Heading Discrimination Threshold (HDT) can quantify human’s ability to discriminate self-motion heading directions in the presence of visual cue, vestibular cue or combined visual-vestibular cues. HDT has been the subject of many studies (e.g. Gu et al. 2008; Fetsch et al. 2009). A Bayesian model had been developed for modelling an optimal visual-vestibular integration process. It could predict a theoretical minimum combined HDTs as functions of visual and vestibular HDTs but suffered a prediction bias. A review of literature indicates that visual-vestibular heading integration has not been studied when vestibular stimuli are weak (peak acceleration of vestibular stimuli 30 mg, 0.2 and 0.5 Hz) (Gap 1). Weak vestibular stimuli are common to passengers or drivers traveling at constant ve...[ Read more ]

Heading Discrimination Threshold (HDT) can quantify human’s ability to discriminate self-motion heading directions in the presence of visual cue, vestibular cue or combined visual-vestibular cues. HDT has been the subject of many studies (e.g. Gu et al. 2008; Fetsch et al. 2009). A Bayesian model had been developed for modelling an optimal visual-vestibular integration process. It could predict a theoretical minimum combined HDTs as functions of visual and vestibular HDTs but suffered a prediction bias. A review of literature indicates that visual-vestibular heading integration has not been studied when vestibular stimuli are weak (peak acceleration of vestibular stimuli < 30 mg, 0.2 and 0.5 Hz) (Gap 1). Weak vestibular stimuli are common to passengers or drivers traveling at constant velocities. More importantly, studies of HDT under weak vestibular cues can test hypotheses related to the possible cause of prediction bias (Section 2.4). Little is known about the possible source of prediction bias (Gap 2). The previous Bayesian model assumed that the distribution of perceived heading directions follows a normal distribution. But actually, the perceived heading distribution involves circular co-ordinates and an angular normal distribution (e.g., von Mises distribution) may be a better choice but has not been studied (Gap 3). In this research, a new optimal visual-vestibular integration model with von Mises distribution assumption is proposed. In order to fill Gap1, Experiment 1 was conducted to examine the full factorial combinations of near-threshold vestibular stimuli (0, 9, 14 and 19 mg) with and without visual stimuli. Results indicated that the near-threshold vestibular stimulus was a disturbance factor rather than a contributing factor in visual-vestibular integration (combined HDTs > visual HDTs / new model prediction, p<0.05). A similar disturbance effect of vestibular stimuli has been reported in neuroscience research (Deutschländer et al. 2002) but has not been considered in visual-vestibular heading integration research. It was hypothesized that if the visual HDTs were corrected with the disturbance effect, the previously reported prediction bias of Bayesian model could be reduced. This hypothesis could not be tested with Experiment 1 data because many subjects’ HDTs could not be measured because the vestibular stimuli were too weak and too closed to thresholds. We need to identify the supra-threshold vestibular stimuli and Experiments 2 were conducted to search for the supra-threshold vestibular stimuli. The effects of frequency (0.25, 0.5 and 1 Hz) and peak acceleration (20, 25 and 30 mg) on vestibular HDTs were studied. Results showed that both the frequency and peak acceleration of the vestibular cues significantly affected vestibular HDTs (p < 0.05). The supra-threshold vestibular stimulus was found to be 0.5 Hz at 30 mg. Experiment 3 was then conducted to search for the parameter of visual stimuli (coherence: 55%, 40% and 25%) that can make subjects’ visual HDTs comparable to these levels of vestibular HDTs. Finally, Experiment 4 tested the combinations of three vestibular stimuli (Peak acceleration: 5, 20 and 30 mg; Frequency: 0.5 Hz) and visual stimuli (none, 0% and 30% coherence). The corrected visual HDT (considering disturbance effect) was estimated in a 5mg combined condition (i.e. 30% visual cues and 5 mg vestibular stimuli). Results showed that the predicted combined HDTs with corrected visual HDTs were not significantly different from the measurement while the prediction with uncorrected visual HDTs were significantly different from measurement (p<0.05). The novel methods of correcting the visual HDTs and the original data of this thesis could provide a better understanding of the visual-vestibular integration of heading perception with weak vestibular stimuli.