Non-linear stimulus processing by the retina
Ulisse Ferrari (Institut de la Vision, Paris)
Understanding how sensory systems process information is an open challenge. This is mostly because these systems are non-linear, making it extremely difficult to model the relation between the stimulus and the sensory response. In this talk I will discuss two strategies to tackle this problem and apply them to the retina.
First, we use ex-vivo multi-electrode array experiments to record the retinal activity and directly model the ganglion cell response to complex stimuli, such as videos of moving objects. Here I will show that standard, nearly-linear, models are not enough and highly non-linear models are required. Then I will present the result of a closed-loop experiment where we adapted the stimulus on-line to investigate how the response changes when the visual stimulation is perturbed.
With this approach we could estimate the optimal performance of a neural decoder and show that the non-linear sensitivity of the retina is consistent with an efficient encoding of stimulus information.
-) U. Ferrari, C. Gardella, T. Mora, O. Marre. eNeuro, vol. 4, 6. 2017.
-) S. Deny, U. Ferrari, P. Yger, R. Caplette, S. Picaud, G. Tkacik, O. Marre, Nature Commun. 8 (1) 2017