Embedded Files
Post date: Oct 21, 2012 6:50:18 PM
Can we find common principles for real-time cognitive processing across multiple domains (perceptual and reward-based choices, lexical decisions, memory tasks, etc.)?
Real-time measurements of cognitive tasks (say, measuring eye movements or kinematics for reaching and pressing a response button during a perceptual decision or a memory task) reveal surprising complex dynamics (Spivey, 2007). Computational models have long recognized that decisions consist in a dynamic and competitive process among (two or more) alternatives (Ratcliff, 1978). However, they leave open the problem if decision and action systems are segregated and arranged serially (decisions first, actions after) or interact.
Most studies have shown that, at the neuronal level, competition among choices (say, concerning a perceptual decision) can be realized as the bottom-up accumulation of evidence up to a threshold (Shadlen et al. 2008) or using a predictive coding scheme with both top-down and bottom-up processes (Friston, 2010). Evidence indicates that when possible (i.e. when each decision maps to a specific actions) the choice is implemented (or at least supported by) the same neuronal structures that control the overt actions. For example, in the choice between two actions, the competition consists in the parallel specification and selection among decision and/or action alternatives, which are maintained in parallel in (premotor) neuronal sub-populations until one wins (Cisek, 2007). This view is compatible with the idea that actions are prepared and even initiated before decisions are completed (the action system could either help computing the choice itself, or receive a "continuous flow" of information from the decision system, see Coles et al., 1985, which is plausibly useful for preparation).
This view can be extended to more complex choices, when decisions do not map directly to actions or actions are delayed. In this case, the continuous flow is not for the effectors but for neuronal systems that implement subsequent decisions (Shadlen et al. 2008). Or, multiple neuronal populations (plausibly representing choices and actions at different levels) could work synergistically rather than being segregated and arranged serially, say "decision first and action specification only after" (Cisek, 2012).
Real-time measurements of cognitive tasks exploit this continuity of decisions and action. Measuring the kinematics of movements during a choice (say, clicking one of two buttons with the mouse) actually "paints" the decisions on the screen. The more uncertain the choice, the more its trajectory is "attracted" by the competing alternative (the non-selected button); and of course velocity and acceleration profiles are highly uncertainty-dependent, too. These methods thus permit to "see" the real-time dynamics of decision much better than traditional methods that only measure response times and errors.
The mousetracker software (Freeman, 2010) is a very simple but powerful way to perform real-time measurements of cognitive tasks (mouse movements when subjects chose by clicking one among two or more buttons on the screen). We are using it to perform human experiments in multiple domains: lexical decisions (Barca and Pezzulo, 2012), categorization of ambiguous figures, competition among object affordances, semantic priming paradigms (more info as new papers get accepted).
Real-time mouse trajectories in a Lexical Decision task (Barca & Pezzulo, 2012) [link]
Ours and other studies of decision-making across several domain reveal that common principles (dynamic, competitive and proactive) could be at work (Song and Nakayama, 2009; Spivey, 2007). They show that cognitive processing is not segregated into the stages of perception, decision and the action. Rather, action starts before the decision is completed, and it produces new perceptions (and proprioceptions) that in turn change the decision.
We have hypothesized that this can be part of a continuous process of "proactive action preparation" (Pezzulo and Ognibene, 2011), and that living organisms take actions that prepare them in the best way to the next choices, even if these choices are unclear, rather than just waiting for the decision to be completed. In turn, this ability capitalizes on prior information and prospective abilities.
Selected Pubs:
Barca, L. and Pezzulo, G. (2012). Unfolding visual lexical decision in time. PLoS ONE, 7(4):e35932. [link]
Ferro, M., Ognibene, D., Pezzulo, G., and Pirrelli, V. (2010). Reading as active sensing: a computational model of gaze planning during word discrimination. Frontiers in Neurorobotics, 4(6). [link]
Pezzulo, G. and Ognibene, D. (2011). Proactive action preparation: Seeing action preparation as a continuous and proactive process. Motor Control, 16(3):386–424.
Other Pubs:
Cisek, P. (2007) Cortical mechanisms of action selection: the affordance competition hypothesis Phil. Trans. R. Soc. B., 362, 1585-1599
Cisek, (2012) P. Making decisions through a distributed consensus. Curr Opin Neurobiol
Coles, M. G.; Gratton, G.; Bashore, T. R.; Eriksen, C. W. & Donchin, E. A (1985) Psychophysiological investigation of the continuous flow model of human information processing. J Exp Psychol Hum Percept Perform, 11, 529-553
Freeman, J.B. & Ambady, N. (2010). MouseTracker: Software for studying real-time mental processing using a computer mouse-tracking method. Behavior Research Methods, 42, 226-241.
Friston, K. (2010). The free-energy principle: a unified brain theory? Nat Rev Neurosci, 11, 127-138
Ratcliff, R. (1978) A theory of memory retrieval Psychological Review, 85, 59-108
Shadlen, M.; Kiani, R.; Hanks, T. & Churchland, A. (2008) Neurobiology of Decision Making: An Intentional Framework. Engel, C. & Singer, W. (Eds.) Better than conscious? decision making, the human mind, and implications for institutions, The MIT Press
Song, J.-H. & Nakayama, K. (2009) Hidden cognitive states revealed in choice reaching tasks. Trends Cogn Sci, 13, 360-366
Spivey, M. (2007) The continuity of mind Oxford University Press, USA
Google Sites
Report abuse