Dynamics of inhibitory networks in the olfactory bulb

Abstract

The representation of an odor goes through multiple transformations in different layers of the olfactory system. At the olfactory receptor neuron layer odor attributes are encoded by the identity of neurons that are activated. In the next layer, the olfactory bulb, this static code is transformed into an elaborate spatiotemporal pattern where both the identity and the timing of neuronal activity, encode the odor. Here we ask, what is the role of the structure of the olfactory bulb network in generating spatiotemporal patterns? In a number of neuronal networks inhibition plays an important role in organizing spatiotemporal patterns. For example, the structure of connections between inhibitory interneurons in the antennal lobe (the insect equivalent of the olfactory bulb) predicts specific patterns of synchrony in the network. Mitral cells, principal neurons in the olfactory bulb, interact only via an inhibitory intermediary. We simulated a model network consisting of mitral cells that inhibit each other. Reciprocal inhibition ensured that neurons that were directly connected to each other were activated at different times while those that did not inhibit each other were synchronously activated. Using a community clustering algorithm, we found clusters of mitral cells that were maximally disconnected from each other. These clusters of neurons fired in a correlated manner, while neurons belonging to different clusters spiked at different times. This structure–dynamics relationship persisted across different instances of random networks. Clusters detected using our algorithm determined the identity of neurons that spiked synchronously. However, it did not specify the temporal order in which these groups were activated. Noisy inputs to the network resulted in different temporal orderings making it an unreliable encoder of odors. We found that when asymmetries were present in the network structure such that some of the clusters detected by our algorithm preferentially connected to others, a particular temporal order was preserved despite noisy variations in the input.