Figure S1 - Synaptic Learning Rules and Sparse Coding in a Model Sensory System

SRDP and the persistence of olfactory representations. Data are represented as in Figure 5. Columns indicate different learning protocols under SRDP. From left to right: “naïve”; learning on a test odor alone (e.g., “1111”); sequential learning on combinations of four different odor blocks (e.g., “2143”); “RANDOM,” learning on the sequence of randomly permuted single 1-s trials of the odors 1, 2, 3, and 4. Rows indicate the test odor for the olfactory circuits above (odors 1–4 were included in the learning sequence; odor 6 was not included in the learning sequence). For each spike count bar plot, the sparseness measure is indicated in the upper left corner (N/A: not available, less than 1/4 of 1-s trials elicited on average at least one spike), whereas the Euclidean distance in 19-D KC odor space from the average representation after learning on a test odor alone is indicated in the upper right corner (multiple odor sequences only).
(0.53 MB EPS)