University of Cambridge > Talks.cam > Computational Neuroscience > Computational Neuroscience Journal Club

Computational Neuroscience Journal Club

Add to your list(s) Download to your calendar using vCal

  • UserZahara Girones and Ivan Tomic
  • ClockTuesday 04 May 2021, 15:00-16:30
  • HouseOnline on Zoom.

If you have a question about this talk, please contact Jake Stroud.

Please join us for our fortnightly journal club online via zoom where two presenters will jointly present a topic together. The next topic is ‘Efficient coding and natural scenes representation’ presented by Zahara Girones and Ivan Tomic.

Zoom information: https://us02web.zoom.us/j/84958321096?pwd=dFpsYnpJYWVNeHlJbEFKbW1OTzFiQT09

The efficient coding hypothesis has been successful in explaining various aspects of sensory neural coding. According to this hypothesis, sensory systems are optimized to maximize the transmitted information (or some other measure of coding fidelity) about the natural environment, under biological resource constraints. Bayesian models of perception, which view perception as inference from noisy sensory evidence merged with prior expectations, have also been successful in accounting for various perceptual biases and illusions. However, some perceptual biases appear to have an anti-Bayesian character. We will review a series of papers that explain this and other perceptual phenomena by combining Bayesian decoding with efficient sensory encoding. In these models both the encoder (via efficient coding) and the Bayesian decoder (via the prior distribution) are adapted to natural stimulus statistics. We will discuss how these theories relate the variations of perceptual bias and psychophysical discriminability over stimulus space to each other and to the distribution of stimuli in the natural environment, and will discuss how different formulations of efficient coding lead to quantitatively different predictions for these relationships. At the end we will look at whether the same principles can be applied to domains beyond low-level perceptual processing, such as subjective valuations.

List of references:

D. Ganguli, E. P. Simoncelli. Efficient Sensory Encoding and Bayesian Inference with Heterogeneous Neural Populations. Neural Comput 2014; 26 (10): 2103–2134. https://doi.org/10.1162/NECO_a_00638

X. Wei, A. Stocker. A Bayesian observer model constrained by efficient coding can explain ‘anti-Bayesian’ percepts. Nat Neurosci 18, 1509–1517 (2015). https://doi.org/10.1038/nn.4105

X. Wei & A.Stocker. Lawful relation between perceptual bias and discriminability. PNAS (2017). https://doi.org/10.1073/pnas.1619153114

M. Morais, J. W. Pillow. Power-law efficient neural codes provide general link between perceptual bias and discriminability. NeurIPS (2018). http://pillowlab.princeton.edu/pubs/Morais18_NeurIPS_powerlawefficientcoding.pdf

R. Polanía, M. Woodford & C.C. Ruff. Efficient coding of subjective value. Nat Neurosci 22, 134–142 (2019). https://doi.org/10.1038/s41593-018-0292-0

This talk is part of the Computational Neuroscience series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2021 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity