Central visual information improves peripheral scene visibility

Past behavioural studies showed that information from an object in a scene could affect the subjective appearance of its scene-context. In these studies, blurred scene-contexts that contained a meaningful intact object were subjectively perceived as sharper than an objectively identical scene-context containing meaningless information (Carrez-Corral et al., 2025). These findings suggest that information from an object can enhance the subjective perception of scene context, especially when the visual signal is noisy. Given that scene context is usually perceived in the peripheral visual field while objects are processed in central vision, this raises the question of whether information from central vision can enhance perception in the periphery. However, these studies used small images of scenes as stimuli and did not control for the central vs. peripheral presentation of objects and scene contexts. The present study aimed to address this question. Participants (n = 29) were presented with stimuli composed of a peripheral ring revealing a scene embedded in 1/f noise at various scene-to-noise ratios and a central disk containing either an intact congruent scene, an incongruent scene or a phase-scrambled scene. Each participant performed two sessions. In the DetectScene session, they were instructed to maintain fixation at the center but attend to the peripheral ring and they had to report whether they perceived a scene within noise in the peripheral ring. In the DetectNoise session, they had to report whether they perceived the presence of noise within the scene in the peripheral ring. In both sessions, the scene-to-noise ratio was adapted trial-by-trial using a staircase procedure (starting with 100% noise in the DetectScene session and 0% noise in the DetectNoise session). These two sessions respectively allowed us to obtain scene and noise detection thresholds, providing complementary measures of scene visibility. Results showed that meaningful central visual information (i.e., congruent or incongruent) facilitates peripheral scene detection in noise and also helps to better resist to noise in scene, relative to meaningless information (i.e., phase-scrambled). We also observed that congruent central information further improves scene visibility in peripheral vision relative to incongruent information. We interpret these findings in the context of predictive processing theories whereby predictions based on information extracted from central vision enhance perception of degraded congruent information in peripheral vision.