The transition from the awake to anesthetized brain state (monitored by the loss of fastwave EEG activity) greatly enhanced odor-evoked ensemble activity: odors elicited stronger mitral cell responses and the density of odor representations increased (Figures 2A and 2B). Under anesthesia, individual mitral

cells respond to more odors (Figure 2C) and responses are stronger (Figure 2D). This increase in mitral cell responsiveness during anesthesia is not due to an increase in sensory input to the bulb (see Figure S1 available online). The effects of anesthesia were indistinguishable with ketamine AZD6244 concentration and urethane, two commonly used and chemically distinct anesthetics (Figure S2), suggesting that the differences in mitral cell activity reflect changes in brain state rather than local pharmacological effects of the drugs. Mitral cell spontaneous firing rates are reportedly higher in the awake versus anesthetized state (Adrian, 1950; Rinberg et al., 2006a). To test whether changes in baseline activity between awake

and anesthetized states could account for the differences in the normalized measure of mitral cell INCB018424 responses (dF/F), we next compared the odor-induced fluorescence changes without normalization between the two states. The enhancement of mitral cell responses with anesthesia was apparent even in this unnormalized measure (Figure S2), indicating that anesthesia increases the absolute amplitudes of mitral cell odor responses. We next examined how differences

in mitral cell ensemble responses in awake and anesthetized states affect odor coding by determining the efficiency of cell ensembles to discriminate between the seven odors. To quantify the efficiency of odor coding, we calculated the fraction of odor trials that are classified correctly using responses for the entire duration of odor stimulation when we randomly sampled different numbers of responsive mitral cell-odor pairs (see Experimental Procedures). In the awake state, fewer mitral cell responses were needed to achieve high levels of correct classification and compared to the anesthetized state (Figure 2E). These results indicate that compared to the anesthetized brain state, the selective odor tuning of mitral cells and sparse odor respresentations during wakefulness are more efficient at odor coding. In addition to the effects of anesthesia on mitral cell odor tuning, there was a marked difference in the temporal dynamics of mitral cell responses between awake and anesthetized brain states. When mice are awake, odor responses are temporally diverse, with the onset timing of different cell-odor pairs fairly evenly tiling the period of odor stimulation and a few seconds after odor offset (Figure 2F, left).