Talk by Dr. James Isbister (CSN Virtual Seminar)

We hereby announce the next talk in 'CSN Virtual Seminar' series

In vivo evidence of time-warped cortical spike time patterns and a large-scale model of the non-barrel primary somatosensory cortex

Speaker: Dr. James Isbister, École Polytechnique Fédérale de Lausanne, Switzerland

Abstract

Neurons communicate rapidly to support survival by working together in groups to create sequences of spikes lasting 10s of milliseconds. The principles by which information is encoded by such sequences, remain to be determined, but can be explored through the study of sequences elicited by sensory stimuli. Such investigation is challenging because the pattern of spikes is different each time a particular stimulus is presented. That is, there isn't a fixed correspondence between a piece of information and its neural representation. Historically, this led many researchers to conclude that information is encoded by the number of spikes rather than their precise timings. Many modern analyses offer an intermediate between spike count and spike time encoding using bins of 10-20ms. The prospect of a finer spike timing code remains attractive, however, due to its potential speed, efficiency, capacity, and the sensitivity of neuronal integration.

In our published [1] and more recent work, we found evidence for a novel spike time code in the rodent barrel cortex, and an explanation for why millisecond precise spike time patterns are rarely reported. Like spike count and dimensionality reduced representations, we showed that precise spike time patterns are modulated by the level of background activity on a single trial. In particular, evoked spike time patterns were compressed or stretched in time (i.e. “time-warped’’) depending on the background activity level. Importantly, the timing differences between the spike times of different neurons, which could be decodable by downstream neurons, were also modulated. We hypothesised that downstream neurons could condition decoding of “time-warped spike time patterns’’ on the background activity level. Our work focuses on spike sequences in which each neuron emits a single spike, which may be an atomic form of information representation in the brain.

The talk will also briefly introduce the Blue Brain Project’s recent and openly available model of the non-barrel primary somatosensory cortex [2], which we will use in parallel to explore how information is parsed into precise representations. Towards this, a novel calibration algorithm was developed which rapidly finds a spectrum of in vivo-like regimes without changing internal model parameters. The algorithm closes the loop on iterative model refinement and creates a mathematical notation for investigation of large-scale simulations. The model exhibits a variety of in vivo-like properties including spatially correlated fluctuations during spontaneous activity, accurate layer-wise population responses to whisker deflections and selective propagation of stimulus-evoked activity to downstream areas.

[1] Isbister, J. B., Reyes-Puerta, V., Sun, J. J., Horenko, I., & Luhmann, H. J. (2021). Clustering and control for adaptation uncovers time-warped spike time patterns in cortical networks in vivo. Scientific Reports.

[2] Isbister, J. B., Ecker, A., Pokorny, C., Bolaños-Puchet, S., Santander, D. E., … & Reimann, M. W. (2023). Modeling and Simulation of Neocortical Micro-and Mesocircuitry. Part II: Physiology and Experimentation. bioRxiv, 2023-05.