Mapping Sensory Stimuli to Behavioral Output

Olfactory perception begins with the recognition of odorants by a large repertoire of olfactory receptors (OR) in the sensory epithelium. In the mouse, each sensory neuron expresses one type of receptor from ~1000 OR genes. Randomly distributed populations of sensory neurons expressing the same receptor converge onto anatomically discrete areas of the bulb called glomeruli, which form a map of odor responses that is stereotyped across animals. At the glomeruli, sensory neurons synapse onto the projection neurons of the olfactory bulb, which, in turn, send axons to several telencephalic areas, including a significant input to the piriform, the primary olfactory cortex.

Olfactory processing is a uniquely tractable system in which to study learning. In sensory systems for vision, touch, and sound, features central to perception are topographically ordered in the sensory organ, and this order is maintained in the primary sensory cortices. Thus, developmentally programmed neural circuits may mediate early processing of these stimuli. In contrast, olfactory features are not topographically ordered in the epithelium or the piriform cortex, and odorant responses in this sensory cortex display no recognizable spatial stereotypy. Therefore it is thought that neurons in the piriform receive convergent input from random collections of glomeruli. And if the representation of an odor in the piriform results from the random convergence of glomerular inputs, then its valence or behavioral meaning cannot be developmentally programmed and, instead, would likely be im¬posed by experience. This renders the piriform an ideal neural substrate for linking an initially neutral stimulus to a behavioral response through learning.

In support of this model, we and others have shown that activation of random ensembles of piriform neurons using channelrhodopsin, photo-activatable ion channel, can drive behaviors of opposing valence, dependent upon leaning, thus implicating the piriform cortex as the neural substrate for associative learning. This observation that an arbitrarily chosen ensemble of neurons can be entrained to multiple, unconditioned stimuli suggests that the piriform has access to output areas that can generate a large repertoire of behavioral responses. Indeed, our initial anatomical tracing studies have identified piriform projections to a number of areas implicated in driving behavior, including the amygdala and ventral striatum. Using these experimentally generated networks of small ensembles of piriform neurons, we are anatomically and functionally mapping the incoming and outgoing piriform circuitry that underlies odor-driven associative learning.