Indeterminacy and cognitive control in neural models of choice under risk

Many neuroscientists have been recently investigating the conditions under which classical expected utility (CEU) fails. One central motivation is that `[…] if we succeed in understanding mechanistically how choices that violate CEU are made at a neural level, then a new global theory of choice will be developed’ (Glimcher and Rustichini, 2005). For example, an ambiguity-sensitive mechanism (associated with the expression of emotion) has been proposed by Damasio and associates in order to explain the so-called Ellsberg’s phenomenon as well as to explain deficiencies in choice exhibited by patients with lesions to the VMPFC. And Tversky and Fox have proposed especial psychological effects, like comparative ignorance to explain Ellsberg.

The first part of the talk focuses on describing two experiments (with CMU students) refuting some of the existing psychological explanations of behavior in the Ellsberg scenario (Arló-Costa and Helzner, 2005). I suggest that a normative theory of choice under risk along the lines suggested by Ellsberg himself is perfectly capable of accommodating the existing data.

The second part focuses on offering a theoretical account of the most salient aspects of cognitivee control linking the process of deliberation and the act of choosing (Arló-Costa, 2003; Sen, 1997), which, I argue, is essential to apply normative accounts of choice. In this talk I focus on learning effects. In this regard I present an experiment in collaboration with NIH/NINDS where VMPFC patients play a sequential version of a trust game (Arló-Costa, et al, 2006). Patients play against an algorithm designed to switch polarity of cooperation/non-cooperation every time that the patient exhibits a robust tendency to non-cooperation/cooperation in the game. This is tantamount to inject a component of reversal learning into a situation of social interaction. The underlying idea is to provide a unified account of choice deficiencies for VMPFC patients (both at the social and non-social levels) in terms of reversal learning effects (compatible with empirical and methodological findings presented in (Fellows and Farah, 2003) and (Maia and McClelland, 2004)). Time permitting, I intend to sketch on-going research focusing on other aspects of cognitive control in choice (like partition formation, heuristics and tie resolution).

Arló-Costa, H., Grafman, J. and Maia, T. `Assessing Reversal Learning in the Social Domain in Patients with Lesions to the Ventromedial Prefrontal Cortex,' CMU-NIH 2006.

Arló-Costa, H., Helzner, J. `Comparative Ignorance and the Ellsberg Phenomenon,' ISIPTA'05, 2005. http://www.sipta.org/isipta05/proceedings/043.html.

Arló-Costa, H. `Integrating Emotion and Rationality in Behavioral Models of Decision Making,’ Recent Advances in Artificial Intelligence, Ingrid Russell and Susan Heller (eds.), 303-09, 2003.

Fellows, L.K. and Farah, M. J. Ventromedial Frontal Cortex Mediates Affective Shifting in Humans: Evidence from a reversal learning paradigm. Brain, 126, 1830-1837, 2003.

Glimcher, P. W. and Rustichini, A. Neuroeconomics: The Consilience of Brain and Behavior. Science, 306, 2004.

Maia, T. and McClelland, J. A Reexamination of the Evidence of the Somatic Marker Hypothesis: What participants really know in the Iowa task. PNAS, 101, 45, 16075–16080, 2004.

Sen, A. Maximization and the Act of Choice. Econometrica 65, 745-779, 1997.