| Popis: |
In this project, we will examine how people make decisions as they interact with other players in a continuous task. In the two-player behavioral task, players are assigned one of five goals to achieve: ATTACK (run into the other player), AVOID (stay away from the other player), INSPECT (stay near the other player but do not run into them), DEFEND (keep the other player away from a prespecified location), or HERD (bring the other player to a prespecified location). Participants will receive points for achieving their assigned goal (all trials) and for identifying what the goal of the other player is (half of the trials). The goal of the project will be to develop and test a model of behavior (joystick movement) and eye tracking data on the task, use the model to infer in real time what a player's goal is from their behavior, compare this modeling approach to human ability to infer intent and to machine learning methods, and connect model parameters to quantify individual differences related to motivation, motor ability, aggression, and anxiety. There are 25 possible combinations of goals on the task (5 for Player 1 x 5 for Player 2). We will divide these combinations up into "high conflict" (players have directly opposing goals) / "medium conflict" (players can both achieve goals, but interaction is competitive) / "low conflict" (both players can achieve goals with no competition) as follows: Low conflict: DEFEND X DEFEND DEFEND X INSPECT INSPECT X INSPECT DEFEND X AVOID AVOID X AVOID INSPECT X HERD Medium conflict: ATTACK X ATTACK DEFEND X HERD AVOID X HERD High conflict: ATTACK X DEFEND ATTACK X INSPECT ATTACK X AVOID ATTACK X HERD HERD X HERD INSPECT X AVOID |
