Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.768038
Title: From features to concepts : tracking the neural dynamics of visual perception
Author: Dima, Diana
ISNI:       0000 0004 7652 2278
Awarding Body: Cardiff University
Current Institution: Cardiff University
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The visual system is thought to accomplish categorization through a series of hi- erarchical feature extraction steps, ending with the formation of high-level cate- gory representations in occipitotemporal cortex; however, recent evidence has chal- lenged these assumptions. The experiments described in this thesis address the question of categorization in face and scene perception using magnetoencephalog- raphy and multivariate analysis methods. The first three chapters investigate neural responses to emotional faces from different perspectives, by varying their relevance to task. First, in a passive view- ing paradigm, angry faces elicit differential patterns within 100 ms in visual cortex, consistent with a threat-related bias in feedforward processing. The next chap- ter looks at rapid face perception in the context of an expression discrimination task which also manipulates subjective awareness. A neural response to faces, but not expressions is detected outside awareness. Furthermore, neural patterns and behavioural responses are shown to reflect both facial features and facial config- uration. Finally, the third chapter employs emotional faces as distractors during an orientation discrimination task, but finds no evidence of expression processing outside of attention. The fourth chapter focuses on natural scene perception, using a passive view- ing paradigm to study the contribution of low-level features and high-level cat- egories to MEG patterns. Multivariate analyses reveal a categorical response to scenes emerging within 200 ms, despite ongoing processing of low-level features. Together, these results suggest that feature-based coding of categories, opti- mized for both stimulus relevance and task demands, underpins dynamic high- level representations in the visual system. The findings highlight new avenues in vision research, which may be best pursued by bridging the neural and behavioural levels within a common computational framework.
Supervisor: Not available Sponsor: Economic and Social Research Council ; Medical Research Council ; Cardiff University
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.768038  DOI: Not available
Keywords: BF Psychology
Share: