Fisher information across the sample. *A*: Fisher information plotted for each stimulus location for all the neurons in the Go condition. The stimulus positions were normalized to the widths (SD) of the best-fit Gaussian function to the neuronal tuning curves. Zero is the center/peak of the tuning curves. The values of Fisher information were normalized to the maximum value at the 50- to 150-ms time interval from the NoGo condition for each neuron. *B*: same as *A* for the NoGo condition data. The dashed lines show binning for *C. C*: average Fisher information after being normalized to the maximum Fisher value in the NoGo condition for each neuron and each stimulus position. Data were binned over locations every 1 SD from −4.5 SD to 4.5 SD, as indicated by the dashed lines (*1–9*) in *A* and *B*. Each circle is the median Fisher information across the sample in a 1 SD bin. *n* = 46, but numbers of neurons in each bin may vary. *D*: Fisher information values (not normalized) for flanking *bins 3*, *4*, *6*, and *7* (−2.5 to −1.5 SD, −1.5 to −0.5 SD, 0.5 to 1.5 SD, and 1.5 to 2.5 SD, as shown in *C*) in the Go condition plotted against those in the NoGo condition. The black line is the unity line. Fisher information values were significantly different between the Go and NoGo conditions (Wilcoxon signed-rank test, *P* = 0.001). *E*: ratios of Fisher information between the Go and NoGo conditions plotted against the ratios of the widths of neuronal tuning curves between the Go and NoGo conditions. The data points are taken from flanking *bins 3*, *4*, *6*, and *7* (−2.5 to −1.5 SD, −1.5 to −0.5 SD, 0.5 to 1.5 SD, and 1.5 to 2.5 SD, as shown in *C*). The two dashed lines represent no change (ratio of 1). Fisher information values in the Go condition were significantly higher than those in the NoGo condition (Wilcoxon signed-rank test, *P* = 0.001). Widths in the two conditions were not significantly different (Wilcoxon signed-rank test, *P* = 0.940).

## PubMed Commons