News Release

Artificial neural network (ANN) systems performed better at predicting future frames of a movie when trained on retinal waves mimicking the spontaneous activity patterns of the retina in animals' eyes, in addition to training on naturalistic movies

Peer-Reviewed Publication

PLOS

Artificial neural network (ANN) systems performed better at predicting future frames of a movie when trained on retinal waves mimicking the spontaneous activity patterns of the retina in animals' eyes, in addition to training on naturalistic movie

image: 

A schematic of the stimuli seen by the mouse after vision onset in a virtual corridor rich with optic flow, created using a 3D animation software and used as the training dataset (Methods 4.1.4). The dataset simulates a mouse’s visual experience when moving forward through space.

view more 

Credit: May et al., 2025, PLOS Computational Biology, CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/)

Artificial neural network (ANN) systems performed better at predicting future frames of a movie when trained on retinal waves mimicking the spontaneous activity patterns of the retina in animals' eyes, in addition to training on naturalistic movies. 

####

Article URL: https://plos.io/4caNWFM

Article Title: Pre-training artificial neural networks with spontaneous retinal activity improves motion prediction in natural scenes

Author Countries: Austria, Germany

Funding: This work was supported by the European Research Council (ERC StG NeuroDevo #804824 to JG) and TUM Innovation Network “Neurotechnology in Mental Health” (to JG). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Journal

PLOS Computational Biology

DOI

10.1371/journal.pcbi.1012830

COI Statement

Competing Interests: The authors have declared that no competing interests exist.

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.