Abstract:
This dissertation investigated the Agent advantage effect in static pictures. The Agent advantage effect referred to the finding that observers respond faster to Agents than to Patients when identifying event roles. Across three studies, this dissertation addressed three related questions: first, what temporal processing dynamics underlay the Agent advantage effect; second, whether low-level visual cues such as motion lines modulated the effect; and third, whether the effect remained stable across different action contexts when visual features were held under strict perceptual control.
The first study used eye-tracking method to investigate the temporal dynamics of the Agent advantage effect in the classical two-fish biting paradigm. The behavioural data replicated the standard finding that participants were faster during Agent search than during Patient search. Importantly, the eye-movement data suggested that this asymmetry did not arise from an early attentional bias toward the Agent. First fixations were not preferentially directed toward either event role. Instead, the critical difference emerged at a later stage of processing: during Patient search, the last fixation before response was more likely to land on the Patient, and this last fixation bias strongly correlated with the size of the reaction time asymmetry. These results suggested that the Agent advantage effect is at least partly related to late-stage verification processes, with identifying the Patient requiring additional visual confirmation.
The second study examined whether the Agent advantage effect was modulated by motion lines, a visual cue often associated with the depiction of motion in static images. Across two experiments, one using parallel motion lines and one using orthogonal motion lines, the Agent advantage effect was consistently replicated when there were no motion lines in the pictures. However, results showed that motion lines did not significantly interact with the search task and therefore did not reliably alter the magnitude of the effect. These results highlight the robustness of the effect and suggest that event role processing and motion line perception might operate independently.
The third study tested whether the Agent advantage effect generalized across action contexts. Using visually indistinguishable abstract fish under strict perceptual control, it compared two action contexts, fighting and chasing, while preserving the same local biting relation between the two event roles. A robust Agent advantage effect emerged during fighting actions, but disappeared during chasing actions. This pattern showed that the Agent advantage effect was not an invariant consequence of any depicted Agent-Patient relation. Instead, its emergence depended on how event role information was extracted within a particular action context. More specifically, the findings suggested that fighting and chasing place different demands on event role processing: fighting may require finer local verification of the overlapping mouth region, whereas chasing may allow a faster global relational parse of the scene.
Taken together, the dissertation showed that the Agent advantage effect in pictures was robust but not unconditional. The findings indicated that the Agent advantage effect was associated with late-stage verification during Patient searches; the presence of motion lines did not impact the effect; and the appearance of the Agent advantage effect was impacted by the action context though the Agent-Patient relation was held constant. More broadly, the dissertation demonstrated that the Agent advantage effect is not a fixed effect of any depicted Agent-Patient relation but rather the outcome of an interaction among event structure, task demands, and the processing dynamics involved in identifying Agents and Patients.
Agent advantage effect