Neural Coding of Real and Implied Motion in the Human Brain

Abstract

Perceiving and processing visual motion is crucial for all animals, including humans. Brain regions in the human brain that are responsive to real motion have been extensively studied with different neuroimaging methods. However, the neural codes that are related to real motion have been primarily addressed using highly reductionist and mostly artificial motion stimuli, mostly using so-called random dot kinematograms. Studies using more natural forms of motion that the brain evolved and developed to deal with are comparably rare. Moreover, real, physical motion is not the only type of stimulus that induces motion perception in humans. Implied motion stimuli also induce motion perception although the stimuli do not carry physical motion information. Implied motion stimuli are for example still images containing a snap-shot of an object in motion. Various contextual cues mediate the percept of motion, including the context of the object in its background, and in particular the object composition and its axial position in the image that mediate both, the impression of implied motion as well as its direction. This means that at the neural level, object processing must be used to generate the implied motion percept. The work described in this thesis investigated the neural coding of real and implied motion in the human brain. The investigation was done using functional brain imaging of human adults and data were collected with a 3-Tesla MRI scanner while the participants viewed a variety of distinct visual stimuli. The visual stimuli contained directional real and implied motion and were created specifically for this study. For real motion stimuli, the aim of was to engage a maximal number of directionally selective units, in order to maximize the overlap to the subset of units potentially involved in coding implied motion. Hence, real motion stimuli were created such that the static component frames had natural image statistics (known to activate neurons more effectively) by using Fourier-scrambled natural images, and motion was presented at a wide range of motion velocities. Similarly, implied motion stimuli were derived from photographs of natural scenes. They were created by placing objects such as airplanes, birds, cars, or snapshots of walking humans on a set of contextual background images such as skylines or streets. For both, real motion and implied motion, stimuli for four directions were created: forwards and backwards,and left- and rightwards.