Simulated self-motion alters perceived time to collision

Many authors have assumed that motor actions required for collision avoidance and for collision achievement (for example, in driving a car or hitting a ball) are guided by monitoring the time to collision (TTC), and that this is done on the basis of moment-to-moment values of the optical variable τ [1-3]. This assumption has also motivated the search for single neurons that fire when τ is a certain value [4-8]. Almost all of the laboratory studies and all the animal experiments were restricted to the case of stationary observer and moving object. On the face of it, this would seem reasonable. Even though humans and other animals routinely perform visually guided actions that require the TTC of an approaching object to be estimated while the observer is moving, τ provides an accurate estimate of TTC regardless of whether the approach is produced by self-motion, object-motion or a combination of both. One might therefore expect that judgements of TTC would be independent of self-motion. We report here, however, that simulated self-motion using a peripheral flow field substantially altered estimates of TTC for an approaching object, even though the peripheral flow field did not affect the value of τ for the approaching object. This finding points to long range interactions between collision-sensitive visual neurons and neural mechanisms for processing self-motion.