Nil Altinordu¹, Geoffrey Boynton², Ione Fine³; ¹University of Washington

Introduction: Recognizing objects in black and white photographs or movies is remarkably easy. Why then is there extensive neural hardware devoted to supporting color vision – what is color vision good for? One alternative to the ‘fruit and berries’ hypothesis is that trichromacy might compensate for optical deterioration as a result of aging in longer-lived primate species. Here we examine whether color information facilitates object recognition under simulated low vision conditions in which high frequency form information is absent. Methods: We measured object-identification performance in real-world scenes (Adobe Stock & BOiS database) presented in grayscale or color across a range of blur levels (0-8 diopters). Participants judged whether a cued object was present. Object presence/absence was randomized across trials, producing a 2 (color/grayscale) × 2 (object present/absent) × 9 (diopter level) within-subject design. In Experiment 1 (N=27) participants viewed scenes and were given unlimited time to search for the object and respond. In Experiment 2 (N=18) the scenes were masked after 2.5 seconds and participants were given an unlimited time to respond. Results: Performance was similar for both experiments. While there was a strong effect of blur, there was little difference in performance between color/grayscale conditions and no interaction between the presence/absence of color and the amount of blur. Conclusions: Color plays a surprisingly limited role in object recognition within complex, real-world scenes even under low vision conditions. It remains possible that we would find an interaction between color and blur for specific tasks based on nature-based stimuli, such as picking out berries or fruit in a bush, or judging the health and emotion of a face. These results have important implications for sight-restoration technologies (e.g., retinal and cortical stimulation, and optogenetics): there may be little added utility in producing colored percepts.

Acknowledgements: Supported by R01EY014645