emotional or motivational states. Consequently, he Proposed that omission of expected food and the presentation of shock may excite some common representation, different from that activated by the omission of expected shock and the presentation of food. Behavioral findings (see, for example, Dickinson & Dearing, 1979) indicate that appropriate reciprocal relations do seem to exist between these opposed motivational systems. Nevertheless, we lack any deep empirical and theoretical analyses of how event nonoccurrence might be effectively encoded by animals and humans -- a gap in our knowledge that demands some filling in the future.
This chapter examined several phenomena collectively linked by their apparent reliance on stimulus absence as a basis for effective learning or performance. Pavlovian trace conditioning with a long unfilled gap between CS and US was successful only when this empty interval was made informative by clearly differentiating it from other segments of conditioning trials, especially the intertrial interval. Studies of the feature-positive effect revealed that, under a variety of discrimination arrangements, animal and human subjects perform considerably better when the presence of some specific feature of a compound stimulus is the signal for a positive event than when absence of the feature signals the positive event. This asymmetry in the way positive and negative information are processed in our specific experiments may arise via different mechanisms for animals vs. humans, because post-training assays revealed that our FN pigeons had apparently learned about the relation between feature presence and nonreinforcoment even though they did not perform appropriately during actual discrimination learning sessions. On the other hand, humans were unlikely to identify the distinguishing feature when it appeared only on negative trials. In our tasks, the pigeon's deficit was presumably one of performance, not learning or detection, whereas the human's difficulty lay in discovery of the relevant cue.
Of course, one should not infer that pigeons "learn" FN discriminations and identify visual features better than human beings. The exact nature of any possible species difference requires considerable research designed to arrange more comparable tasks (e.g., types of features, the consequences on negative trials) and a variety of performance assays, including those associated with extinction and discrimination reversal procedures. Successful human subjects seem to formulate a rather flexible, verbal rule that enables them to quickly learn a reversal of the original discrimination. In contrast, pigeons do not show analogous transfer; they appear unduly controlled by characteristics of and relations between particular external stimuli.
I briefly discussed some issues pertinent to the conceptualization and potential encoding of "absence," and mentioned the possibility that organisms capable of learning may possess a predisposition to more easily form and/or utilize associations between the occurrences of two events than between the nonoccurrence of one event and the occurrence of another. Whether or not this broad speculation is warranted, trace