Decision of learning status based on modeling of the information measurement of social behavioral tasks in rhesus monkeys

We are interested in identifying the learning status of the social behavioral tasks in the rhesus monkey. In addition, we define the characteristic of stimulus with a numerical quantification. We allow monkeys to interact with individuals of different social status, while we monitor the viewer monkey's behavior by tracking its scan paths. With these observations, we can understand the learning status of this animal via looking behavior analysis on the stimulus. First, the viewer monkey shows different looking patterns among six different classes. Therefore, we can generate different data descriptors of these classes and observe the classification performance of the machine learning algorithm. Second, we design the ground truth model based on the characteristic of each stimulus. We define the distribution of information from the ratio of the face, body, and background area in the stimulus. Lastly, we link them to figure out whether the viewer monkey learned enough about the information in the stimulus.