UCL logo
skip to navigation. skip to content.

Gatsby Computational Neuroscience Unit




UCL Home
  • UCL Home
  • UCL Gatsby Computational Neuroscience Unit
UCL Gatsby Unit
  • introduction
  • people
  • research
  • publications
  • courses
  • phd programme
  • events
  • directions
  • greater gatsby
  • vacancies
  • Internal
  • ucl

 

 

  • Home
  • Staff & Students
  • Vacancies

 

Gasper Tkacik

 

IST(Institute of Science and Technology) Austria

(http://wp.ist.ac.at/group_tkacik/)

 

Wednesday 18th April 2012

2 - 3pm

 

B10 Seminar Room, Basement,

Alexandra House, 17 Queen Square, London, WC1N 3AR

 

 

Retinal metric: a stimulus distance measure derived from population neural responses

 

 

Abstract:

On a single neuron level, the mapping from stimulus to spikes is often described as linear filtering in the neuron's receptive field followed by a non-linear response function, indicating that distinguishable responses are produced only when the stimuli differ along a small number of directions in the stimulus space. It is unclear how this picture generalizes to a population of interacting neurons encoding the same stimulus. Does the population use the diversity of neural sensitivities to distinguishably represent all possible stimuli, or is it inherently able to discriminate much better between some pairs of stimuli than between others? Here we address this question by inferring a distance measure on the stimulus space directly from the recorded activity of 100 neurons in the salamander retina. In contrast to previously used measures of stimulus similarity, the ``retinal distance'' tells us precisely how distinguishable, given the noise in population neural responses, a pair of stimulus clips is to the retina. We find that the retinal distance strongly deviates from Euclidean, and indeed from any distance with a static metric, yet it nevertheless has a simple structure that relies on a small number of stimulus projections. By constructing a mathematical model for the retinal distance we show that its non-Euclideanity has important consequences for decoding and inferring feature sensitivity of whole neural populations.

 

 

 

 

 

 

  • Disclaimer
  • Freedom of Information
  • Accessibility
  • Privacy
  • Advanced Search
  • Contact Us
Gatsby Computational Neuroscience Unit - Alexandra House - 17 Queen Square - London - WC1N 3AR - Telephone: +44 (0)20 7679 1176

© UCL 1999–20112011