I see moving people: Expectations drive detection of biological motion in noisy point-light displays

While biological motion processing has been extensively studied, little is known about the top-down impact of expectations in this context. We tested whether expectations about the likelihood of encountering a human walker influence the detection of biological motion in point-light displays, particularly when perceptual information is unreliable. Seventy-four participants completed a signal detection task, responding to stimuli featuring either a human walker or scrambled biological motion, each masked with one of four levels of visual noise. Participants were randomly assigned to the high or low expectations group and were told that 75% or 25% of the displays would feature a human walker, although the actual proportion was 50%. Participants in the high expectation group showed a greater tendency to respond "yes," with the largest group difference emerging at the highest level of noise. These findings suggest that expectations can bias biological motion detection, particularly under conditions of sensory unreliability. The results also support the predictive processing model of agency detection, which proposes that false-positive perceptions of (supernatural) agents arise from expectations combined with ambiguous input.