Iris van Rooij

TU EINDHOVEN, Mens Techniek Interactie (MTI)
IPO 0.15, Postbus 513, 5600 MB EINDHOVEN
I.v.Rooij@tm.tue.nl

Wanted: A Tractable Theory of Rationality! Embodiment and Embeddedness Welcome, But No Free Lunch Included.

Abstract
Humans are generally able to make reasonable judgments, inferences and decisions, be it in the common sense, scientific, social or economic domain. This general ability we call human rationality. Computational attempts to account for human rationality have been plagued by the problem of computational intractability. Informally, a computational theory is said to be computationally intractable if it assumes an unrealistic amount of computational resources (e.g., space, time or speed). In this talk I address the following question: Can the conceptual framework of Embodied Embedded Cognition (EEC) help provide computationally tractable accounts of human rationality. I see at least two ways in which it could:
(1) EEC assumes strong ecological constraints on the set of inputs. Such constraints may allow for internal-processing to be simplified (cf. Gigerenzer's notion of 'fast and frugal heuristics') or even absent (cf. Gibson's notion of 'direct perception').
(2) EEC allows the implementation of human computation to extend beyond the 'biological skin bag' (cf. Clark's notion of the 'extended mind', and Hutchins' notion of 'distributed cognition'). As a consequence less computational demand may be imposed on the biological brain. Although I believe that (1) can significantly aid our understanding of human rationality, I will argue that (2) does little to circumvent the problem of computational intractability. If EEC researchers wish to pursue along the lines of (1), I do call on them to develop detailed models of what constitute 'ecological inputs' and perform the required tractability analyses.

Research Description
My research interests span the cognitive sciences¾including cognitive psychology, AI, philosophy and computer science. My main tools for research are mathematical modeling, conceptual analysis, experimentation and simulation. I obtained my Masters degree in experimental psychology at the Radboud University Nijmegen in 1998. My MA research involved a conceptual and empirical investigation of the dynamical systems approach to human cognition (van Rooij, Bongers, & Haselager, 2002; Haselager, Bongers, & van Rooij, 2003). After teaching for a year at the Radboud University of Nijmegen, I started my Ph.D. studies at the University of Victoria, Canada, in September 1999. My dissertation research involved an interdisciplinary project on the role of computational complexity theory in cognitive theory formation (van Rooij, 2003, van Rooij, Stege, & Kadlec, in press). During my four-year stay in Canada, I collaborated with researchers in psychology and computer science on the Traveling Salesperson problem (van Rooij, Stege, Schactman, 2003), double dissociation methodology (Kadlec & van Rooij, 2003) and parameterized complexity theory (Stege, van Rooij, Hertel & Hertel, 2002).
Since October 2003, I hold a position as postdoctoral researcher in the department of Human Technology Interaction at the Eindhoven University of Technology. Research topics I am currently working on include: Belief-desire modeling and the frame problem (with Catholijn Jonker and Pim Haselager); the nature of psychological explanation (with Cory Wright); limiting roles of publication bias and null-hypothesis testing (with Marc van Buiten and Tibor van Rooij); and the Tractable Cognition thesis.