Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Mar 1.
Published in final edited form as: J Exp Psychol Hum Percept Perform. 2017 Aug 10;44(3):356–366. doi: 10.1037/xhp0000457

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

Douglas A Addleman 1, Jinyi Tao 1, Roger W Remington 1, Yuhong V Jiang 1
PMCID: PMC5809231  NIHMSID: NIHMS879535  PMID: 28795835

Abstract

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.

Keywords: spatial attention, goal-driven attention, incidental learning, probability cuing

Influential models of selective attention explain covert shifts of attention in terms of a priority map, where attentional priority is determined by bottom-up stimulus salience, top-down goals, and in some cases selection history (Awh, Belopolsky, & Theeuwes, 2012; Bisley & Goldberg, 2010; Fecteau & Munoz, 2006; Itti & Koch, 2001; Wolfe, 2012). While disagreement surrounds the relative contributions of these factors to attentional selection (Nobre & Kastner, 2014), most theories support the idea that a single priority map determines attentional allocation. The majority of studies informing these theories draw their conclusions from single-task experiments (e.g., Folk, Remington, & Johnston, 1992; Hopfinger, Buonocore, & Mangun, 2000; Luck, Chelazzi, Hillyard, & Desimone, 1997; Theeuwes, 1994). In many real world contexts, though, humans must juggle multiple simultaneous tasks. When someone needs to attend to one region for a task and another, different region for other purposes, how would this affect spatial attention?

One possibility is that the attentional priority map is unitary across tasks. On this account, the spotlight of spatial attention for one task may inevitably spread1 to affect performance in all tasks (task-general account). This view is consistent with the idea that the cognitive system encourages coherence across the many tasks that it confronts (Cowan, 1988; Duncan, 1996), the observation of cross talks between concurrent tasks (Jiang & Swallow, 2014), and the cost of switching between tasks (Monsell, 2003). The task-general account also finds empirical support in studies on spatial attention; as reviewed next, attending to a spatial region in one task biases attention to that region in a concurrent task.

Participants in Awh, Jonides, and Reuter-Lorenz (1998) maintained in working memory either an object’s spatial location or its identity. They simultaneously performed a secondary character identification task. Rehearsing the object’s location in working memory facilitated the identification of a character in that location. The spread of spatial attention from the working memory task to the character identification task provides evidence that people do not maintain separate spatial attentional preferences for two simultaneous tasks. Analogous findings have been reported in research on multimodal allocation of attention. Spence and Driver (1996) asked participants to localize targets in two modalities. They provided a cue predictive only of upcoming auditory targets and examined responses when the target was visual instead. Even though the visual target was less likely to appear on the cued side, participants were faster detecting it when it appeared there. This finding supports the claim that spatial attention is task-general across different modalities. Similar results have been observed when the tasks involved visual and tactile modalities (Spence, Pavini, & Driver, 2000) and when participants performed two visual search tasks (Burnett, Close, D’Avossa, & Sapir, 2016). These results support the idea that goal-driven spatial attention acts through activation within a priority map common to all tasks.

A second possibility is that, at least in select circumstances, people are capable of maintaining independent attentional priority for concurrent tasks (task-specific account). One such circumstance is when one task involves spatial selection and a concurrent task does not. In Awh and colleagues (1998), for example, rehearsing an object’s identity in working memory does not affect spatial attention in a secondary task. More broadly, a recent theory of attention dissociates two forms of top-down attentional control, suggesting that spatial attention may not be implemented in a unitary attentional priority map (Jiang, Swallow, & Capistrano, 2013). According to this view, “declarative” goal-driven attention operates much like priority map theories suggest, while incidental learning is a form of “procedural” attention that is habitual, rather than explicitly guided (p. 9). In the first, an explicit cue, such as an arrow or verbal instructions indicating a likely target location in visual search, provides information about the most likely target location (Posner, 1980). In the second, task history (such as a target’s high probability of occurrence in one region of space) facilitates search in that region even when participants are not explicitly aware of the target’s location probability (Geng & Behrmann, 2005; Jiang, Swallow, Rosenbaum, & Herzig, 2013). Evidence strongly suggests that location probability learning affects attentional guidance and differs from goal-driven attention (Jiang, Sha, & Remington, 2015; Jiang, Swallow, & Capistrano, 2013; Twedell, Koutstaal, & Jiang, 2016; Won & Jiang, 2015).

Even though goal-driven attention may act on a task-general priority map, habitual guidance of spatial attention may primarily affect performance for the relevant task. This is because incidentally learned attention is hypothesized to induce habitual search behavior (Jiang et al., 2015), which is deployed only once stimuli have appeared and search has commenced (Jiang, Sigstad, & Swallow, 2013). In contrast, endogenous attention can be shifted in anticipation of the search task (Egeth & Yantis, 1997; Kastner, Pinsk, De Weerd, Desimone, & Ungerleider, 1999; Luck et al., 1997). It is plausible, then, that implicitly learned attentional biases do not act through a representation common across concurrent tasks, as seems to be the case for endogenous cuing, but through biasing spatial attentional shifts within a specific task context. If true, this would provide evidence for the dual-system view and entail that unitary spatial map theories require revision to accommodate the unique effects of habitual learning on spatial attention.

To examine how endogenous and implicit spatial attentional biases affect the spread of attention between concurrent tasks, we conducted three dual-task experiments. On each trial, participants viewed an array of letters presented against a background of four natural scenes, one in each quadrant (Figure 1). The display was presented briefly to curtail saccadic eye movements. In the primary task, participants searched for the target (the letter T) and reported its orientation. In the secondary task, participants encoded the scenes for a later memory task. We introduced a spatial attentional bias in the search task either endogenously (through task instructions) or incidentally (through location probability learning). We then examined how spatial selection in the search task affected memory of the background scenes. Even though the memory task requires an equal allocation of attention to all scenes, any spread of spatial attention from the search task should result in superior memory for scenes presented in the cued quadrant relative to the other quadrants.

Figure 1.

Figure 1

Open in a new tab

A sample search array. Participants were shown arrays like this for 216ms and were asked to report the orientation of the ‘T’. Participants also needed to remember the background scenes for a memory test.

The combination of a visual search task and a scene memory task also enables us to test the interaction between attention and memory. Some theories of attention and memory suggest that shifting attention to a location should enhance memory at that location (Cowan et al., 2005). Because these theories do not distinguish endogenous attention from implicitly guided attention, they make predictions similar to those of the unitary priority map theory. In contrast, other studies on attention and memory suggest that attending to a location does not necessarily lead to better memory for that location. For example, although spatial attention is deployed to an object undergoing substitution masking (as indexed by the N2pc component), it is insufficient for the encoding of the object into working memory (Woodman & Luck, 2003). This latter finding suggests that the link between attention and memory may be dependent on various factors. By testing two types of attentional guidance (explicit vs. implicit), this study may shed light on the relationship between spatial attention for one task and visual memory for another.

Experiment 1

Experiment 1 used an explicit, endogenous cuing paradigm to introduce an attentional bias toward one quadrant for the visual search task. The experiment was divided into 40 blocks of trials. At the beginning of each block participants were told that the target T was likely to appear in one specific quadrant, such as the upper left. They were asked to prioritize search in this quadrant. At the same time, we informed them that there would be a future memory test on all of the background scenes. The cued quadrant changed from block-to-block, preventing participants from acquiring consistent location probability learning. Spatial selection was therefore endogenously driven in the search task. In the first 24 blocks, the letter T indeed appeared in the instructed quadrant more often than chance; it was there on 50% of the trials. In the last 16 blocks, the letter T appeared in random locations despite the instructions. Periodically, we probed memory for the background scenes to examine whether participants prioritized the encoding of scenes in the quadrant cued for visual search.

Method

Participants

Participants in this study were students from the University of Minnesota. They were healthy adults naïve to the purpose of the study. All participants reported having normal or corrected-to-normal visual acuity and normal color vision. Participants signed an informed consent form prior to participation and were compensated with extra course credit or $10/hour.

Sixteen participants completed this experiment. This sample size was predetermined to be the same as previous studies on location probability learning (e.g., Jiang & Swallow, 2014). The 16 participants included 13 females and 3 males with a mean age of 23 years.

Equipment

Participants were tested individually in a room with fluorescent overhead lighting. They sat an unconstrained distance, approximately 60cm, from a 19-inch CRT monitor (1024x768 resolution; 75 Hz). Experiments were run using Psychtoolbox (Brainard, 1997; Pelli, 1997) implemented in MATLAB (www.mathworks.com).

Stimuli

A red fixation dot subtending 0.15° was presented in the center of the display throughout the experiment. The search stimuli consisted of an array of eight white letters that ranged in size from 0.7º to 2.7º. Items farther from the fixation point were larger, scaled according to the cortical magnification factor (Carrasco, Evert, Chang, & Katz, 1995). There were 32 possible item locations, evenly divided into four eccentricities (approximately 1.5º, 4º, 6.5º, and 12º). The eight letters included one letter T and seven letter Hs; each quadrant always contained 2 letters. Hs were created by combining a T with a second T rotated 180° to increase feature similarity to the target. The letter T had a random orientation of 0º, 90º, 180º, or 270º. The Hs could be upright or rotated 90º. The background of the search display contained four natural scenes, one in each quadrant (11.5ºx11.5º). The distance between the center of a scene and fixation was 8.75º. Scenes were randomly drawn from a bank of 660 images taken from the Internet. To make sure the letters were visible against any part of the scene, the letters were inscribed in black circles (Figure 1).

Procedure

The experiment was divided into 40 blocks. Each block contained 12 dual-task trials in which participants searched for the T and encoded the scenes to memory, followed by 8 memory test trials. On each dual-task trial, participants initiated the task by clicking on the red fixation dot. The task required eye-hand coordination and ensured that fixation was centralized. The mouse click caused the search array and scenes to appear. The display disappeared after 216ms, leaving just the red fixation point. We used the brief presentation to limit eye movements. The participants’ task was to report the orientation of the T (up, down, left or right) using the keyboard. A practice block of 16 trials was administered to familiarize participants with the task. The target’s location was random during the practice block. We emphasized accuracy in performing the task. Participants were asked to also remember the scenes for an old/new recognition task. Participants were told to treat the T/H search task as the primary task, and remember the scenes as best they could without sacrificing search accuracy. Upon response, an auditory tone provided accuracy feedback; following each block, participants were given their average block accuracy. To ensure that memory accuracy for the scenes was not at chance, we presented the same four scenes within a block of 12 dual-task trials. These scenes were randomly chosen from the database of 660 scenes with the constraint that scenes shown in earlier blocks would not appear again. Their spatial location was maintained across the 12 trials, ensuring that spatial selection biases induced by the search task would be consistently directed to the same scene.

Following the 12 dual-task trials, participants were shown 8 trials testing scene memory. These included the 4 scenes they had just seen and 4 new scenes they had not seen before. These scenes were presented one by one, in a random order, at the center of the display. Participants pressed ‘o’ or ‘n’ to indicate whether the scene was old or new, with accuracy feedback provided.

Design

We manipulated spatial attention for the visual search task through explicit instructions. At the beginning of each block, a yellow frame (13.5ºx13.5º, border thickness 0.05º) surrounded one quadrant. A computer voice instructed participants to prioritize that region during the search task. The cued quadrant changed from block-to-block, preventing participants from acquiring consistent location probability learning. In the first 24 blocks (the ‘biased phase’) the instruction was informative – the target T appeared in the instructed quadrant 50% of the time, significantly above chance. It appeared in each of the other three quadrants 16.7% of the time. In the last 16 blocks (the ‘neutral phase’) the instruction was uninformative – the target T appeared in each quadrant 25% of the time. Participants were not informed of this change. This ‘neutral’ phase allows comparisons between the extinction of attentional bias in the memory advantage in different experiments; it is possible that data from biased phases would be similar, and yet different attentional cues would yield different extinction profiles during the neutral phase. We expect that participants would form a spatial bias toward the cued quadrant, and this bias should be stronger when the cue was informative (the biased phase) than when it was uninformative (the neutral phase).

Results

Visual search

Because each block contained a total of just 12 trials in the dual-task phase, we binned every 4 blocks into an epoch. This yielded 6 epochs in which the target T’s location was biased toward the cued quadrant (“biased” phase), and 4 epochs in which the target T’s location was unbiased (“neutral” phase).

We first examined how successful participants were in establishing a spatial bias in the visual search task. Figure 2 shows search accuracy as a function of whether the target was in the cued quadrant for the two phases of the experiment.

Figure 2.

Figure 2

Open in a new tab

Results from the visual search task of Experiment 1. Participants were instructed (cued) to prioritize one quadrant. The target was frequently in the cued quadrant (which changed from block to block) during the biased phase, and was equally likely to be in all quadrants during the neutral phase. Error bars show +/− 1 standard error of the mean.

Repeated-measures ANOVAs showed that participants were significantly more accurate when the target was in the instructed (cued) quadrant rather than in the other quadrants, both in the biased phase, F(1, 15) = 43.43, p < .001, ηp2 = .74, and in the neutral phase, F(1, 15) = 17.13, p < .001, ηp2 = .53. In neither phase did the cue effect interact with epoch, Fs < 1. The spatial bias was weaker in the neutral phase. This was confirmed in an analysis that contrasted all data from the biased phase with those from the neutral phase. A significant interaction between cue validity and phase was found, F(1, 15) = 4.90, p < .05, ηp2 = .25.

Scene memory

The 8 test trials of each block included 4 new scenes, providing an index of false alarm rates. It also included 4 old scenes, one in the cued quadrant and three in the uncued quadrants. This yielded just a single observation per block for the cued quadrant. To achieve sufficient statistical power, we combined data from all 24 blocks of the “biased” phase, and all 16 blocks of the “neutral” phase. Figure 3 shows the memory hit rates, separately for scenes in the cued quadrant and those in the uncued quadrants.

Figure 3.

Figure 3

Open in a new tab

Recognition rates for scenes presented in Experiment 1. Horizontal dotted lines represent false alarm rate (classifying new scenes as old ones) for that phase. Error bars show +/−1 S.E. of the mean.

First, we verified that, despite the difficulty of performing both tasks, participants’ memory was above chance. The overall hit rate (correctly identifying an old scene) was significantly higher than the false alarm rate, t(15) = 3.71, p < .002. Next, we examined hit rates for scenes presented in the cued and uncued quadrants, separately for biased and neutral phases.

In the biased phase, scenes presented in the quadrant cued for visual search were identified more accurately than those in the other quadrants, t(15) = 3.08, p < .008. This effect was not significant in the neutral phase, t(15) = 1.20, p > .20. An ANOVA combining data across both phases showed a significant main effect of the scene’s quadrant, F(1, 15) = 4.86, p < .05, ηp2 = .25. The interaction between scene quadrant and phase was not significant, F(1, 15) = 1.12, p > .30.

All scene memory analyses in all experiments were also analyzed using d′ as a measure of sensitivity. In all cases, results were comparable when analyzing either hit rates as reported above or sensitivity. Note especially that, for within-phase analyses, d′ differences come exclusively from differences in hit rates, as the false alarm rate is the same for all quadrants within a given phase.

Discussion

In this experiment, endogenous spatial cuing yielded both a search advantage for the relevant task and a memory advantage for the secondary task. Given that all scenes were tested regardless of their encoding location, the memory task provided no incentive to preferentially attend to any region of space. The spatial bias in the memory task, therefore, provides strong evidence that spatial selection from the search task spread to the memory task. This finding is consistent with the idea that endogenous spatial attention is task-general and supports the idea that attention may operate in conjunction with working memory and long-term memory (Cowan et al., 2005). Specifically, attending to a location induces better memory for scenes from that location. Analogous effects were previously reported in cross-modal attention (Spence & Driver, 1996) and in spatial working memory (Awh & Jonides, 2001; Awh et al., 1998). Here, unlike in past studies, our results demonstrate spread of attention using a search task and a secondary background memory task. They support the conceptualization of goal-driven spatial attention as an amodal and task-general mechanism that enhances processing in the attended location of stimuli from all tasks.

Experiment 1 also suggests that the spread across tasks was not complete. Spatial cuing appeared to have a larger effect on the relevant task—visual search—than on the secondary memory task. Search performance improved by 30% in the cued quadrant relative to the uncued quadrants, but memory recognition improved by less than 10%. In addition, in the neutral phase when the cue no longer validly predicted the target’s location, participants continued to maintain a significant (albeit weaker) spatial bias toward the cued quadrant in search. Yet this was not sufficient to yield a strong cuing effect in the memory task. This is consistent with the findings of Burnett et al. (2016), in which a cue informative for only one of two simultaneous search tasks improved performance on both tasks, but did so to a greater degree in the relevant task. On balance, Experiment 1 shows that endogenous spatial attention can spread to secondary tasks.

Experiment 2

The first experiment shows that endogenously guided spatial attention can spread from the search task to the memory task. In Experiment 2, we used a different form of endogenous cuing for the search task. Rather than indicating a high-probability quadrant for a full block of trials, an arrow appearing before each trial indicated a likely target region for that trial. In addition, the arrow was presented briefly (100ms) and was followed immediately by the search display. This design extends our results to an additional form of endogenous cuing.

Method

Participants

Thirty-two new participants completed Experiment 2, including 24 females and 8 males, with a mean age of 22 years.

Procedure

This experiment was similar to Experiment 1 except for the following changes. We removed the block-wise spatial cue, replacing it with a trial-specific central arrow cue. On each trial a white arrow (4.5º in length) was presented at the center of the display, pointing at one of the four quadrants. The arrow appeared immediately upon the participant clicking the fixation dot to initiate the trial, and after 100ms was replaced by the search array, which appeared for 216ms as before. In the first 24 blocks (the biased phase), the arrow was 50% predictive of the search target’s quadrant (i.e., the target T appeared in the cued quadrant on 6 of the 12 trials). Counterbalancing ensured that for a given block of trials the arrow pointed equally often to each quadrant. These two constraints make it impossible to ensure that the target occurs equally often in each quadrant in each block; instead, the target’s location was counterbalanced in every set of two blocks. In the next 16 blocks (the neutral phase), the arrow was not informative of the target’s quadrant, as the target appeared in the cued quadrant only 25% of the time. For counterbalancing purposes, we increased the number of search trials per testing block to 16. As in Experiment 1, participants were given no indication that the utility of the cue for the search task would change; they were only told that the arrow was 50% predictive of target location.

Similar to Experiment 1, the same four natural scenes were presented within a block of trials. However, though the spatial cue changed directions from trial to trial, it was necessary to place the same scene in the cued quadrant on each trial. One scene was assigned to be the “cued” scene and it always appeared in the quadrant to which the arrow pointed on a given trial. The position of the other three scenes was shuffled among the remaining quadrants.

Results

Visual search

The central arrow cue successfully induced a spatial bias toward the cued quadrant (Figure 7). Accuracy was higher when the target appeared in the cued rather than the uncued quadrants. This effect was significant in the biased phase, F(1, 31) = 42.17, p < .001, ηp2 = .58. It diminished, though remained significant, in the neutral phase, F(1, 31) = 13.98, p < .001, ηp2 = .31. An ANOVA that directly compared data across the two phases showed a significant interaction between the target’s quadrant and phase, F(1, 31) = 15.11, p < .001, ηp2 = .33. Thus endogenous spatial attention was induced in the biased phase and weakened in the neutral phase.

Figure 7.

Figure 7

Open in a new tab

Scene memory for unaware participants in Experiment 3. Horizontal dotted lines represent false alarm rate (classifying new scenes as old ones) for that phase. Error bars show +/− 1 S.E. of the mean.

Scene memory

The overall hit rate for old scenes was higher than the false alarm rate for new scenes, t(31) = 6.47, p < .001, suggesting that participants were able to perform both tasks at the same time (Figure 5). In the biased phase when spatial attention was directed to the cued quadrant, scenes presented in the cued quadrant were recognized better than scenes presented in the other quadrants, t(31) = 2.97, p < .01. As the spatial bias weakened in the neutral phase, the memory advantage for scenes in the cued quadrant became non-significant, t(31) = 1.52, p > .10. Similar to Experiment 1, the two-way ANOVA of phase and quadrant showed a significant main effect of quadrant, F(1, 31) = 10.37, p < .01. The interaction between quadrant and phase did not reach significance, F(1, 31) = 1.16, p > .20.

Figure 5.

Figure 5

Open in a new tab

Scene memory from Experiment 2. Horizontal dotted lines represent false alarm rate (classifying new scenes as old ones) for that phase. Error bars show +/− 1 S.E. of the mean.

Discussion

Experiment 2 replicated results from Experiment 1. Both experiments showed that endogenous spatial cuing in a visual search task yielded an advantage for scenes presented in the cued quadrant. As in Experiment 1, the spatial bias weakened in the neutral phase for visual search. The memory bias also weakened in the neutral phase. However, the existence of a significant memory effect during the biased phase is evidence that spread of attention between the two tasks occurred.

Experiment 3

Experiment 3 tests whether the spread of a spatial bias seen with endogenous cuing would also be observed with implicit location probability learning. In the location probability learning version of the search task, the target was presented more often in one quadrant of the screen. To promote implicit learning, subjects were not informed of the probability manipulation, and there was no demarcation between quadrants. Biasing a target’s location in space makes it possible that some participants would become aware of the location asymmetry on their own, resulting in goal-driven attentional guidance to the location believed to contain the target most often. Since this would confound the effects of endogenous guidance of attention and location probability learning, self-report questions following the experiment gauged participants’ awareness of the target location probabilities. Participants demonstrating awareness of the high-probability quadrant were excluded from the analyses.

Method

Participants

To ensure a high enough number of participants unaware of the high-probability locations, we increased participant numbers for Experiment 3. 48 new participants, 33 females and 15 males, completed Experiment 3. Their mean age was 20.

Design and Procedure

This experiment was similar to Experiment 1 except for the following changes. First, participants were not instructed to prioritize any region of space, nor were they informed of the target’s potential locations. Second, to introduce location probability learning, we manipulated the target’s location probability in the first 24 blocks (the “biased” phase). During this phase, the target appeared in one quadrant 50% of the time, and in each of the other three quadrants 16.7% of the time. The high-probability quadrant was randomly determined and counterbalanced across participants. Crucially, this quadrant was consistent during the entire biased phase, allowing participants to acquire location probability learning. In the next 16 blocks (the “neutral” phase), the target appeared in each quadrant 25% of the time.

For 16 of 48 participants, the neutral phase had 12 search trials per block as in Experiment 1. The other 32 participants in Experiment 3 had the same block structure as Experiment 2: 16 search trials in all blocks. The difference in trial number didn’t lead to significant differences in performance, so data were combined across all 48 participants.

Results

To examine location probability learning separately from goal-driven attention, we divided participants into two groups: one group that was ‘aware’ of the target location probability structure and one group that was ‘unaware’. Two self-report questions determined the groups: the first asked if the search target occurred equally often in all locations, and the second informed participants that the target occurred in some regions more often than others and asked them to choose one quadrant as the higher-probability region. We considered participants aware, and therefore likely to have guided attention explicitly, only if they answered both questions correctly. Answering only the first correctly would suggest that participants did not have the correct awareness and may have answered “yes” due to demand characteristics; answering only the second correctly would suggest that participants were able to retroactively identify the high-probability quadrant, but were not using that knowledge actively prior to being told a high-probability region existed.

26 of the 48 participants answered at least one question incorrectly, and were thus considered unaware. The remaining 22 participants answered both questions correctly and were considered aware of the target probability structure. All analyses reported here consider only unaware participants who likely show implicit location probability learning rather than endogenous attentional guidance; results from aware participants are in the Appendix, and bear strong resemblance to results of the endogenous cuing of Experiments 1 and 2.

Visual search

Replicating previous studies, the target’s uneven location probability induced a strong spatial bias toward the high-probability quadrant. Unaware participants found the target more accurately when it appeared in the high-probability quadrant rather than the other quadrants (Figure 6). Search accuracy was higher in the high probability quadrant in the biased phase, F(1, 25) = 30.28, p < .001, ηp2 = .55, and in the neutral phase, F(1, 25) = 11.39, p < .01, ηp2 = .31. Comparing the biased and neutral phases for unaware participants showed a main effect of quadrant, F(1, 25) = 23.57, p < .001, ηp2 = .49, and an interaction between quadrant and phase, F(1, 25) = 7.64, p < .05, ηp2 = .23, suggesting that the accuracy advantage decreased in the neutral phase.

Figure 6.

Figure 6

Open in a new tab

Visual search accuracy for unaware participants from Experiment 3. During the biased phase, the target occurred in a single high probability quadrant on 50% of trials (17% for each other quadrant). During the neutral phase, the target occurred equally often in each quadrant. Error bars show +/− 1 S.E. of the mean.

Scene memory

Participants were able to recognize the scenes at above-chance levels. The overall hit rate for old scenes was higher than the false alarm rate for new scenes for unaware participants, t(25) = 4.61, p < .001. Implicit cuing participants showed a different pattern of memory results than those in the endogenous cuing experiments (Figure 7). In the biased phase, there was no accuracy difference for scenes based on quadrant, t(25) = .45, p > .5. In the neutral phase, accuracy was significantly greater in the low probability quadrants than the high-probability quadrant, t(25) = 2.26, p < .05. Analyzing the two phases together yielded no significant main effect of quadrant, F(1, 25) = 2.33, p > .1. The phase by quadrant interaction was not significant, F(1, 25) = 3.05, p = .09.

Discussion

Participants in Experiment 3 developed a preference for the high-probability quadrant during visual search. This preference persisted into the neutral phase as in other research (Jiang, Swallow, Rosenbaum, et al., 2013). However, this selection bias did not extend to scene memory for participants not reporting awareness of the high-probability target region. No memory advantage for the high-probability quadrant emerged in the biased phase, with accuracy regardless of quadrant being nearly identical (54% in the low-probability quadrants versus 55% in the high-probability quadrant). This is true even though the search advantage from arrow cuing (Experiment 2) is similar to that from location probability cuing (Experiment 3), t(56) = 1.25, p > .20. Results from Experiment 3 suggest that there is little to no spread of attention when location probability learning occurs implicitly. This supports the dual-system view of attention: location probability learning relies on different mechanisms than does endogenous guidance.

Combined Analyses

In three experiments, we found that goal-driven spatial attention for one task spread to a secondary task, while implicit location probability learning did not. To investigate this difference more directly, we ran additional analyses combining data from the endogenous cuing experiments (1 and 2) and comparing them to those of the unaware probability cuing group. We did not consider aware participants in these analyses, as the search advantage of aware participants would be affected both by implicit location probability learning (before participants became aware of the bias) and goal-driven attentional guidance (once they became aware).

First, we compared the strength of the memory advantage for scenes in the cued quadrant relative to uncued quadrants between experiments (Figure 8a). Using data from the scene memory task in the biased phase, we entered quadrant condition (visual search high- vs. low- probability) as a within-subject factor and type of cuing (endogenous cuing of Experiments 1 and 2 vs. implicit cuing of Experiment 3) as a between-subject factor in an ANOVA. This analysis showed a significant interaction between quadrant condition and type of cuing, F(1, 72) = 4.22, p < .05, ηp2 = .06. Specifically, the memory advantage for the cued quadrant was stronger in endogenous cuing experiments than in unaware probability cuing. The fact that endogenous cuing led to a greater attentional bias in the memory task than did implicit location probability learning supports the claim that goal-driven guidance and implicit guidance rely on different mechanisms.

Figure 8.

Figure 8

Open in a new tab

Results from comparisons between incidental unaware participants and endogenous cuing participants (Experiments 1 and 2). A. Memory advantages for high-probability quadrants during the biased phases of unaware participants in probability cuing experiments and endogenous cuing participants, as measured by the memory accuracy for the cued quadrant minus the average memory accuracy for the uncued quadrant. Error bars show +/− 1 S.E. of the mean. B. Scatterplot of search advantage (search accuracy advantage for the high probability quadrant) and memory advantage (memory accuracy advantage for the high probability quadrant) for endogenous cuing participants during the biased phase (first 6 epochs). Filled dots are from Experiment 1, open dots from Experiment 2. Correlation line (r = .35) is also shown. C. Scatterplot of search advantage and memory advantage for 26 unaware probability cuing participants during the biased phase. Correlation line (r = −.02) is also shown.

Additionally, if the memory advantage for the high-probability quadrant is in fact an index of the spread of spatial attention from the search task, the strength of one participant’s memory advantage should correlate with the strength of the corresponding search advantage. We examined this effect during the biased phase for the endogenous cuing experiment (Figure 8b) and for unaware participants in Experiment 3 (Figure 8c). For endogenous cuing participants, the correlation between biased phase search and memory advantage was significant, r = .35, t(46) = 2.57, p < .02. This correlation was apparent even when data from Experiments 1 and 2 were separately analyzed: r = .45, t(14) = 1.92, p < 0.1, in Experiment 1 and r = .37, t(30) = 2.19, p < .05, in Experiment 2. In contrast, no significant correlation was found between search advantage and memory advantage in the biased phase for unaware participants, r = −0.02, t(24) = 0.11, p > .50. This pattern of results provides additional evidence that the memory advantage in endogenous cuing experiments reflects the spread of attention between tasks, and that this spread does not occur with location probability learning.

General Discussion

These experiments revealed important differences between endogenous, goal-driven attention and implicitly learned attentional biases in the spread of spatial attention between concurrent visual search and scene encoding tasks. Goal-driven attention enhances processing in a region of space for both tasks, as indicated by the spread of attention to the memory task in Experiments 1 and 2 (and for aware participants in Experiment 3, Appendix). On the other hand, probability cuing primarily affects selection in the relevant task: attentional biases to one region for visual search did not affect performance on the scene memory task. This suggests that these two forms of attention recruit different processes.

Could the spread of attention across tasks in the endogenous guidance experiments have occurred after the identification of the target, rather than when people respond to the spatial cue? On this account, spatial attention may dwell on the quadrant where the target is found, yielding a memory advantage. The results from our experiments, when considered together, do not support this view. First, since the display is presented briefly, by the time the search target is found the scenes are no longer present. Dwelling on the target quadrant is unlikely to provide any significant advantage for memory encoding. Second, and more critically, if dwelling contributes to a scene memory advantage, it should not be restricted to endogenous cuing. In the probability cuing experiment, the search target is more often found in the high-probability quadrant in the biased phase. Any dwelling should have facilitated memory for scenes presented in the high-probability quadrant. These results suggest that spreading of spatial selection occurs relatively early and is not a consequence of dwelling on the target’s quadrant.

The present results show that contemporary theories of spatial attention inadequately explain habitual attentional biases. Unitary priority map theories (Bisley & Goldberg, 2010; Itti & Koch, 2001) suggest that any effects of spatial attention are determined by top-down and bottom-up factors through the influence of a priority map. Similarly, the premotor theory of attention (Rizzolatti, Riggio, Dascola, & Umilta, 1987) proposes that all covert shifts of attention are subthreshold activations of the oculomotor system. Cowan’s (1988) theory of memory and attention argues that shifting attention to a region should improve memory for that region. In each case, implicit guidance and goal-driven guidance should have the same effects on scene memory, contrary to our results.

Unlike most theories of attention, the dual-system view of spatial attention explains these findings well (Jiang, Swallow, and Capistrano, 2013). In this view, goal-driven attention is a form of declarative attention that guides spatial attention through an explicit cue prior to appearance of the search array. In contrast, incidentally acquired spatial preferences alter habitual search behavior, a procedural form of attention that occurs while performing the search task. Its generalization therefore depends not on where objects are, but on whether two tasks share the same procedure of moving attention. Consistent with this idea, previous studies have shown that changing viewer perspective interfered with habitually learned attentional biases but not explicit search (Jiang & Swallow, 2014; Jiang, Swallow, & Capistrano, 2013), and furthermore that probability cuing transfers between visual search tasks, even ones that use different stimuli (Jiang et al., 2015).

Our results demonstrate the importance of incorporating habitual attention as an additional form of spatial attentional guidance. Some calls for incorporating explanations of habit into theoretical models of attentional guidance exist (Anderson, 2013; Awh et al., 2012; Wolfe & Horowitz, 2017), due to the growing evidence for the ability for environmental statistics and history of behavior to affect attentional guidance (Chun & Jiang, 1998; Geng & Behrmann, 2005; Jiang, Swallow, Rosenbaum, et al., 2013; Kunar, Flusberg, & Wolfe, 2008). For example, Awh and colleagues (2012) suggest that selection history is another way of affecting the priority map, adding a third category of influence alongside goal-driven controls and bottom-up salience. Our interpretation goes beyond some of these recent proposals. The present results demonstrate that habitual attention does not influence spatial attentional guidance with the same ultimate effect on attentional shifts; it guides attention in a way that has different effects once it rests on a region than does either goal-driven or salience-driven attention.

The difference between habitual and endogenous guidance of attention may involve the relationship of each to the process of segmenting the search array from overlapping visual information irrelevant to search. Such segmentation occurs preattentively, with search behavior restricted to the segmented search array for the duration of search (Wolfe et al., 2002). Because endogenous cuing engages spatial attention “off-line”, prior to stimulus onset, a task-general priority map is already deployed prior to the onset of the array (Egeth & Yantis, 1997; Kastner et al., 1999; Luck et al., 1997). This facilitates encoding of scenes in the cued quadrant, resulting in better memory for those scenes. Habitual attentional biases, on the other hand, act within the search process—that is, after the preattentive segmentation of scenes from background. Targets in the high-probability locations are more quickly found because people have acquired a search habit to shift attention in that direction, but the spread of attention to scenes is limited because implicit learning shifts spatial attention among the already segmented search elements.

Existing empirical evidence for the difference between goal-driven and habitual attention is primarily based in research on target probability learning, including both location probability learning and feature probability learning. Such habitual guidance has been shown to differ from endogenous guidance in a variety of ways including temporal profile, decline in aging, and resilience to high working memory load (Jiang et al., 2015; Jiang, Sigstad, et al., 2013; Twedell et al., 2016; Won & Jiang, 2015). Further investigation into potential differences between habitual guidance of attention and other forms of guidance is needed to verify that they implement spatial attention differently, rather than merely recruit the same spatial attentional process through different causes.

Overall, our endogenous cuing results fit with the large literature on multitasking in demonstrating a limit in people’s ability to exert parallel and independent attentional controls (Marti, Sigman, & Dehaene, 2012; Pashler, 1994). Although the data on endogenous cuing support the idea that humans have limited abilities in controlling spatial attention differently across two tasks, data from probability cuing suggest that not all forms of spatial attention are task-general. This finding is consistent with the broader literature on implicit learning. Adding secondary tasks typically interferes with explicit, but not implicit, learning (Hayes & Broadbent, 1988; Waldron & Ashby, 2001; Zeithamova & Maddox, 2006). More generally, incidentally learned attention is less likely to interfere with a secondary task, making management of various strategies between tasks more realistic. This finding is both useful for informing theories of goal-driven and habitual attentional control and for developing applications of attention research in everyday settings.

Conclusion

The present study has demonstrated, in a novel paradigm, an important difference between spatial attention driven by top-down goals and incidentally acquired spatial preferences. Whereas top-down modulation improves both search performance and search-irrelevant memory for spatially proximate scenes, implicit learning improves only search performance. Based on this, we have argued for a dual-system theory of spatial attention that dissociates habitual attention from other forms of attention. Future research may elaborate on this theory by identifying specific cognitive and neural mechanisms uniquely supporting habitual spatial attention.

Figure 4.

Figure 4

Open in a new tab

Visual search accuracy from Experiment 2. A central arrow instructed (cued) participants to prioritize one quadrant. The target was frequently in the cued quadrant (which changed from trial to trial) during the biased phase and was equally likely to be in each quadrant during the neutral phase. Error bars show +/− 1 S.E. of the mean.

Public Significance Statement.

Spatial attention can be guided either explicitly through task instructions or implicitly through making search targets more probable in one region of a visual display. Although the means of guiding attention differ, most theories suggest that the form of guidance should not alter the effects of spatial attention. Here we show that the two means of shifting attention are dissociable in their effects on a secondary task. Whereas explicit guidance affects all concurrent tasks, implicit guidance is more easily contained to a given task. In light of our results, we argue that contemporary models of spatial attention should be revised. These findings have implications for how people deploy attention in tasks such as driving, spatial navigation, and team sports.

Appendix

Results of Aware Participants in Location Probability Learning

Not all participants in Experiment 3 were oblivious to the manipulation in target location probability; 22 of 48 participants correctly identified that the target had a high-probability region and what that region was. Here we report results for these participants. The experimental design is identical to Experiment 3 described above; the only difference is in the participants’ responses to self-report questions. Note that results reported here are highly similar to the results of endogenous cuing in Experiments 1 and 2.

Visual Search

A search advantage for the high probability quadrant developed for aware participants (Figure A1). This was the case in the biased phase, F(1, 21) = 46.01, p < .001, ηp2 = .67, and persisted in the neutral phase, F(1, 21) = 21.01, p < .001, ηp2 = .50. A direct comparison between the biased and neutral phases revealed a significant main effect of the target’s quadrant, F(1, 21) = 38.44, p < .001, ηp2 = .6467, without a phase by quadrant interaction, F = 1.44, p > .2.

Scene Memory

Aware participants from Experiment 3 were able to perform the memory task at above-chance levels (Figure A2). The overall hit rate for old scenes was higher than the false alarm rate for aware participants, t(21) = 7.82, p < .001. Furthermore, scene memory was significantly better for scenes presented in the high-probability quadrant than the low-probability quadrants. This was the case in the biased phase, t(21) = 2.54, p < .02, but not in the neutral phase, t(21) = 0.64, p > .5. Combining data across the two phases, we observed a significant effect of quadrant, F(1, 21) = 4.71, p < .05, ηp2 = .1832, and no significant interaction between phase and quadrant, F(1, 21) = 1.81, p > .1.

Figure A1.

Figure A1

Open in a new tab

Visual search accuracy for aware participants in Experiment 3. Error bars show +/− 1 S.E. of the mean.

Figure A2.

Figure A2

Open in a new tab

Scene memory for aware participants in Experiment 3. Horizontal dotted lines represent false alarm rate (classifying new scenes as old ones) for that phase. Error bars show +/− 1 S.E. of the mean.

Footnotes

1

Note that we use “spread” as a general term to describe the processing of stimuli not relevant to the task for which spatial attention is allocated. Our study does not directly test whether the spread constitutes a slippage of spatial attention (e.g., owing to shared visual properties; Lachter, Forster, & Ruthruff [2004]) or a leakage of attention (e.g., owing to failure of executive control).

References

  1. Anderson BA. A value-driven mechanism of attentional selection. Journal of Vision. 2013;13(2013):1–16. doi: 10.1167/13.3.7. http://doi.org/10.1167/13.3.7.doi. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Awh E, Belopolsky AV, Theeuwes J. Top-down versus bottom-up attentional control: A failed theoretical dichotomy. Trends in Cognitive Sciences. 2012;16(8):437–443. doi: 10.1016/j.tics.2012.06.010. http://doi.org/10.1016/j.tics.2012.06.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Awh E, Jonides J. Overlapping mechanisms of attention and spatial working memory. Trends in Cognitive Sciences. 2001;5(3):119–126. doi: 10.1016/s1364-6613(00)01593-x. http://doi.org/10.1016/S1364-6613(00)01593-X. [DOI] [PubMed] [Google Scholar]
  4. Awh E, Jonides J, Reuter-Lorenz PA. Rehearsal in spatial working memory. Journal of Experimental Psychology: Human Perception & Performance. 1998;24(3):780–790. doi: 10.1037//0096-1523.24.3.780. http://doi.org/http://dx.doi.org/10.1037/0096-1523.24.3.780. [DOI] [PubMed] [Google Scholar]
  5. Bisley JW, Goldberg ME. Attention, intention, and priority in the parietal lobe. Annual Review of Neuroscience. 2010;33:1–21. doi: 10.1146/annurev-neuro-060909-152823. http://doi.org/10.1146/annurev-neuro-060909-152823. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Brainard DH. The psychophysics toolbox. Spatial Vision. 1997;10:433–436. http://doi.org/10.1163/156856897X00357. [PubMed] [Google Scholar]
  7. Burnett KE, Close AC, D’Avossa G, Sapir A. Spatial attention can be biased towards an expected dimension. The Quarterly Journal of Experimental Psychology. 2016;69(11):2218–2232. doi: 10.1080/17470218.2015.1111916. http://doi.org/10.1080/17470218.2015.1111916. [DOI] [PubMed] [Google Scholar]
  8. Carrasco M, Evert DL, Chang I, Katz SM. The eccentricity effect: Target eccentricity affects performance on conjunction searches. Perception & Psychophysics. 1995;57(8):1241–1261. doi: 10.3758/bf03208380. http://doi.org/10.3758/BF03208380. [DOI] [PubMed] [Google Scholar]
  9. Chun MM, Jiang Y. Contextual cueing: Implicit learning and memory of visual context guides spatial attention. Cognitive Psychology. 1998;36(1):28–71. doi: 10.1006/cogp.1998.0681. http://doi.org/10.1006/cogp.1998.0681. [DOI] [PubMed] [Google Scholar]
  10. Cowan N. Evolving conceptions of memory storage, selective attention, and their mutual constraints within the human information-processing system. Psychological Bulletin. 1988;104(2):163–191. doi: 10.1037/0033-2909.104.2.163. http://doi.org/http://dx.doi.org/10.1037/0033-2909.104.2.163. [DOI] [PubMed] [Google Scholar]
  11. Cowan N, Elliott EM, Saults JS, Morey CC, Mattox S, Conway ARA. On the capacity of attention: Its estimation and its role in working memory and cognitive aptitudes. Cognitive Psychology. 2005;51(1):42–100. doi: 10.1016/j.cogpsych.2004.12.001. http://doi.org/10.1016/j.cogpsych.2004.12.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Duncan J. Cooperating brain systems in selective perception and action. In: Toshio I, McClelland JL, editors. Attention and Performance XVI: Information Integration in Perception and Communication. Boston, MA: The MIT Press; 1996. pp. 549–578. [Google Scholar]
  13. Egeth HE, Yantis S. Visual attention: Control, representation, and time course. Annual Review of Psychology. 1997;48:269–297. doi: 10.1146/annurev.psych.48.1.269. http://doi.org/10.1146/annurev.psych.48.1.269. [DOI] [PubMed] [Google Scholar]
  14. Fecteau JH, Munoz DP. Salience, relevance, and firing: A priority map for target selection. Trends in Cognitive Sciences. 2006;10(8):382–390. doi: 10.1016/j.tics.2006.06.011. http://doi.org/10.1016/j.tics.2006.06.011. [DOI] [PubMed] [Google Scholar]
  15. Folk CL, Remington RW, Johnston JC. Involuntary covert orienting is contingent on attentional control settings. Journal of Experimental Psychology: Human Perception and Performance. 1992:1030–1044. [PubMed] [Google Scholar]
  16. Geng JJ, Behrmann M. Spatial probability as an attentional cue in visual search. Perception & Psychophysics. 2005;67(7):1252–1268. doi: 10.3758/bf03193557. http://doi.org/10.3758/BF03193557. [DOI] [PubMed] [Google Scholar]
  17. Hayes NA, Broadbent DE. Two modes of learning for interactive tasks. Cognition. 1988;28(3):249–276. doi: 10.1016/0010-0277(88)90015-7. http://doi.org/10.1016/0010-0277(88)90015-7. [DOI] [PubMed] [Google Scholar]
  18. Hopfinger JB, Buonocore MH, Mangun GR. The neural mechanisms of top- down attentional control. Nature Neuroscience. 2000;3(3):284–291. doi: 10.1038/72999. http://doi.org/doi:10.1038/72999. [DOI] [PubMed] [Google Scholar]
  19. Itti L, Koch C. Computational modelling of visual attention. Nature Reviews Neuroscience. 2001;2(3):194–203. doi: 10.1038/35058500. http://doi.org/10.1038/35058500. [DOI] [PubMed] [Google Scholar]
  20. Jiang YV, Sha LZ, Remington RW. Modulation of spatial attention by goals, statistical learning, and monetary reward. Attention, Perception, & Psychophysics. 2015;77(7):2189–2206. doi: 10.3758/s13414-015-0952-z. http://doi.org/10.3758/s13414-015-0952-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Jiang YV, Swallow KM, Rosenbaum GM, Herzig C. Rapid acquisition but slow extenction of an attentional bias in space. Journal of Experimental Psychology: Human Perception and Performance. 2013;39(1):87–99. doi: 10.1037/a0027611. http://doi.org/10.1037/a0027611. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jiang YV, Sigstad HM, Swallow KM. The time course of attentional deployment in contextual cueing. Psychonomic Bulletin & Review. 2013;20(2):282–288. doi: 10.3758/s13423-012-0338-3. http://doi.org/10.3758/s13423-012-0338-3. [DOI] [PubMed] [Google Scholar]
  23. Jiang YV, Swallow KM. Changing viewer perspectives reveals constraints to implicit visual statistical learning. Journal of Vision. 2014;14(12):1–16. doi: 10.1167/14.12.3. http://doi.org/10.1167/14.12.3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Jiang YV, Swallow KM, Capistrano CG. Visual search and location probability learning from variable perspectives. Journal of Vision. 2013;13(6):1–13. doi: 10.1167/13.6.13. http://doi.org/10.1167/13.6.13. [DOI] [PubMed] [Google Scholar]
  25. Kastner S, Pinsk MA, De Weerd P, Desimone R, Ungerleider LG. Increased activity in human visual cortex during directed attention in the absence of visual stimulation. Neuron. 1999;22(4):751–61. doi: 10.1016/s0896-6273(00)80734-5. http://doi.org/http://dx.doi.org/10.1016/S0896-6273(00)80734-5. [DOI] [PubMed] [Google Scholar]
  26. Kunar MA, Flusberg SJ, Wolfe JM. Time to guide: Evidence for delayed attentional guidance in contextual cuing. Visual Cognition. 2008;16(6):804–825. doi: 10.1080/13506280701751224. http://doi.org/10.1080/13506280701751224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lachter J, Forster KI, Ruthruff E. Forty-five years after Broadbent (1958): Still no identification without attention. Psychological Review. 2004;111(4):880–913. doi: 10.1037/0033-295X.111.4.880. http://doi.org/10.1037/0033-295X.111.4.880. [DOI] [PubMed] [Google Scholar]
  28. Luck SJ, Chelazzi L, Hillyard SA, Desimone R. Neural mechanisms of spatial selective attention in areas V1, V2, and V4 of macaque visual cortex. Journal of Neurophysiology. 1997;77(1):24–42. doi: 10.1152/jn.1997.77.1.24. [DOI] [PubMed] [Google Scholar]
  29. Marti S, Sigman M, Dehaene S. A shared cortical bottleneck underlying attentional blink and psychological refractory period. NeuroImage. 2012;59(3):2883–2898. doi: 10.1016/j.neuroimage.2011.09.063. http://doi.org/10.1016/j.neuroimage.2011.09.063. [DOI] [PubMed] [Google Scholar]
  30. Monsell S. Task switching. Trends in Cognitive Sciences. 2003;7(3):134–140. doi: 10.1016/s1364-6613(03)00028-7. http://doi.org/10.1016/S1364-6613(03)00028-7. [DOI] [PubMed] [Google Scholar]
  31. Nobre AC, Kastner S, editors. The Oxford Handbook of Attention. Oxford: Oxford University Press; 2014. [Google Scholar]
  32. Pashler H. Dual-task interference in simple tasks: Data and theory. Psychological Bulletin. 1994;116(2):220–244. doi: 10.1037/0033-2909.116.2.220. http://doi.org/10.1037/0033-2909.116.2.220. [DOI] [PubMed] [Google Scholar]
  33. Pelli DG. The VideoToolbox software for visual psychophysics: transforming numbers into movies. Spatial Vision. 1997;10(4):437–442. http://doi.org/10.1163/156856897X00366. [PubMed] [Google Scholar]
  34. Posner MI. Orienting of attention. The Quarterly Journal of Experimental Psychology. 1980;32(1):3–25. doi: 10.1080/00335558008248231. http://doi.org/10.1080/00335558008248231. [DOI] [PubMed] [Google Scholar]
  35. Rizzolatti G, Riggio L, Dascola I, Umilta C. Reorienting attention across the horizontal and vertical meridians: evidence in favor of a premotor theory of attention. Neuropsychologia. 1987;25(1A):31–40. doi: 10.1016/0028-3932(87)90041-8. http://doi.org/https://doi.org/10.1016/0028-3932(87)90041-8. [DOI] [PubMed] [Google Scholar]
  36. Spence C, Driver J. Audiovisual links in endogenous covert spatial attention. Journal of Experimental Psychology: Human Perception and Performance. 1996;22(4):1005–1030. doi: 10.1037//0096-1523.22.4.1005. http://doi.org/10.1037/0096-1523.22.4.1005. [DOI] [PubMed] [Google Scholar]
  37. Spence C, Pavini F, Driver J. Crossmodal links between vision and touch in covert endogenous spatial attention. Journal of Experimental Psychology: Human Perception and Performance. 2000;26(4):1298–1319. doi: 10.1037//0096-1523.26.4.1298. http://doi.org/10.3758/BF03194823. [DOI] [PubMed] [Google Scholar]
  38. Theeuwes J. Endogenous and exogenous control of visual selection. Perception. 1994;23(4):429–440. doi: 10.1068/p230429. http://doi.org/10.1068/p230429. [DOI] [PubMed] [Google Scholar]
  39. Twedell E, Koutstaal W, Jiang YV. Aging affects the balance between goal-guided and habitual spatial attention. Psychnomic Bulletin & Review. 2016:1–7. doi: 10.3758/s13423-016-1214-3. http://doi.org/10.3758/s13423-016-1214-3. [DOI] [PubMed]
  40. Waldron EM, Ashby FG. The effects of concurrent task interference on category learning: evidence for multiple category learning systems. Psychonomic Bulletin & Review. 2001;8(1):168–176. doi: 10.3758/bf03196154. http://doi.org/doi:10.3758/BF03196154. [DOI] [PubMed] [Google Scholar]
  41. Wolfe JM. Guided search 4.0: Current progress with a model of visual search. In: Gray WD, editor. Integrated Models of Cognitive Systems. Oxford: Oxford University Press; 2012. pp. 99–120. http://doi.org/10.1093/acprof:oso/9780195189193.003.0008. [Google Scholar]
  42. Wolfe JM, Horowitz TS. Five factors that guide attention in visual search. Nature Human Behaviour. 2017;1(3):58. doi: 10.1038/s41562-017-0058. http://doi.org/10.1038/s41562-017-0058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Wolfe JM, Oliva A, Horowitz TS, Butcher SJ, Bompas A. Segmentation of objects from backgrounds in visual search tasks. Vision Research. 2002;42(28):2985–3004. doi: 10.1016/s0042-6989(02)00388-7. http://doi.org/10.1016/S0042-6989(02)00388-7. [DOI] [PubMed] [Google Scholar]
  44. Won B, Jiang YV. Spatial working memory interferes with explicit, but not probabilistic cuing of spatial attention. Journal of Experimental Psychology: Learning, Memory and Cognition. 2015;41(3):1–30. doi: 10.1037/xlm0000040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Zeithamova D, Maddox WT. Dual-task interference in perceptual category learning. Memory & Cognition. 2006;34(2):387–398. doi: 10.3758/bf03193416. http://doi.org/10.3758/BF03193416. [DOI] [PubMed] [Google Scholar]

RESOURCES