Pyramidal cells also activate GABAergic interneurons that form powerful inhibitory Proteases inhibitor synapses onto nearby pyramidal cells to counter, and often overwhelm, the recurrent excitation. The recurrent circuitry in the piriform cortex therefore produces global excitation that recruits strong local inhibition, which scales with the excitatory drive. This allows temporal

pairing of bulbar input with activation of the recurrent network to alter piriform responses, thereby shaping the odor representation. Projections from individual glomeruli are distributed throughout the piriform cortex without any obvious topographic order, and individual pyramidal cells receive convergent input from a random collection of glomeruli. This afferent information is then redistributed across the piriform by the diffuse and apparently random selleck compound recurrent network. Nevertheless, an odor will consistently activate the same ensemble of piriform

neurons in an individual (Poo and Isaacson, 2009 and Stettler and Axel, 2009). We consider two distinct models for the activation of a cortical odor ensemble. In one model, an odorant may activate a sufficient number of mitral and tufted cell inputs to generate a direct suprathreshold synaptic response in all of the piriform neurons responsive to the odorant. In this case, the long-range recurrent excitation would mainly serve to recruit inhibitory neurons to generate a strong, diffuse feedback inhibition. Alternatively, an odorant may evoke suprathreshold input from the olfactory bulb in a small subset of odor-responsive neurons. This small fraction of spiking piriform cells would then generate

sufficient recurrent excitation to recruit a larger population of neurons that receive subthreshold afferent input. The strong feedback inhibition resulting from activation of this larger population of neurons would then suppress further spiking and prevent runaway recurrent excitation. In the extreme, some cells could receive enough recurrent input to fire action potentials without receiving afferent input. Two recent studies lend support to the second model. First, Davison and Ehlers (2011) observed robust responses in piriform neurons upon activation of a secondly set of glomeruli that were not synaptically connected to the recorded cell. Second, Poo and Isaacson (2011 [this issue of Neuron]) observed that, in a subpopulation of neurons, afferent LOT input only accounts for a small fraction of the odor-evoked excitatory drive onto layer II pyramidal cells. Our studies demonstrate that pairing weak bulbar inputs with recurrent inputs can dramatically increase the activation of piriform neurons. These effects are observed even though we expressed ChR2 in less than 1% of piriform neurons. Thus, the spiking of only a small fraction of piriform cells by direct input from the bulb could activate the recurrent circuitry to recruit the ensemble of odor-responsive neurons.