Evaluation and neurocomputational modelling of visual adaptation to optically induced distortions

Abstract

Spatial geometrical distortions are major artefacts in vision aid optical spectacles. Progressive additional lenses (PALs) are among such spectacles incurring inherent distortions. Distortions alter perceived features of the natural environment and are one of the causes for visual discomforts, such as apparent motion perception and spatial disorientation, experienced by novice spectacle wearers. Thus, fast and efficient visual adaptation to the distortions is a necessity to increase the users’ comfort and consequently overcome the related problems, e.g. risk of fall in the elderly when using PALs. Inspired by this necessity, the work is targeted to investigate the visual mechanisms underlying adaptation to distortions, in particular in PALs. Psychophysical procedures are employed to probe the characteristics of the neural mechanisms underlying the adaptation process in natural viewing conditions. With psychophysical approaches, three main properties of distortion adaptation are revealed; its cortical origin, the reference frame in which it is achieved and its long-term temporal dynamics. In order to discern how the functional organization of neurons enables the visual system to carry out a robust distortion adaptation in a natural environment, biologically plausible recurrent neural network models are utilized. Prediction performance of model variants with different neural network complexity and temporal dynamics of operation were assessed. From the model simulations, major functional roles of recurrent bottom-up and top-down cortical interactions in neural response tuning and in mediating adaptation at different time scales were depicted. The outcomes would further contribute to suggest a solution for facilitating adaptation. The relevance of the research within these aforementioned studies is not restricted to PALs but extends to distortions in other daily used optical utilities, such as virtual reality (VR) displays. Optical distortions are also artefacts in artificial sensory systems, like lens distortions in cameras used in machine vision. Understanding the neural correlates of distortion adaptation in human vision will thereby elicit characteristic features of robust and flexible neural systems to be implemented in brain inspired artificial vision.