Attention, Perception, & Psychophysics

Set-size effects in change detection is often used to investigate the capacity limits of dividing attention. Such capacity limits have been attributed to a variety of processes including perception, memory encoding, memory storage, memory retrieval, comparison and decision. In this study, we investigated the locus of the effect of increasing set size from 1 to 2. To measure purely attentional effects and not other phenomena such as crowding, a precue was used to manipulate relevant set size and keep the display constant across conditions. The task was to detect a change in the orientation of 1 or 2 Gabor patterns. The locus of the capacity limits was determined by varying when observers were cued to the only stimulus that was relevant. We began by measuring the baseline set-size effect in an initial experiment. In the next experiment, a 100% valid postcue was added to test for an effect of decision. This postcue did not change the set-size effects. In the critical experiments, a 100% valid cue was provided during the retention interval between displays, or only one stimulus was presented in the second display (local recognition). For both of these conditions, there was little or no set-size effect. This pattern of results was found for both hard-to-discriminate stimuli typical of perception experiments and easy-to-discriminate stimuli typical of memory experiments. These results are consistent with capacity limits in memory retrieval, and/or comparison. For these set sizes, the results are not consistent with capacity limits in perception, memory encoding or memory storage. Significance SectionThe change detection paradigm is often used to demonstrate effects of divided attention. But it is not clear whether these effects are due to perception, memory, or judgment and decision. In this article, we present new evidence that the divided attention effect in change detection is due to limits in memory retrieval or comparison processes. These results are not consistent with limits in perception, memory encoding or memory storage.

Attention is automatically guided towards stimuli that match the contents of working memory. This has been studied extensively using simplified computer tasks, but it has never been investigated whether (yet often assumed that) memory-driven guidance also affects real-life search. Here we tested this open question in a naturalistic environment that closely resembles real life. In two experiments, participants wore a mobile eye-tracker, and memorized a color, prior to a search task in which they looked for a target word among book covers on a bookshelf. The memory color was irrelevant to the search task. Nevertheless, we found that participants gaze was strongly guided towards book covers that matched the memory color. Crucially, this memory-driven guidance was evident from the very start of the search period. These findings support that attention is guided towards working-memory content in real-world search, and that this is fast and therefore likely reflecting an automatic process. Significance statementA core concept in the field of visual working memory (VWM) is that visual attention is automatically guided towards things that resemble the content of VWM. For example, if you hold the color red in VWM, your attention and gaze would automatically be drawn towards red things in the environment. So far, studies on such memory-driven guidance have only been done with well-controlled computer tasks that used simplified search displays. Here we address the crucial and open question of whether attention is guided by the content of VWM in a naturalistic environment that closely resembles real life. To do so, we conducted two experiments with mobile eye tracking. Crucially, we found strong memory-driven guidance from the very early phase of the search, reflecting that this is a fast, and therefore likely automatic, process that also driven visual search in real life.

Recent research indicates there is overlap in the neural resources used during imagery and visual short-term memory. But do visual short-term memory and visual imagery operate on similar representations during recall? Here we investigated this question by asking participants to perform a delayed match to sample task for the contrast of visual gratings as cues. In the "Imagery" condition, participants were asked to form an accurate mental image of the visual cue, and at the end of the trial, perform a matching task on the mental image contrast. In the "Memory" condition, participants were not required to perform visual imagery but merely instructed to perform the delayed contrast-matching task. In Experiment 1, participants were told at the beginning of each block whether to engage in memory or imagery. The results showed that for the relevant contrast feature, matching judgments were more accurately in the "Memory" than in the "Imagery" condition. Thus imagery did not maintain an accurate representation of the encoded image, even when the visual features could still be maintained in visual short-term memory. In Experiment 2, participants were required to engage in memory and imagery simultaneously, and were told which to base their judgment on after each trial. The key finding was that the superior accuracy for memory over imagery remained, indicating that the contents of VSTM and imagery are based on distinct representations.

In visual search, improved performance when a target appears at a recently cued location is taken as strong evidence that attention was shifted to this cue. Here, we provide evidence challenging the canonical interpretation of spatial-cueing (or cue-validity) effects and supporting the Priority Accumulation Framework (PAF). According to PAF, attentional priority accumulates over time at each location until the search context triggers selection of the highest-priority location. Spatial-cueing effects reflect how long it takes to resolve the competition and can thus be observed even when attention was never shifted to the cue. Here, we used a spatial-cueing paradigm with abruptly onset cues and search displays varying in target-distractor similarity. We show search performance on valid-cue trials deteriorated the more difficult the search, a finding that is incompatible with the standard interpretation of spatial-cueing effects. By using brief displays (Experiment 1) and by examining the effect of search difficulty on the fastest trials (Experiment 2), we invalidate alternative accounts invoking post-perceptual verification processes (Experiment 1) or occasional failures of the onset cue to capture attention (Experiment 2). In Experiment 3, we used a combination of the spatial-cueing and dot-probe paradigms. We show that the events that occurred in both the cue and search displays affected attentional distribution, and that the relative attentional priority weight that accumulated at the target location determined how easily the competition was resolved. These findings fully support PAFs predictions. Public significance statementMany studies aim at establishing whether certain objects mandatorily capture our attention. Here, we show that there is no "yes-or-no" answer to this question because the context in which an object appears determines whether this object captures attention. We show that our attention is not shifted to the highest-priority object at any given time: instead, information about priority is collected across time until some signal indicates that the appropriate moment for deploying our attention has arrived. Striking failures to notice conspicuous events routinely illustrate how limited our attentional system is: we can attend to very few objects at any given time, and probably to just one. In natural conditions, when we move the focus of our attention from one object to another, we also shift our gaze towards the attended location: this allows us to place the object of most interest in the center of our fovea, which maximizes the quality of its perceptual processing. Tracking the locus of such overt attention is easily achieved by using eye-tracking devices. However, in order to isolate the benefits of attention from the benefits of visual acuity, one must study covert attention - that is, attentional shifts in the absence of eye movements. These shifts are not directly observable and must therefore be inferred using indirect measures of processing.

Confidence in a perceptual decision is a subjective estimate of the accuracy of ones choice. As such, confidence is thought to be an important computation for a variety of cognitive and perceptual processes, and it features heavily in theorizing about conscious access to perceptual states. Recent experiments have revealed a "positive evidence bias" (PEB) in the computations underlying confidence reports. A PEB occurs when confidence, unlike objective choice, over-weights the evidence for the chosen option, relative to evidence against the chosen option. Accordingly, in a perceptual task, appropriate stimulus conditions can be arranged that produce selective changes in confidence reports but no changes in accuracy. Although the PEB is generally assumed to reflect the observers perceptual and/or decision processes, post-decisional accounts have not been ruled out. We therefore asked whether the PEB persisted under novel conditions that eliminated two possible post-decisional accounts: 1) post-decision evidence accumulation that contributes to a confidence report solicited after the perceptual choice, and 2) a memory bias that emerges in the delay between the stimulus offset and the confidence report. We found that even when the stimulus remained on the screen until observers responded, and when observers reported their choice and confidence simultaneously, the PEB still emerged. Signal detection-based modeling also showed that the PEB was not associated with changes to metacognitive efficiency, but rather to confidence criteria. We conclude that once-plausible post-decisional accounts of the PEB do not explain the bias, bolstering the idea that it is perceptual or decisional in nature.

Static statistical regularities in the placement of targets and salient distractors within the search display can be learned and used to optimize attentional guidance. Whether statistical learning also extends to dynamic regularities governing the placement of targets and distractors on successive trials has been less investigated. Here, we applied the same dynamic cross-trial regularity (one-step shift of the critical item in clock-/counterclockwise direction) either to the target or a distractor, and additionally varied whether the distractor was defined in a different (color) or the same dimension (shape) as the target. We found robust learning of the predicted target location: processing of the target at this (vs. a random) location was facilitated. But we found no evidence of proactive suppression of the predictable distractor location. Facilitation of the anticipated target location was associated with explicit awareness of the dynamic regularity, whereas participants showed no awareness of the distractor regularity. We propose that this asymmetry arises because, owing to the targets central role in the task set, its location is explicitly encoded in working memory, enabling the learning of dynamic regularities. In contrast, the distractor is not explicitly encoded; so, statistical learning of distractor locations is limited to static regularities. Public significance statementCan we learn the cross-trial dynamic regularity of a target or a task-irrelevant salient distractor (e.g., one-step shift of the critical item in clock-/counterclockwise direction) to boost search performance? The present study found robust learning of the predicted target location, but no evidence of proactive suppression of the predictable distractor location. Facilitation of the anticipated target location was associated with explicit awareness of the dynamic regularity. This asymmetry highlights the important role of the target-centered task set in the learning of dynamic regularities.

The decoy effect occurs when adding an inferior third option biases choice between two others, even though the decoy is rarely chosen. While robust in value-based decisions, evidence in perceptual tasks is mixed. Using the rhesus-macaque paradigm from Parrish et al. (2015), we tested whether a perceptual decoy effect generalizes to humans. Participants (n = 50) completed 400 trials. Contrary to our preregistered prediction, we found no reliable decoy effect. Accuracy improved on the hardest trials (Level 1) when a decoy was present, response times were slower in decoy conditions than baseline, and accuracy was higher for tall versus wide rectangles, consistent with the vertical-horizontal asymmetry. The relatively wide spacing of stimuli may have reduced grouping and attentional clustering; because spacing was not manipulated, this remains a hypothesis for future tests. Results suggest that context effects in perceptual choice operate under narrower boundary conditions than in value-based domains.

Perception across time is limited: phenomena like masking and temporal crowding reveal perceptual interference between sequential stimuli, even when separated by hundreds of milliseconds. However the underlying causes of these suppressive temporal interactions remain unclear. A potential explanatory mechanism is temporal normalization, a form of divisive suppression proposed to operate across time in the visual cortex, but largely untested behaviorally. Here in three experiments we tested a key perceptual prediction of temporal normalization: contrast-dependent suppression between sequential stimuli. By independently manipulating the contrasts of two sequential gratings presented 250 ms apart, we found that perceptual sensitivity to a targets orientation decreased when a high-vs. low-contrast non-target appeared either before or after it. Reduced sensitivity arose from both decreased orientation precision and increased "swapping" errors. The results provide perceptual support for a temporal normalization computation operating across hundreds of milliseconds, offering a theoretical framework for perceptual limits across time. Public significance statementTemporal context strongly affects perception, but a general theoretical framework that captures such effects is lacking. Neural studies support the existence of temporal normalization, the divisive suppression of information across time. However the behavioral impact of temporal normalization on perception remains unclear. Here we tested a key behavioral prediction of temporal normalization: contrast-dependant suppression between sequential stimuli. We found evidence for such suppression across hundreds of milliseconds both forward and backward in time. Our results provide perceptual evidence for temporal normalization, linking neural data to behavior, and support a theoretical framework that could bridge a range of temporal phenomena such as visual masking, adaptation, and temporal crowding.

Continuous recognition has long been used to study the recency effect in recognition memory. In continuous recognition, response time (RT) increases as a logarithmic function of the lag since a probe was last presented. Although this could simply be due to decaying trace strength, careful examination of response time (RT) distributions showed that the onset of RT distributions changed with the logarithm of the lag since a probe was originally presented. Each doubling of lag resulted in a shift of roughly 20 ms in the rise time of the RT distributions. To test the hypothesis that this increase was simply due to increased facility in processing the probe item, Experiment six repeated items six times. Repetition resulted in faster RTs, but did not change the effect of log lag on RT. In light of recent neurophysio-logical evidence, we consider the hypothesis that memory requires a recovery of temporal context, and the time to retrieve a prior temporal context goes up with the logarithm of the time since it was experienced.

Viewpoint effects on object recognition interact with object-scene consistency effects. While recognition of objects seen from "accidental" viewpoints (e.g., a cup from below) is typically impeded compared to processing of objects seen from canonical viewpoints (e.g., the string-side of a guitar), this effect is reduced by meaningful scene context information. In the present study we investigated if these findings established by using photographic images, generalise to 3D models of objects. Using 3D models further allowed us to probe a broad range of viewpoints and empirically establish accidental and canonical viewpoints. In Experiment 1, we presented 3D models of objects from six different viewpoints (0{degrees}, 60{degrees}, 120{degrees}, 180{degrees} 240{degrees}, 300{degrees}) in colour (1a) and grayscaled (1b) in a sequential matching task. Viewpoint had a significant effect on accuracy and response times. Based on the performance in Experiments 1a and 1b, we determined canonical (0{degrees}-rotation) and non-canonical (120{degrees}-rotation) viewpoints for the stimuli. In Experiment 2, participants again performed a sequential matching task, however now the objects were paired with scene backgrounds which could be either consistent (e.g., a cup in the kitchen) or inconsistent (e.g., a guitar in the bathroom) to the object. Viewpoint interacted significantly with scene consistency in that object recognition was less affected by viewpoint when consistent scene information was provided, compared to inconsistent information. Our results show that viewpoint-dependence and scene context effects generalize to depth rotated 3D objects. This supports the important role object-scene processing plays for object constancy.

The Selective Tuning model proposed that attention to a visual stimulus suppresses interfering portions of the processing hierarchy. This has been shown in spatial and featural domains and the proposal extends to the temporal dimension. We investigated whether attending to a point in time leads to processing suppression of nearby time points. We presented a sequence of letters with an embedded target, and observers indicated the targets orientation. In the neutral condition, the target appeared randomly in one of the frames. In the informative condition, the target appeared in the same frame on most of the trials ( expected trials), and the expected frame varied between blocks (Experiments 1-3) or groups (Experiment 4). On the remaining trials, it appeared before or after the most-probable frame ( unexpected trials). Four different experiments were conducted to adjust the method and in the final one we found higher accuracy in expected trials than in neutral trials, indicating the allocation of temporal attention to the expected frame. When the target appeared after the expected frame, accuracy was lower than in the same neutral condition frame, suggesting an attentional suppressive wake in time (because the effect of attention cannot be observed until after stimulus onset).

This study aimed to investigate how working memory interacts with perception at the behavioral level. We examined whether working memory modulates subjective perception when the general capacity for perceptual processing is kept constant. This possibility is in part motivated by the neuropsychological phenomenon of blindsight, in which subjective perception and general perceptual capacity are dissociated. To test this, in this pilot study (n = 8), we manipulated the congruency between perceptual and working memory content and examined its effects on sensitivity in a Yes/No (Y/N) detection task. To match perceptual processing capacity across conditions, we titrated stimulus contrast in a two-interval forced choice (2-IFC) task. There was a trend that working memory selectively modulated subjective perception, as measured by detection sensitivity (da) in the Y/N task, even though processing capacity in the 2-IFC task was matched. Specifically, when the working memory content was categorically incongruent with the perceptual stimulus, Y/N sensitivity was selectively impaired. These preliminary findings suggest that working memory modulates subjective perception in ways that are less general than some other manipulations, such as visual masking, which generally impairs both objective and subjective aspects of perception. We hereby treat this preprint as a form of pre-registration, specifying our hypotheses, experimental designs, and analysis methods before examining whether the preliminary results will be confirmed in a larger sample.

We provide evidence that the brain searches for salient stimuli below the level of conscious awareness. We show that, while targets are found very efficiently in Rapid Serial Visual Presentation (RSVP), distractors do not leave a strong memory trace, especially if the memory probe is unexpected. This fits with traditional theories of late attentional selection, whereby "correct rejections" of non-targets are performed with very little processing cost. Our findings are also consistent with the tokenized percept theory of conscious perception, one element of which is that conscious awareness is required to provide sustained representations.

Recent studies show that cross-modal semantic congruence plays a role in spatial attention orienting and visual search. However, the extent to which these cross-modal semantic relationships attract attention automatically is still unclear. At present the outcomes of different studies have been inconsistent. Variations in task-relevance of the cross-modal stimuli (from explicitly needed, to completely irrelevant) and the amount of perceptual load may account for the mixed results of previous experiments. In the present study, we addressed the effects of audio-visual semantic congruence on visuo-spatial attention across variations in task relevance and perceptual load. We used visual search amongst images of common objects paired with characteristic object sounds (e.g., guitar image and chord sound). We found that audio-visual semantic congruence speeded visual search times when the cross-modal objects are task relevant, or when they are irrelevant but presented under low perceptual load. Instead, when perceptual load is high, sounds fail to attract attention towards the congruent visual images. These results lead us to conclude that object-based crossmodal congruence does not attract attention automatically and requires some top-down processing.

How well can one perceive simultaneous stimuli at two widely spaced visual locations? Are the stimuli processed independently? If not, does the dependency affect perception, disrupt signals in later stages, or both? To address these questions, we measured effects of divided attention using a dual-task paradigm with stimuli presented in noise on either side of fixation. This paradigm was applied to detecting Gabor patches and to the semantic categorization of words. We measured dual-task deficits which are a decline in mean performance for a dual task compared to a single task. There was such a deficit for categorizing two words but relatively little deficit for detecting two Gabors. We also measured congruency effects which are when performance at one location depends on whether the stimulus at the other location requires the same response. There was such a congruency effect for detecting two Gabors but relatively little congruency effect for categorizing two words. Further experiments were consistent with the dual-task deficit in word categorization being perceptual, but the congruency effect in Gabor detection being due to later processes. Results of additional experiments showed that the congruency effect was consistent with either graded selection errors or with all-or-none selection followed by graded interactive processing. To answer our opening question: for Gabor detection, perceptual processes were largely independent but later processes caused congruency effects; for word categorization, perceptual processes had capacity limits but even in combination with later processes caused relatively little congruency effects. In summary, there was evidence for two different kinds of dependency. Such complementary dependencies are inconsistent with theories of divided attention that depend on a single dependency such as a single resource or single source of interactive processing.

Serial dependence - the bias from recent experience on present response - is often linked to shared memory representations. Yet, it remains unclear whether this bias across tasks when stimulus features are shared but response modes differ. To test this, we interleaved temporal reproduction and bisection tasks using a post-cue design that held duration encoding constant while varying motor output across trials. Using structural equation modeling (SEM), we dissociated perceptual (stimulus-driven) and decisional (response-driven) components of serial dependence. With consecutive same tasks, we replicated repulsive perceptual serial dependence and attractive decisional carryover. Critically, these effects vanished across tasks, despite identical stimulus processing - suggesting that consistent motor responses, not shared memory alone, drive sequential biases. Moreover, SEM uncovered repulsive perceptual influences that standard regression missed, highlighting its power to isolate overlapping effects. Together, these findings reveal that response-specific reactivation underpins serial dependence, pointing to motor-context binding as a key factor in temporal decision-making.

Our attentional resources are allocated to the various aspects of the environment based on the context, and predictive coding has been used as a model to explain the interaction between sensory-based information and top-down expectations in visual attention (Spratling, 2008; Rauss et al., 2011). On the other hand, the saliency of the environmental stimuli is also hypothesized to be capturing the attentional resources of the individuals involuntarily, and thus, it is thought to be playing a crucial role in attentional resource allocation. The current study investigates the role of predictive processing of task difficulty in selective visual attention in the presence of various distractors. Utilizing a letter search task, we provided brief cues about the upcoming tasks difficulty, and participants were asked to detect the target letters. We investigated whether predictive processing about task demands may cause a difference in behavioral measures in the presence of semantically less salient distractors in Experiment 1 (Gabor patches) and semantically more salient distractors in Experiment 2 (faces). Results showed that unmet expectations about the task demands caused longer reaction times in both studies. We observed that all independent variables, which are task difficulty, cue congruency, and distractor presence, affected reaction times in both experiments, but cue congruency interacted with distractor presence only in Experiment 2. Here, we argue that though predictive processing plays a role in attentional resource allocation and, distractors characteristics are also crucial as the saliency level interacts with the cue congruency.

Over the years a number of researchers have reported enhanced performance of targets located horizontally to a cued location relative to those located vertically. However, many of these reports could stem from a known meridian asymmetry in which stimuli on the horizontal meridian show a performance advantage relative to those on the vertical meridian. Here we show a horizontal advantage for target and cue locations that reside outside the zone of asymmetry; that is, targets that appear horizontal to the cue, but above or below the horizontal meridian, are more accurate than those that appear vertical to the cue, but again either above or below the horizontal meridian (Experiments 1 and 4). This advantage does not extend to non-symmetrically located targets in the opposite hemifield (Experiment 2), nor horizontally located targets within the same hemifield (Experiment 3). These data raise the possibility that display designs in which the target and cue locations are positioned symmetrically across the vertical midline may be underestimating the cue validity effect.

The ability to store information in visual working memory is essential to plan and successfully execute memory-guided actions in natural human behavior. Typically, visual working memory research investigates how storing affects subsequent action. In doing so, however, the importance of how the action affects prior storing remains underappreciated. Therefore, we here question how the required precision for an action to succeed, affects how relevant visual information is encoded, maintained and finally acted on. To this end, we had participant memorize 1, 2 or 4 colors for delayed continuous report. Crucially, we manipulated how (im)precise the report was allowed to still be marked correct. Behavioral results showed that for actions with higher required precision, reports became more precise, but only when one or two colors were memorized. Also, reports became slower with higher required precision, regardless of the number of colors. By leveraging pupillometry, we further showed that with higher required precision, 1) colors were encoded deeper (since pupils constricted more during presentation), and 2) more effort was exerted to maintain the colors (since pupils dilated more during retention). Moreover, we found that participants kept exerting more effort to be precise (with increasing precision requirements), even when additional effort did not result in better performance anymore. Our findings demonstrate that humans consider their intended actions when encoding and maintaining information in visual working memory. Our results highlight the essential role of action in understanding how visual information is stored during natural goal-directed behavior. 242 words

The nature of working memory (WM) limitations has been a topic of long-standing debate, with several models proposed to elucidate this issue. In this study, we conducted a systematic comparison of seven visual WM models to assess their ability to account for target consolidation during the attentional blink (AB). The AB phenomenon refers to where participants often fail to encode the second of two targets when there is a short time interval of [~]500 msec or less between them, providing an opportunity to evaluate commensurate WM limitations. Despite the growing consensus on the applicability of some WM models, such as the standard mixture model and the variable precision model, to the AB domain, no study has systematically evaluated these models in this context. We compared the performance of seven widely adopted visual WM models in four different AB datasets, drawn from three separate laboratories. We fitted each model and computed the Akaike Information Criterion (AIC) values at an individual level, across different conditions and experiments, based on which we compared the models. Slot-family models most often minimized AIC for second-target reports at short lags, while variable-precision models improved at longer lags and with color targets, indicating predominantly discrete consolidation during the AB, with feature- and lag-dependent graded components. These patterns imply that failure-to-encode (guessing) dominates over low-precision encoding, except when feature content or lag affords partial consolidation, refining theories of episodic tokenization and WM consolidation during the AB.