Abstract
A prevalent idea in "perception of physics (i.e., kinematics/dynamics)" experiments is that in order to predict the future position of an object, humans employ "naive physics" models. These models are essentially reduced simulations of the physical environment, in which some diminished form of the laws of motion are implemented. However, simulations are costly and an alternative, yet effective, strategy would be to implement a simple, orthogonal set of heuristics that would mimic a "naive physics" model if implemented well. Here, as our initial attempt to distinguish between the two possibilities, we tested the capacity of observers to extrapolate the effects of a force. Observers viewed videos of a flag situated outside a building and flapping in the wind. The building had a glass window, through which a stationary object hanging from a pendulum could be viewed. During each trial, observers indicated the height to which the object would swing if the walls of the building were removed and the wind could act directly on it. We varied the material (wood, glass, metal), size (small/big), and shape (cone, cube, sphere) of the objects, since all of these are relevant to how the wind would effect the motion, and the strength of the wind was varied over five roughly perceptually equally spaced steps. Observers responses increased mostly linearly with increases in wind strength and there were effects of material, shape, and size, with an interaction between material and wind strength. We find that when an object is medium size, glass and wood are seen as equally heavy and lighter than metal, but when the object is larger, responses shift and glass is perceived as heavier, as much as metal. Observers extrapolate forces, but in a way that is not always consistent with physics.