Meeting link:
https://meet.google.com/hze-ccsx-hhx
PoYu Chou,
Chi Keung ChenAnticipation is important for living organisms to survive. The anticipative dynamics is found in retina [1], which can compensate for the delay of visual signals during transmission and processing. Responses of retinas (r(t)) from bull frogs have been investigated for anticipative properties in a multi- electrode array system by using whole field stochastic stimulations (S(t)) generated by an Ornstein–Uhlenbeck (OU) process. Time correlated S(t) can be then created by passing the OU signal through a low pass filter with various cutoff frequencies. Anticipative properties of the elicited spikes from the retinas are then characterized by the method of time lag mutual information (TLMI) between r(t) and S(t) [1]. We find that only stimulations with long enough correlations can elicit anticipative responses from the retinas; similar to the finding of [1] in which information in the stimulation was time coded. However, information is being rate coded in the present experiment. Recently, Voss [2] proposed that a negative group delay filter can produce anticipative response to low-pass filtered random signals. To test this idea, it is shown that an NGD filter with appropriate parameters can indeed be used to produce TLMI from S(t) similar to those observed in experiments. Furthermore, experiments with dark and bright Gaussian light pulses further confirmed that retina can be considered as an NGD filter; but only for the dark pulses. This last finding and the NGD capability of the retina suggest that there is a delayed negative feedback in the off-pathway of the retina. In fact, a two neuron-model with delayed negative feedback can be shown to produce properties of an NGD filter. Presumably, such mechanism might also exist in a retina.
Reference
1\. Chen, Kevin Sean, Chun-Chung Chen, and C. K. Chan. "Characterization of predictive behavior of a retina by mutual information." _Frontiers in computational neuroscience_ 11 (2017): 66.
2\. Voss, Henning U. "Signal prediction by anticipatory relaxation dynamics." _Physical Review E_ 93.3 (2016): 030201.
