Figure 3
Fluoxetine affected information dependence with noise correlation and impaired robust encoding of the stimulus. (a) Dependence (d) of pairwise entropy on noise correlation computed as the Spearman’s correlation between measures for several timescales (from 50 to 500 ms) centered at successive times after cue onset (− 1 to + 6 s) for the three treatments. (b) Using a timescale of 400 ms the mean Spearman’s correlation between the measures (± SEM), from − 1 to + 6 s, is shown. The mean correlation was reduced for the acute and chronic fluoxetine treatments respect to control group (***p < 0.001, U Mann–Whitney test). We did not find a significant difference between acute and chronic fluoxetine treatments (p > 0.05). (c) Example of the exponential fitting curve computed for the pairwise mutual information (MI) as a function of the flip probability (pflip) of the binary state of one neuron of the pair across trials (i.e., state 0 flips to 1, and state 1 flips to 0) using a timescale of 400 ms at 3.8 s after the tone. The intersection point of the estimated exponential function with the y axis (at pflip = 0) is the uncorrupted MI (MIdata). The chronic fluoxetine group of neurons conveyed more information about the occurrence of the stimulus, but it decayed faster with pflip when compared with the other treatments. (d) Parameters MIdata and the robustness factor (r) of the exponential fitting curves are shown for several timescales (from 50 to 500 ms) centered at successive times after the tone (− 1 to + 6 s) for the three treatments.
