Scaling effects in a model of the olfactory bulb

Most computational models of the olfactory bulb are much smaller than any biological olfactory bulb—usually because the number of granule cells is much lower. The resulting subsampling of the inhibitory input may distort network dynamics and processing. We have constructed a large-scale model of the zebrafish olfactory bulb, as well as two smaller models, using the efficient parallellizing neural simulator SPLIT and data from a previously existing GENESIS model. We are studying several characteristics—among them overall behaviour, degree of synchrony of mitral cells and the timescale of appearance of synchrony—using cross-correlation plots and synthesized EEGs. Larger models with higher proportions of granule cells to mitral cells appear to give more synchronized output, especially for stimuli with shorter timescales.