Modulation of spatial attention by goals, statistical learning, and monetary reward

Graded prioritisation of targets in search: reward diminishes the low prevalence effect

In many real-life contexts, observers are required to search for targets that are rarely present (e.g. tumours in X-rays; dangerous items in airport security screenings). Despite the rarity of these items, they are of enormous importance for the health and safety of the public, yet they are easily missed during visual search. This is referred to as the prevalence effect. In the current series of experiments, we investigate whether unequal reward can modulate the prevalence effect, in a multiple target search task. Having first established the impact of prevalence (Experiment 1) and reward (Experiment 2) on how efficiently participants can find one of several targets in the current paradigm, we then combined the two forms of priority to investigate their interaction. An unequal reward distribution (where lower prevalence items are more rewarded; Experiment 3) was found to diminish the effect of prevalence, compared to an equal reward distribution (Experiment 4) as indicated by faster response times and fewer misses. These findings suggest that when combined with an unequal reward distribution, the low prevalence effect can be diminished.

Target voice probability influences enhancement in auditory selective attention

Auditory selective attention is thought to consist of two mechanisms: an enhancement mechanism that boosts the target signal, and a suppression mechanism that attenuates concurrent distracting signals. The current study explored the conditions necessary to observe enhancement of predictable auditory objects. Participants heard scenes consisting of three voices and a distracting noise. They were asked to find the gender singleton (target) and report whether it was saying even or odd numbers. One of the voices appeared as the high-probability target (70%) across trials. We expected responses to be faster when the high-probability target was in the scene, and results from Experiment 1 supported that prediction. However, this target enhancement effect was substantially weakened when a distracting noise was also in the scene, suggesting that the distractor captured attention and interfered with enhancement. Experiment 2 tested the hypothesis that distractor predictability modulates target enhancement by varying the probability of the distractor. Although this hypothesis was not supported, the results of Experiment 1 were replicated. Findings support the existence of an easily disruptable enhancement mechanism that boosts the representation of highly probable target objects.

Statistical learning of target location and distractor location rely on different mechanisms during visual search

More studies have demonstrated that people have the capacity to learn and make use of environmental regularities. This capacity is known as statistical learning (SL). Despite rich empirical findings, it is not clear how the two forms of SL (SL of target location and SL of distractor location) influence visual search and whether they rely on the shared cognitive mechanism. In Experiment 1 and Experiment 2, we manipulated the probability of target location and the probability of distractor location, respectively. The results suggest that attentional guidance (they referred to overt attention) may mainly contribute to the SL effect of the target location and the distractor location, which is in line with the notion of priority mapping. To a small extent, facilitation of response selection may also contribute to the SL effect of the target location but does not contribute to the SL effect of the distractor location. However, the main difference between the two kinds of SL occurred in the early stage (it involved covert attention). Together, our findings indicate that the two forms of SL reflect partly shared and partly independent cognitive mechanisms.

Top-down knowledge surpasses selection history in influencing attentional guidance

Visual attention is influenced by the characteristics of the stimuli (bottom-up), their task relevance (top-down), and prior experience (e.g., selection history and learning). However, it is largely unclear how learning and selection history interact with top-down attentional guidance. We combined trial-and-error learning with a spatial cueing protocol to test whether previously learned target-defining features continued to capture attention if participants were instructed to search for a new target feature (Experiment 1) or had to learn a new target feature (Experiment 2). It turned out that the previously learned feature quickly stopped capturing attention when the target feature changed (Experiment 1; even before participants learned the new target-defining feature, in Experiment 2). Finally, in Experiment 3, in which participants learned to search for targets defined by two redundant features (color and orientation), we found possible reasons for the dominance of the instructed feature over learning. Participants reported using only the target color for their search. Consequently, only cues with a target color captured attention. The unused target orientation only captured attention in participants aware of both target-defining features (13 out of 23) and only if the orientation was presented in the target color. We conclude that knowledge of target-defining features and their use as search criterion is critical for attentional guidance, while previously learned target features either influence attentional guidance only contingent on such deliberately selected top-down based attentional control settings or may influence visual search but not attentional guidance.

Learning to suppress a distractor may not be unconscious

The additional singleton task has become a popular paradigm to explore visual statistical learning and selective attention. In this task, participants are instructed to find a different-shaped target among a series of distractors as fast as possible. In some trials, the search display includes a singleton distractor with a different color, making search more difficult. This singleton distractor appears more often in one location than in the remaining locations. The typical results of these experiments show that participants learn to ignore the area of the screen that is more likely to contain the singleton distractor. It is often claimed that this learning takes place unconsciously, because at the end of the experiment participants seem to be unable to identify the location where the singleton distractor appeared most frequently during the task. In the present study, we tested participants’ awareness in three high-powered experiments using alternative measures. Contrary to previous studies, the results show clear evidence of explicit knowledge about which area of the display was more likely to contain the singleton distractor, suggesting that this type of learning might not be unconscious.

Is probabilistic cuing of visual search an inflexible attentional habit? A meta-analytic review

In studies on probabilistic cuing of visual search, participants search for a target among several distractors and report some feature of the target. In a biased stage the target appears more frequently in one specific area of the search display. Eventually, participants become faster at finding the target in that rich region compared to the sparse region. In some experiments, this stage is followed by an unbiased stage, where the target is evenly located across all regions of the display. Despite this change in the spatial distribution of targets, search speed usually remains faster when the target is located in the previously rich region. The persistence of the bias even when it is no longer advantageous has been taken as evidence that this phenomenon is an attentional habit. The aim of this meta-analysis was to test whether the magnitude of probabilistic cuing decreases from the biased to the unbiased stage. A meta-analysis of 42 studies confirmed that probabilistic cuing during the unbiased stage was roughly half the size of cuing during the biased stage, and this decrease persisted even after correcting for publication bias. Thus, the evidence supporting the claim that probabilistic cuing is an attentional habit might not be as compelling as previously thought.

The past, present, and future of selection history

The last ten years of attention research have witnessed a revolution, replacing a theoretical dichotomy (top-down vs. bottom-up control) with a trichotomy (biased by current goals, physical salience, and selection history). This third new mechanism of attentional control, selection history, is multifaceted. Some aspects of selection history must be learned over time whereas others reflect much more transient influences. A variety of different learning experiences can shape the attention system, including reward, aversive outcomes, past experience searching for a target, target‒non-target relations, and more. In this review, we provide an overview of the historical forces that led to the proposal of selection history as a distinct mechanism of attentional control. We then propose a formal definition of selection history, with concrete criteria, and identify different components of experience-driven attention that fit within this definition. The bulk of the review is devoted to exploring how these different components relate to one another. We conclude by proposing an integrative account of selection history centered on underlying themes that emerge from our review.

Gotcha: Working memory prioritization from automatic attentional biases

Attention is an important resource for prioritizing information in working memory (WM), and it can be deployed both strategically and automatically. Most research investigating the relationship between WM and attention has focused on strategic efforts to deploy attentional resources toward remembering relevant information. However, such voluntary attentional control represents a mere subset of the attentional processes that select information to be encoded and maintained in WM (Theeuwes, Journal of Cognition, 1[1]: 29, 1-15, 2018). Here, we discuss three ways in which information becomes prioritized automatically in WM-physical salience, statistical learning, and reward learning. This review integrates findings from perception and working memory studies to propose a more sophisticated understanding of the relationship between attention and working memory.

Awareness is necessary for attentional biases by location-reward association

Many studies have reported attentional biases based on feature-reward associations. However, the effects of location-reward associations on attentional selection remain less well-understood. Unlike feature cases, a previous study that induced participants' awareness of the location-reward association by instructing them to look for a high-reward location has suggested the critical role of goal-driven manipulations in such associations. In this study, we investigated whether the reward effect occurred without goal-driven manipulations if participants were spontaneously aware of the location-reward association. We conducted three experiments using a visual search task that included four circles where participants received rewards; one possible target location was associated with a high reward, and another with a low reward. In Experiment 1, the target was presented among distractors, and participants had to search for the target. The results showed a faster reaction time in the high-reward rather than the low-reward locations only in participants aware of the location-reward association, even if they were not required to look for the association. Moreover, in Experiment 2, we replicated the main findings of Experiment 1, even when the target had an abrupt visual onset to restrict goal-driven manipulations. Furthermore, Experiment 3 confirmed that the effect observed in Experiment 2 could not be attributed to the initial eye position. These findings suggest that goal-driven manipulations are unnecessary for inducing reward biases to high-reward locations. We concluded that awareness of the association rather than goal-driven manipulations is crucial for the location-reward effect.

Statistical regularities cause attentional suppression with target-matching distractors

Visual search may be disrupted by the presentation of salient, but irrelevant stimuli. To reduce the impact of salient distractors, attention may suppress their processing below baseline level. While there are many studies on the attentional suppression of distractors with features distinct from the target (e.g., a color distractor with a shape target), there is little and inconsistent evidence for attentional suppression with distractors sharing the target feature. In this study, distractor and target were temporally separated in a cue–target paradigm, where the cue was shown briefly before the target display. With target-matching cues, RTs were shorter when the cue appeared at the target location (valid cues) compared with when it appeared at a nontarget location (invalid cues). To induce attentional suppression, we presented the cue more frequently at one out of four possible target positions. We found that invalid cues appearing at the high-frequency cue position produced less interference than invalid cues appearing at a low-frequency cue position. Crucially, target processing was also impaired at the high-frequency cue position, providing strong evidence for attentional suppression of the cued location. Overall, attentional suppression of the frequent distractor location could be established through feature-based attention, suggesting that feature-based attention may guide attentional suppression just as it guides attentional enhancement.

The priority of goal-relevant information and evolutionarily threatening information in early attention processing:Evidence from behavioral and ERP study

Previous studies have demonstrated that evolutionarily threatening information and goal-relevant information can both capture attention. However, some studies have suggested that goal-relevant information is prioritized over evolutionarily threatening information, while some studies have shown the opposite conclusion. The aim of the present study was to investigate the attention advantage by presenting evolutionarily threatening information and goal-relevant information simultaneously. Three conditions were presented in this study: evolutionarily threatening information + an irrelevant stimulus, goal-relevant information + an irrelevant stimulus, and evolutionarily threatening information + goal-relevant information. The behavioral results showed no attentional bias toward evolutionarily threatening information in the two conditions including evolutionarily threatening information; in the two conditions including goal-relevant information, participants showed attentional bias toward goal-relevant information in both. However, the ERP results showed that in the two conditions including evolutionarily threatening information, a significantly stronger N2pc response was seen for evolutionarily threatening information than for the other types of pictures, and goal-relevant information produced a significantly stronger N2pc response than that for an irrelevant stimulus. The abovementioned results indicated that in the earlier stage of attention, both evolutionarily threatening information and goal-relevant information have attention processing advantages over irrelevant stimuli; furthermore, attention was captured by evolutionarily threatening information faster than it was by goal-relevant information.

Specificity and persistence of statistical learning in distractor suppression

Statistical regularities in distractor location trigger suppression of high-probability distractor locations during visual search. The degree to which such suppression reflects generalizable, persistent changes in a spatial priority map has not been examined. We demonstrate that suppression of high-probability distractor locations persists after location probabilities are equalized and likely reflects a genuine reshaping of the priority map rather than more transient effects of selection history. Statistically learned suppression generalizes across contexts within a task during learning but does not generalize between task paradigms using unrelated stimuli in identical spatial locations. These findings suggest that stimulus features do play a role in learned spatial suppression, potentially gating the weights applied to a spatial priority map. However, the binding of location to context during learning is not automatic, in contrast to the previously reported interaction of location-based statistical learning and stimulus features. (PsycINFO Database Record (c) 2020 APA, all rights reserved).

Attentional Orienting by Non-informative Cue Is Shaped via Reinforcement Learning

It has been demonstrated that a reward-associated stimulus feature captures attention involuntarily. The present study tested whether spatial attentional orienting is biased via reinforcement learning. Participants were to identify a target stimulus presented in one of two placeholders, preceded by a non-informative arrow cue at the center of the display. Importantly, reward was available when the target occurred at a location cued by a reward cue, defined as a specific color (experiments 1 and 3) or a color-direction combination (experiment 2). The attentional bias of the reward cue was significantly increased as trials progressed, resulting in a greater cue-validity effect for the reward cue than the no-reward cue. This attentional bias was still evident even when controlling for the possibility that the incentive salience of the reward cue color modulates the cue-validity effect (experiment 2) or when the reward was withdrawn after reinforcement learning (experiment 3). However, it disappeared when the reward was provided regardless of cue validity (experiment 4), implying that the reinforcement contingency between reward and attentional orienting is a critical determinant of reinforcement learning-based spatial attentional modulation. Our findings highlight that a spatial attentional bias is shaped by value via reinforcement learning.

Reward learning biases the direction of saccades

The role of associative reward learning in guiding feature-based attention and spatial attention is well established. However, no studies have looked at the extent to which reward learning can modulate the direction of saccades during visual search. Here, we introduced a novel reward learning paradigm to examine whether reward-associated directions of eye movements can modulate performance in different visual search tasks. Participants had to fixate a peripheral target before fixating one of four disks that subsequently appeared in each cardinal position. This was followed by reward feedback contingent upon the direction chosen, where one direction consistently yielded a high reward. Thus, reward was tied to the direction of saccades rather than the absolute location of the stimulus fixated. Participants selected the target in the high-value direction on the majority of trials, demonstrating robust learning of the task contingencies. In an untimed visual foraging task that followed, which was performed in extinction, initial saccades were reliably biased in the previously rewarded-associated direction. In a second experiment, following the same training procedure, eye movements in the previously high-value direction were facilitated in a saccade-to-target task. Our findings suggest that rewarding directional eye movements biases oculomotor search patterns in a manner that is robust to extinction and generalizes across stimuli and task.

Experience-Driven Auditory Attention

In addition to conscious goals and stimulus salience, an observer's prior experience also influences selective attention. Early studies demonstrated experience-driven effects on attention mainly in the visual modality, but increasing evidence shows that experience drives auditory selection as well. We review evidence for a multiple-levels framework of auditory attention, in which experience-driven attention relies on mechanisms that acquire control settings and mechanisms that guide attention towards selected stimuli. Mechanisms of acquisition include cue-target associative learning, reward learning, and sensitivity to prior selection history. Once acquired, implementation of these biases can occur either consciously or unconsciously. Future research should more fully characterize the sources of experience-driven auditory attention and investigate the neural mechanisms used to acquire and implement experience-driven auditory attention.

Statistical learning of anomalous regions in complex faux X-ray images does not transfer between detection and discrimination

The visual environment contains predictable information - "statistical regularities" - that can be used to aid perception and attentional allocation. Here we investigate the role of statistical learning in facilitating search tasks that resemble medical-image perception. Using faux X-ray images, we employed two tasks that mimicked two problems in medical-image perception: detecting a target signal that is poorly segmented from the background; and discriminating a candidate anomaly from benign signals. In the first, participants searched a heavily camouflaged target embedded in cloud-like noise. In the second, the noise opacity was reduced, but the target appeared among visually similar distractors. We tested the hypothesis that learning may be task-specific. To this end, we introduced statistical regularities by presenting the target disproportionately more frequently in one region of the space. This manipulation successfully induced incidental learning of the target's location probability, producing faster search when the target appeared in the high-probability region. The learned attentional preference persisted through a testing phase in which the target's location was random. Supporting the task-specificity hypothesis, when the task changed between training and testing, the learned priority did not transfer. Eye tracking showed fewer, but longer, fixations in the detection than in the discrimination task. The observation of task-specificity of statistical learning has implications for theories of spatial attention and sheds light on the design of effective training tasks.

Neurobiology of value-driven attention

What we pay attention to is influenced by reward learning. Converging evidence points to the idea that associative reward learning changes how visual stimuli are processed in the brain, rendering learned reward cues difficult to ignore. Behavioral evidence distinguishes value-driven attention from other established control mechanisms, suggesting a distinct underlying neurobiological process. Recently, studies have begun to explore the neural substrates of this value-driven attention mechanism. Here, I review the progress that has been made in this area, and synthesize the findings to provide an integrative account of the neurobiology of value-driven attention. The proposed account can explain both attentional capture by previously rewarded targets and the modulatory effect of reward on priming, as well as the decoupling of reward history and prior task relevance in value-driven attention.

On the representational nature of value-driven spatial attentional biases

Reward learning biases attention toward both reward-associated objects and reward-associated regions of space. The relationship between objects and space in the value-based control of attention, as well as the contextual specificity of space-reward pairings, remains unclear. In the present study, using a free-viewing task, we provide evidence of overt attentional biases toward previously rewarded regions of texture scenes that lack objects. When scrutinizing a texture scene, participants look more frequently toward, and spend a longer amount of time looking at, regions that they have repeatedly oriented to in the past as a result of performance feedback. These biases were scene specific, such that different spatial contexts produced different patterns of habitual spatial orienting. Our findings indicate that reinforcement learning can modify looking behavior via a representation that is purely spatial in nature in a context-specific manner. NEW & NOTEWORTHY The representational nature of space in the value-driven control of attention remains unclear. Here, we provide evidence for scene-specific overt spatial attentional biases following reinforcement learning, even though the scenes contained no objects. Our findings indicate that reinforcement learning can modify looking behavior via a representation that is purely spatial in nature in a context-specific manner.

The Risks of Downplaying Top-Down Control

Is top-down control necessarily scarce, slow, and hence unimportant in visual selection? Here we outline the risks of downplaying top-down control. Contrary to Theeuwes' review, we suggest that not all sources of attention map onto a unitary attentional priority map. Goals and search habits may influence where and how people deploy attention, respectively. Because goals have modulatory effects on sensory processing, their impact on attention is broad and not always deliberate. In addition, when multiple sources influence attention, top-down control often dominates over less deliberate forms of attention. We agree with Theeuwes that selection history can drive attention independent of explicit goals. Nonetheless, top-down control remains a cornerstone of visual selection.

Habitual versus goal-driven attention

Recent research has expanded the list of factors that control spatial attention. Beside current goals and perceptual salience, statistical learning, reward, motivation and emotion also affect attention. But do these various factors influence spatial attention in the same manner, as suggested by the integrated framework of attention, or do they target different aspects of spatial attention? Here I present evidence that the control of attention may be implemented in two ways. Whereas current goals typically modulate where in space attention is prioritized, search habits affect how one moves attention in space. Using the location probability learning paradigm, I show that a search habit forms when people frequently find a visual search target in one region of space. Attentional cuing by probability learning differs from that by current goals. Probability cuing is implicit and persists long after the probability cue is no longer valid. Whereas explicit goal-driven attention codes space in an environment-centered reference frame, probability cuing is viewer-centered and is insensitive to secondary working memory load and aging. I propose a multi-level framework that separates the source of attentional control from its implementation. Similar to the integrated framework, the multi-level framework considers current goals, perceptual salience, and selection history as major sources of attentional control. However, these factors are implemented in two ways, controlling where spatial attention is allocated and how one shifts attention in space.

Explicit goal-driven attention, unlike implicitly learned attention, spreads to secondary tasks

To what degree does spatial attention for one task spread to all stimuli in the attended region, regardless of task relevance? Most models imply that spatial attention acts through a unitary priority map in a task-general manner. We show that implicit learning, unlike endogenous spatial cuing, can bias spatial attention within one task without biasing attention to a spatially overlapping secondary task. Participants completed a visual search task superimposed on a background containing scenes, which they were told to encode for a later memory task. Experiments 1 and 2 used explicit instructions to bias spatial attention to one region for visual search; Experiment 3 used location probability cuing to implicitly bias spatial attention. In location probability cuing, a target appeared in one region more than others despite participants not being told of this. In all experiments, search performance was better in the cued region than in uncued regions. However, scene memory was better in the cued region only following endogenous guidance, not after implicit biasing of attention. These data support a dual-system view of top-down attention that dissociates goal-driven and implicitly learned attention. Goal-driven attention is task general, amplifying processing of a cued region across tasks, whereas implicit statistical learning is task-specific. (PsycINFO Database Record

Altering spatial priority maps via statistical learning of target selection and distractor filtering

The cognitive system has the capacity to learn and make use of environmental regularities - known as statistical learning (SL), including for the implicit guidance of attention. For instance, it is known that attentional selection is biased according to the spatial probability of targets; similarly, changes in distractor filtering can be triggered by the unequal spatial distribution of distractors. Open questions remain regarding the cognitive/neuronal mechanisms underlying SL of target selection and distractor filtering. Crucially, it is unclear whether the two processes rely on shared neuronal machinery, with unavoidable cross-talk, or they are fully independent, an issue that we directly addressed here. In a series of visual search experiments, participants had to discriminate a target stimulus, while ignoring a task-irrelevant salient distractor (when present). We systematically manipulated spatial probabilities of either one or the other stimulus, or both. We then measured performance to evaluate the direct effects of the applied contingent probability distribution (e.g., effects on target selection of the spatial imbalance in target occurrence across locations) as well as its indirect or "transfer" effects (e.g., effects of the same spatial imbalance on distractor filtering across locations). By this approach, we confirmed that SL of both target and distractor location implicitly bias attention. Most importantly, we described substantial indirect effects, with the unequal spatial probability of the target affecting filtering efficiency and, vice versa, the unequal spatial probability of the distractor affecting target selection efficiency across locations. The observed cross-talk demonstrates that SL of target selection and distractor filtering are instantiated via (at least partly) shared neuronal machinery, as further corroborated by strong correlations between direct and indirect effects at the level of individual participants. Our findings are compatible with the notion that both kinds of SL adjust the priority of specific locations within attentional priority maps of space.

Knowing where is different from knowing what: Distinct response time profiles and accuracy effects for target location, orientation, and color probability

When a location is cued, targets appearing at that location are detected more quickly. When a target feature is cued, targets bearing that feature are detected more quickly. These attentional cueing effects are only superficially similar. More detailed analyses find distinct temporal and accuracy profiles for the two different types of cues. This pattern parallels work with probability manipulations, where both feature and spatial probability are known to affect detection accuracy and reaction times. However, little has been done by way of comparing these effects. Are probability manipulations on space and features distinct? In a series of five experiments, we systematically varied spatial probability and feature probability along two dimensions (orientation or color). In addition, we decomposed response times into initiation and movement components. Targets appearing at the probable location were reported more quickly and more accurately regardless of whether the report was based on orientation or color. On the other hand, when either color probability or orientation probability was manipulated, response time and accuracy improvements were specific for that probable feature dimension. Decomposition of the response time benefits demonstrated that spatial probability only affected initiation times, whereas manipulations of feature probability affected both initiation and movement times. As detection was made more difficult, the two effects further diverged, with spatial probability disproportionally affecting initiation times and feature probability disproportionately affecting accuracy. In conclusion, all manipulations of probability, whether spatial or featural, affect detection. However, only feature probability affects perceptual precision, and precision effects are specific to the probable attribute.