Fig. 9.

Odor discrimination using a varying number of principal components. Time-averaged fraction of trials correctly classified by the network during the fixed point period of the odor response as a function of the number of principal components used in the classification task. The discrimination task was performed for 15 odors coded in a combinatorial manner (top row) and 10 odors coded as intensity distributions (bottom row) using both the intact network (left column) and the completely decoupled network (right column). In each discrimination task, odor trajectories were all projected onto the same principal components and the classification rate was determined as described in the Methods (except using the principal component trajectories rather than the original firing rate trajectories) using a varying number of principal components