Spatiotemporal Localization of Efficient Neural Coding Correlates Along the Somatosensory Pathway
OPEN ACCESS
Loading...
Author / Producer
Date
2022
Publication Type
Doctoral Thesis
ETH Bibliography
yes
Citations
Altmetric
OPEN ACCESS
Data
Rights / License
Abstract
Efficient coding of the sensory stream is critical for navigating and thriving in complex environments. Such an efficient coding strategy should minimize the amount of neural activity required to transmit maximal peripheral information. Devoting neural resources to peripheral stimuli that are salient, novel, or surprising may align with this strategy. Here we investigate the correlates of efficient coding through extracellular recordings made along the somatosensory pathway combined with peripheral stimulation protocols targeting novel and surprising stimuli. We first probed the neural response to Bayesian surprise, identifying that sensitivity to Bayesian surprise first emerges cortically, and displays a precise spatiotemporal representation in the different layers of the primary and secondary somatosensory cortices. Correlates of Bayesian surprise were identified in the multiunit activity, evoked potentials, and time-frequency representations in time windows ranging from ten to two-hundred milliseconds after the arrival of the surprising stimulus. Further, we found surprise sensitivity across wakefulness states, indicating that even the more complex computations underlying Bayesian inference may be fundamentally embedded in cortical circuits. We then investigated the effects of several specific forms of sensory adaptation on neural activity. Such adaptation describes a reduction in neural responses to frequently presented stimuli, which could promote a sparser, more efficient encoding of the sensory environment with an emphasis on novel stimuli. Here we specifically investigate stimulus-specific and stimulus-non-specific adaptation to quantify the extent to which stimulation of a primary stimulus adapts the neural response to a secondary stimulus. We find that, unlike the finer computations captured in Bayesian surprise representations, both aforementioned forms of sensory adaptation are observed already in the somatosensory thalamic nuclei and are further amplified in the multiunit activty and evoked potentials recorded in the distinct layers of the primary somatosensory cortex. Given these findings, our experiments indicate that the use of anaesthetized subjects for electrophysiology experiments can lead to advances in understanding of the fundamental properties of efficient neural processing. We therefore further investigated depth of anaesthesia monitoring strategies in mice. Here we identified interhemispheric somatosensory coherence as a strong indicator of anaesthetic depth and established a proof of concept monitoring system for adult mice which could support optimally controlled, repeatable experiments while also reducing the burden on laboratory animals. Overall, our findings provide important insights into the modulation of neural responses elicited by the surprise generated by Bayesian inference, the sensory statistical environment, and the administration of anaesthesia and advance the understanding of efficient neural coding strategies across brain states.
Permanent link
Publication status
published
External links
Editor
Contributors
Examiner: Yanik, Mehmet Fatih
Examiner: von der Behrens, Wolfger
Examiner : Stephan, K. E.
Examiner : Mante, V.
Book title
Journal / series
Volume
Pages / Article No.
Publisher
ETH Zurich
Event
Edition / version
Methods
Software
Geographic location
Date collected
Date created
Subject
Bayesian Surprise; Cortical Computation; Efficient Coding; Anaesthesia
Organisational unit
09474 - Yanik, Mehmet Fatih / Yanik, Mehmet Fatih