Pivots for pointing: visually-monitored pointing has higher arm elevations than pointing blindfolded.
Pointing gestures are a vital aspect of human communication. Nevertheless, observers consistently fail to determine the exact location to which another person points when that location lies in the distance. Here we explore the reasons for this misunderstanding. Humans usually point by extending the arm and finger. We show that observer's interpret these gestures by nonlinear extrapolation of the pointer's arm-finger line. The nonlinearity can be adequately described as the Bayesian-optimal integration of a linear extrapolation of the arm-finger line and observers' prior assumptions about likely referent positions. Surprisingly, the spatial rule describing the interpretation of pointing gestures differed from the rules describing the production of these gestures. In the latter case, the eye, index finger, and referent were aligned. We show that the differences in the production and interpretation of pointing gestures accounts for the systematic spatial misunderstanding of pointing gestures to distant referents. No evidence was found for the hypotheses that action-related processes are involved in the perception of pointing gestures. How participants interpreted pointing gestures was independent of how they produce these gestures and whether they had practiced pointing movements before. By contrast, both the production and interpretation seem to be primarily determined by salient visual cues.