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: a doubly Bayesian method for inf
 erring mental representations - Máté Lengyel  (University of Cambridge)
DTSTART:20101118T140000Z
DTEND:20101118T150000Z
UID:TALK26521@talks.cam.ac.uk
CONTACT:Michal Kosinski
DESCRIPTION:A central challenge in cognitive science is to measure and qua
 ntify the mental representations humans develop -- in other words\, to ‘
 read’ subject’s minds. In order to eliminate potential biases in repor
 ting mental contents due to verbal elaboration\, subjects’ responses in 
 experiments are often limited to simple binary decisions or discrete choic
 es that do not require conscious reflection upon their mental contents. Ho
 wever\, it is unclear what such impoverished data can tell us about the po
 tential richness and dynamics of subjects’ mental representations. To ad
 dress this problem\, we used ideal observer models that formalise choice b
 ehaviour as (quasi-) Bayes-optimal\, given subjects’ representations in 
 long-term memory\, acquired through prior learning\, and the stimuli curre
 ntly available to them. Bayesian inversion of such ideal observer models a
 llowed us to infer subjects’ mental representation from their choice beh
 aviour in a variety of simple psychophysical tasks. The inferred mental re
 presentations also allowed us to predict future choices of subjects with r
 easonable accuracy\, even in tasks that were different from those in which
  the representations were estimated. These results demonstrate a significa
 nt potential in standard binary decision tasks to recover detailed informa
 tion about subjects’ mental representations. Extensions of this framewor
 k to inferring dynamical changes in mental representations\, and to optima
 l experimental design will also be discussed.\n\nThis is joint work with F
 erenc Huszar\, Neil Houlsby (CBL\, Dept Engineering\, U Cambridge) and Uta
  Noppeney (MPI Tuebingen).\n
LOCATION:Seminar Room\, PPSIS\, New Museums Site
END:VEVENT
END:VCALENDAR