Abstract
Predictive mechanisms for target interception based on ‘tau’ (Lee, Perception, 1976) have been revealed in fronto-parietal networks (Field and Wann, Curr. Biol., 2005; Merchant and Georgopoulos, J. Neurophysiol., 2006). The mechanisms underlying interception of falling targets are less understood. A recent fMRI study indicated that neural populations in the vestibular network encode visual gravitational acceleration (Indovina et al., Science, 2005). Here we presented 1g (natural gravity), 0g (constant speed), −1g (reversed gravity) targets in smooth motion (RM) or long-range apparent motion (AM) during fMRI to test whether the vestibular network provides a sensory code or an internal model of visual gravity. We expected that a sensory mechanism selective for downward image accelerations should be engaged only during interception of 1g targets, but an internal model should be engaged also by 0g targets which are mistaken for gravitational. Moreover, finding similar activity patterns with either RM or AM would exclude that the activation merely reflects differences in low-level spatiotemporal properties of the stimuli, and would indicate generalization of the internal model across low-level and high-level motion. We found that both 1g and 0g targets engaged the vestibular network in RM and AM, and in addition engaged a higher-level motion region of inferior parietal lobule in AM (Claeys et al., Neuron, 2003). Therefore, internalized gravity constrains the matching process used to interpret and intercept targets' vertical motion.