When people see a dog bite a man, they will focus on the dog (i.e., the agent performing the action) first and the man (i.e., the patient being acted upon) second. Human observers are faster in responding toward the agent than the patient information, which is called the agent advantage effect (Segalowitz, 1982). The agent advantage effect can be found in sentences, manual signs, and pictures (Cohn & Paczynski, 2013). A pilot experiment from our lab suggests that participants’ slower responses to patients than agents in pictures might be caused by participants showing increased gaze toward the patient just before their response. Within the present project, we aim at studying the research question of whether it is an agent benefit or rather a patient cost that leads to the agent advantage effect in pictures. We will use an eye tracker to collect rich eye-movement data, to uncover the process of how the participants search, extract, process, and finally identify the agent (or patient), thus determining the main factors that lead to the agent advantage (or patient cost) effect when comprehending pictures.