Turing's Three Senses of “Emotional”

Introduction

¹ Philosophical and psychological theories of emotion attempt to explain the nature and origin of emotion; in particular, they focus on the rationality of emotion and the role of emotion in deliberative decision-making. This is complicated by the fact that in everyday language the word “emotion” has multiple uses. These different senses have prompted conflicting analyses of emotion—as a feeling (quale or somatic state), as behaviour, or as a cognitive state. Consistent with the multiplicity of everyday uses of “emotion” and its cognates, Turing used the expression “emotional” in three quite distinct ways—to discuss respectively emotional concepts, emotional arguments, and emotional communication. In the first case he used this expression to mean (what philosophers call) response-dependent,² in the second case he used it to mean irrational, and in the third case he used it (tongue-in-cheek) to mean feeling.

Turing’s notions of emotional concepts, arguments, and communication are central to his philosophy of machine intelligence. What is intelligence? Here Turing suggested a novel and intriguing approach to the concept of intelligence that is yet to be developed. Can machines think? Turing dissected and countered objections to the possibility of machine intelligence—objections that are still found today. How are we to build an intelligent machine? Here Turing set out a research programme for AI—to build a “child machine”—that is now pursued in social and developmental robotics. In this paper I analyse Turing’s comments on emotion and their significance for affective computing in the 21^st century.