RT Journal Article
SR Electronic
T1 Efficient coding of natural scenes improves neural system identification
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 2022.01.10.475663
DO 10.1101/2022.01.10.475663
A1 Qiu, Yongrong
A1 Klindt, David A.
A1 Szatko, Klaudia P.
A1 Gonschorek, Dominic
A1 Hoefling, Larissa
A1 Schubert, Timm
A1 Busse, Laura
A1 Bethge, Matthias
A1 Euler, Thomas
YR 2022
UL http://biorxiv.org/content/early/2022/01/11/2022.01.10.475663.abstract
AB Neural system identification aims at learning the response function of neurons to arbitrary stimuli using experimentally recorded data, but typically does not leverage coding principles such as efficient coding of natural environments. Visual systems, however, have evolved to efficiently process input from the natural environment. Here, we present a normative network regularization for system identification models by incorporating, as a regularizer, the efficient coding hypothesis, which states that neural response properties of sensory representations are strongly shaped by the need to preserve most of the stimulus information with limited resources. Using this approach, we explored if a system identification model can be improved by sharing its convolutional filters with those of an autoencoder which aims to efficiently encode natural stimuli. To this end, we built a hybrid model to predict the responses of retinal neurons to noise stimuli. This approach did not only yield a higher performance than the “stand-alone” system identification model, it also produced more biologically-plausible filters. We found these results to be consistent for retinal responses to different stimuli and across model architectures. Moreover, our normatively regularized model performed particularly well in predicting responses of direction-of-motion sensitive retinal neurons. In summary, our results support the hypothesis that efficiently encoding environmental inputs can improve system identification models of early visual processing.Significance Computational models use experimental data to learn stimulus-response functions of neurons, but they are rarely informed by normative coding principles, such as the idea that sensory neural systems have evolved to efficiently process natural stimuli. We here introduce a novel method to incorporate natural scene statistics to predict responses of retinal neurons to visual stimuli. We show that considering efficient representations of natural scenes improves the model’s predictive performance and produces biologically-plausible receptive fields. Generally, our approach provides a promising new framework to test various (normative) coding principles using experimental data for understanding the computations of biological neural networks.Competing Interest StatementThe authors have declared no competing interest.