Fig 2

Fig 2 Latent dynamics in the optic tecta of zebrafish larvae in response to visual stimulation.
A. Illustration of our approach to data analysis. Datasets such as the one shown on the left consisted of fluorescence traces from n neurons. These were trial averaged and the dimensionality of the trial averaged data was reduced using PCA. The new k dimensions are referred to as neural modes. B. The first 5 neural modes explain ~90% of the variance in the data. The shaded area represents the standard deviation, n = 9). C. Shown are the latent dynamics of an individual wild type larva for the first 5 neural modes. The latent dynamics were calculated independently for each of the three trials. The mean correlation of the dynamics among the individual trials to the trial average is shown as r̄. The presentation of each stimulus is shown as a gray bar, and the direction of each stimulus is shown at the top. D. Graph summarizing the mean correlations across trials along each mode for the wild type larvae (n = 9). E. Latent dynamics determined from trial averaged data. Traces (purple) are the dynamics from individual larvae and the shaded area (blue) is the variance for the wild type group. The value, r̄, represents the mean correlation of each individual neural mode to the mean dynamics averaged across the group (see f). F. Summary of the correlation between individual larvae and the group average along neural modes (n = 9). G,H. Three-dimensional representation of latent dynamics for two, individual larvae. Each color represents a direction of motion for the visual stimulus. I. A three-dimensional representation of latent dynamics averaged across the wild type group.