BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Mind reading by machine learning: an ideal observer based analysis
  of cognitive scientific experiments - Ferenc Huszar (Budapest University 
 of Technology and Economics &amp\; Collegium Budapest)
DTSTART:20090224T100000Z
DTEND:20090224T110000Z
UID:TALK17105@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:A central challenge in cognitive science is to measure and qua
 ntify experimentally the mental representations humans (and other animals)
  develop -- in other words\, to "read" subjects' minds. In order to elimin
 ate potential biases in reporting mental contents due to verbal elaboratio
 n\, subjects' responses in experiments are often limited to simple binary 
 decisions or discrete choices that do not require conscious reflection upo
 n their mental contents. However\, it is unclear what such impoverished da
 ta can tell us about the potential richness and dynamics of subjects' ment
 al representations. To address this problem\, we used ideal observer model
 s that formalise choice behaviour as a quasi-optimal (stochastic) function
  of subjects' representations in long-term memory\, acquired through prior
  learning\, and the information currently available to them. Bayesian inve
 rsion of such models allowed us to infer subjects' mental representation f
 rom their choice behaviour in a task as simple as the standard one-back ta
 sk -- in which successively presented items have to be judged as being the
  same or different. In comparison with earlier methods developed along sim
 ilar lines (eg. Sanborn & Griffiths\, NIPS 2008)\, our method does not req
 uire the introduction of several trials of a special-purpose psychophysics
  task and thus has sufficient temporal resolution to track as mental repre
 sentations develop through learning. \nIn this talk I will introduce our s
 tatistical generative model for subject's behaviour in cognitive scientifi
 c experiments and present Markov chain Monte Carlo inference under a simpl
 ified model\,where the subject's mental representation is assumed to remai
 n constant through the analysed segment of the experiment.\nAs subjects ar
 e assumed to learn during the experiment\, it is desired to allow for chan
 ges in mental representation during the course of the experiment. I will p
 resent the challenge of incorporating temporal dynamics of mental represen
 tations in the present statistical model.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR