The neural substrates of object individuation

Abstract

Despite the constant influx of visual information, observers are nonetheless able to segment this input into discrete objects and events. The perceptual system does so on the basis of spatial and temporal properties, thus allowing one to keep track of visual objects as they move to different locations across time. This process of object individuation is integral for visual awareness; when it is disrupted, stimuli are no longer perceived. Behavioural studies that have investigated object individuation in temporal and spatial domains converge on the idea that object individuation is a capacity-limited process that gates which items proceed for further analysis. Although there has been considerable theoretical and behavioural work on object individuation, we know relatively little about its neural substrates. The experiments in this thesis investigate the brain regions that support object individuation across multiple episodic contexts and processing stages, characterise its capacity limits and relationship to identification, and isolate the stage of processing at which individuation arises.Chapter 2 describes a functional magnetic resonance imaging (fMRI) study that isolated the neural bases of temporal individuation during perception, and the consequences that arise when its processing limit is reached (repetition blindness, RB). RB is a rapid serial visual presentation phenomenon characterised by reduced performance on trials with a target repetition compared with those in which the two targets have different identities (non-repetition trials). This failure of perceptual consciousness is thought to reflect a capacity limit of temporal individuation. I first verified that my RB paradigm elicited the standard behavioural effect (Experiment 1) and was specific to the temporal limits of individuation, rather than identification (Experiment 2). Using fMRI (Experiment 3), I found that multivariate patterns of blood-oxygen-level-dependent (BOLD) activity across a large number of occipital, parietal and frontal regions could discriminate between trials in which a repetition was correctly reported (a demanding individuation scenario), compared with correct non-repetition trials (a relatively easy individuation scenario). Consistent with current models of consciousness, and contrary to existing work on spatial individuation at the level of memory encoding, these findings suggest that temporal individuation is supported by a distributed set of brain regions. In terms of RB itself, I found greater activity in the left premotor cortex for incorrect versus correct repetition trials. This result suggests that the left premotor region is critical for the processing limitations that give rise to RB.In Chapter 3, I tested whether object individuation and identification can be dissociated in the brain at the level of visual short term memory (VSTM) encoding and beyond, as proposed in the neural object file theory. Participants completed a delaye VSTM task in which they had to remember the identity and spatial locations of one object, four identical objects or four different objects. To isolate object individuation regions, BOLD activity was compared between one object and four identical objects. By contrast, to identify object identification regions, BOLD activity was contrasted between four identical objects and four different objects. Across univariate and multivariate analyses, I found brain regions that were specific to individuation or identification processes, and others that were common between the two. These findings challenge the neural object file theory, and instead suggest that object individuation and identification processes have distributed and overlapping neural substrates.The aim of the experiments reported in Chapter 4 was to characterise the timecourse of object individuation for attended and unattended objects, and determine the extent to which this operation draws on early sensory cortices. Previous event-related potential (ERP) studies were unable to show definitive evidence of object individuation at early perceptual stages of analysis, because the paradigms in these studies confounded manipulations of individuation load (i.e., number of targets) with low-level visual features (e.g., luminance). I first developed a novel enumeration paradigm involving items defined by illusory contours, which held all physical properties constant across conditions (Experiment 1), and then used electroencephalography (EEG) to investigate the timecourse of object individuation for attended and unattended stimuli (Experiment 2). Both P1 (100-140 ms) and N2 (185-250 ms) amplitudes increased with the number of attended targets, but the number of unattended non-targets only modulated the N2. An fMRI study (Experiment 3) showed that early visual cortex (including V2) was sensitive to individuation load, which I hypothesised might underpin the observed P1 effect. These findings demonstrated that task-relevant individuation occurs at a relatively early stage of visual information processing, and voluntary spatial attention modulates the timecourse of this operation.Taken together, the experiments reported in this thesis offer novel insights into our understanding of the neural underpinnings of object individuation across various stages of processing. These findings have implications for current theoretical accounts of object individuation and its associated processes in the brain, and contribute to models of how individuation should be operationalised as a construct.