Rapid depth perception in hunting archerfish II. An analysis of potential cues

Abstract

Based on the initial movement of falling prey hunting archerfish select a C-start that turns them right to where their prey is going to land and lends the speed to arrive simultaneously with prey. Our preceding study suggested that the information sampled in less than 100 ms also includes the initial height of falling prey. Here we examine which cues the fish might be using to gauge height so quickly. First, we show that binocular cues are not required: C-starts that either could or could not have used binocular information were equally fast and precise. Next, we explored whether the fish were using simplifying assumptions about the absolute size of their prey or its distance from a structured background. However, experiments with unexpected changes from the standard conditions failed to cause any errors. We then tested the hypothesis that the fish might infer depth from accommodation or from cues related to blurring in the image of their falling prey. However, the fish determined also the height of 'fake-flies' correctly, whose image could never be focused and whose combined size and degree of blurring should have mislead the fish. Our findings are not compatible with the view that the fish uses a flexible combination of cues. They also do not support the view that height is gauged relative to structures in the vicinity of starting prey. We suggest that the fish use an elaborate analysis of looming to rapidly gauge initial height.