BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Optimal encoding and decoding in sensory populations - Eero Simonc
 elli (New York University\, HHMI)
DTSTART:20120502T090000Z
DTEND:20120502T100000Z
UID:TALK37374@talks.cam.ac.uk
CONTACT:Prof Máté Lengyel
DESCRIPTION:Experimental evidence suggests that human judgments of many pe
 rceptual attributes are consistent with Bayesian estimation\, in which noi
 sy sensory measurements are combined with prior knowledge of the environme
 ntal distribution of those attributes.  How are such computations achieved
  in the brain? I'll first describe a means by which populations of neurons
  can efficiently encode a scalar sensory variable.  The structure of this 
 representation\, which can be derived in closed form based on the prior di
 stribution of the relevant sensory variable\, provides an implicit encodin
 g of the prior\, and is consistent with both physiological and perceptual 
 data for a variety of sensory  attributes.  I'll then describe a novel dec
 oder that can approximate the Bayes least squares estimate\, converging to
  the true value as the neural population size increases.  The decoder is n
 eurally plausible\, and requires knowledge only of the preferred stimuli a
 nd a fixed filter\, and not the prior distribution or family of tuning cur
 ves.  I'll end by discussing the means by which the neural populations mig
 ht go about learning these representations.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR