Bearing, heading and the perceptual geometry of collective motion

Abstract

Traditional models of collective motion commonly assume that animals turn to match their neighbours' directions of travel. In the harmonic theory, decisions follow from a neural representation of direction, and a neighbour enters only as a stimulus at a bearing, the direction in which it lies. Here we show that a neighbour's spatial extent allows its orientation to enter through projection geometry alone, and the theory generates a matrix of coupling coefficients indexed by the neighbour's bearing and aspect. Responses to bearing alone form one edge of this matrix, and classical heading matching is a special case on its diagonal, one ray through a two-parameter family. Independently, an animal that responds only to neighbour position, perceived with a delay, acquires an alignment coefficient set by the slope of its distance-weighted attraction profile, so that it can align, anti-align or reverse sign. Alignment need not be a rule to be measured. In schooling fish and walking pigeons with body orientation measured independently of travel, social responses are recovered, including bearing responses in the fish. Heading excess depends on how turning is measured, appearing in the fish with forward but not centred responses and remaining unresolved in pigeons. Controls show why neither outcome identifies a direct alignment rule. In fish recordings with motion-assigned body polarity, heading-dependent structure departs from Vicsek form and fails the coefficient test for a common angular potential. Whether animals align thereby becomes a measurable question, with geometric, perceptual and behavioural origins that can be distinguished.