Object recognition is shaped by the dynamic interplay between low-level contrast and high-level context.

Human perception relies on bottom-up sensory input and top-down prior knowledge, and object recognition is known to be modulated by the semantic relationship between objects and their surrounding context. However, the dynamic processing mechanism during object recognition and the resulting interaction between top-down and bottom-up factors remains unclear. The present study investigated how stimulus contrast modulates the effect of scene-object semantic congruency. Participants performed an object-recognition task in which target objects of high or low contrast appeared within semantically congruent or in congruent natural scenes. In Experiment 1, behavioral results (N= 73) on recognition accuracy revealed a congruency benefit for high-contrast objects and an incongruency benefit for low-contrast objects, indicating a contrast-dependent shift in how scene context influences recognition. Experiment 2 (N= 19) employed event-related potentials (ERPs) to examine the neural activity of these effects. For high-contrast targets, congruency effects emerged in the N300 and N400 over central-parietal regions, and in the P600 over frontal sites. For low-contrast targets, effects were observed in the late component. These findings provide novel evidence that low-level visual contrast determines whether prior knowledge facilitates or interferes with object recognition, offering critical insights into the mechanisms underlying the dynamic interaction between bottom-up and top-down processes in visual perception.