Researchers tested the hypothesis by measuring the spatial contrasts in natural images and quantifying the distribution of lightness and darkness. At all scales, the authors found that natural images contain relatively more dark contrasts than light.
"Photoreceptors respond to light," Vijay Balasubramanian, professor of physics and the study's lead author, said. "But a couple of layers deeper down in the retina, cells are responding to changes and differences in the amount of light across the image. The eye is not a digital camera, recording little pixels. The eye doesn't do that. The eye tells the brain that there are differences in light between neighboring points. The brain learns about contrast. And in this case, there is about twice as much brain activity responding to darker spots."
The team confirmed this across a range of spatial scales and traced the origin of this phenomenon to the statistical structure of natural scenes. Researchers showed that the optimal mosaics for encoding natural images are also asymmetric, with off elements smaller and more numerous, thus matching retinal structure. Finally, the concentration of synapses within a dendritic field matches the information content, suggesting a simple principle to connect a concrete fact of neuroanatomy with the abstract concept of "information": equal synapses for equal bits.
Researchers were interested in how the visual system is adapted to the physical structure of the world, an assumption that makes sense from an evolutionary standpoint. The physics of the natural world should correspond to the processing capabilities of the brain and there are many observable incidences of this phenomenon in the human and animal world. For example, frogs eat flies. That fact predicted that frog's eye contains "fly detectors." Flies, in turn, track potential mates in mid-air, predicting specialized
|Contact: Jordan Reese|
University of Pennsylvania