Contextual Influences in Visual Perception: Behavioural and Neural Insights into Size Representations

In order to interpret incoming sensory input effectively, the visual system integrates multiple sources of information from the surrounding environment. Although many perceptual processes are often assumed to operate automatically, accumulating evidence suggests that they rely on complex and dynamic mechanisms. Building on previous work employing behavioural measures, neuroimaging, and predictive coding frameworks, this thesis investigates how contextual influences modulate size representations, with a particular focus on ensemble summary statistics. Study I examined whether implicitly coded objects contribute to perceived average size by manipulating contextual size information using Ebbinghaus inducers. The results demonstrated that rescaled objects influenced ensemble representations regardless of whether they were explicitly or implicitly coded, suggesting that contextual size modulation occurs relatively early in the visual processing stream and remains available for subsequent statistical computations. Study II extended these findings by showing that when two sets of stimuli are presented simultaneously, their perceived average sizes mutually influence each other through contrast-like interactions. Behavioural and functional results indicated that ensemble representations are processed in parallel but remain susceptible to contextual modulation from neighbouring ensembles, even when they are task-irrelevant. Together, these findings challenge models assuming independent or pooled ensemble processing and instead support a framework in which ensemble summary statistics interact dynamically through contextual comparison mechanisms. Overall, this thesis demonstrates that ensemble representations are not mere summaries of stimulus features but actively serve as contextual cues in size perception, highlighting the importance of context in shaping perceptual representations.