Attention Capture by Angry Faces Depends on the Distribution of Attention

<p>The threat-capture hypothesis posits a threat-detection system that automatically directs visual attention to threat-related stimuli (e.g., angry facial expressions) in the environment. Importantly, this system is theorised to operate preattentively, processing all input across the visual field in parallel, prior to the operation of selective attention. The threat-capture hypothesis generates two predictions. First, because the threat-detection system directs attention to threat automatically, threat stimuli should capture attention when they are task-irrelevant and the observer has no intention to attend to them. Second, because the threat-detection system operates preattentively, threat stimuli should capture attention even when it is engaged elsewhere. This thesis tested these predictions using behavioural measures of attention capture in conjunction with the N2pc, an event-related potential (ERP) index of attention selection. Experiment 1 tested the first prediction of the threat-capture hypothesis – that threat stimuli capture attention when they are task-irrelevant. Participants performed a dot-probe task in which pairs of face cues – one angry and one neutral – preceded a lateral target. On some trials, the faces were Fourier phase-scrambled to control for low-level visual properties. Consistent with the threat-capture hypothesis, an N2pc was observed for angry faces, suggesting they captured attention despite being completely task-irrelevant. Interestingly, this effect remained when faces were Fourier phase-scrambled, suggesting it is low-level visual properties that drive attention capture by angry faces. Experiments 2A and 2B tested the second prediction of the threat capture hypothesis – that threat stimuli capture attention when it is engaged elsewhere. Participants performed a primary task in which they searched a column of letters at fixation for a target letter. The perceptual load of this task was manipulated to ensure that attentional resources were consumed by this task. Thus there were high and low perceptual load conditions in these experiments. Task-irrelevant angry faces interfered with task performance when the perceptual load of the task was high but not when it was low (Experiment 2A). Similarly, angry faces elicited an N2pc, indicating that they captured attention, but only when perceptual load was high and when faces were phase-scrambled (Experiment 2B). These experiments further suggest that low-level visual factors are important in attention capture by angry faces. These results appear to be inconsistent with the threat-capture hypothesis, and suggest that angry faces do not necessarily capture attention when it is engaged elsewhere.</p>

Attention Capture by Angry Faces Depends on the Distribution of Attention

<p>The threat-capture hypothesis posits a threat-detection system that automatically directs visual attention to threat-related stimuli (e.g., angry facial expressions) in the environment. Importantly, this system is theorised to operate preattentively, processing all input across the visual field in parallel, prior to the operation of selective attention. The threat-capture hypothesis generates two predictions. First, because the threat-detection system directs attention to threat automatically, threat stimuli should capture attention when they are task-irrelevant and the observer has no intention to attend to them. Second, because the threat-detection system operates preattentively, threat stimuli should capture attention even when it is engaged elsewhere. This thesis tested these predictions using behavioural measures of attention capture in conjunction with the N2pc, an event-related potential (ERP) index of attention selection. Experiment 1 tested the first prediction of the threat-capture hypothesis – that threat stimuli capture attention when they are task-irrelevant. Participants performed a dot-probe task in which pairs of face cues – one angry and one neutral – preceded a lateral target. On some trials, the faces were Fourier phase-scrambled to control for low-level visual properties. Consistent with the threat-capture hypothesis, an N2pc was observed for angry faces, suggesting they captured attention despite being completely task-irrelevant. Interestingly, this effect remained when faces were Fourier phase-scrambled, suggesting it is low-level visual properties that drive attention capture by angry faces. Experiments 2A and 2B tested the second prediction of the threat capture hypothesis – that threat stimuli capture attention when it is engaged elsewhere. Participants performed a primary task in which they searched a column of letters at fixation for a target letter. The perceptual load of this task was manipulated to ensure that attentional resources were consumed by this task. Thus there were high and low perceptual load conditions in these experiments. Task-irrelevant angry faces interfered with task performance when the perceptual load of the task was high but not when it was low (Experiment 2A). Similarly, angry faces elicited an N2pc, indicating that they captured attention, but only when perceptual load was high and when faces were phase-scrambled (Experiment 2B). These experiments further suggest that low-level visual factors are important in attention capture by angry faces. These results appear to be inconsistent with the threat-capture hypothesis, and suggest that angry faces do not necessarily capture attention when it is engaged elsewhere.</p>

High-Density Electrophysiological Recordings to Assess the Dynamic Properties of Attention

The brain has limited processing capacities. Attention selection processes are continuously shaping humans’ world perception. Understanding the mechanisms underlying such covert cognitive processes requires the combination of psychophysical and electrophysiological investigation methods. This combination allows researchers to describe how individual neurons and neuronal populations encode attentional function. Direct access to neuronal information through innovative electrophysiological approaches, additionally, allows the tracking of covert attention in real time. These converging approaches capture a comprehensive view of attentional function.

Contextual Cueing Accelerated and Enhanced by Monetary Reward: Evidence From Event-Related Brain Potentials

The vital role of reward in guiding visual attention has been supported by previous literatures. Here, we examined the motivational impact of monetary reward feedback stimuli on visual attention selection using an event-related potential (ERP) component called stimulus-preceding negativity (SPN) and a standard contextual cueing (CC) paradigm. It has been proposed that SPN reflects affective and motivational processing. We focused on whether incidentally learned context knowledge could be affected by reward. Both behavior and brain data demonstrated that contexts followed by reward feedback not only gave rise to faster implicit learning but also obtained a larger CC effect.

Dynamics of Internal Attention and Internally-Directed Cognition: The Attention-to-Thoughts (A2T) Model

The propensity to focus inward – internal attention – is fundamental to human mental life and internally-directed cognition (IDC) such as mindwandering and (mal)adaptive self-reflection. Yet, our understanding of the mechanisms through which internal attention shapes IDC is limited. We argue that our capacity to predict and model (mal)adaptive IDC may be significantly facilitated through understanding the complexity and dynamics of how internal attention interacts with other cognitive processes from which higher-level IDC emerges. We therefore introduce the Attention-to-Thoughts (A2T) model – a dynamic systems theory and computational model of internal attention in IDC. Through the model we aim to, first, conceptually and computationally define momentary states of this dynamic system; and, second, to simulate and predict differential temporal trajectories of this dynamic system through which IDC emerge. Through a series of experimental simulations, we explore how A2T may be used to better understand how internal attention selection is expressed from moment-to-moment; how the dynamic system of internal attention unfolds by documenting how, as a function of contextual demands for focused attention, internal attentional selection iteratively transacts with working-memory and emotion; and, in turn, how higher-level maladaptive IDC (e.g., repetitive negative thinking, cognitive dyscontrol) emerges from temporal trajectories of the dynamic system of internal attention. Finally, we highlight key conceptual, computational and methodological directions for the study of internal attention, IDC and related phenomena (e.g., mindfulness).

Scalable Person Re-Identification by Harmonious Attention

Existing person re-identification (re-id) deep learning methods rely heavily on the utilisation of large and computationally expensive convolutional neural networks. They are therefore not scalable to large scale re-id deployment scenarios with the need of processing a large amount of surveillance video data, due to the lengthy inference process with high computing costs. In this work, we address this limitation via jointly learning re-id attention selection. Specifically, we formulate a novel harmonious attention network (HAN) framework to jointly learn soft pixel attention and hard region attention alongside simultaneous deep feature representation learning, particularly enabling more discriminative re-id matching by efficient networks with more scalable model inference and feature matching. Extensive evaluations validate the cost-effectiveness superiority of the proposed HAN approach for person re-id against a wide variety of state-of-the-art methods on four large benchmark datasets: CUHK03, Market-1501, DukeMTMC, and MSMT17.

Perception and misperception of bodily symptoms from an Active Inference perspective: Modelling the case of panic disorder.

We advance a novel computational model that characterizes formally the ways we perceive or misperceive bodily symptoms, in the context of panic attacks. The computational model is grounded within the formal framework of Active Inference, which considers top-down prediction and attention dynamics as key to perceptual inference and action selection. In a series of simulations, we use the computational model to reproduce key facets of adaptive and maladaptive symptom perception: the ways we infer our bodily state by integrating prior information and somatic afferents; the ways we decide whether or not to attend to somatic channels; the ways we use the symptom inference to make decisions about taking or not taking a medicine; and the ways all the above processes can go awry, determining symptoms misperception and ensuing maladaptive behaviors, such as hypervigilance or excessive medicine use. While recent existing theoretical treatments of psychopathological conditions focus on prediction-based perception (predictive coding), our computational model goes beyond them, in at least two ways. First, it includes action and attention selection dynamics that are disregarded in previous conceptualizations but are crucial to fully understand the phenomenology of bodily symptoms perception and misperception. Second, it is a fully implemented model that generates specific (and personalized) quantitative predictions, thus going beyond previous qualitative frameworks.