BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Craik Club
SUMMARY:Bayesian inference with probabilistic population c
odes - Alexandre Pouget\, University of Rochester
DTSTART;TZID=Europe/London:20060321T130000
DTEND;TZID=Europe/London:20060321T140000
UID:TALK4754AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/4754
DESCRIPTION:Recent psychophysical experiments indicate that hu
mans perform near-optimal Bayesian inference in a
wide variety of tasks\, ranging from cue integrati
on to decision-making to motor control. This impli
es that neurons both represent probability distrib
utions and combine those distributions according t
o a close approximation to Bayes rule. At first si
ght\, it would appear that the high variability in
the responses of cortical neurons would make it d
ifficult to implement such optimal statistical inf
erence in cortical circuits. I will show that\, in
fact\, this variability generates probabilistic p
opulation codes which represent probability distri
butions over the encoded stimulus. Moreover\, when
the neural variability is Poisson-like\, as is th
e case in cortex\, a broad class of Bayesian infer
ence\, such as cue integration\, or integrating ev
idence for decision-making \, can be closely appro
ximated with simple linear combinations of probabi
listic population codes. Therefore\, this theory s
uggests that the Poisson-like variability in the c
ortex greatly simplifies Bayesian inference in neu
ral circuits.
LOCATION:Seminar Room (ground floor)\, Craik-Marshall Build
ing
CONTACT:Cordula Becker
END:VEVENT
END:VCALENDAR