Hu, Dan
(2024)
Exploring the common mechanisms of motion-based visual prediction.
PhD thesis, University of Nottingham.
Abstract
The ability of the visual system to predict the spatiotemporal properties of a moving object (e.g., when and where a moving ball will be) is essential to daily life. Motion-based visual prediction may serve not only for explicit anticipation about the future but also for compensation for neural transmission delays in visual processing when perceiving motion. We asked whether there are generic mechanisms of motion-based visual prediction for motion perception in the thesis.
In Chapter 3, we exploited the Motion Induced Spatial Conflict (MISC) effect. The striking illusory jitter occurs when two borders, with different perceived speeds, move smoothly as a combined rigid stimulus. MISC jitter frequency has invariably been shown to be at around 10 Hz within a subject and has been considered to reflect the rate of processing of motion-based spatial pattern prediction in the brain. We employed an individual difference approach to explore the link between MISC and other tasks addressing visual prediction, assuming that an inter-subject task performance correlation suggests a common mechanism for the tasks. We showed that the characteristic individual MISC rate was positively associated with the individual accumulation speed in the Motion Induced Position Shift (MIPS) effect, where carrier drifting motion within a stationary envelope induces a perceived position shift of both carrier and envelope spatial patterns along the motion path. Furthermore, we demonstrated that the individual differences in the MIPS rate were reliable across speed and eccentricity conditions that affected the MIPS illusory shift magnitude, in the range of 1 to around 4 deg/s for 4- and 7-deg eccentricity conditions, adding evidence that the individual MIPS rate variations arise from intrinsic brain mechanisms rather than random noise. In addition, Chapter 4 used the EEG technique to investigate relationships between brain oscillation and MISC illusory jitter perception. The results confirmed an alpha-band power enhancement in the brain while perceiving the MISC jitter versus a physical jitter at the same frequency suggested by previous MEG research. Taken together, the observations across the two chapters support a shared and periodic motion-based spatial pattern prediction mechanism underlying motion perception, which operates at approximately 10 Hz and connects with the alpha-band brain activity.
In Chapter 5, we addressed the role of motion-based visual prediction in motion deblurring in another line of research. It has been suggested that perceptual blur might be caused by the visual persistence of object patterns in motion and that the visual system may aggregate and summate the object’s spatial patterns along the motion pathway to prevent motion blur. Prior studies have proposed a large number of space-time-oriented passive filters in the early visual cortex to achieve this deblurring process, with the filter’s behaviour stimulated by motion but independent of motion parameters, such as speed. However, herein, we assumed that the visual prediction mechanism underpins motion deblurring by shifting forward object patterns on the motion trajectory leading to alignment in space and spatiotemporal summation, utilising specific motion signals. After conducting a series of psychophysical experiments, we have shown that the motion deblurring effect persistently depended on stimulus speed, while manipulating stimulus contrast, duration and while interrupting the motion. The results are less likely to be accounted for by the performance of passive filters but are compatible with our visual prediction assumption for motion deblurring sensitive to image speeds. Moreover, we found that the motion-based visual prediction effect was concomitant with the motion deblurring effect for the same stimuli and that there was a positive correlation between the magnitude of these effects, indicating a firm linkage of mechanisms. Results were accommodated by a simple mathematical model described in this chapter. Overall, these findings converged to advance the fundamental role of motion-based visual prediction in motion deblurring by ‘displacing object spatial patterns forward along the motion path to allow for the summation of the spatially aligned patterns, consistent with the motion’.
To summarise, the thesis has made important contributions to understanding motion-based visual prediction by revealing two common mechanisms in general motion perception scenarios about spatial pattern prediction at a characteristic rate of ~10 Hz and about motion-prediction-based deblurring relying on a spatial pattern forward shift guided by motion.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
![[thumbnail of Thesis Dan Hu - corrected.pdf]](https://eprints.nottingham.ac.uk/style/images/fileicons/application_pdf.png "Thesis Dan Hu - corrected.pdf")
![[thumbnail of Movies 1-3 in a Zip file]](https://eprints.nottingham.ac.uk/style/images/fileicons/video.png "Movies 1-3 in a Zip file")