In the previous chapter, Prinz introduced various theories of emotion. There it became clear that one of the major problems in the emotional research is whether emotions are cognitive. In chapters 2 and 3, Prinz tries to reconcile cognitive theories with noncognitive ones that he supports.

　　First of all, Prinz shows differences and similarities between cognitive theories in philosophy and psychology, thus making clear what the cognitive theory is. Cognitive theorists in the two disciplines share three assumptions: first, they presume that emotions require having the concepts that correspond to the content of propositional attitudes or appraisals (conceptual hypothesis). Second, both cognitive theorists hold that the cognitive components bound to emotions cannot be equated with bodily changes or inner states related to bodily changes (disembodied hypothesis). Third, both cognitive theories assume that emotions are bounded to “representations of an organism-environment relationship that bears on well-being”, i.e. appraisals (appraisal hypothesis) (p. 25).

　　Are cognitive theories successfully defended by philosophical arguments or psychological evidences? Prinz examines three types of philosophical arguments, and concludes that none of these are decisive. Then he goes on to examine psychological evidences. He focuses on Richard Lazarus’s dimensional appraisal theory. Lazarus and his colleagues performed several experiments, but the results were self-evident or indecisive. An innovative study by Klaus Scherer could not solve these problems entirely. Their failure exposes an underlying methodological problem: the experiments rely on introspective reports of subjects. Subjects would make various assessments of emotional experiences they have gone through, but we cannot conclude from their reports that they actually made such inner judgments when emotions are caused. Their reports might be not about inner judgments but about the external conditions that caused emotions. Experimenters need to extract the former and to see if it is compatible with their theory, but their method cannot isolate the former.

　　Thus cognitive theorists failed to provide decisive evidence. Furthermore, Robert Zajonc offered evidence against cognitive theories. He proposed five points in favor of noncognitive theories. First, emotions emerge both phylogenetically and ontogenetically before the development of cognitions. As evidence for this claim, Zajonc refers to emotion-indicative facial expressions and behavior in human infants and non-human animals. Second, neuroscientific studies show that emotional and　cognitive processes are localized in different structures of the brain. Third, appraisal and emotions do not always correlate with each other; sometimes we cannot change our emotion even if we discover that it occurred because of our misjudgment. Fourth, the establishment of emotional reactions need not be via appraisal. For instance, unconscious learning is possible. The fifth point, the most direct evidence, is that emotional states can be induced with no preceding mental states. Emotions can be induced by direct physical means such as drugs or the intentional change of facial expressions.

　　Evidence offered by Zajonc seems to show various conditions under which emotions deviate from cognition. Is his interpretation of the results correct? Prinz assesses Lazarus’s responses to Zajonc’s argument. We focused on Lazarus’s replies to Zajonc’s second and fifth points, which we found more problematic.

　　As to Zajonc’s second point, i.e. differences between neuroanatomical structures, Lazarus replies that there remains a question. According to him, it is unanswered whether the activity of a neural correlate of an emotional response qualifies as noncognitive. Prinz points out that appraisals supposed in cognitive theories are quite unlikely to be represented at the subcortical level. An emotional response is correlated with an activity of the amygdala, which is a mere body control center. It is difficult to consider that activity in such a structure might represent even simple appraisals. As a counterargument, Lazarus could hold that the true correlates of emotions are activities of both subcortical structures and higher cognitive centers. But, according to Prinz, bodily responses can be started earlier than cognitive centers get involved.

　　Surprisingly, Lazarus offers no response to Zajonc’s fifth point. Instead, Gerald Clore has made a counterargument to Zajonc. Clore claimed that the feelings and bodily changes induced by changing facial expressions constitute a pseudo-emotion, although they are often mistaken as a real one. Assessing this counterargument, Prinz mentions the fact that emotions and induced feelings play a very similar role. Clore would riposte to Prinz by claiming that induction cases actually involve cognition and hence it is qualified as an emotion. This is mere speculation, however, and thus far from decisive.

　　In sum, Lazarus’s dimensional appraisal theory has serious flaws: it has no neuroanatomical basis and has trouble in explaining cases of physical induction. The situation seems to be of great advantage to Zajonc and his noncognitive view of emotion. In fact we cannot yet conclude that all emotions are noncognitive. If cognitions include some nonconceptual and embodied states, such states might constitute cognitive emotions. Thus Prinz tries to determine what cognitions are.

　　Before presenting his own definition, Prinz takes up six candidate definitions. We focused on three of them. The first candidate is that cognition involves access consciousness. This is inappropriate, because according to modern theories of the mind, our cognition includes information processing not available for access consciousness. The second candidate defines cognition by appealing to rational mental states, but this definition has the problem of defining the rational. The final candidate is that cognitions are states containing concepts. But what are concepts? We cannot take concepts as mental representations that can constitute thoughts. Such a definition is circular, because in daily terms, we call cognitions as thoughts.

　　Instead of these failing definitions, Prinz proposes his own definition. According to him, cognitions are states containing “representations that are under the control of organism” (p. 45). A representation is under the control of an organism “if the organism has activated it or maintains it in working memory” (ibid.). This definition allows us to draw a distinction between cognitions and the act of cognition. Being cognitions or thoughts requires being made up of representations that we can willfully control, i.e. form thoughts using them. In contrast, to do a cognitive act is to actually produce a thought under top-down control. Here “top-down control” means the control by executive centers of the brain. Prinz claims that his definition has two advantages: first, it avoids circularity, because it does not contain the term cognition or its cognates. Second, unlike the access consciousness definition, Prinz’s one does not seem too narrow or demanding.

　　Are emotions cognitive in Prinz’s definition? If there are emotions that can be put under organismic control, then mental states containing such emotions might be qualified as cognitive. Willfully imagined emotions (i.e. imagined fear of riding a roller-coaster) might be such kind of emotions. But the emotions caused by events in our daily life are under exogenous control, and thus they are noncognitive. Prinz concludes that our emotions are usually not cognitive at all.

　　Prinz’s conclusion is in line with Zajonc’s claim. But Prinz holds that Lazarus’s theory is not entirely wrong. Actually, Prinz equates emotions with appraisals, just as does Lazarus. The main difference is that, according to Prinz, emotions are embodied appraisals. This will be the theme of the next chapter.

Haruka Tsutsui (The University of Tokyo)