this volume, by Wasserman)]. Such speculations provoked two kinds of reactions from psychologists who argued against reference to cognitive entities in the analysis of animal behavior.
Watson ( 1914) dismissed them on the grounds that they could not be observed objectively. Accordingly, "psychology must discard all references to consciousness." Skinner ( 1938, 1950) simply questioned their explanatory value. Saying that an animal did X because it "believed" that X would lead to a reward was, for Skinner, a vacuous exercise in directing attention away from observable independent variables to an inner belief whose features coincided exactly with those needed to account for X.1
While questions about the nature of consciousness in animals still arise (e.g., Griffin, 1976, 1978), it has not been a major issue in the recent revival of interest in animal cognition. Just as the modern rationale for using human cognitive terms is not based upon arguments that appeal to consciousness or to introspective reports, the rationale for the study of cognitive processes in animals requires no reference to animal consciousness. Both in human and animal cognition it is assumed that the normal state of affairs is unconscious activity and thought. Indeed, additional processes must be postulated in order to account for one's consciousness of activities, perceptions, thoughts, and so on (cf. Pylyshyn, 1973, 1981; Skinner, 1969, Chapter 6). In this respect it is interesting to note a rare point of agreement between such dissimilar psychologists as Freud, Skinner, and modern investigators of human cognition.
Exorcising the ghosts of consciousness and introspection has proved to be a much easier task than demonstrating and defining the unconscious processes that constitute the subject matter of animal cognition. It is not too great an oversimplification to observe that many of the major recent advances in animal cognition are best characterized as convincing demonstrations of the inadequacies of so-called S-R models of animal behavior, -models which rely exclusively on observable (or potentially observable) stimuli and responses. On the other hand, less progress has been made in characterizing the nature of the cognitive processes whose existence has been demonstrated clearly by a wide variety of recent studies (see below).
One obvious obstacle to progress in defining the nature of animal cognition is the difficulty of inferring the critical features of unobservable processes, -a problem shared by the study of human cognition. Even though that difficulty has been reduced by focusing on unconscious processes and by the use of objective criteria for postulating their existence, our understanding of animal cognition has been impeded by an____________________