Neural correlates of contextually modulated depth perception in humans

Abstract

Binocular disparity is one of the most powerful cues for retrieving depth information from object structures. Classical literature in the study of depth perception proposed that disparity processing primarily involves dorsal areas. However, recent neuroimaging studies present findings which indicate that both ventral and dorsal areas contribute to disparity processing. Interestingly, visual processing appears to be affected by object context. For instance, race modulates luminance judgements of face stimuli. The mechanisms underlying depth processing under complex contexts remain unclear. This thesis presents a series of studies that examines human disparity processing in terms of behavioural sensitivity and neural mechanisms under various contexts. Chapter Two documents a study that used both behavioural and fMRI paradigms to measure behavioural thresholds and neural responses to depth information of complex 3D objects. Stimuli were disparity-defined geometric objects rendered as random dot stereograms (RDS), and presented in plausible and implausible variations (e.g., a normal versus a physically implausible Penrose triangle). The behaviour experiment involved two tasks that targeted different aspects of depth perception: 1) a signal-in-noise (SNR) depth task that involved judging the depth position of a target embedded in noise, and 2) a feature depth discrimination task that involved discriminating the nearer of two consecutively presented targets. Multivariate representations for plausible and implausible objects were notably distinguishable along depth-relevant region V7, in addition to object-relevant lateral occipital complex/cortex (LOC). Chapter Three examines the functional causality of contextual modulation by applying transcranial magnetic stimulation (TMS) to brain areas previously identified to be V7 and LOC. Observers completed a SNR task whilst receiving repeated-TMS. Observers were stimulated in either left and right V7, or left and right LOC, alongside a control site across separate sessions. Results suggested that disruption of the control site and LOC did not alter the threshold pattern for plausible and implausible objects (i.e., thresholds for implausible objects were lower than for plausible objects). However, the trend was reversed during disruption of V7. Chapter Four examines whether the contexts of more biological relevance than physical plausibility, specifically differing levels of animacy and status (i.e., living versus non-living) of daily objects, affect depth processing. Observers were asked to complete either a SNR or feature task. Results from the SNR task indicated heightened depth sensitivities towards living, low-animacy objects as compared to living, high-animacy objects which were not found in the feature task. The above findings reveal the susceptibility of human depth processing to object contexts, specifically object plausibility, level of animacy and status. It further speaks to the intricate relationship between higher, object-level processing and lower-level depth processing. Major findings include the implication of both LOC and V7 in the representation and retrieval of complex 3D shape information. Particularly, V7 may be causally involved in the modulation of signal weights during depth processing. Lastly, contexts of daily objects are relevant to depth sensitivity, further implying that experience and learning of seemingly unrelated features may impact lower perceptual processes. Future research could address how the representation of various contexts of daily objects affects depth processing.