Fig. 7. Demonstration that archer fish base their prediction on an extraction of the horizontal speed and direction of dislodged prey and not on an extrapolation of the spatial trajectory. Fish were confronted with prey that moved only in a horizontal plane. Yet the fish turned towards the virtual point of incidence at which a natural prey of the same take-off speed and direction would have hit the water surface. This is shown for both heights, h=20 cm (N=83 responses) (A) and 40 cm (N=58 responses) (B) by displaying the errors e made with respect to the virtual point of incidence. At both heights the means did not deviate significantly from zero (t-tests). Prey velocity was adjusted to be within the naturally occurring range at the two heights (see Figs 4 and 5).