What is a computational emotion?

Abstract

ABSTRACT Much recent work in cognitive science and psychology has rediscovered the James-Lange hypothesis, which grounds affect in physiological states and processes. This article complements such an approach in a very different manner, through the computational analysis of large historical datasets. Rather than focusing attention upon the singular affective body […]’. By so doing, it seeks to extend James’s theory of emotion across much larger temporal and spatial scales. Rather than focusing attention upon the singular affective body, I examine a much wider cultural body – the hundreds of thousands of texts collected under Eighteenth Century Collections Online – in order to demonstrate the changing discourse on physiological affect. Computational modelling, I demonstrate, enables us to extend scholarly attention beyond canonical texts so as to consider more obscured or miscellaneous literature, ranging from surgical journals to midwifery manuals. By focusing upon a series of key words and terms – ‘irritability’, ‘sensibility’, ‘mental emotion’ – I demonstrate that such overlooked material anticipated the James-Lange hypothesis in unsuspected ways. I conclude by making a number of speculative analogies between the distributed nature of computational enquiry, on the one hand, and the distributed nature of embodied feeling, on the other.