Neural mechanisms of selective visual attention

The specificity of sequential statistical learning: Statistical learning accumulates predictive information from unstructured input but is dissociable from (declarative) memory for words

Learning statistical regularities from the environment is ubiquitous across domains and species. It might support the earliest stages of language acquisition, especially identifying and learning words from fluent speech (i.e., word-segmentation). But how do the statistical learning mechanisms involved in word-segmentation interact with the memory mechanisms needed to remember words - and with the learning situations where words need to be learned? Through computational modeling, we first show that earlier results purportedly supporting memory-based theories of statistical learning can be reproduced by memory-less Hebbian learning mechanisms. We then show that, in a memory recall task after exposure to continuous, statistically structured speech sequences, participants track the statistical structure of the speech sequences and are thus sensitive to probable syllable transitions. However, they hardly remember any items at all, with 82% producing no high-probability items. Among the 30% of participants producing (correct) high- or (incorrect) low-probability items, half produced high-probability items and half low-probability items - even while preferring high-probability items in a recognition test. Only discrete familiarization sequences with isolated words yield memories of actual items. Turning to how specific learning situations affect statistical learning, we show that it predominantly operates in continuous speech sequences like those used in earlier experiments, but not in discrete chunk sequences likely more characteristic of early language acquisition. Taken together, these results suggest that statistical learning might be specialized to accumulate distributional information, but that it is dissociable from the (declarative) memory mechanisms needed to acquire words and does not allow learners to identify probable word boundaries.

Biased competition between action representations

We propose a generalized version of the biased competition account of attention that may be applied to all domains of cognition. Based on our Generalized Biased Competition account, we propose a formal race model of selection of action representations. The model explains how action representations stored in long-term memory are competing for selection based on their match to the current environmental context and their importance weight. We then present results and model fits from three experiments using a recently developed multiple cue paradigm where several attention shifts with different associated reward values are competing. We show that participants were surprisingly efficient at selecting both when the number of cues and the number of possible reward values were increased. Only when we manipulated reward contingencies and knowledge of these, did the participants show suboptimal performance. The new Generalized Biased Competition account can also explain failures of executive control exemplified by goal neglect where instructions fail to influence behavior despite being retrievable. Finally, we argue that our model may provide a unified understanding of intentions, routines, and habits. Specifically, intentions, routines, and habits may be understood as a continuous range of the same fundamental form of action representation but with variation in their strength of long-term memory traces and importance weights.

Visual search is relational without prior context learning

The most prominent models of visual attention assume that we tune attention to the specific feature value of a sought-after object (e.g., a specific colour or orientation) to aid search. However, subsequent research has shown that attention is often tuned to the relative feature of the target, that the target has in relation to other items in the surround (e.g., redder/greener, darker/lighter, larger/smaller), in line with a Relational Account of Attention. Previous research is still limited though, as it used repeated-target designs and relatively sparse displays. With this, it is still unknown whether we can indeed tune attention to relative features prior to the first eye movement, or whether this requires context knowledge gained from experience. Moreover, it is unclear how search progresses from one item to the next. The present study tested these questions in a 36-item search display with multiple distractors and variable target and non-target colours. The first fixations on a trial showed that these displays still reliably evoked relational search, even when observers had no knowledge of the context. Moreover, the first five fixations within a trial showed that we tend to select the most extreme items first, followed by the next-extreme, until the target is found, in line with the relational account. These findings show that information about the relative target feature can be rapidly extracted and is used to guide attention in the first fixation(s) of search, whereby attention only hones in on the target colour after multiple fixations on relatively more extreme distractors.

Evidence from pupillometry, fMRI, and RNN modelling shows that gain neuromodulation mediates task-relevant perceptual switches

Perceptual updating has been hypothesised to rely on a network reset modulated by bursts of ascending neuromodulatory neurotransmitters, such as noradrenaline, abruptly altering the brain's susceptibility to changing sensory activity. To test this hypothesis at a large-scale, we analysed an ambiguous figures task using pupillometry and functional magnetic resonance imaging (fMRI). Behaviourally, qualitative shifts in the perceptual interpretation of an ambiguous image were associated with peaks in pupil diameter, an indirect readout of phasic bursts in neuromodulatory tone. We further hypothesised that stimulus ambiguity drives neuromodulatory tone, leading to heightened neural gain, hastening perceptual switches. To explore this hypothesis computationally, we trained a recurrent neural network (RNN) on an analogous perceptual categorisation task, allowing gain to change dynamically with classification uncertainty. As predicted, higher gain accelerated perceptual switching by transiently destabilising the network's dynamical regime in periods of maximal uncertainty. We leveraged a low-dimensional readout of the RNN dynamics to develop two novel macroscale predictions: perceptual switches should occur with peaks in low-dimensional brain state velocity and with a flattened egocentric energy landscape. Using fMRI, we confirmed these predictions, highlighting the role of the neuromodulatory system in the large-scale network reconfigurations mediating adaptive perceptual updates.

The 'Task' of Mind-Wandering Splits Both Multiple Demand and Default Mode Regions and Ramps-up the Deactivating Regions

The activation of multiple demand (MD) regions to diverse tasks has been linked to the demands of making task-related cognitive control changes - keeping it focussed on task, controlling attention and working memory, organizing and maintaining a task model that will control the sequence and identity of what is to be done when, etc. Demanding tasks that require such control are also accompanied by a deliberative cognition whereby cognitive changes do not occur automatically and have to be made deliberately. We investigated whether the deliberativeness of cognition activates MD regions regardless of task-related demands. When not engaged in demanding tasks, the mind wanders. We asked participants to do the same during task periods, and to differentiate from rests, we asked them to deliberately and intensely wander their minds across random thoughts. We found that a set of MD regions - pre-supplementary motor area (preSMA), anterior insula, and posterior part of the middle frontal gyrus - activated during these periods, and another set - intraparietal sulcus, right anterior prefrontal cortex - deactivated. In fact, some of the activating regions (e.g., preSMA) activated more during this task than in response to robust working memory updating demands. Dissociations were also present in the Default Mode Network (DMN). Parts of the temporoparietal junction deactivated while posterior cingulate and medial prefrontal regions activated. Lastly, we found that the deactivating regions ramped-up their activity across the 'task' duration, showing that this ramp-up, previously linked to demands of sequentially organizing extended tasks, occurs during any construed task, including those without such demands.

Distribution of attention in three-dimensional space

The distribution of spatial attention has mostly been studied for visual events presented within a two-dimensional space. In this study, we examined the distribution of spatial attention in a three-dimensional space (i.e., across the z-axis). Much previous research suggests that attention is universally biased toward stimuli appearing in near space compared to far space. However, the results of some studies suggest this 'near advantage' is task-specific, with some tasks instead producing an attention bias toward stimuli in far space. The current study investigated whether two tasks that differ in attentional priorities (i.e., target localization vs target discrimination) differentially bias attention across near and far depth. Across three experiments, we compared target localization and target discrimination tasks when a single target appeared as the stimulus (Experiment 1) and then, for a cue-target task, compared target localization (Experiment 2A) and target discrimination tasks (Experiment 2B). Our results support the proposal that the near advantage is task-specific. For target localization, reaction times (RTs) were shorter for near-targets than for far-targets, however, for target discrimination, RTs were shorter for far-targets than for near-targets. This result was revealed in both uncued and cue-target paradigms. The cue-target paradigm also showed that relative to same-depth conditions, the cueing effect pointed to greater facilitation when orienting attention from far-to-near space for target localization but from near-to-far space for target discrimination. These findings argue against a universal near advantage. Overall, the results were consistent with the notion that different task demands can differentially bias the distribution of attention across near and far depth, a proposal that has implications for the potential involvement of the dorsal and ventral visual processing streams.

Stimulus representations in visual cortex shaped by spatial attention and microsaccades

Microsaccades (MSs) are commonly associated with covert spatial attention, yet their impact on cortical processing of visual objects remains unclear. Rhesus macaques, randomly cued to attend to a target object amid distracters, were rewarded for detecting a color change in the target. While spatial attention does not affect the object tuning curves of V4 cells, the direction of MS significantly influenced object representations in V4 throughout the entire trial. Specifically, intervals following an MS toward the target exhibited superior stimulus decoding and sharper tuning curves compared to intervals following an MS away from the target. Furthermore, MSs directed toward the target enhanced neuronal responses to behaviorally relevant color changes, leading to faster reaction times. This sharpening effect stems from both a refreshing of the initial sensory response and an amplification of attention effects. The firing rate enhancement associated with spatial attention is delayed until the occurrence of the first MS directed toward the target. Subsequently, a positive effect of attention on firing rate, influenced by MS direction, was found throughout the trial across deep and superficial layers of V4, lateral pulvinar, and IT cortex. In summary, these findings underscore a crucial link between covert attention, object processing, and their coordination with MSs.

What roles matter? An explorative study on bullying and cyberbullying by using the eye-tracker

BACKGROUND

Bullying and cyberbullying are serious public health concerns that involve more roles beyond the bully and the victim (pro-bullies, defenders, bystanders). However, students often perceive the phenomena as dyadic.

AIM

The purpose was to examine students' perceptions of different roles when observing bullying and cyberbullying scenes combining implicit (attention by using the eye-tracker) and explicit (verbal reports) measures.

SAMPLE

We included 50 Italian students (aged 10-11).

METHODS

Students watched 12 drawings of different types of bullying and cyberbullying while their gaze was tracked, and subsequently described each drawing verbally. We ran repeated measure ANOVAs to compare attentional indexes (fixation count, visit count and total fixation duration) in observing roles and Cochran's Q test to evaluate differences in the verbal identification of roles.

RESULTS

Overall, the victim and bully were the most observed and identified roles in every type of bullying and cyberbullying scenario. Concerning the other roles, a discrepancy was observed between the implicit and explicit measures since although it was greatly identified, the pro-bully received less attention, and while the bystander received great attention, it was mentioned less. Finally, the defender was more observed and identified in physical bullying and cyberbullying.

CONCLUSIONS

Our study points out for the first time the dyadic perception of the phenomena among adolescents using implicit and explicit measures and sheds light on differences among the roles in different forms of bullying. Further research including the eye-tracker would be worthwhile given the possibility of exploring the phenomena from different perspectives.

Systematic review and meta-analysis of early visual processing, social cognition, and functional outcomes in schizophrenia spectrum disorders

Non-affective psychotic disorders are marked by cognitive and sensory processing abnormalities, including in early visual processing and social cognition. Understanding the relationships between these deficits and their impact on daily-life functional outcomes may help to improve outcomes in affected individuals. This systematic review and meta-analysis aimed to summarise the existing evidence on the relationships between early visual processing, social cognition, and functional outcomes, and to assess the evidence regarding the mediating role of social cognition in the association between early visual processing and functional outcomes in individuals with schizophrenia spectrum disorders. A comprehensive search across five databases identified 364 potentially eligible studies, with eight articles meeting all inclusion criteria. Meta-analytic techniques were employed to synthesise effect sizes and assess a meta-mediation model. Three random-effects meta-analyses revealed significant associations between all three domains of interest. Social cognition partially mediated the relationship between early visual processing and functional outcomes. The direct effect of early visual processing on functional outcomes remained significant, albeit with a reduced effect size. The findings suggest that interventions targeting both early visual processing and social cognition concurrently may improve functional outcomes more effectively than focusing on either domain alone.

Large-scale functional networks underlying visual attention

Attention networks are loosely defined as the regions of the brain which interact to control behaviour during attentional tasks, but the specific definition of attention networks varies between research programs based on task demands and modalities. The Attention Network Task was designed to exemplify three aspects of attention, alerting, orienting, and executive control, using a visual cueing paradigm. Its proponents propose a system of networks which underlies these aspects. It is debated whether there exists a unified system of networks which underlies attention independently of other cognitive and sensory processing systems. We review the evidence for an attention system within the domain of visual attention. Neuroimaging research using fMRI, EEG, MEG, and others across a variety of tasks attributed to attention, visual cueing, visual search, and divided attention, is compared. This concludes with a discussion on the limitations of an independent "attention system" for describing how the brain flexibly controls many abilities attributed to visual attention.

Neuroelectrical and behavioral correlates of constructed action recognition in Finnish Sign language

Language can be processed with varying levels of attentional involvement; consequently, the interplay between the language and attentional systems in the brain has been extensively studied in spoken languages. However, in signed languages (SLs), this interplay is less well understood. Here, we use Constructed Action (CA) - a meaning-making strategy based on enactment - as a window into the attentional mechanisms recruited in signed language comprehension. We explored the attentional processing of CA by identifying the sequence of processes involved and in which stage CA and its types might be processed differently. Finally, we investigated the associations between the brain mechanisms of CA detection and their behavioral manifestations, as well as with components of attention of the Attention Network Test (ANT). We also measured the electrophysiological correlates of performance on an oddball CA detection task in deaf and hearing L1 signers. We found that processes involved in all signers' active detection of CA involved automatic (indexed by N1 and P2) and attention-based processes (indexed by N2s and P3s). N2 posterior bilateral were also more negative for tokens of overt CA than for PT-only signs, while P3a was more positive for all types of CA than for PT. No significant results were found regarding the ANT. We conclude that specific attentional involvement in CA detection is triggered by the increasing enacting elements and saliency involved in CA. This study yielded new insights into the functional interaction between the neural mechanisms underlying attentional control and those mediating CA processing in SL.

Agile DQN: adaptive deep recurrent attention reinforcement learning for autonomous UAV obstacle avoidance

Unmanned Aerial Vehicle (UAV) obstacle avoidance in 3D environments demands sophisticated handling of high-dimensional inputs and effective state representations. Current Deep Reinforcement Learning (DRL) algorithms struggle to prioritize salient aspects of state representations and manage extensive state and action spaces, particularly in partially observable environments. Addressing these challenges, this paper proposes Agile DQN (AG-DQN), a novel algorithm that dynamically focuses on key visual features and robust Q-value estimation to enhance learning. The AG-DQN architecture synergizes several components-Glimpse Network, LSTM Recurrent Network, Emission Network, and Q-Network-to dynamically and selectively process crucial visual features, optimizing decision-making without processing the entire state. AG-DQN's adaptive temporal attention strategy also adjusts to environmental changes, maintaining a balance between recent and past observations. Experimental results demonstrate AG-DQN's improved performance over existing DRL methods, highlighting its potential in advancing autonomous UAV navigation and robotics.

Spatial and feature-selective attention interact to drive selective coding in frontoparietal cortex

Attention enables the selective processing of relevant information. Two types of selective attention, spatial and feature-selective attention, have separable neural effects but in real life are often used together. Here, we asked how these types of attention interact to affect information coding in a frontoparietal 'multiple-demand' (MD) network, essential for attentional control. Using functional magnetic resonance imaging (fMRI) with multivariate pattern analysis, we examined how covert attention to object features (colour or shape) and spatial locations (left or right) influences coding of task-related stimulus information. We found that spatial and feature-selective attention interacted multiplicatively on information coding in MD and visual regions, such that there was above-chance decoding of the attended feature of the attended object and no detectable coding of visually equivalent but behaviourally irrelevant aspects of the visual display. The attended information had a multidimensional neural representation, with stimulus information (e.g., colour) and discrimination difficulty (distance from the categorical decision boundary) reflected in separate dimensions. Rather than boosting processing of whole objects or relevant features across space, our results suggest neural activity reflects precise tuning to relevant information, indicating a highly selective control process that codes behaviourally relevant information across multiple dimensions.

Age-invariant benefits of spatiotemporal predictions amidst distraction during dynamic visual search

Visual search tasks are widely used to study attention amidst distraction, often revealing age-related differences. Research shows older adults typically exhibit poorer performance and greater sensitivity to distraction, reflecting declines in goal-driven attention. However, traditional search tasks are static and fail to capture the challenges and opportunities in natural environments, which include predictive structures within extended contexts. We designed a search variation where targets and distractors compete over time and embedded spatiotemporal regularities afford prediction-led guidance of attention. Critically, we manipulated the number of distractors to chart how benefits of expectations and deficits from distraction varied with age. Younger and older adults searched for multiple targets as they faded in and out of the display while varying the number of distracting elements between trials. Half the targets appeared at the same time and approximate locations and could be predicted. While we found evidence for decrement and elevated sensitivity to distraction with increasing age, benefits from predictions occurred in all groups. Interestingly, regardless of age, effects of predictions were only significant during periods of high distraction. This work extends our understanding of attention control through ageing to dynamic settings and indicates a dissociation between goal-directed and learning-driven attentional guidance.

Brain mechanisms underlying the inhibitory control of thought

Controlling action and thought requires the capacity to stop mental processes. Over the past two decades, evidence has grown that a domain-general inhibitory control mechanism supported by the right lateral prefrontal cortex achieves these functions. However, current views of the neural mechanisms of inhibitory control derive largely from research into the stopping of action. Whereas action stopping is a convenient empirical model, it does not invoke thought inhibition and cannot be used to identify the unique features of this process. Here, we review research that addresses how organisms stop a key process that drives thoughts: memory retrieval. This work has shown that retrieval stopping shares right dorsolateral and ventrolateral prefrontal mechanisms with action stopping, consistent with a domain-general inhibitory control mechanism, but also recruits a distinct fronto-temporal pathway that determines the success of mental control. As part of this pathway, GABAergic inhibition within the hippocampus influences the efficacy of prefrontal control over thought. These unique elements of mental control suggest that hippocampal disinhibition is a transdiagnostic factor underlying intrusive thinking, linking the fronto-temporal control pathway to preclinical models of psychiatric disorders and fear extinction. We suggest that retrieval-stopping deficits may underlie the intrusive thinking that is common across many psychiatric disorders.

Effort and salience jointly drive saccade selection

Choosing where to move the eyes ('saccade selection') is one of the most frequent human decisions and fundamentally shapes perception. Currently, saccade selection is thought to be predominantly driven by the observer's goals, selection history, and by the physical salience of stimuli. Recent work demonstrates that the inherent effort associated with planning and executing saccades ('saccade costs') also drives saccade selection: participants prefer making affordable over costly saccades. Do saccade costs still affect saccade selection when other factors such as salience attract gaze? Here, we addressed if, and how, saccade costs and salience together drive saccade selection by having participants freely choose between two potential saccade targets in different directions. Saccade targets either differed in salience or not, allowing us to disentangle the effects of saccade costs and salience. We observed that salience predicted saccade selection: participants chose salient over non-salient targets. Furthermore, saccade costs predicted saccade selection when equally salient targets were presented. When the possible targets differed in salience, the effect of saccade costs on saccade selection was reduced but not eliminated. Further analyses demonstrate that saccade costs and salience jointly drive saccade selection. Together, our results are in line with an accumulating body of work, and show that the role of effort in saccade selection is robust to salience. We conclude that effort must be considered a fundamental factor that drives where the eyes are moved.

The inferior fronto-occipital fasciculus: bridging phylogeny, ontogeny and functional anatomy

The inferior-fronto-occipital fasciculus (IFOF) is a long-range white matter tract that connects the prefrontal cortex with parietal, posterior temporal and occipital cortices. First identified in the 19th century through the pioneering studies of Mayo and Meynert using blunt dissection, its anatomy and function remain contentious topics. Structurally, its projections are well documented in human blunt dissection and tractography literature, yet its existence has been questioned by tract-tracing studies in macaques. Functionally, while traditional results from direct white matter stimulation during awake surgery suggested a contribution to language, recent evidence from stimulation and lesion data may indicate a broader role in executive control, extending to attention, motor cognition, memory, reading, emotion recognition and theory of mind. This review begins by examining anatomical evidence suggesting that the IFOF evolved in non-human primates to connect temporal and occipital cortices to prefrontal regions involved in context-dependent selection of visual features for action. We then integrate developmental, electrophysiological, functional and anatomical evidence for the human IFOF to propose it has a similar role in manipulation of visual features in our species-particularly when inhibition of overriding but task-irrelevant stimuli is required to prioritize a second, task-relevant stimulus. Next, we introduce a graded model in which dorsal (orbitofrontal, superior and middle frontal to precuneal, angular and supero-occipital projections) and ventral (inferior frontal to posterotemporal, basal temporal and infero-occipital) projections of the IFOF support perceptual or conceptual control of visual representations for action, respectively. Leveraging this model, we address controversies in the current literature regarding language, motor cognition, attention and emotion under the unifying view of cognitive control. Finally, we discuss surgical implications for this model and its impact on predicting and preventing neurological deficits in neurosurgery.

A research on cross-age facial recognition technology based on AT-GAN

Currently, predicting a person's facial appearance many years later based on early facial features remains a core technical challenge. In this paper, we propose a cross-age face prediction framework based on Generative Adversarial Networks (GANs). This framework extracts key features from early photos of the target individual and predicts their facial appearance at different ages in the future. Within our framework, we designed a GAN-based image restoration algorithm to enhance image deblurring capabilities and improve the generation of fine details, thereby increasing image resolution. Additionally, we introduced a semi-supervised learning algorithm called Multi-scale Feature Aggregation Scratch Repair (Semi-MSFA), which leverages both synthetic datasets and real historical photos to better adapt to the task of restoring old photographs. Furthermore, we developed a generative adversarial network incorporating a self-attention mechanism to predict age-progressed face images, ensuring the generated images maintain relatively stable personal characteristics across different ages. To validate the robustness and accuracy of our proposed framework, we conducted qualitative and quantitative analyses on open-source portrait databases and volunteer-provided data. Experimental results demonstrate that our framework achieves high prediction accuracy and strong generalization capabilities.

Object relations are processed with, but not without, awareness

The scope of unconscious integration is widely debated. Here, we examined this question, focusing specifically on deciphering the relations between two associatively related objects, in a set of five behavioral and electrophysiological experiments. Participants were presented with masked pairs of related and unrelated objects and were asked to judge their relatedness. When the masked pairs were visible, we found both a behavioral priming effect and a difference in the magnitude of the electrophysiological N400 component for unrelated compared with related pairs. In sharp contrast, when the pairs were invisible (validated using both subjective and objective awareness measures), no convincing evidence was found for relatedness processing: with electroencephalography, no difference in N400 amplitude nor above-chance decoding of pair relations was found in two separate experiments. Based on these results, we conclude that the data do not support unconscious relatedness processing, suggesting that consciousness might have a prominent role in enabling relational integration beyond the single object level, which is in line with leading theories of consciousness.

Top-down categorical information can be utilized in distractor suppression

Studies of attention have found behavioral benefits when observers are provided with top-down information related to distractor items. This effect is known as distractor suppression and is thought to be an independent process from target guidance. Most studies have focused on salient or singleton cues to elucidate upon the top-down versus bottom-up debate. Here, we examined if distractor suppression applies to categorical cues through three experiments. In Experiment 1, we modeled previously reported paradigms and replicated suppression effects to salient stimuli but with categorical objects. In Experiment 2, stimulus salience was kept constant, and participants were informed that a category of objects was always a distractor. We found distractor suppression effects where response times (RTs) were faster on trials that included the cued distractor category. In Experiment 3, distractor suppression effects were preserved when we utilized a slightly modified target set embedded in categorical objects. The present set of studies indicate that categorical information can be suppressed. The results suggest that attentional guidance may not be reliant on only the weighting of target features when categorical information is known, but also the use of relevant distractor information.

Focusing attention in working and long-term memory through dissociable mechanisms

We developed an experimental approach to compare how attentional orienting facilitates retrieval from spatial working memory (WM) and long-term memory (LTM), and how selective attention within these two memory types impacts incoming sensory information processing. In three experiments with healthy young adults, retrospective attention cues prioritize an item represented in WM or LTM. Participants then retrieve a memory item or perform a perceptual task. The retrocue is informative for the retrieval task but not for the perceptual task. We show that attentional orienting benefits performance for both WM and LTM, with stronger effects for WM. Eye-tracking reveals significant gaze shifts and microsaccades correlated with attention in WM, but no statistically significant gaze biases were found for LTM. Visual discrimination of unrelated visual stimuli is consistently improved for items matching attended WM locations. Similar effects occur at LTM locations but less consistently. The findings suggest at least partly dissociable attention-orienting processes for different memory types. Although our conclusions are necessarily constrained to the type of WM and LTM representations relevant to our task, they suggest that, under certain conditions, attentional prioritization in LTM can operate independently from WM. Future research should explore whether similar dissociations extend to non-spatial or more complex forms of LTM.

The geometry of efficient codes: How rate-distortion trade-offs distort the latent representations of generative models

Living organisms rely on internal models of the world to act adaptively. These models, because of resource limitations, cannot encode every detail and hence need to compress information. From a cognitive standpoint, information compression can manifest as a distortion of latent representations, resulting in the emergence of representations that may not accurately reflect the external world or its geometry. Rate-distortion theory formalizes the optimal way to compress information while minimizing such distortions, by considering factors such as capacity limitations, the frequency and the utility of stimuli. However, while this theory explains why the above factors distort latent representations, it does not specify which specific distortions they produce. To address this question, here we investigate how rate-distortion trade-offs shape the latent representations of images in generative models, specifically Beta Variational Autoencoders ([Formula: see text]-VAEs), under varying constraints of model capacity, data distributions, and task objectives. By systematically exploring these factors, we identify three primary distortions in latent representations: prototypization, specialization, and orthogonalization. These distortions emerge as signatures of information compression, reflecting the model's adaptation to capacity limitations, data imbalances, and task demands. Additionally, our findings demonstrate that these distortions can coexist, giving rise to a rich landscape of latent spaces, whose geometry could differ significantly across generative models subject to different constraints. Our findings contribute to explain how the normative constraints of rate-distortion theory shape the geometry of latent representations of generative models of artificial systems and living organisms.

Emotion-Guided Attention Impacts Deliberate Multi-Evidence Emotion-Related Perceptual Decision-Making

Emotion-guided endogenous attention (e.g., attending to fear) may play a crucial role in determining how humans integrate emotional evidence from various sources when assessing the general emotional tenor of the environment. For instance, what emotion a presenter focuses on can shape their perception of the overall emotion of the room. While there is an increasing interest in understanding how endogenous attention affects emotion perception, existing studies have largely focused on single-stimulus perception. There is limited understanding of how endogenous attention influences emotion evidence integration across multiple sources. To investigate this question, human participants (N = 40) were invited to judge the average emotion across an array of faces ranging from fearful to happy. Endogenous attention was manipulated by instructing participants to decide whether the face array was "fearful or not" (fear attention), "happy or not" (happy attention). Eye movement results revealed an endogenous attention-induced sampling bias such that participants paid more attention to extreme emotional evidence congruent with the target emotion. Computational modeling revealed that endogenous attention shifted the decision criterion to be more conservative, leading to reduced target-category decisions. These findings unraveled the cognitive and computational mechanisms of how endogenous attention impacts the way we gather emotional evidence and make integrative decisions, shedding light on emotion-related decision-making.

Capacity and architecture of emotional face-ensemble coding

The ability to process emotion in ensembles of faces is essential for social functioning and survival. This study investigated the efficiency and underlying architecture of this ability in two contrasting tasks: (a) extracting the mean emotion from a set of faces, and (b) visually searching for a single, redundant-target face within an ensemble. I asked whether these tasks rely on similar or distinct processing mechanisms. To address this, I applied the capacity coefficient-a rigorous measure based on the entire response time distribution. In Experiment 1, participants judged the average emotion of face ensembles. In Experiments 2 and 3, participants searched for a predefined emotional target among multiple faces. In both tasks, workload was manipulated by varying the number of faces in the display. Results revealed that ensemble averaging is a super-capacity process that improves with increased workload, while visual search is capacity-limited and impaired by greater workload. These findings suggest that averaging is a preattentive process supported by a coactive, summative architecture, whereas visual search is attention-dependent and governed by a serial or parallel architecture with inhibitory interactions between display items.

Emotional conflict affects microsaccade dynamics in the emotional face-word Stroop task

Achieving optimal performance requires effectively resolving emotional conflict arising from the interference of task-irrelevant, emotionally salient stimuli. While microsaccade behavior has been linked to various cognitive and emotional processes, whether emotional conflict affects microsaccade responses remains to be determined. Additionally, pupil dilation is known to be modulated by emotional conflict signals, and both microsaccades and pupil dilation are arguably mediated by the superior colliculus (SC). However, the relationship between microsaccades and pupil dilation remains poorly understood. In this study, we investigated the effects of emotional conflict on microsaccade rates and metrics by presenting an emotional face-word stimulus in the face-word Stroop task. Larger microsaccade amplitudes (or higher peak velocities) were observed in the incongruent condition compared to the congruent condition, while microsaccade rates were similar between the two conditions. Additionally, microsaccade amplitudes were larger in incongruent trials following congruent trials than in those following incongruent trials. Furthermore, interindividual correlations between differences in microsaccade responses and State-Trait Anxiety Inventory scores were observed. Finally, trials with lower microsaccade rates were associated with larger pupil dilation. These results demonstrate the modulation of microsaccade metrics by emotional conflict, implicating the SC in integrating signals from the locus coeruleus network to coordinate these responses.

Limited Evidence for Probabilistic Cueing Effects on Grating-Evoked Event-Related Potentials and Orientation Decoding Performance

We can rapidly learn recurring patterns that occur within our sensory environments. This knowledge allows us to form expectations about future sensory events. Several influential predictive coding models posit that, when a stimulus matches our expectations, the activity of feature-selective neurons in the visual cortex will be suppressed relative to when that stimulus is unexpected. However, after accounting for known critical confounds, there is currently scant evidence for these hypothesized effects from studies recording electrophysiological neural activity. To provide a strong test for expectation effects on stimulus-evoked responses in the visual cortex, we performed a probabilistic cueing experiment while recording electroencephalographic (EEG) data. Participants (n = 48) learned associations between visual cues and subsequently presented gratings. A given cue predicted the appearance of a certain grating orientation with 10%, 25%, 50%, 75%, or 90% validity. We did not observe any stimulus expectancy effects on grating-evoked event-related potentials. Multivariate classifiers trained to discriminate between grating orientations performed better when classifying 10% compared to 90% probability gratings. However, classification performance did not substantively differ across any other stimulus expectancy conditions. Our findings provide very limited evidence for modulations of prediction error signaling by probabilistic expectations as specified in contemporary predictive coding models.

Human V4 size predicts crowding distance

Visual recognition is limited by both object size (acuity) and spacing. The spacing limit, called "crowding", is the failure to recognize an object in the presence of other objects. Here, we take advantage of individual differences in crowding to investigate its biological basis. Crowding distance, the minimum object spacing needed for recognition, varies 2-fold among healthy adults. We test the conjecture that this variation in psychophysical crowding distance is due to variation in cortical map size. To test this, we make paired measurements of brain and behavior in 49 observers. We use psychophysics to measure crowding distance and calculate λ, the number of letters that fit into each observer's visual field without crowding. In the same observers, we use functional magnetic resonance imaging (fMRI) to measure the surface area A of retinotopic maps V1, V2, V3, and V4. Across observers, λ is proportional to the surface area of V4 but is uncorrelated with the surface area of V1 to V3. The proportional relationship of λ to area of V4 indicates conservation of cortical crowding distance across individuals: letters can be recognized if they are spaced by at least 1.4 mm on the V4 map, irrespective of map size and psychophysical crowding distance. We conclude that the size of V4 predicts the spacing limit of visual perception.

Dynamic categorization rules alter representations in human visual cortex

Everyday tasks often require stimuli to be categorized dynamically, such that an identical object can elicit different responses based on the current decision rule. Traditionally, sensory regions have been viewed as separate from such context-dependent processing, functioning primarily to process incoming inputs. However, an alternative view suggests sensory regions also integrate inputs with current task goals, facilitating more efficient information relay to higher-level areas. Here we test this by asking human participants to visually categorize novel shape stimuli based on different decision boundaries. Using fMRI and multivariate analyses of retinotopically-defined visual areas, we show that cortical shape representations become more distinct across relevant decision boundaries in a context-dependent manner, with the largest changes in discriminability observed for stimuli near the decision boundary. Importantly, these modulations are associated with improved task performance. These findings demonstrate that visual cortex representations are adaptively modulated to support dynamic behavior.

Constraints on multi-item working memory access: performance costs and retrieval dynamics

To support goal-directed behavior, working memory (WM) must flexibly access relevant information. While the mechanisms underlying single-item WM access are comparatively well-studied, less is known about the principles governing multi-item access. Some studies have suggested that dual-item retrieval can be as efficient as single-item access, but it remains unclear whether this reflects reduced inhibitory demands or truly parallel, cost-free retrieval. In Experiment 1, we manipulated the number of relevant vs. irrelevant items in a pre-and retro-cuing WM task. The rationale was that if reduced inhibitory demands benefit multi-item access, then having fewer irrelevant items to suppress would enhance performance. Instead, we found that selecting two out of three items was slower and less accurate than selecting one, arguing against the idea that diminished inhibition underlies multi-item retrieval efficiency. Experiments 2a and 2b further probed retrieval efficiency using a modified dual-access paradigm that leveraged object repetition benefits. By including a control condition to prevent temporal associations between repeated targets and non-targets, we observed that repetition benefits for each item were additive-consistent with serial or limited parallel retrieval-rather than overadditive, which would be expected under fully parallel, cost-free retrieval. These findings clarify key limitations of multi-item WM, with important implications for complex tasks such as language comprehension, decision-making, and problem solving.

Effort drives saccade selection

What determines where to move the eyes? We recently showed that pupil size, a well-established marker of effort, also reflects the effort associated with making a saccade ('saccade costs'). Here, we demonstrate saccade costs to critically drive saccade selection: when choosing between any two saccade directions, the least costly direction was consistently preferred. Strikingly, this principle even held during search in natural scenes in two additional experiments. When increasing cognitive demand experimentally through an auditory counting task, participants made fewer saccades and especially cut costly directions. This suggests that the eye-movement system and other cognitive operations consume similar resources that are flexibly allocated among each other as cognitive demand changes. Together, we argue that eye-movement behavior is tuned to adaptively minimize saccade-inherent effort.

Neural dynamics underlying the cue validity effect in target conflict resolution

Cue validity significantly influences attention guidance, either facilitating or hindering the ability for conflict resolution. Previous studies have demonstrated that the validity effect and conflict resolution are associated with better/worse behavioral performance and specific neural activations; however, the underlying neural mechanism of their interaction remains unclear. We hypothesized that the effect of cue validity might sustain specific sequences of neural activities until target occurrence and throughout the subsequent conflict resolution. In this study, we recorded the scalp electroencephalography during the Attention Network Test paradigm to investigate their interactions in neural dynamics. Specifically, we performed a cluster-level channel-time-frequency analysis to explore significant time-frequency neural activity patterns associated with these interactions, in scalp regions of interest determined by a data-driven strategy. Our results revealed a string of significant neural dynamics in the frontal and parietal regions, including initial broad-band (especially the gamma-band) activations and subsequent complex cognitive processes evoked/effected by the invalid cue, that were firstly elicited. Finally, the resolution of conflict was completed by the frontal behavior-related theta-band power reduction. In summary, our findings advanced the understanding of the temporal and spectral sequences of neural dynamics, with the key regions involved in the resolution of conflict after invalid cueing.

Allocation of spatial attention in human visual cortex as a function of endogenous cue validity

Several areas of visual cortex contain retinotopic maps of the visual field, and neuroimaging studies have shown that covert attentional guidance will result in increases of activity within the regions representing attended locations. However, little research has been done to directly compare neural activity for different types of attentional cues. Here, we used fMRI to investigate how retinotopically-specific cortical activity would be modulated depending on whether we provided deterministic or probabilistic spatial information. On each trial, a four-item memory array was presented and participants' memory for one of the items would later be probed. Critically, trials began with a foveally-presented endogenous cue that was either 100% valid (deterministic runs), 70% valid (probabilistic runs), or neutral. By dividing visual cortex into quadrant-specific regions of interest (qROIs), we could examine how attention was spatially distributed across the visual field within each trial, depending on cue type and delay. During the anticipatory period prior to the memory array, we found increased activation at the cued location compared to noncued locations, with surprisingly comparable levels of facilitation for both deterministic and probabilistic cues. However, we found significantly greater facilitation on deterministic relative to probabilistic trials following the onset of the memory array, with only deterministic cue-related facilitation persisting through the presentation of the probe. These findings reveal how cue validity can drive differential allocations of neural resources over time across cued and noncued locations, and that the allocation of attention should not be assumed to invariably scale alongside the validity of a cue.

The Role of Fronto-Central Theta Oscillations in Inter-Sensory Selective Attention

Selective attention supports top-down control by biasing information processing toward stimuli that are potentially relevant to the immediate goal. It has been recently proposed that theta band oscillations (~4-8 Hz) in the frontal midline regions are a key mechanism of endogenous selective attention. The current electroencephalography study investigated theta oscillatory dynamics using an inter-sensory cueing paradigm in which a symbolic cue indicated, on a trial-by-trial basis, the modality (visual or auditory) of the upcoming discrimination task. Time-frequency analyses were used to quantify phase- (evoked) and non-phase-locked (induced) fronto-central theta activity during preparatory attentional states. In a sample of 20 young adult participants, we found that those who relied on the cues to selectively attend to the sensory modality of the discrimination task performed more efficiently (i.e., faster and with greater accuracy) and presented greater non-phase-locked fronto-central theta power 200-400 ms post-cue onset. Moreover, greater non-phase-locked theta oscillations were associated with better behavioral performance. Secondary analyses on alpha oscillations revealed concomitant brain activity to theta with a pronounced decrease in alpha power in fronto-central regions, without significant effect on task performance. These findings suggest that increased non-phase-locked fronto-central theta oscillations are a neuronal correlate of preparatory attentional control and that the interplay of theta-alpha rhythms differentially contributes to attentional and perceptual aspects.

The fate of visual working memory items after their job is done

Visual working memory is a competitive, capacity-limited system for the storage of feature and object-based information. In change-detection tasks, items are encoded into memory and, after a retention period, are compared against a test set. Loss of information can occur from attentional interference or prioritizing some items over others. But what happens to the memory representations after the change-detection task is completed? The current article examines the fate of a memory item after its behavioral purpose has been fulfilled. Participants encoded a single item in memory for a difficult change-detection task. Visual search trials were presented both before and after the memory test was completed. Singleton distractors were present in these search trials that could match or not the memory item. In Experiment 1, memory-driven capture (the memory-matching distractors led to longer search response times than the unrelated distractor) was observed in the pre-memory test and, in a weaker form, the post-test search trials. In Experiment 2, we introduced cues that indicated the memory test would not occur on a subset of trials, controlling for re-exposure to the memory stimulus. Memory-driven capture was again observed for these post-cue search trials, but only at a short time interval, at a longer interval this effect was attenuated. These results suggest that the memory representations only linger briefly in the visual system.

Inhibition in large set sizes depends on search mode, not salience

Attention can be attracted to salient items in a visual scene. Recent studies have shown that when the feature of an irrelevant salient item is known, it can be suppressed below baseline leading to facilitated search. Wang and Theeuwes (Experimental Psychology: Human Perception and Performance, 46(10), 1051-1057, 2020) criticised previous inhibition studies by claiming that the sparse displays attenuated the salience of the distractors. In their study they increased the number of display items (i.e., set size), and found that an irrelevant salient distractor captured attention. The current paper argues that the displays used by Wang and Theeuwes encouraged participants to use a singleton search mode, in which participants actively look for salient regions to find the target and consequently do not inhibit salient items. Specifically, their displays included multiple repeated non-target shapes, so that the target became a singleton. We used two search displays with ten items, one with repeated non-targets (R-NT displays), allowing a singleton search mode, and one with heterogeneous non-targets, encouraging a feature search mode. In Experiment 1 the singleton distractor was inhibited in the heterogeneous condition, but not in the R-NT condition. Experiment 2 intermixed the two display types in unbalanced blocks. When the majority of trials had heterogeneous non-targets, inhibition was observed for both the heterogeneous displays and the R-NT displays. Conversely, when R-NT displays were the majority, inhibition was attenuated for both display types. These results show that distractor features can be suppressed at large set sizes dependant on the search strategy promoted by the displays.

Recognition memory fluctuates with sustained attention regardless of task relevance

Sustained attention fluctuates over time, affecting task-related processing and memory. However, it is less clear how attentional state affects processing and memory when images are accompanied by irrelevant visual information. We first quantify behavioral signatures of attentional state in an online sample (N1=92) and demonstrate that images presented in high attentional states are better remembered. Next, we test how sustained attention influences memory in two online samples (N2=188, N3=185) when task-irrelevant images are present. We show that high attention leads to better memory for both task-relevant and task-irrelevant images. This suggests that sustained attentional state does selectively affect processing for task-relevant information, but rather affects processing broadly, regardless of task relevance. Finally, we show that other components of attention such as selective attention contribute to the mnemonic fate of stimuli. Our findings highlight the necessity of considering and characterizing attention's unique components and their effects on cognition.

The role of alpha oscillations in resisting distraction

The role of alpha oscillations (8-13 Hz) in suppressing distractors is extensively debated. One debate concerns whether alpha oscillations suppress anticipated visual distractors through increased power. Whereas some studies suggest that alpha oscillations support distractor suppression, others do not. We identify methodological differences that may explain these discrepancies. A second debate concerns the mechanistic role of alpha oscillations. We and others previously proposed that alpha oscillations implement gain reduction in early visual regions when target load or distractor interference is high. Here, we suggest that parietal alpha oscillations support gating or stabilization of attentional focus and that alpha oscillations in ventral attention network (VAN) support resistance to attention capture. We outline future studies needed to uncover the precise mechanistic role of alpha oscillations.

The role of cerebellar-cortical connectivity in modulating attentional abilities: insight from football athletes

Neuroplasticity, a phenomenon present throughout the lifespan, is thought to be influenced by physical training. However, the relationship between neuroplastic differences and attentional abilities remains unclear. This study explored the differences in brain function and attentional abilities between professional football athletes and novices, and further investigated the relationship between the two. To address this question, we included 49 football athletes and 63 novices in our study, collecting data on resting-state functional connectivity and Attention Network Test (ANT). Behavioral results from the ANT indicated that football experts had superior orienting attention but weaker alerting functions compared to novices, with no difference in executive control attention. fMRI results revealed that football experts exhibited higher fractional Amplitude of Low-Frequency Fluctuations (fALFF) values in the bilateral anterior cerebellar lobes, bilateral insula, and left superior temporal gyrus. Functional connectivity analysis showed increased connectivity between the left anterior cerebellar lobe and various cortical regions, including the right supramarginal gyrus, left precuneus, left superior frontal gyrus, bilateral posterior cerebellar lobes, and bilateral precentral gyri in experts compared to novices. More importantly, in the expert group but not in novice group, functional connectivity differences significantly predicted attentional orienting scores. Graph theoretical analysis showed that experts exhibited higher betweenness centrality and node efficiency in the right cerebellar lobule III (Cerebelum_3_R) node. Our findings demonstrate that long-term professional football training may significantly affect neuroplasticity and attentional functions. Importantly, our analysis reveals a substantive connection between these two aspects, suggesting that the integration of neuroplastic and attentional changes is likely mediated by cerebellar-cortical connectivity.

The effect of expectancy on conditioned pain modulation: evidence from functional near-infrared spectroscopy

Background and objective

The psychological mechanisms that make Conditioned Pain Modulation (CPM) an effective non-pharmacological intervention are still not fully understood. Expectancy is believed to be a critical psychological factor affecting CPM effects, but its specific role has yet to be fully clarified. This study aims to explore the relationship between expectancy and CPM while providing physiological evidence using functional near-infrared spectroscopy (fNIRS).

Method

A standardized CPM induction paradigm was implemented, with verbal guidance used to induce expectancy. The Numeric Rating Scale (NRS) assessed the intensity of the test stimulus (TS), while an 11-point scale evaluated participants' attentional focus on the TS and the effect of expectancy. fNIRS was employed to monitor changes in prefrontal cortex (PFC) activity.

Results

Expectancy significantly amplified the CPM effect (p = 0.036) while markedly reducing attention to the experimental stimulus (p = 0.004). fNIRS findings indicated significant reductions in activity within the left frontal eye field, left dorsolateral prefrontal cortex, and left frontal pole regions. In the post-test, the control group demonstrated significantly higher cortical activity in the right frontal pole region compared to the expectancy group (p < 0.05). Within the expectancy group, bilateral frontal pole cortical activity was significantly lower in the post-test compared to the pre-test (p < 0.05).

Conclusion

Expectancy represents a key psychological mechanism underlying the CPM effect, potentially modulating its magnitude through attention regulation and accompanied by a reduction in oxygenated hemoglobin activity in the frontal pole region and introduced the Expectancy-Attention-CPM Modulation Model (ECAM).

Altered resting-state network connectivity in internet gaming disorder

BACKGROUND

The growing popularity of internet gaming among adolescents and young adults has driven an increase in both casual and excessive gaming behavior. Nevertheless, it remains unclear how progressive increases in internet gaming engagement led to changes within and between brain networks. This study aims to investigate these connectivity alterations across varying levels of gaming involvement.

METHODS

In this cross-sectional study, 231 participants were recruited and classified into three groups according to Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria for Internet Gaming Disorder (IGD): IGD group, highly engaged gaming(HEG) group, and lowly engaged gaming (LEG) group. Resting-state fMRI data from 217 participants (143 males, 74 females) were included in the final analysis. Independent component analysis was used to examine differences in intra- and inter-network functional connectivity (FC)across the three groups.

RESULTS

No significant differences were found in intra-network FC across the three groups. However, significant inter-network differences between the dorsal attention network(dAN)and the visual network (VN) among the three groups were observed. The HEG group exhibited significantly higher dAN-VN functional network connectivity (FNC) compared to the LEG group. Linear correlation analyses showed no significant correlation between the dAN-VN FNC values and IGD-20T scores.

CONCLUSION

Throughout the development of IGD, increasing levels of engagement are associated with a rise and subsequent decline in FNC of DAN-VN. This pattern may reflect top-down attentional regulation in the early stages of addiction, followed by attentional bias as addiction progresses.

The attentional boost effect: current landscape and future directions

Cognitive functions such as attention and memory significantly impact performance in daily life and in various professions, including driving vehicles and providing healthcare services. Driven by the importance of understanding attention, early studies have explored the attentional theories and discovered the attentional boost effect (ABE). In experiments studying the ABE, participants are required to engage in two concurrent tasks: (1) memorising a sequence of briefly displayed stimuli (e.g. images or words) for a later memory test and (2) concurrently detecting a simultaneously presented target signal (e.g. pressing a button when seeing a target white square and taking no action for a distractor black square). Surprisingly, attending to a target boosts memory encoding for the concurrently presented information, contrary to the typical expectation of lowered performance owing to dual-task interference. This effect has been documented not only in behavioural experiments across different materials and modalities but also in neuroimaging investigations. This review paper is divided into several main sections, covering the behavioural evidence supporting the ABE, interpretations of the effect from neuroimaging studies, individual differences, consensus and controversies in ABE research as well as prospective future research in this area. The discussion in this review might also offer helpful insights to researchers for translating this phenomenon into real-world practical applications.

Understanding speech in "noise" or free energy minimization in the soundscapes of the anthropocene

Listening to speech in the presence of irrelevant sounds is ubiquitous in the modern world, but is generally acknowledged to be both effortful and unpleasant. Here we argue that this problem arises largely in circumstances that our human auditory system has not evolved to accommodate. The soundscapes of the Anthropocene are frequently characterized by an overabundance of sound sources, the vast majority of which are functionally irrelevant to a given listener. The problem of listening to speech in such environments must be solved by an auditory system that is not optimized for this task. Building on our previous work linking attention to effortful listening and incorporating an active inference approach, we argue that the answers to these questions have implications not just for the study of human audition. They are also significant for the development and broad awareness of hearing aids and cochlear implants, as well as other auditory technologies such as earbuds, immersive auditory environments, and systems for human-machine interaction.

Multiple item representations in visual working memory simultaneously guide attention

Visual working memory (VWM) is a subject of ongoing debate regarding whether multiple item representations can simultaneously guide attention. The Single Item Template hypothesis (SIT) posits that VWM representations only allow a single item to guide attention, while the Multiple Item Template hypothesis (MIT) suggests that multiple items in VWM representations can guide attention simultaneously. This study further investigates this through a dual-task paradigm. Participants were required to complete memory-search tasks under different memory and match types, with memory items and distractors being of the same category (Experiment 1) and different categories (Experiment 2). Results show: (1) In Experiment 1, when the memory type was color, multiple item representations in VWM simultaneously guided attention, providing support for the MIT hypothesis; however, when the memory type was graphic, multiple item representations in VWM did not simultaneously guide attention, providing support for the SIT hypothesis. (2) In Experiment 2, whether the memory type was color or graphic, multiple item representations in VWM did not simultaneously guide attention, providing support for the SIT hypothesis. Whether multiple item representations in VWM can simultaneously guide attention is significantly influenced by memory types and the relationship between memory items and categories of distractors.

Coarse matching was sufficient to capture attention by working memory representations unless matching features with the target

BACKGROUND

In most theoretical frameworks, the effectiveness of attentional selection relies significantly on the perceptual similarity between the target template and visual input. Nevertheless, ambiguity exists surrounding whether attentional capture triggered by irrelevant representations in Working Memory (WM) is influenced by the perceptual similarity levels of features between WM content and its matching distractors.

METHODS

We designed a hybrid WM and visual search task, varying such perceptual similarity of colors across three levels: exact, high-similar, and low-similar matching. To quantify the extent of the capture effect, we compared these conditions against a neutral baseline (i.e., completely different color) using eye movement and behavioral data in two experiments.

RESULTS

We consistently observed robust attentional capture effects across two experiments, evident in both eye movement indices and manual reaction times. In Experiment 1, where WM representations solely matched features to visual search distractors (task-irrelevant scenario), we found that changes in perceptual similarity did not influence attentional capture. Conversely, in Experiment 2, where WM representations had the potential to match the visual search target (task-relevant scenario), we observed a significantly more robust attentional capture effect for high-similar matching compared to low-similar matching conditions.

CONCLUSIONS

These findings imply that coarse matching between distractors and WM contents is sufficient to capture attention, unless the matching features potentially correspond to the visual target. Furthermore, task relevance sharpens perceptual sensitivity to visual input, highlighting distinct mechanisms underlying attentional capture by irrelevant representations and target templates within WM.

EEG evidence for spatial selectivity in feature-based preparation for visual search

In many visual search tasks, the detection of target objects in visual search requires feature-selective attentional guidance and space-based attentional selection. Feature-based attention is often assumed to operate in a spatially global fashion across the entire visual field, but there is also evidence that it can be restricted to task-relevant locations under some conditions. Here, we investigated whether such spatial filtering processes are already evident when representations of target-defining features (attentional templates) are activated during the preparation for an upcoming search episode. We measured N2pc components (an electrophysiological index of attentional allocation) in response to a rapid series of lateral task-irrelevant but template-matching colour probes that appeared while participants prepared for an upcoming search task with colour-defined targets. Critically, search targets would either always appear in the same lateral regions of visual space as the probes, or at different locations (near fixation or in lateral areas that never contained probes), thus rendering the probed locations either task-relevant or irrelevant. N2pc components triggered by target-colour probes during the preparation period emerged later and were attenuated when probes were presented at irrelevant locations. This demonstrates that the effects of preparatory feature-based attentional templates can be modulated by spatial expectations. However, this type of spatial filtering during search preparation only attenuates but not completely eliminates feature-based attentional modulations.

Temporal integration as an adaptive process in visual perception, attention, and working memory

I propose that temporal integration is ubiquitous in visual perception, because it serves an adaptive role. To support this idea, I draw together evidence from historically separated research fields that target different timescales. At one extreme, this concerns the detection and discrimination of successive stimuli within intervals of less than a quarter of a second. At an intermediate level, associated with attentional episodes, intervals between half a second up to a few seconds are considered. Finally, at the other extreme, this involves high-level, conceptual events across intervals of multiple seconds or even minutes. Across such varying intervals, the nature of temporal integration and the resultant perceptual events are clearly different. I nevertheless propose that temporal integration should be understood as a continuous process that serves a common adaptive goal: To maximize the amount of useful information, at minimal costs, tailored to the observer's current needs and circumstances. Emerging from this viewpoint are several research directions that might be pursued on the topic of temporal integration, and on its consequences for perception and memory.

Physically activated modes of attentional control

As we navigate through the day, our attentional control processes are constantly challenged by changing sensory information, goals, expectations, and motivations. At the same time, our bodies and brains are impacted by changes in global physiological state that can influence attentional processes. Based on converging lines of evidence from brain recordings in physically active humans and nonhumans, we propose a new framework incorporating at least two physically activated modes of attentional control in humans: altered gain control and differential neuromodulation of control networks. We discuss the implications of this framework for understanding a broader range of states and cognitive functions studied both in the laboratory and in the wild.

Aversive conditioning, anxiety, and the strategic control of attention

What we pay attention to is influenced by both reward learning and aversive conditioning. Although early attention tends to be biased toward aversively conditioned stimuli, sustained ignoring of such stimuli is also possible. How aversive conditioning influences how a person chooses to search, or the strategic control of attention, has not been explored. In the present study, participants learned an association between a colour and an aversive outcome during a training phase, and in a subsequent test phase searched for one of two targets presented on each trial; one target was rendered in the aversively conditioned colour (CS+) and the other in a neutral colour (CS-). Given the distribution of colour stimuli in the search array, it was more optimal to search for and report a target in one of the two colours on some trials. Our results demonstrate that participants were biased away from the CS+ target, which resulted in non-optimal search on some trials. Surprisingly, rather than accentuate this bias, greater state anxiety was associated with a stronger tendency to find and report the CS+ target. Our findings have implications for our understanding of the learning-dependent control of attention and abnormal attentional biases observed in high-anxious individuals.

Neural mechanisms for the attention-mediated propagation of conceptual information in the human brain

The visual environment is complicated, and humans and other animals accordingly prioritize some sources of information over others through the deployment of spatial attention. Cognitive theories propose that one core purpose of this is to gather information that can be used in downstream cognitive processes, including the development of concepts and categories. However, neuroscientific investigation has focused closely on the identification of the systems and algorithms that support attentional control or that instantiate the effect of attention on sensation and perception. Much less is known about how attention impacts the acquisition and activation of concepts. Here, we use machine learning of EEG and concurrently recorded EEG/MRI to temporally and anatomically characterize the neural network that abstracts from attended perceptual information to activate and construct semantic and conceptual representations. We find that variance in the amplitude of N2pc-an event-related potential (ERP) component closely linked to selective attention-predicts the emergence of conceptual information in a network including prefrontal, posterior parietal, and anterior insular cortex. This network appears to play a key role in the attention-mediated translation of perceptual information to concepts, semantics, and action plans.