Fig. 6: Expanding orientation bandwidth improves stimulus discriminability in V1 neurons.

a Example field of view of an experiment with the central spatial frequency of 0.16 cpd and narrow or broad orientation bandwidth. Stimuli are shown at the bottom left of the plane, scale bar: 100 μm. b Number of responsive cells to broad orientation bandwidth (45°) with either 0.04 cpd or 0.16 cpd central spatial frequency, normalized to the respective number of cells responding to the narrow condition (5°). Data for 0.04 cpd are the same as in Fig. 1 and shown here for reference in gray. c Orientation modulation index, calculated as mean broad minus narrow orientation bandwidth responses, divided by their sum. Shown are all positively narrow and broad responding neurons for each central spatial frequency (n = 1333 responsive neurons for central SF = 0.04 cpd in gray, similar to data in Supplementary Fig. S1c, n = 1345 responsive neurons for central SF = 0.16 cpd in blue. n = 4892 neurons in total across 18 sessions from 9 mice). d Discriminability between neural responses to stimuli with a central spatial frequency of either 0.04 cpd or 0.16 cpd. Shown is the discrimination index for narrow and broad orientation bandwidth in light and dark blue, respectively (n = 2096 neurons from 18 sessions). Shown are the results for all cells that responded to any of the presented stimuli. The horizontal black dotted line shows the median discriminability for the narrow band. e Example field of view of an experiment with 90° central orientation and narrow or broad frequency bandwidth, scale bar: 100 μm. f–h Same as panels (b–d) but for narrow and broad frequency bandwidth and 0° and 90° central orientation. Box plots indicate the median (horizontal line), interquartile range (box bounds: 25th–75th percentiles), and whiskers (1.5 × interquartile range). Stars in all panels mark significant (Bonferroni correction for two tests, α = 0.025) differences from a two-sided LME test against 1 (panels c, g) or the narrow bandwidth (panels b, d, f, h). For visualization only, outliers were excluded from distributions in panels (b–d, f–h).