Beyond the computational-representational brain: why affective neuroscience tells us attitudes must be explained on multiple levels

A popular conception in the mind/brain sciences today is the metaphor of the brain as a computer. According to the computational theory of mind, all relevant cognition consists of executing some welldefined rules over representations which model (and have a direct correspondence to) the external world (Chalmers, 1996). I would like to challenge this notion and advocate a pluralistic/biological approach to cognition, specifically with respect to explaining emotions. In the following, I take explanatory pluralism to be the program of explaining complex phenomena by theories operating on multiple levels of measurement/explanation. I furthermore suggest that theories should be systematically connected (reduced by showing implementation, if possible) to theories at other levels. The necessity of biological modeling is probably self-evident to most neuroscientists. However, many theories in cognitive science and psychology present a certain functional constellation of which it is not clear how/if it relates to lower levels. These theories fall under the algorithmoriented framework of cognitivism, which guides the study of thought in terms of abstract processes in some architecture of the mind/brain. A framework which emphasizes the body in place of abstract processes— embodied cognition—has gained traction in the last decades and seems to have an indispensable advantage over cognitivism in that it allows for explanations which span multiple levels. Furthermore, it allows these explanations to be embedded in the body and environment. I argue that affective neuroscience has revealed that emotions are such an embodied phenomenon requiring description on multiple levels—including (but not limited to) the neurochemical, the systemic, and behavioral. I conclude that the use of algorithmic explanation (as opposed to simulation) should be limited and subject to verification within such a pluralistic epistemological landscape. In the following, drives are intended to denote homeostatic behavioral processes which are dependent on interoceptive representation. Emotions are intended to denote an affective state with cognitive, behavioral, and embodied components, thus subsuming drives. Attitudes are intended to denote long-term behavioral (especially object-directed) tendencies resulting from a history of emotional experiences.