The Theory of Mind in Strategy Representations

Abstract

The Theory of Mind in Strategy Representations Andrew S. Gordon (gordon@ict.usc.edu) Institute for Creative Technologies, University of Southern California 13274 Fiji Way, Marina del Rey CA USA Abstract Many scientific fields continue to explore cognition related to Theory of Mind abilities, where people reason about the mental states of themselves and others. Experimental and theoretical approaches to this problem have largely avoided issues concerning the contents of representations employed in this class of reasoning. In this paper, we describe a new approach to the investigation of representations related to Theory of Mind abilities that is based on the analysis of commonsense strategies. We argue that because the mental representations of strategies must include concepts of mental states and processes, the large-scale analysis of strategies can be informative of the representational scope of Theory of Mind abilities. The results of an analysis of this sort are presented as a description of thirty representational areas that organize the breadth of Theory of Mind concepts. Implications for Theory Theories and Simulation Theories of Theory of Mind reasoning are discussed. Investigating the Theory of Mind One of the most challenging areas of research in the cognitive sciences has concerned the Theory of Mind, in reference to the abilities humans have to perceive and reason about their own mental states and the mental states of other people. Along with the inherent difficulties in investigating behavior that is largely unobservable, researchers in this area are required to be extremely interdisciplinary. Many research fields contribute evidence that influences our understanding of these human abilities, although the methods used to gather this evidence are diverse. Researchers in developmental psychology largely choose to investigate the Theory of Mind as a set of abilities that progressively emerge in normal child development (Wellman & Lagattuta, 2000). By the last half of their second year, toddlers demonstrate an understanding of the role of intentionality in action, and that other people have subjective experiences. By the age of four and five, children comprehend and use vocabulary to refer to mental states such as thoughts, imaginations, and knowledge. As children advance into grade-school years and adulthood, there is a growing appreciation of people as active constructors and interpreters of knowledge, and awareness that others have ongoing thoughts. There is evidence that Theory of Mind capabilities continue to improve into the later adult years, even while non-social reasoning abilities begin to degrade (Happe et al., 1998). In the research area of abnormal psychology, compelling cases have been made relating illnesses such as autism (Baron-Cohen, 2000) and schizophrenia (Corcoran, 2001) to deficits in Theory of Mind abilities. Neuropathology studies of stroke patients have provided evidence that Theory of Mind mechanisms may be localized in the brain (Happe et al., 1999), and ongoing functional neuroimaging studies continue to provide further evidence for localization (Frith & Frith, In search of a more process-oriented understanding of Theory of Mind abilities, it is the philosophy community that has made the most contributions, proposing two classes of process theories that have been extensively debated. First, the Theory Theory hypothesizes that Theory of Mind abilities are computed by prediction and explanation mechanisms by employing representation-level knowledge about mental attitudes (Gopnik & Meltzoff, 1997; Nichols & Stich, forthcoming). The opposing view is that of Simulation Theory (Goldman, 2000), which argues that Theory of Mind abilities are computed by imagining that you are in the place of the other person, then inferring their mental states by monitoring the processing that is done by your own cognitive mechanisms. While some high-level process-oriented cognitive models have been proposed (e.g. Nichols and Stich, 2000), there are many unanswered questions that prohibit the creation of detailed, computational models of Theory of Mind abilities. Most lacking in our theoretical understanding of Theory of Mind abilities is a description of the specific c o n t e n t s of the mental representations that are employed in this reasoning. There is general agreement that these representational elements must include concepts such as beliefs and desires (e.g. Harris, 1996), and these two concepts in particular have taken a privileged role in the cognitive models that have been proposed. A potential benefit of the focus on these concepts is that this representational area (beliefs, desires, intentionality) is among the very few where established axiomatic theories have been developed in the artificial intelligence community (Cohen & Levesque, 1990). Continued artificial intelligence progress in developing axioms for inference concerning