Figure 7.
Change in synaptic input (defined as the difference in input between the blend response and the average response to the single components at low concentration, multiplied by the corresponding synaptic weights) for PNs and LNs that respond with excitation of different blend interaction types (μ ± SEM for 100 model realizations). (A) Change in input to PNs arriving from LNs (left panel), PNs (middle panel), and OSNs (right panel). (B) Change in input to LNs arriving from LNs (left panel), PNs (middle panel), and OSNs (right panel). The p values for cases in which we found significant differences in the t-test (p ≤ 0.01, after Bonferroni correction) are listed in the following. For the change in input arriving from LNs to PNs that respond with excitation (left panel in A): suppression vs. hypoadditivity p < 1 × 10−6; suppression vs. linear addition p < 1 × 10−6, suppression vs. synergy p < 1 × 10−6; hypoadditivity vs. linear addition p < 1 × 10−6; hypoadditivity vs. synergy p = 1 × 10−6. For the change in input arriving from PNs to PNs that respond with excitation (middle panel in A): suppression vs. hypoadditivity p < 1 × 10−6; suppression vs. linear addition p < 1 × 10−6; suppression vs. synergy p < 1 × 10−6; hypoadditivity vs. linear addition p < 1.41 × 10−3. For the change in input arriving from OSNs to PNs that respond with excitation (right panel in A) suppression vs. hypoadditivity p = 2.8 × 10−5; suppression vs. linear addition p < 1 × 10−6; suppression vs. synergy p = 1 × 10−6. For the change in input arriving from LNs to LNs that respond with excitation (left panel in B): suppression vs. hypoadditivity p < 1 × 10−6; suppression vs. linear addition p < 1 × 10−6; suppression vs. synergy p < 1 × 10−6; hypoadditivity vs. linear addition p = < 1 × 10−6, hypoadditivity vs. synergy p < 1 × 10−6, linear addition vs. synergy p = 7.9 × 10−5.