Catalog Home Page

Visual processing in the human brain: Investigating deviance detection from a predictive coding perspective

Male, Alie Gabriella (2019) Visual processing in the human brain: Investigating deviance detection from a predictive coding perspective. PhD thesis, Murdoch University.

[img]
Preview
PDF - Whole Thesis
Download (8MB) | Preview

Abstract

According to predictive coding, the brain gives extra processing to unpredicted events that disrupt anticipated patterns. To adapt to these events, the brain continually extracts statistical regularities about sensory input from past input. When something unpredicted occurs, it produces an error. In vision, this can be shown by the visual mismatch negativity (vMMN) in event-related potentials (ERPs). The vMMN reaches its maximum amplitude between 150 and 300 ms after the onset of an irregular, deviant event in a sequence of otherwise regular, standard events and it is usually measured from areas on the scalp closest to the visual cortices (e.g., parieto-occipital areas). Attention toward a deviant is not necessary to generate the vMMN, suggesting that regularities and irregularities are pre-attentively encoded and detected, respectively.

Although vMMN research continues to grow, there are still unanswered questions about it. This thesis focuses on clarifying some of these issues, asking whether the type or size of the difference between predicted and unpredicted visual input (i.e., the magnitude of deviance) or visual field in which deviance occurs can affect the vMMN. To remedy this, I manipulated these facets across four studies. My thesis was that local aspects of change detection, such as the magnitude of deviance, affect the brain’s error response to unpredicted input, evidenced by the vMMN.

A conclusion regarding the effect of magnitude of deviance, the type of change, or visual field on the vMMN was not possible given that (1) ERPs to rule-based deviants and standards did not differ where participants found it difficult to detect irregularities in visual input, and (2) changes in basic properties of well-controlled visual stimuli do not evoke the vMMN. Subsequently, my thesis became that isolated changes in basic properties of visual input do not evoke the vMMN, perhaps because these changes are detected and resolved prior to the vMMN.

Instead, this thesis provides evidence for an earlier deviant-related positivity for changes in low-level features of visual input. This is the first report of a possible pre-vMMN positive prediction error and represents a significant and original contribution to the wider field.

Item Type: Thesis (PhD)
Murdoch Affiliation: Psychology, Counselling, Exercise Science and Chiropractic
Supervisor(s): Drummond, Peter, O'Shea, R. and Roeber, U.
URI: http://researchrepository.murdoch.edu.au/id/eprint/52071
Item Control Page Item Control Page

Downloads

Downloads per month over past year