Check out the newest Neuroforum issue of the German Neuroscience Society and learn more about the members of the MODOLFOR research unit 5424 "Modulation in Olfaction" and their scientific interests (in German):
nwg-info.de/sites/nwg-in...
7/7
This project has been in the works for quite some time, so we’re very happy to finally share it. Huge thanks to everyone involved — and very curious to hear people’s thoughts and feedback!
#neuroskyence #olfaction #computationalneuroscience
6/7 These findings reframe how we think about early olfactory processing.
Rather than relying primarily on complex recurrent circuitry, much of olfactory bulb decorrelation may emerge from scalable feedforward gain control, with lateral interactions further refining representations.
1/7 In olfaction, odour representations are thought to become decorrelated from sensory input to output. But how systematic is this transformation, and what circuitry generates it?
Using 2P imaging, we recorded OB input and output across a diverse odour panel to map transformations in odour space.
5/7 The model suggests a simple principle:
Channels carrying unique information are amplified, while redundant channels are attenuated.
This feedforward scaling mechanism alone reproduced a substantial fraction of the experimentally observed decorrelation.
4/7 We then asked: What circuit motifs are needed to generate this transformation?
Surprisingly, modelling showed that much of the decorrelation could already be reproduced without lateral connectivity. Simple channel-specific gain modulation performed nearly as well as unconstrained network models.
3/7 This transformation had direct functional consequences: linear classifiers decoded odour identity more accurately from output than input activity.
By computationally transforming the representational geometry, we found that improved decoding accuracy was directly linked to decorrelation itself.
2/7 Odour representations became substantially less correlated from input to output. Response variance was preserved, dimensionality increased, and output activity occupied a higher-dimensional space.
Thus, the bulb decorrelates odour representations without simply suppressing activity.
🚨 Preprint alert! 🚨
How does the olfactory bulb transform overlapping sensory inputs into representations that are easier to distinguish?
New study by Sina Tootoonian, Yikai Yang, @andreas-t-schaefer.bsky.social @crick.ac.uk and myself.‬
www.biorxiv.org/content/10.6...
A short 🧵 below
