BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian learning of visual chunks by human observers - Dr Mate Le
 ngyel (University of Cambridge)
DTSTART:20080213T140000Z
DTEND:20080213T150000Z
UID:TALK10492@talks.cam.ac.uk
CONTACT:Philip Sterne
DESCRIPTION:Efficient and versatile processing of any hierarchically struc
 tured information requires a learning mechanism that combines lower level 
 features into higher level chunks. We investigated this chunking mechanism
  in humans with a visual pattern-learning paradigm. Based on Bayesian mode
 l comparison\, we developed an ideal learner that extracts and stores only
  those chunks of information that are minimally sufficient to encode a set
  of visual scenes. Our ideal Bayesian chunk learner not only reproduced th
 e results of a large set of previous empirical findings in the domain of h
 uman pattern learning\, but also made a key prediction that we confirmed e
 xperimentally. In accordance with Bayesian learning but contrary to associ
 ative learning\, human performance was well above chance when pair-wise st
 atistics in the exemplars contained no relevant information. Thus\, humans
  extract chunks from complex visual patterns by generating accurate yet ec
 onomical representations and not by encoding the full correlational struct
 ure of the input.
LOCATION:TCM Seminar Room\, Cavendish Laboratory\, Department of Physics
END:VEVENT
END:VCALENDAR