Flexible population coding of odors in the olfactory bulb: A biologically plausible multi-odor spiking network model with e-prop

Abstract

Understanding the neural mechanisms underlying odor representation and memory in the olfactory bulb (OB) is essential for advancing our knowledge of sensory processing and plasticity. Building on previous work and in vivo findings [1,2], we have extended our recurrent spiking neural network (SNN) model — implemented in the NEST simulator [3] and trained with eligibility propagation (e-prop) plasticity [4] — from binary (two-odor) discrimination to the biologically more realistic case of n distinct odors.

Our new model architecture incorporates both regular and adaptive leaky integrate-and-fire neurons, and allows simulation of cholinergic and GABAergic modulation as observed experimentally. By presenting the network with multiple odor cues, we can explore capacity, generalization, and interference phenomena reminiscent of real olfactory circuits. The model reproduces distinct, odor-specific activation patterns in simulated glomerular populations and captures modulatory effects of cholinergic and GABAergic inputs, in line with experimental observations.

Moving beyond stimulus-specific activity, we investigate the structure and dimensionality of population activity patterns evoked by multiple odors in the generalized network. Our current analyses focus on whether the network’s responses to different odors are organized along low-dimensional manifolds or within dedicated neural subspaces, and how these structures relate to the coding capacity and flexibility of the olfactory bulb. We present this biologically plausible modeling framework together with first results on the organization of odor representations at the neural population level.

References:
[1] Inna Schwarz and Monika Müller, Irina Pavlova, Jens Schweihoff, Fabrizio Musacchio, Manuel Mittag, Martin Fuhrmann, Martin K. Schwarz, The diagonal band of Broca regulates olfactory-mediated behaviors by modulating odor-evoked responses within the olfactory bulb, 2021, bioRxiv 2020.11.07.372649, doi: 10.1101/2020.11.07.372649ꜛ
[2] Monika Müller, Odor-evoked modulation of the olfactory bulb by projections from the horizontal limb of the diagonal band of Broca, Bonn, 2021, Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn. Online-Ausgabe in bonndoc: nbn-resolving.org/urn:nbn:\de:hbz:5-63028ꜛ
[3] Marc-Oliver Gewaltig and Markus Diesmann, NEST (NEural Simulation Tool), 2007 Scholarpedia, 2(4), 1430, doi: 10.4249/scholarpedia.1430ꜛ
[4] Guillaume Bellec, Franz Scherr, Anand Subramoney, Elias Hajek, Darjan Salaj, Robert Legenstein, Wolfgang Maass, A solution to the learning dilemma for recurrent networks of spiking neurons, 2020, Nature Communications, Vol. 11, Issue 1, pages n/a, doi: 10.1038/s41467-020-17236-yꜛ