The many ways of building collision-sensitive neurons

Publication Summary

problems raised by the initial formulation of and suggested that looming could provide a first-order approximation of ttc for use in fast interceptive actions, although alone could not possibly account for many of the accurate ttc judgements made by humans. If looming is not sufficient to account for the temporal precision of some human behaviour, what other data are available? One useful source of information is available from binocular vision: it is possible for binocular cues to provide ttc information in the same way as looming (via their rate of change). Rushton and Wann explored the use of looming and binocular information in catching in humans by perturbing either or both of these cues inside a virtual-reality display. They demonstrated that the human nervous system used both cues and that interfering with either cue causes timing errors. These data could be accounted for using a simple scheme called the ‘dipole’ model. This model copes with ‘cue drop out’ (for example, losing a binocular view of the ball) by implicitly switching to the remaining cue and returning to dual weighting if both cues become available. If a conflict arises between the looming and the binocular information then the model simply increases the weight of the most-immediate cue (that is, the one that specifies the shortest ttc). In summary, Rind and Simmons have provided an admirable description of how neurones respond to looming stimuli and they thus provide an excellent review of one of the neural systems subserving ttc judgements. In humans, looming can provide useful information on ttc but it cannot explain the precision of many timing judgements. The weight of evidence suggests that looming is combined with binocular cues for fast interceptive actions in humans. Furthermore, analysis of extant studies suggests that the perceptual information used for timing action is both task and situation specific. Tresilian argued that there can be no general theory of ttc perception because of the dynamic nature of the visual system: a noisy adaptive neural system that converges on complicated and task-specific solutions to ‘get the job done’. Timing judgements might involve looming and binocular information, but the use of these cues is neither necessary nor sufficient for a range of tasks. Future studies will hopefully elucidate the other neural mechanisms involved in providing the requisite perceptual information for such precise behaviour.