Affective computing works mostly under a vision of emotions based on a functionalist conception of the mind in which emotions, as any other mental state, are understood as functional relations of information processing. The way in which these functional relations are achieved, whether through neuronal activity and organization or by artificial computer programming, is irrelevant to what emotions essentially are. These ideas are in stark contrast to the positions of embodied cognitive science, especially those emerging from the 4E approach to cognition (Embodied, Ecological, Embedded, Enactive), to which, in general, affectivity is seen as constitutive to cognition and cognition is always embodied.

 

In this workshop we discuss how relevant is embodiment for the synthesis of affectivity based in AI or other forms of implementation. The workshop is open to the widest possible disciplinary audience to tackle both the theoretical and philosophical aspects of synthetic affectivity, and how this is relevant for real-world implementations. We believe that this discussion is not only relevant in terms of advancing technology –which is exciting all by itself–, but it is a great opportunity to put the embodiment of emotions and affectivity in sharper relief by considering if and how this affective life can be shared with synthetic systems or even artificially implemented. We thus propose a dialogue in which the AI concern with artificial affectivity and the embodied methodologies of ALife can meet.

 

Possible topics include but are not limited to:

• What is the notion of affectivity or emotion at play when trying to synthesize it? What kind of affectivity, if any, could AI have?

• What kind of body if any is required for synthesizing affectivity?

• What if any is the role of affectivity in cognitive systems that should instantiate a normative point of view?

• What are the technical limitations of computing mechanisms to recognize, predict or implement affectivity?

• What strategies could and should be used for implementing artificial affectivity? Is it preferable to use bottom-up or top-down strategies?