Emotion in Cognitive Systems Architectures

Abstract

Inspired by theories of situated and embodied cognition, much AI and cognitive systems research today, autonomous robotics in particular, is biologically inspired, although at very different levels of abstraction. This ranges from explicitly biomimetic approaches, concerned with biological plausibility and scientific modeling, to the use of abstractly biologically inspired techniques such as evolutionary and neural computing in engineering. However, it is quite clear that even today the state of the art in cognitive systems research hardly lives up to the original (e.g. Brooksian) ambition to build complete robotic autonomous agents. The ‘incompleteness’ is particularly evident in the somewhat superficial treatment of the issue of embodiment (and its reduction to the physical body’s sensorimotor interaction with the environment) as well the widespread neglect of emotion (or more broadly, affect) and its grounding in homeostatic/bioregulatory processes. This talk discusses work in the European project ICEA - Integrating Cognition, Emotion and Autonomy (cf. www.iceaproject.eu). The twofold hypothesis behind the project is (1) that the emotional and bioregulatory mechanisms that come with the organismic embodiment of living cognitive systems also play a crucial role in the constitution of their “high-level” cognitive processes, and (2) that models of these mechanisms can be usefully integrated in artificial cognitive systems architectures, which will constitute a significant step towards more autonomous robotic cognitive systems capable of dealing with issues such as energy management, selfmonitoring, self-repair, etc. A crucial question addressed in the ICEA project is how in hierarchically organized, layered embodied cognitive architectures lower-level (e.g. bodily, homeostatic) mechanisms can modulate processing at higher levels through emotional/affective mechanisms.