Sharpening of hierarchical visual feature representations of blurred images

Abstract The robustness of the visual system lies in its ability to perceive degraded images. This is achieved through interacting bottom-up, recurrent, and top-down pathways that process the visual input in concordance with stored prior information. The interaction mechanism by which they integrate visual input and prior information is still enigmatic. We present a new approach using deep neural network (DNN) representation to reveal the effects of such integration on degraded visual inputs. We transformed measured human brain activity resulting from viewing blurred images to the hierarchical representation space derived from a feedforward DNN. Transformed representations were found to veer towards the original non-blurred image and away from the blurred stimulus image. This indicated deblurring or sharpening in the neural representation, and possibly in our perception. We anticipate these results will help unravel the interplay mechanism between bottom-up, recurrent, and top-down pathways, leading to more comprehensive models of vision. Significance statement One powerful characteristic of the visual system is its ability to complement visual information for incomplete visual images. It operates by projecting information from higher visual and semantic areas of the brain into the lower and mid-level representations of the visual stimulus. We investigate the mechanism by which the human brain represents blurred visual stimuli. By decoding fMRI activity into a feedforward-only deep neural network reference space, we found that neural representations of blurred images are biased towards their corresponding deblurred images. This indicates a sharpening mechanism occurring in the visual cortex.