BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Neural representations of uncertainty: the right tool for the righ
 t job - Cristina Savin (IST Austria)
DTSTART:20161117T121500Z
DTEND:20161117T131500Z
UID:TALK69229@talks.cam.ac.uk
CONTACT:Prof Máté Lengyel
DESCRIPTION:In many situations humans and animals seem to use uncertainty 
 information to guide close to optimal behaviour. The neural underpinnings 
 of such probabilistic computation remain highly controversial (Fiser et al
 \, 2010\; Pouget et al\, 2013). Contrary to the traditional view of probab
 ilistic population codes (PPCs) and sampling-based representations as mutu
 ally exclusive alternatives\, here we argue that both are useful data stru
 ctures\, needed at different computational stages for (approximately) opti
 mal decision making. First\, sampling allows neural circuits to represent 
 the joint statistics over many features and to marginalise out nuisance va
 riables (Savin and Deneve\, 2014). Second\, PPCs enable a compact\, quasi-
 instantaneous representation of the marginals which can be easily combined
  with cost-related information to yield approximatively optimal decisions.
 \n\nWe propose a model in which the two are linked via a probabilistic for
 m of evidence integration akin to that in (Boerlin et al\, 2011)\, with th
 e ‘evidence’ given by samples from the target distribution. We use the
  same spike-based code (Boerlin\, 2011) for representing both the sampling
 \, and the evidence-integration modules at the level of neural activity. T
 he core idea is that the circuit evolves through recurrent dynamics such t
 hat the relevant signal (either samples from a target distribution\, or a 
 PPC-like representation of any of its marginals) can be linearly decoded f
 rom neural responses. This shared encoding means that many classic measure
 s of single cell responses are consistent with experimental data and\, mor
 eover\, are preserved across processing stages. Thus the model provides a 
 computationally well-justified reconciliation between competing probabilis
 tic neural codes and points to population-level analyses as tools for vali
 dating probabilistic computation experimentally.  \n
LOCATION:Cambridge University Engineering Department\, CBL\, BE-438 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR