| Summary: | Studies of interception have shown that participants expect free-falling targets to descend vertically and accelerate at g (9.81 m/s2). Visuomotor delays render motor commands obsolete in such rapidly changing situations, so predictive control is necessary. Internal models have been proposed as a mechanism to predict the future position of free-falling objects. Internal models are predictive systems that provide input to the motor and sensory systems, and are acquired by the CNS from experience. The overall aim of this thesis was to investigate the existence of a possible internal model of gravity and to explore its directional and acceleration amplitude aspects. This thesis is the first study to investigate directional and amplitude aspects of such a model by systematically varying stimulus target kinematics. In the first two experiments, participants performed visually guided tracking of realistic virtual stimuli. In experiment 1, stimuli accelerated at g but moved in different straight-line directions. In experiment 2, stimuli always descended vertically but with different accelerations (as percentages of g). Tracking was by finger pointing, laser pointing, gaze tracking in isolation or gaze tracking with concomitant manual or laser tracking. In the absence of on-screen feedback of indicated position (i.e. non-laser trials), tracking was generally tuned for targets that a) descended vertically (or tended increasingly to vertically downwards over time) and b) accelerated at g or slightly less than g. In the third experiment, participants judged whether a target accelerating at g was moving vertically, or to the left or right of the vertical. The stimuli from experiment 1 were used. I found that a) participants’ reports were influenced by an expectation that gravity would affect the trajectory even when there is visual information to the contrary (i.e. an effect of gravity was expected on observed trajectory); and b) physics expertise yielded no difference in performance suggesting that performance, based on an internal model of gravity, is determined by experience of seeing things fall in the real-world under gravity, rather than by intellectual understanding of gravity. In the fourth experiment, participants judged whether a target accelerated at less than g, at g, or greater than g. The stimuli from experiment 2 were used. As with experiment 3, I found no effect of physics expertise and only weak evidence in support of an internal model of gravity. I conclude that the CNS is likely construct and maintain an internal model of gravity that predicts the future position of free-falling targets to be in a ‘real-world' manner – acceleration at 99% or 100% of g and vertically downwards or in a direction tending increasingly to vertically downwards over time. I found no evidence for ‘direct perception’ (observation yielding first-order time-to-contact information). In the final conclusion I briefly discuss alternative possibilities for predictive motor control.
|
|---|
