Increased pattern similarity in two major olfactory cortices despite higher sparseness levels

Abstract

Pattern separation is a fundamental process that enhances discrimination of similar stimuli and can be achieved by sparsening the neural activity and expanding the coding space. Odor stimuli evoke patterns of activity in the olfactory bulb (OB) and these activity patterns are projected to several cortical regions that contain larger numbers of neurons and show sparser activity levels. However, whether these projected patterns are better separated is still unclear. Here we compared odor responses in the OB, anterior piriform cortex (aPC) and anterior olfactory nucleus (AON) to the exact same odor stimuli. We found that odor representations are more similar, noisier and less distinctive in aPC and AON than in the OB. The increase in similarity was observed for both similar and dissimilar odors. Modeling odor transformation from the OB to the olfactory cortex using simulated as well as actual OB odor responses as inputs, demonstrates that the observed rise in odor representation similarity can be explained by assuming biologically plausible variation in the number of OB inputs each cortical neuron receives. We discuss the possible advantages of our findings to odor processing in the aPC and AON. HighlightsO_LIOdor representations in the aPC and AON are more correlated despite increase in sparseness levels. C_LIO_LIOdor identity is best represented in the OB. C_LIO_LIVariance in the number of inputs from OB can explain the reduction in odor separation. C_LI