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

  Browse A-Z

Log in

Cambridge users (raven)details Other usersdetails No account?details

Information on

Subscribing to talksdetails Finding a talkdetails Adding a talkdetails Disseminating talksdetails Help and Documentationdetails

Computational Neuroscience Journal Club

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

If you have a question about this talk, please contact Luke Johnston.

Please join us for our fortnightly Computational Neuroscience journal club on Tuesday 20th June at 2pm UK time in the CBL seminar room, or online on zoom. The title is ‘Geometry of neural population responses’, presented by Yashar Ahmadian and Puria Radmard.

Zoom information: https://eng-cam.zoom.us/j/84204498431?pwd=Um1oU284b1YxWThObGw4ZU9XZitWdz09 Meeting ID: 842 0449 8431 Passcode: 684140

Summary:

Efficient coding theories have long hypothesized that neural representations should minimize redundancy and correlations between neurons’ responses (at least when noise does not dominate). Low correlations are equivalent to a high dimensionality of the coding subspace (the subspace of the neural response space spanned by signals). On the other hand, in many experiments stimulus representations and task relevant variables seem to furnish a low-dimensional representation. However, this might simply reflect the artificially low complexity of typical experiments (e.g. due to a small or low-dimensional set of stimuli, or simple tasks). In the first paper we will present, Stringer et al. (2019) present a large ensemble of natural images to mice and record simultaneously from thousands of neurons in their V1. They find a high-dimensional representation of natural images, with a signal covariance spectrum that drops as a scale-invariant power-law 1/n. Moreover they mathematically derive a bound on the decay of this spectrum for smooth population codes on d-dimensional stimulus manifolds, and conclude that the mouse V1 operates close to this limit for large d. Wang and colleagues (2023) extend this analysis to whole brain activity in zebrafish. In this case, they find that dimensionality remains well above the critical limit for spontaneous and behaviour related activity, both for the whole brain and for random subsets of neurons. They further use a Euclidean Random Matrix model to provide a functional embedding for the neural population, showing anatomical clustering in this space.

References:

[1] Stringer et al., 2019: High-dimensional geometry of population responses in visual cortex (https://www.nature.com/articles/s41586-019-1346-5)

[2] Wang et al., 2023: The scale-invariant covariance spectrum of brain-wide activity (https://www.biorxiv.org/content/10.1101/2023.02.23.529673v1)

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.

Other lists

farmasiz miTalks Cambridge Global Food Security

Other talks

Statistics Clinic Easter 2023 V Tea and Coffee Break Some analytic questions arising in quantum information Modeling traffic jam and growth process of neurons using isogeometric analysis and physics-informed neural network Spin model of collective animal behavior: from whole group to a single animal and back Data-driven modeling of collective behavior in schooling fish
 

© 2006-2024 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity