Item

Abstract

Two models are presented as examples to discuss the computational tasks, sensory environments, and computational mechanisms in early olfactory processing. One model proposes that the odor inputs during a sniff cycle are detected and coded by the emergence of odor-dependent and coherent neural oscillatory patterns observed in the mammalian bulb, and that the odor mixtures are segmented by the selective olfactory adaptation to the preexisting and detected odors. The other model argues that the early olfactory processing is to separate the odor mixtures in the input environment into individual odor sources before the odor identities are known, and proposes a fast synaptic learning algorithm that uses the temporal fluctuation structures in the receptor cells for source separation.