Are emotion-producing processes modular? Jerry Fodor, in his classic introduction of the notion of modularity (Fodor 1983), holds that its most important feature is cognitive impenetrability or information encapsulation. If a process possesses this feature, then, as standardly understood (Currie and Sterelny 2000), “what we want or believe makes no difference to how [it] works” (147).

In this paper, we will start with the issue of the cognitive impenetrability of emotion-producing processes. It turns out that, while there is abundant evidence of emotion-producing processes that are not cognitively impenetrable, some nonetheless are. We will look at two sorts of case. The first concerns emotional reactions to observed faces, and the second involves what we can call “primitive emotions,” emotions that can be activated by non-doxastic input into regions of the brain we share with more primitive animals.

In seeing how some emotion-producing processes can be cognitively impenetrable while others are not, we need to use two commonsensical theses. First, a discussion of modularity must in general operate with a taxonomy that allows for sub-processes or stages of processes.