News Release

Neuro filter sharpens visual processing

Findings provide a more comprehensive account of how we 'see' the world

Peer-Reviewed Publication

Society for Neuroscience

Visual Perception

image: Blurry and clear versions of an image are represented similarly in the brain, showing how the visual system fills in missing information to maintain perception when visibility is low. view more 

Credit: Abdelhack and Kamitani, <i>eNeuro</i> (2018)

Blurry and clear versions of an image are represented similarly in the brain, according to a neuroimaging study published in eNeuro. The research shows how the visual system fills in missing information to maintain perception when visibility is low.

Vision arises from a "bottom-up" process that transduces light into neural signals and a "top-down" process in which the brain assembles that information into a coherent visual representation of the environment. The interaction between these two pathways is not fully understood.

Mohamed Abdelhack and Yukiyasu Kamitani investigated top-down processing of degraded visual information with an artificial neural network that translates human brain activity as participants view blurry and progressively sharper images. They found that the representations of the blurrier images were skewed toward that of the original, unaltered images. This effect was enhanced when participants were informed that they would be viewing images from a distinct set of categories. These findings provide a more comprehensive account of how individuals perceive their world through vision by combining visual input with prior knowledge. A better understanding of the top-down and bottom-up pathways could help to explain how their disruption may, for example, generate hallucinations.

###

Article: Sharpening of hierarchical visual feature representations of blurred images
DOI: https://doi.org/10.1523/ENEURO.0443-17.2018
Corresponding author: Yukiyasu Kamitani (Kyoto University, Japan), kamitani@i.Kyoto-u.ac.jp

About eNeuro

eNeuro, the Society for Neuroscience's open-access journal launched in 2014, publishes rigorous neuroscience research with double-blind peer review that masks the identity of both the authors and reviewers, minimizing the potential for implicit biases. eNeuro is distinguished by a broader scope and balanced perspective achieved by publishing negative results, failure to replicate or replication studies. New research, computational neuroscience, theories and methods are also published.

About The Society for Neuroscience

The Society for Neuroscience is the world's largest organization of scientists and physicians devoted to understanding the brain and nervous system. The nonprofit organization, founded in 1969, now has nearly 37,000 members in more than 90 countries and over 130 chapters worldwide.


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.