?. Circular Receptive Field Structures for Flow Analysis and Heading Detection

Recent years have brought forward different models on how the brain might encode heading from optic flow. Neurons in these models can encode heading for a variety of self-motion conditions, while responding to optic flow stimuli similarly as found in electrophysiological studies. Yet, little attention has been given to the receptive field structure of neurons that integrate local motion signals to analyse the optic flow. Intuitively, radial structures might seem suited for the task of heading detection, since pure observer translation causes flow to eminate from the point of heading (Figure 1a). However, rotational flow (Figure 1b) during simultaneous eye rotation can cause the retinal flow to be shifted away from the heading (Figure 1c). Moreover, variation in point distances with respect to the translating eye results in retinal motion differences, called motion parallax, that can cause the flow during eye rotation to deviate even more from being purely radial. This poses the question what would be the optimal receptive field structure to deal with complex flow fields for retrieving heading.