasoning requires a basis, which is plausibly provided by a link-based architecture. Moreover, in their attempt to compare two general classes of models (link-based and propositional), Mitchell et al. refer to only two generic models and ignore the large variety of different models within each class.

Behavioral momentum theory has evolved within the realm of operant conditioning. The thought-provoking momentum metaphor equates the strength of an operant response with its resistance to change and preference (i.e., choice) for that response over other available responses. Whereas baseline response rate (velocity in the metaphor) is assumed to be largely influenced by the response-reinforcer contingency, resistance to change and preference are assumed to reflect an intervening variable called behavioral mass, which is determined primarily by the stimulus-reinforcer relationship. This invites the question of how well the momentum metaphor applies to the stimulus-reinforcer relationships of traditional Pavlovian paradigms. Presumably, a correspondence exists between behavioral mass and the notion of associative strength in the associative learning literature. Although response rate has little meaning in the trialwise structure of classical (i.e., Pavlovian) conditioning, response probability or magnitude might be regarded metaphorically as velocity. Momentum theory suggests that resistance to change (e.g., extinction) is a better indicator of associative strength than is response probability or magnitude. Therefore, variables that strengthen Pavlovian learning should influence resistance to extinction of conditioned responding in a similar manner. Moreover, it is important to assess momentum theory outside of strictly operant paradigms, particularly because in clinical settings many common disorders (e.g., phobias) and their therapies (e.g., cue exposure) are thought to be classically conditioned.

Rachlin's substantive points about the relationship between altruism and self-control are obscured by simplistic and outdated portrayals of evolutionary psychology in relation to learning theory.

We emphasize the feature of Webb's presentation that bears most directly on contemporary research with real animals. Many neuroscience modelers erroneously conclude that a model that performs like an animal must have achieved this goal through processes analogous with those used by the animal. A simulation failure justifies rejecting a model, but success does not justify acceptance. However, an important benefit of models, successful or otherwise, is to stimulate new research.