BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian integration in sensorimotor learning - Keith Vertanen (Un
 iversity of Cambridge)
DTSTART:20091030T110000Z
DTEND:20091030T120000Z
UID:TALK19972@talks.cam.ac.uk
CONTACT:Emli-Mari Nel
DESCRIPTION:This paper is by Konrad K&ouml\;rding and Daniel Wolpert\, Nat
 ure (2004).  Available here:\nhttp://homepages.inf.ed.ac.uk/svijayak/teach
 ing/MLSC/HW2papers/Wolpert.pdf\n\nABSTRACT\n\nWhen we learn a new motor sk
 ill\, such as playing an approaching tennis ball\, both our sensors and th
 e task possess variability. Our sensors provide imperfect information abou
 t the ball’s velocity\, so we can only estimate it. Combining informatio
 n from multiple modalities can reduce the error in this estimate1–4. On 
 a longer time scale\, not all velocities are a priori equally probable\, a
 nd over the course of a match there will be a probability distribution of 
 velocities. According to bayesian theory an optimal estimate results from 
 combining information about the distribution of velocities—the prior—w
 ith evidence from sensory feedback. As uncertainty increases\, when playin
 g in fog or at dusk\, the system should increasingly rely on prior knowled
 ge. To use a bayesian strategy\, the brain would need to represent the pri
 or distribution and the level of uncertainty in the sensory feedback. Here
  we control the statistical variations of a new sensorimotor task and mani
 pulate the uncertainty of the sensory feedback. We show that subjects inte
 rnally represent both the statistical distribution of the task and their s
 ensory uncertainty\, combining them in a manner consistent with a performa
 nce-optimizing bayesian process. The central nervous system therefore empl
 oys probabilistic models during sensorimotor learning.\n
LOCATION:TCM Seminar Room\, Cavendish Laboratory\, Department of Physics
END:VEVENT
END:VCALENDAR