Abstract
Watching others act can cause unintentional biases in the observer's next action. This "contagion" occurs due to the shared processes engaged during observation and execution, interfering with performance after observation. Biases can be imitative or compensatory, with the direction thought to be dependent on the presence of observed-induced prediction errors (PE: difference between predicted/expected and observed performance). To replicate and test the proposed PE mechanism, we compared golfers watching "on-target" and errorful putts. Twenty-three experienced golfers alternated putting a ball to a centre target on a nine-square grid (without outcome feedback) and watching videos of an actor putting to the same grid. Across four conditions, we covaried the location of the actor's putts (centre, corner) and expectations about the actor's aiming location, to manipulate the presence of a PE. As expected, imitative execution biases emerged after watching "correct" putts to corner squares, but not when these same corner putts were perceived as errors. Compensatory biases from PEs were also absent after watching "misses" to the centre square. These data provide evidence for different behaviours after the observation of correct versus errorful actions but raise questions about the types and/or stability of errors that underpin these effects.