Human peripheral blur is optimal for object recognition

TL;DR

This study investigates whether the human peripheral blur pattern is evolutionarily optimized for object recognition by training neural networks on foveated images with varying blur profiles.

Contribution

It demonstrates that a peripheral blur profile matching human vision yields optimal object recognition performance, suggesting evolution shaped peripheral vision for this purpose.

Findings

Neural networks trained with human-like peripheral blur outperform other profiles.

Humans show decreased categorization accuracy with steeper peripheral blur.

Peripheral blur may be an adaptation for optimal object recognition.

Abstract

Our vision is sharpest at the center of our gaze and becomes progressively blurry into the periphery. It is widely believed that this high foveal resolution evolved at the expense of peripheral acuity. But what if this sampling scheme is actually optimal for object recognition? To test this hypothesis, we trained deep neural networks on 'foveated' images with high resolution near objects and increasingly sparse sampling into the periphery. Neural networks trained using a blur profile matching the human eye yielded the best performance compared to shallower and steeper blur profiles. Even in humans, categorization accuracy deteriorated only for steeper blur profiles. Thus, our blurry peripheral vision may have evolved to optimize object recognition rather than merely due to wiring constraints.