Schematic of the optomotor assay. (a) Stimuli were passed through a BITS++ video processor and displayed on a flat-screen cathode ray tube (CRT) monitor facing upwards. The five-lane zebrafish arena with a transparent base was positioned above the monitor screen, and a camera fixed above the arena obtained digital images of larvae positions before and after each trial. (b) Overhead view of the zebrafish arena, with inner walls dividing the arena into multiple lanes. (c) Stimuli were narrow-band filtered Gaussian-noise textures drifting parallel to the long axis of the arena lanes. (d) Between trials, larvae were corralled to the centre of the lane by high-contrast compound gratings drifting from each end, converging at the midline (dashed white line).

Spatial-frequency tuning from the optomotor response of visually naïve and experienced larval zebrafish from 5 to 7 dpf. The top row shows spatial-frequency tuning functions (mean ± SEM) at stimulus speeds of (a) 25 °/s (b), 50 °/s, and (c) 100 °/s. Black solid, red dashed and blue dashed lines show tuning functions for visually naïve 5, 6 and 7 dpf larvae, respectively. Red and blue solid lines show spatial-frequency tuning functions for visually experienced 6 and 7 dpf larvae, respectively. Functions are a three-parameter log-Gaussian, fit by minimising the least-squares error. The bottom row shows normalised amplitude at stimulus speeds of (d) 25 °/s (e), 50 °/s, and (f) 100 °/s. Error bars show the 95% confidence intervals. P < 0.05 after Bonferroni correction was considered to be statistically significant.

Spatial-frequency tuning from the optomotor response of visually naïve and experienced larval zebrafish at 7 dpf. The top row shows spatial-frequency tuning functions (mean ± SEM) at stimulus speeds of (a) 25 °/s (b), 50 °/s, and (c) 100 °/s. Black solid, red dashed, blue dashed, red solid and blue solid lines show functions for naïve, 1-session static, 1-session moving, 2-session static and 2-session moving groups, respectively. Tuning functions are a three-parameter log-Gaussian, fit by minimising the least-squares error. The bottom row shows normalised amplitude at stimulus speeds of (d) 25 °/s (e), 50 °/s, and (f) 100 °/s. Error bars show the 95% confidence intervals. P < 0.05 after Bonferroni correction was considered to be statistically significant.

PSD-95 in the retinal IPL of visually naïve and experienced larvae from 5 to 7 dpf. (a) Example confocal images of 7 dpf retinae stained with DAPI (blue) and PSD-95 (green) in naïve (top row), 2-session static experienced (middle row) and 2-session moving experienced (bottom row) larvae. The larger scale bar (bottom right) represents 50 µm. Example cropped images for puncta analysis are also shown in the middle column, with the smaller scale bar representing 5 µm. Sections show the ganglion cell layer (GCL), inner nuclear layer (INL), inner plexiform layer (IPL), and outer plexiform layer (OPL). (b) Violin plots of PSD-95 density in 5 to 7 dpf visually naïve and experienced larvae. Points are individual retinae, white lines represent medians and dark bands indicate the inter-quartile range. Bayes Factors (BF) >3, considered as evidence in favour of a difference between groups, are shown.