B1 Transformation of input correlations to V1 spike correlations and behavioural choice in the awake monkey

(a) Example of a natural image used as stimulus with receptive field outlines of 13 simultaneously recorded neurons. (b) Using probabilistic models we will map out how network interactions in neuronal populations change with differences in the stimulus statistics.

Neural representations in the early visual system are assumed to be adapted to the statistics of the sensory input. For example, functional properties of V1 neurons have been related to normative coding principles suggesting that neural responses should be as statistically independent as possible. The goal of this project is (1) to derive precise predictions from the redundancy reduction principle for input driven correlations in neural population responses and (2) to determine empirically how correlations in the sensory input are transformed into correlations across populations of neurons in V1 in awake, behaving macaques. To this end, we record from populations of well-isolated single neurons using chronically implanted tetrode arrays while a monkey is viewing natural images or textures with precisely controlled correlation structure. Subsequently, we apply probabilistic models to the population activity to quantitatively link input and output correlations.


Project leaders: Matthias Bethge, Andreas Tolias
Coworkers: Niklas Luedtke*, Philipp Berens, Alexander Ecker
* funded by BMBF


Key publications:

  • Ecker AS, Berens P, Keliris GA, Bethge M, Logothetis NK, Tolias AS (2010). Decorrelated Neuronal Firing in Cortical Microcircuits. Science 327:584-­‐587
  • Eichhorn J, Sinz F, Bethge M (2009) Natural image coding in V1: how much use is orientation selectivity? PLoS Comput. Biol 5:e1000336
  • Macke JH, Berens P, Ecker AS, Tolias AS, Bethge M (2009). Generating Spike Trains with Specified Correlation Coefficients. Neural Computation 21:1-­‐27
 
Sponsored by the Federal Ministry of Education and Research