27 maggio 2019: M.L. Cappuccio: Robots that “choke”. Human performance and artificial intelligence
Talk organized by the Neurophilosophy Research Groupfor the Cognition in Action Seminar 2019
Massimiliano L. Cappuccio
Robots that “choke”. Human performance and artificial intelligence
27 maggio 2019, h. 14:30-16:30.
Room Enzo Paci, Directorate of the Department of Philosophy, Via Festa del Perdono, 7, Milan
For a long time, sport and performance psychologists have been studying the so-called “choking effect”, i.e. the well-known tendency of experts athletes to underperform in pressured-filled situations. Qualitative reports portray the experience of choking as a form of semi-paralyzing incapability to re-enact familiar motor routines. The most prominent theories (skill focus, execution focus, etc.) claim that, at the behavioral/functional level, choking is caused neither by distraction nor overload of the available attentional resources, but by the compulsive self-monitoring of well-trained sensorimotor skills. When this happens, athletes attend too closely the component processes of their own movements, reducing the fluidity and flexibility of their familiar action routines. According to dispositional reinvestment theory, compulsive tendencies to monitor one’s own actions are motivated by the involuntary, temporary regression of the agent to the beginner stage, due to performance anxiety and insecurity. Assuming that this is the psychological background of choking, what kinds of cognitive mechanisms are responsible for this phenomenon at the level of the underlying information processing? Some authors speculate that choking could be determined by an over-sensitivity of the error prediction system when motor commands are generated. In other words, because the threshold for what constitutes an error is set too low (so that the slightest deviations are tagged as errors), the motor system issues too many corrective commands, leading to freezing or a sputtered execution. In psycho-linguistics, a similar excessive issuing of corrective actions has been proposed as the cause of stuttering during speech production, a condition that presents many analogies with choking at the neuro-cognitive and behavioral level. In this paper, I suggest that modelling the phenomenon of skill disruption with synthetic methods, based on a combination of predictive processing and dynamical systems theory, can improve our understanding of the fundamental nature of skill and expertise, in multiple ways. First, from the point of view of neuro-cognitive and psychological research, such model can help us identify the characteristic features of the athletes that are most prone to choking, tailoring therapeutic intervention on their specific needs. Second, from the point of view of a biomimetic approach to robotics, it tells us what it would take for an artificial system to simulate the remarkable adaptivity and flexibility, but also the fragility, of the human expert sensorimotor skills. Third, from the point of view of phenomenology and theoretical research, it allows us to distinguish the specific roles played by the notion of consciousness, awareness, intention, and attention in the control of automatized action.
The Lectures will be held in English.
Participation is strongly recommended to students of the Doctoral School in Philosophy and Human Sciences.
Everyone interested is welcome to attend.