Need for cross-level iterative re-entry in models of visual processing.

Two main hypotheses regarding the directional flow of visual information processing in the brain have been proposed: feed-forward (bottom-up) and re-entrant (top-down). Early theories espoused feed-forward principles in which processing was said to advance from simple to increasingly complex attributes terminating at a higher area where conscious perceptions occur. That view is disconfirmed by advances in neuroanatomy and neurophysiology, which implicate re-entrant two-way signaling as the predominant form of communication between brain regions. With some notable exceptions, the notion of re-entrant processing has had a relatively modest effect on computational models of perception and cognition, which continue to be predominantly based on feed-forward or within-level re-entrant principles. In the present work we describe five sets of empirical findings that defy interpretation in terms of feed-forward or within-level re-entrant principles. We conclude by urging the adoption of psychophysical, biological, and computational models based on cross-level iterative re-entrant principles.

Alerting effects occur in simple-But not in compound-Visual search tasks.

Alerting (e.g., a brief brightening of the screen just before a target display) is known to facilitate visual search in simple tasks that involve the single step of detecting a pop-out item within a stimulus array. What is not known is whether alerting facilitates performance also in compound search tasks which involve two steps: First, locate the pop-out item, then identify a detail of that item. In a series of five experiments, we show that alerting facilitates performance of each component of a compound task when tested separately, (Experiments 2a and 2b) but not when the components are combined in a compound task (Experiment 1). Yet, alerting does facilitate performance in a compound task when the pop-out item is displayed in the same location on successive trials (Experiment 3). We hypothesized that such spatial repetition allows attention to linger at that location, thus allowing the first component (locate the pop-out item) to be bypassed. In practice, this turns the compound task into a simple task. That hypothesis was confirmed in Experiment 4 using a reorienting cue to shift the focus of attention to another location. An overall account of the absence of alerting effects in compound search tasks is proposed in terms of the temporal relationship between a period of enhancement rendered as an ex-Gaussian function and the hypothesized sequence of processing stages in visual search. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Metacontrast masking reduces the estimated duration of visible persistence.

A brief visual display can give rise to a sensation that outlasts the duration of the physical stimulus. The duration of this visible persistence has been estimated with paradigms that require the temporal integration of two brief sequential displays (frames) separated by a blank temporal gap. Temporal integration is said to occur when the visible persistence generated by the first frame is sufficiently long to bridge the inter-frame temporal gap. The longest gap at which integration still occurs is taken as an estimate of the duration of visible persistence. In the present work, we show that the duration of visible persistence has been underestimated in at least some of the experiments involving the temporal integration of successive displays. This is because the trailing frame can act as a metacontrast mask that foreshortens the visibility of the leading frame. Specifically, we show that operations that reduce the strength of metacontrast masking yield longer estimates of visible persistence. The relationship between metacontrast masking and visible persistence had been mentioned in some individual studies, but a comprehensive examination of that relationship is currently unavailable. Finally, we show that estimates based on single displays (e.g., the Sperling paradigm) also fail to provide untainted estimates because, in single displays, visible persistence is confounded with informational persistence.

T1 difficulty does not modulate the magnitude of the attentional blink.

When the visual system is busy processing one stimulus, it has problems processing a subsequent stimulus if it arrives soon after the first. Laboratory studies of this second-stimulus impairment-known as attentional blink (AB)-have employed two targets (T1, T2) presented in rapid sequence, and have found identification accuracy to be nearly perfect for T1, but impaired for T2. It is commonly believed that the magnitude of the AB is related directly to the difficulty of T1: the greater the T1 difficulty, the larger the AB. A survey of the experimental literature disconfirms that belief showing it to have arisen from artificial constraints imposed by the 100% limit of the response scale. Removal of that constraint, either using reaction time (RT) instead of accuracy as the dependent measure, or in experiments in which the functions of T2 accuracy over lags do not converge to the limit of the response scale, reveals parallel functions for the easy-T1 and the hard-T1 conditions, consistent with the idea that T1 difficulty does not modulate AB magnitude. This finding is problematic for all, but the Boost and Bounce (B&B) and the Locus Coeruleus-Norepinephrine (LC-NE) theories in which T1 acts merely as a trigger for an eventual refractory period that leads to the failure to process T2, rendering T1 difficulty and its relationship to the AB an irrelevant consideration.

What factors influence the switch from unitary to divided attention?

The focus of attention can be either unitary or divided and can transition from unitary to divided while performing a task. In Experiment 1, we investigated whether alerting hastens the transition from unitary to divided attention. To this end, we employed a dual-RSVP-stream Attentional Blink task (AB; impaired perception of the second of two rapidly sequential targets) with two pairs of letter targets (T1-pair and T2-pair). One component of the AB known as Lag-1 sparing (unimpaired perception of the T2-pair when it is presented directly after the T1-pair) occurs only when the T2-pair falls in an attended location. When the T2-pair falls in an unattended location, the converse pattern occurs (Lag-1 deficit). Accordingly, we used the incidence of Lag-1 sparing/deficit to index whether a location was attended or unattended. We found that presenting a brief brightening flash of the screen (alerting) just before the T1-pair hastened the transition from the initial unitary focus to a divided focus. In Experiment 2, we pitted the hastening account against an alternative hypothesis that the flash triggers phasic activation of the Locus Coeruleus-norepinephrine neuromodulatory system, thus resetting the underlying neural networks that mediate the distribution of attention, triggering a switch from unitary to divided attention. The results of Experiment 2 were incompatible with the hastening account, but consistent with the network-reset account.

Are the inverse-duration and inverse-proximity effects in temporal integration subserved by independent mechanisms?

Displays shorter than about 100 ms are normally seen as lasting longer than their physical duration. This visible persistence can bridge a temporal gap between two sequential stimuli causing them to be temporally integrated into a single percept. We investigated two findings in the temporal-integration literature: the inverse duration effect (temporal integration is progressively impaired as the duration of the first stimulus is increased) and the inverse proximity effect (temporal integration is progressively impaired as the spatial proximity between the stimuli is increased). In two experiments we asked whether the two effects are separable (i.e., whether they are subserved by independent mechanisms) or interact with one another. To estimate the duration of visible persistence we used the missing element paradigm in Experiment 1 and directional stroboscopic motion between two lines in Experiment 2. In both experiments we manipulated the duration of the leading stimulus and the spatial gap between the elements of the two sequential displays. Additive-factors logic was employed to examine the separability of the effects of duration and proximity. Independence (separability) of the two factors would be evidenced in a graph in which the functions of duration over proximity are parallel. The results pointed uniformly to separability. A plausible mechanism for the inverse duration effect is the burst of processing activity time-locked to stimulus onset. A plausible mechanism for the inverse proximity effect is lateral inhibition that acts to reduce the visible persistence of the leading stimulus.

The attentional blink: why does Lag-1 sparing occur when the dependent measure is accuracy, but Lag-1 deficit when it is RT?

Perception of the second of two targets (T1, T2) displayed in rapid sequence is impaired if it comes shortly after the first (attentional blink, AB). In an exception, known as Lag-1 sparing, T2 is virtually unimpaired if it is presented directly after T1. Three experiments examined the seemingly inconsistent findings that Lag-1 sparing occurs in accuracy but Lag-1 deficit occurs in RT. Experiment 1 pointed to masking of T2 as the critical factor. When T2 was not masked, the results replicated the conventional findings. The novel finding was that Lag-1 sparing occurred in RT, provided that T2 was masked. An account was provided by a psychological refractory period-based model in which processing was said to occur in two broadly sequential stages: stimulus selection and response planning. Experiments 2 and 3 tested predictions from the PRP-based model regarding Lag-1 sparing/Lag-1 deficit. In Experiment 2, we increased T2 salience, notionally reducing the duration of the T2 selection stage, with corresponding reduction in Lag-1 sparing. In Experiment 3, we manipulated the compatibility between the T1 stimulus and the response to notionally decrease/increase the duration of the T1 response-planning stage with corresponding increment/decrement in Lag-1 sparing. The results of both experiments confirmed predictions from the PRP-based model.

The exogenous and endogenous control of attentional focusing.

Selective visual attention involves prioritizing both the location (orienting) and distribution (focusing) of processing. To date, much more research has examined attentional orienting than focusing. One of the most well-established findings is that orienting can be exogenous, as when a unique change in luminance draws attention to a spatial location (e.g., Theeuwes in Atten Percept Psychophys 51:599-606, 1992; Yantis and Jonides in J Exp Psychol Hum Percept Perform 10:601, 1984), and endogenous, as when a red distractor shape diverts attention when one is looking for a red target (e.g., Bacon and Egeth in Percept Psychophys 55:485-496, 1994; Folk et al. in J Exp Psychol Hum Percept Perform 18:1030, 1992). Here we ask whether attentional focusing-the broadening and contracting of prioritized processing-is influenced by the same two factors. Our methodology involved a dual-stream attentional blink task; participants monitored two spatially separated streams of items for two targets that could appear unpredictably either in the same stream or in opposite streams. The spatial distribution of attention was assessed by examining second-target accuracy in relation to inter-target lag and target location (same or opposite streams). In Experiment 1, we found that attentional contracting was more rapid when the targets differed in luminance from the distractor items. In Experiments 2 and 3, we found that the rate of attentional contracting was slower when there were task-relevant distractors in the stream opposite the first target. These results indicate that the rate of attentional focusing, like orienting, can be modulated by both exogenous and endogenous mechanisms.

Sudden Events Change Old Visual Objects Into New Ones: A Possible Role for Phasic Activation of Locus Coeruleus.

We report a novel visual phenomenon called the rejuvenation effect. It causes an "old" object that has been on view for some time to acquire the properties of a suddenly appearing new object. In each experiment, a square outline was displayed continuously on one side of fixation. The target (an asterisk) was presented either inside the square or on the opposite side of fixation. On half of the trials, a transient visual or auditory event preceded the target. In Experiment 1a ( N = 139), response times were faster when the target appeared inside the square, but only when it was preceded by a transient event, consistent with the network-reset theory of locus coeruleus-norepinephrine (LC-NE) phasic activation. Three further experiments confirmed the predictions of network-reset theory, including the absence of rejuvenation in participants with atypical LC-NE functioning (individuals with symptoms of autism spectrum disorder). These findings provide new perspectives on what causes a visual object to be perceived as new.

Attention is a sterile concept; iterative reentry is a fertile substitute.

Attention has been defined as a filter, a limited resource, a spotlight, a zoom lens, and even as a glue that binds disconnected visual features into a coherent object. Here, I claim that all of these metaphor-based explanations are circular. As such, they fail to provide adequate accounts of the phenomena they are purported to explain. In contrast, those very phenomena can be explained on the idea that perceptions emerge from iterative exchanges between cortical regions linked by two-way pathways. Processing can occur in one of two modes: feed-forward and reentrant. In feed-forward mode, the system is configured optimally for the expected input, and perception occurs on the feed-forward sweep. This form of processing corresponds to what is commonly referred to as "preattentive". If the system cannot be configured appropriately, perceptions emerge from iterative reentrant processing, which is slower, and corresponds to what is commonly referred to as "attentive".

Deployment of spatial attention to a structural framework: exogenous (alerting) and endogenous (goal-directed) factors.

The main question examined in the present work was whether spatial attention can be deployed to an appropriate structural framework not only endogenously when the framework is displayed continuously, as in previous work, but also exogenously, when it is displayed transiently 100 ms before the target. The results of five experiments answered that question in the negative. We found that the onset transient triggered by a brief presentation of the structural framework did enhance the response to the upcoming target. That enhancement, however, was due not to the framework itself but to the alerting effect produced by its sudden onset, witness the finding that the same enhancement was produced by an onset transient triggered by a featureless stimulus (i.e., by a brief dimming of the entire screen, in the absence of a structural framework). We conclude that spatial attention can be deployed to the region demarcated by a structural framework when it is deployed endogenously but not when it is deployed exogenously. A theoretical account of the results is proposed in terms of the temporal dynamics of the locus cœruleus/norepinephrine neuromodulatory system.

Perception of temporal order during the attentional blink: Using stimulus salience to modulate prior entry.

When multiple targets are presented in rapid sequence, observers frequently confuse the order in which they were presented. The probability of order reversals is known to vary throughout the period of the attentional blink (AB), which refers to impairment in the perception of the second of two targets when it is presented within approximately 500 ms from the first. Our objective was to examine the principle of prior entry (in which perception of temporal order is said to be affected by the relative latency at which each target is processed) as a determinant of the perception of temporal order throughout the AB. In two experiments, three letter targets (T1, T2, T3) were inserted in a stream of digit distractors, with T3 always presented directly after T2. The T1-T2 lag was varied to assess the perception of T2-T3 temporal order throughout the period of the AB. Processing latency was manipulated by means of salience. In Experiment 1, salience of T2 and T3 was manipulated exogenously by coloring the salient target red with all other stimuli being green. In Experiment 2, salience was modulated endogenously by manipulating which of the two targets matched the contents of working memory. Consistent with the principle of prior entry, perception of temporal order in both experiments was enhanced throughout the period of the AB when T2 was salient, and impaired when T3 was salient. Simulations based on the Episodic Simultaneous Type, Serial Token (eSTST) model that incorporates prior-entry, matched the empirical results.

Visual search is postponed during the period of the AB: An event-related potential study.

In the phenomenon known as the attentional blink (AB), perception of the second of two rapidly sequential targets (T2) is impaired when presented shortly after the first (T1). Studies in which T2 consisted of a pop-out search array provided evidence suggesting that visual search is postponed during the AB. In the present work, we used behavioral and electrophysiological measures to test this postponement hypothesis. The behavioral measure was reaction time (RT) to T2; the electrophysiological measure was the onset latency of an ERP index of attentional selection, known as the N2pc. Consistent with the postponement hypothesis, both measures were delayed during the AB. The delay in N2pc was substantially shorter than that in RT, pointing to multiple sources of delay in the chain of processing events, as distinct from the single source postulated in current theories of the AB. Finally, the finding that the N2pc was delayed during the AB strongly suggests that attention is involved in the processing of pop-out search arrays.

Individual differences in rate of encoding predict estimates of visual short-term memory capacity (K).

The capacity of visual short-term memory (VSTM) is commonly estimated by K scores obtained with a change-detection task. Contrary to common belief, K may be influenced not only by capacity but also by the rate at which stimuli are encoded into VSTM. Experiment 1 showed that, contrary to earlier conclusions, estimates of VSTM capacity obtained with a change-detection task are constrained by temporal limitations. In Experiment 2, we used change-detection and backward-masking tasks to obtain separate within-subject estimates of K and of rate of encoding, respectively. A median split based on rate of encoding revealed significantly higher K estimates for fast encoders. Moreover, a significant correlation was found between K and the estimated rate of encoding. The present findings raise the prospect that the reported relationships between K and such cognitive concepts as fluid intelligence may be mediated not only by VSTM capacity but also by rate of encoding.

Is pop-out visual search attentive or preattentive? Yes!

Is the efficiency of "pop-out" visual search impaired when attention is preempted by another task? This question has been raised in earlier experiments but has not received a satisfactory answer. To constrain the availability of attention, those experiments employed an attentional blink (AB) paradigm in which report of the second of 2 targets (T2) is impaired when it is presented shortly after the first (T1). In those experiments, T2 was a pop-out search display that remained on view until response. The main finding was that search efficiency, as indexed by the slope of the search function, was not impaired during the period of the AB. With such long displays, however, the search could be postponed until T1 had been processed, thus allowing the task to be performed with full attention. That pitfall was avoided in the present Experiment 1 by presenting the search array either until response (thus allowing a postponement strategy) or very briefly (making that strategy ineffectual). Level of performance was impaired during the period of the AB, but search efficiency was unimpaired even when the display was brief. Experiment 2 showed that visual search is indeed postponed during the period of the AB, when the array remains on view until response. These findings reveal the action of at least 2 separable mechanisms, indexed by level and efficiency of pop-out search, which are affected in different ways by the availability of attention. The Guided Search 4.0 model can account for the results in both level and efficiency.

Escape from temporal-integration masking: the roles of visible persistence and input filtering.

A brief target embedded in­and coterminating with­a noise mask is identified easily when the duration of the mask is long but not when it is short (Di Lollo, 1980; inverse-duration effect). Identification has been said to be mediated by the visible persistence of the target, which outlasted that of the mask. We tested an alternative account based on input filtering triggered by the onset and offset of the target, relative to those of the mask, without recourse to visible persistence. The results of Experiment 1 could not be explained wholly in terms of visible persistence but were entirely consistent with input filtering. Identification suffered in Experiment 2 when transient responses were attenuated by "ramping." In Experiment 3, accuracy improved gradually as a function of leading-mask duration. All results were consistent with a modified version of von Holst's (1954) hypothesis that a new stimulus (e.g., the present mask) establishes an input filter within the system. Any sudden onsets or offsets then lead to the perception of a new object only when they do not match the input filter, thus becoming segregated from the temporally leading stimulus.

On the time course of attentional focusing in older adults.

Many sensory and cognitive changes accompany normal ageing, including changes to visual attention. Several studies have investigated age-related changes in the control of attention to specific locations (spatial orienting), but it is unknown whether control over the distribution or breadth of attention (spatial focus) also changes with age. In the present study, we employed a dual-stream attentional blink task and assessed changes to the spatial distribution of attention through the joint consequences of temporal lag and spatial separation on second-target accuracy. Experiment 1 compared the rate at which attention narrows in younger (mean age 22.6, SD 4.25) and older (mean age 66.8, SD 4.36) adults. The results showed that whereas young adults can narrow attention to one stream within 133 ms, older adults were unable to do the same within this time period. Experiment 2 showed that older adults can narrow their attention to one stream when given more time (266 ms). Experiment 3 confirmed that age-related changes in retinal illuminance did not account for delayed attentional narrowing in older adults. Considered together, these experiments demonstrate that older adults can narrow their attentional focus, but that they are delayed in initiating this process compared to younger adults. This finding adds to previously reported reductions in attentional dynamics, deficits in inhibitory processes, and reductions in posterior parietal cortex function that accompany normal ageing.

When can spatial attention be deployed in the form of an annulus?

Can spatial attention be deployed as an annulus? Some studies have answered this question in the positive, others in the negative. We tested the hypothesis that annular deployment depends on the presence of a suitable structural framework to which attention can be anchored. To this end, we added a structural framework to the displays of a study that failed to find an annular distribution of attention. The targets were displayed in an annular region around a central stream of task-irrelevant distractors which captured attention and impaired target identification. This design was replicated in our No-Anchors condition. In the Anchors condition in Experiments 1 and 2, a square outline was displayed at each of the four possible target locations. Consistent with the idea that attention can be deployed as an annulus only when a visual framework is present, the targets were identified more accurately (Experiment 1) and more rapidly (Experiment 2) when anchors were present than when they were absent. The number of anchors was increased to eight in Experiment 3. In Experiment 4 the central stream was omitted to verify that the enhanced performance did not arise from intrinsic properties of the anchors themselves. In Experiment 5, targets were presented in a blank annular region delimited by two concentric circles, thus obviating the possibility that attention was deployed as four or eight separate foci in Experiments 2 and 3, respectively.

Reentrant processing mediates object substitution masking: comment on Põder (2013).

Object-substitution masking (OSM) occurs when a target stimulus and a surrounding mask are displayed briefly together, and the display then continues with the mask alone. Target identification is accurate when the stimuli co-terminate but is progressively impaired as the duration of the trailing mask is increased. In reentrant accounts, OSM is said to arise from iterative exchanges between brain regions connected by two-way pathways. In an alternative account, OSM is explained on the basis of exclusively feed-forward processes, without recourse to reentry. Here I show that the feed-forward account runs afoul of the extant phenomenological, behavioral, brain-imaging, and electrophysiological evidence. Further, the feed-forward assumption that masking occurs when attention finds a degraded target is shown to be entirely ad hoc. In contrast, the evidence is uniformly consistent with a reentrant-processing account of OSM.

The flexible focus: whether spatial attention is unitary or divided depends on observer goals.

The distribution of visual attention has been the topic of much investigation, and various theories have posited that attention is allocated either as a single unitary focus or as multiple independent foci. In the present experiment, we demonstrate that attention can be flexibly deployed as either a unitary or a divided focus in the same experimental task, depending on the observer's goals. To assess the distribution of attention, we used a dual-stream Attentional Blink (AB) paradigm and 2 target pairs. One component of the AB, Lag-1 sparing, occurs only if the second target pair appears within the focus of attention. By varying whether the first-target-pair could be expected in a predictable location (always in-stream) or not (unpredictably in-stream or between-streams), observers were encouraged to deploy a divided or a unitary focus, respectively. When the second-target-pair appeared between the streams, Lag-1 sparing occurred for the Unpredictable group (consistent with a unitary focus) but not for the Predictable group (consistent with a divided focus). Thus, diametrically different outcomes occurred for physically identical displays, depending on the expectations of the observer about where spatial attention would be required.