Abstract

When engaged in dynamic visuospatial tasks, the brain copes with perceptual and cognitive processing challenges. During multiple-object tracking (MOT), the number of objects to be tracked (i.e., load) imposes attentional demands, but so does spatial interference from irrelevant objects (i.e., close encounters). Presently, it is not clear whether the effect of load on accuracy solely depends on the number of close encounters. If so, the same cognitive and physiological mechanisms deal with increasing load by preparing for and dealing with spatial interference. However, this has never been directly tested. Such knowledge is important to understand the neurophysiology of dynamic visual attention and resolve conflicting views within visual cognition concerning sources of capacity limitations. We varied the processing challenge in MOT task in two ways: the number of targets and the minimum spatial proximity between targets and distractors. In a first experiment, we measured task-induced pupil dilations and saccades during MOT. In a separate cohort, we measured fMRI activity. In both cohorts, increased load and close encounters (i.e., close spatial proximity) led to reduced accuracy in an additive manner. Load was associated with pupil dilations, whereas close encounters were not. Activity in dorsal attentional areas and frequency of saccades were proportionally larger both with higher levels of load and close encounters. Close encounters recruited additionally ventral attentional areas that may reflect orienting mechanisms. The activity in two brainstem nuclei, ventral tegmental area/substantia nigra and locus coeruleus, showed clearly dissociated patterns. Our results constitute convergent evidence indicating that different mechanisms underlie processing challenges due to load and object spacing.

INTRODUCTION

When faced with a dynamic visuospatial task that requires top–down control, that is, a task that cannot be solved by automatic processes alone, the logistics of attention become engaged. The brain continuously assesses the appropriate level of processing resources and optimally recruits the mechanisms that deal with current demands. In the case of multiple-object tracking (MOT; Pylyshyn & Storm, 1988), the brain flexibly handles the attentional challenge posed by the number of moving objects that should be tracked simultaneously (“load”; Meyerhoff, Papenmeier, & Huff, 2017; Scimeca & Franconeri, 2015). Tracking multiple objects requires dynamic attention and engages a frontoparietal brain network, which shows increased activity with increasing load (Alnæs et al., 2014; Jahn, Wendt, Lotze, Papenmeier, & Huff, 2012; Shim, Alvarez, Vickery, & Jiang, 2009; Tomasi, Ernst, Caparelli, & Chang, 2004; Culham, Cavanagh, & Kanwisher, 2001; Jovicich et al., 2001; Culham et al., 1998). This is consistent with findings that implicate this network in dealing with attentional demands across tasks and modalities (Ptak, 2012; Duncan, 2010; Corbetta & Shulman, 2002). In addition to this cortical system, activity in the noradrenergic systems positively correlates with the effort imposed by an increasing number of targets to track (Alnæs et al., 2014; Wright, Boot, & Morgan, 2013).

In addition to the number of targets, other task-specific factors influence accuracy in the MOT: the speed of the objects (Meyerhoff, Papenmeier, Jahn, & Huff, 2016; Feria, 2013; Tombu & Seiffert, 2008), the duration of tracking (Oksama & Hyönä, 2004), the distance the objects travel (Franconeri, Jonathan, & Scimeca, 2010), and the frequency of close encounters between targets and distractors (Vater, Kredel, & Hossner, 2017; Feria, 2013; Bae & Flombaum, 2012; Iordanescu, Grabowecky, & Suzuki, 2009; Franconeri, Lin, Pylyshyn, Fisher, & Enns, 2008; Shim, Alvarez, & Jiang, 2008; Alvarez & Franconeri, 2007; He, Cavanagh, & Intriligator, 1996). In particular, brief events of close encounters constitute moments of perceptual confusion and likely engage an effort-related system. However, the relation between the challenges posed by close encounters and by load to dynamic attention remains unclear.

A prominent hypothesis about what is the actual source of visuospatial limitations in MOT proposes that the number of close encounters (i.e., events with reduced interobject spacing) is the only limiting factor, because a greater number of targets, larger speed, and longer tracking time could be associated with higher chance for close encounters (Scimeca & Franconeri, 2015; Franconeri et al., 2010). Thus, attentional capacity may be limited by the ability of the attentional networks to keep mental representations of the target objects separate from each other and uninterrupted by distractors (Franconeri, Alvarez, & Cavanagh, 2013). A different view would be that load and close encounters constitute different sources of challenge to dynamic attention, and distinct brain systems would be engaged in keeping track of the targets and in dealing with close encounters. Several findings in MOT are indicative of this. Higher load represents the need to initially encode and then sustain a large number of objects in working memory (Lapierre, Cropper, & Howe, 2017; Allen, McGeorge, Pearson, & Milne, 2006) and may recruit a general attentional resource related to arousal (Kahneman, 1973). Close encounters are dealt with by dynamically increasing the local attentional resolution for targets requiring higher precision processing (Meyerhoff et al., 2016; Srivastava & Vul, 2016; Iordanescu et al., 2009). The brain can achieve this by either altering the spatial distribution of attention covertly or by directing the eyes to gain higher spatial resolution in retinotopic areas undergoing close encounters (e.g., rescue saccades; Meyerhoff, Schwan, & Huff, 2018; Zelinsky & Todor, 2010). In summary, the capacity limitations to dynamic attention appear to be, in addition to keeping a mental representation of the targets, the ability to flexibly allocate resources under events of close proximity (Vul, Alvarez, Tenenbaum, & Black, 2009), but this has never been tested directly.

Here, we examined whether dynamically tracking few or many targets is equivalent behaviorally and physiologically when perceptual limitations (the number of close encounters) are controlled for. For this, in addition to the load manipulation (two, three, and four targets), we controlled the minimum interobject spacing within the trials. In trials of low object spacing (i.e., more difficult), we counted instances of close encounters between targets and distractors. The number of close encounters was equalized across load levels; that is, there was the same number of close encounters at each load level. In trials of high object spacing, we expected little confusion of targets by distractors. This allowed us to test whether spatial interference is the actual source of load effects during MOT. To evaluate whether specific physiological responses were associated with each type of challenge, we leveraged ocular measures in a first experiment and brain activity in a separate experiment with independent cohorts. The two experiments provide complementary information about the brain mechanisms that deal with tracking difficulty. In the fMRI experiment, we measured whole-brain activity and directly assessed the networks associated to the manipulations. In the eye-tracking experiment, we measured the pupil size and the frequency of saccades during tracking. The brainstem locus coeruleus–norepinephrine (LC-NE) system has been strongly tied to pupil dilations, effort, and arousal and has been suggested to act as an attentional filter that selects for the temporal occurrence of relevant stimuli (Joshi, Li, Kalwani, & Gold, 2016; Aston-Jones & Cohen, 2005). In MOT, the LC-NE system is recruited with higher load (Alnæs et al., 2014). A still open question is whether instances of close encounters during tracking also activate this system. In addition to NE, other neuromodulatory systems are involved in effortful attention (Petersen & Posner, 2012). Recent research reveals that dopamine (DA; another cathecolamine produced in the brainstem nuclei ventral tegmental area [VTA] and substantia nigra [SN]) may be implicated not only in executive/attentional control but also in spatial attention (Thiele & Bellgrove, 2018; Noudoost & Moore, 2011) and effort (Husain & Roiser, 2018; Pessiglione, Le Bouc, & Vinckier, 2018; Westbrook & Braver, 2015, 2016; Varazzani, San-Galli, Gilardeau, & Bouret, 2015). Differences in the recruitment of NE and DA effort-related systems may lead to individual differences in capacity (Unsworth & Robison, 2015). Differences in capacity have been linked to interindividual variability in performance in MOT (Störmer, Li, Heekeren, & Lindenberger, 2013; Drew & Vogel, 2008; Oksama & Hyönä, 2004). Assessing the activity in the LC-NE and VTA/SN-DA systems allowed us to evaluate the dynamic allocation of attention during MOT due to increasing load, number of close encounters, and tracking capacity.

In summary, we pursued the question of how the brain deals with challenges imposed by load and close encounters. In particular, if load effects are to be accounted for solely by close encounters, there should be no effect of the load manipulation when the number of close encounters is kept constant (Figure 1A). Furthermore, trials with higher number of close encounter instances should elicit larger pupil dilations and brain frontoparietal activation, but there should be no effect of load on these measures when the number of close encounters is constant. Alternatively, if load and close encounters impose at least partially distinct limitations to attention, we would expect performance to drop with load even though the number of close encounters is held constant (Figure 1B, C). In this case, we would also expect both load and number of close encounters to activate frontoparietal top–down networks, because both challenges should elicit increased attentional engagement, but given distinct mechanisms associated with each challenge, the patterns of brain activation should not overlap completely. For example, we would expect that the manipulation of object spacing should promote differences in saccade frequency and FEF activity. In contrast, because load may be particularly associated with effort, arousal, and the NE neuromodulatory system, we would expect that the load manipulation should generate differences in pupil dilation and LC activity.

Figure 1. 

Hypothesized effects of load and number of close encounters on accuracy. If the amount of encounters (i.e., events of close interobject spacing between a target and a distractor) is held constant across load levels, the predicted patterns differ whether load and close encounters rely on the same or different neural mechanisms: (A) expected pattern for the load effect being given by the amount of close encounters; (B) expected pattern for a model with independent, additive effects of load and close encounters; (C) expected pattern for a model with shared mechanisms. This would be expected because the two loading factors would act as “dual tasks” (Strobach, Wendt, & Janczyk, 2018; Williges & Wierwille, 1979).

Figure 1. 

Hypothesized effects of load and number of close encounters on accuracy. If the amount of encounters (i.e., events of close interobject spacing between a target and a distractor) is held constant across load levels, the predicted patterns differ whether load and close encounters rely on the same or different neural mechanisms: (A) expected pattern for the load effect being given by the amount of close encounters; (B) expected pattern for a model with independent, additive effects of load and close encounters; (C) expected pattern for a model with shared mechanisms. This would be expected because the two loading factors would act as “dual tasks” (Strobach, Wendt, & Janczyk, 2018; Williges & Wierwille, 1979).

METHODS

MOT Task

In each trial of the MOT task (Figure 2), participants are presented with 10 objects (disks). Some of them were cued as targets by changing their color for a short period. Afterward, these targets acquired the same appearance as the other objects (or distractors), and then all the disks started to move independently in random directions. The participant's task was to track the targets as the objects moved around and, at the end of the trial, to indicate which objects in the display belonged to the target set. We parametrically varied the number of target objects or cognitive load so that the participants should correspondingly adjust the required degree of attentional effort. In the present experiments, we used three levels of load, with trials of two, three, or four objects to be tracked. In addition, we included two levels of minimum interobject spacing for each of the load levels. A close encounter was defined as an instance where one distractor approached a target within a certain close range distance (see description below). This proximity increases confusion between targets and distractors and therefore increases the risk of swapping the target with a distractor or losing the target (Drew, Horowitz, & Vogel, 2013).

Figure 2. 

MOT task. In the assignment period of a trial (A), 10 black disks are presented. Then, a number of them (the targets) is highlighted in red, so as to obtain different load levels (i.e., 2, 3, or 4). Then, all objects turn blue and start moving (B). Participants are instructed to track the target objects while they move. At the end of the tracking period (10 sec), participants indicate the targets by clicking on them with the mouse (C). Close encounters (i.e., spatial interference) were varied as the minimum distance at which distractors approached targets: reduced interobject spacing (>15 pixels) rendered close encounters, and large interobject spacing (>80 pixels) rendered far encounters.

Figure 2. 

MOT task. In the assignment period of a trial (A), 10 black disks are presented. Then, a number of them (the targets) is highlighted in red, so as to obtain different load levels (i.e., 2, 3, or 4). Then, all objects turn blue and start moving (B). Participants are instructed to track the target objects while they move. At the end of the tracking period (10 sec), participants indicate the targets by clicking on them with the mouse (C). Close encounters (i.e., spatial interference) were varied as the minimum distance at which distractors approached targets: reduced interobject spacing (>15 pixels) rendered close encounters, and large interobject spacing (>80 pixels) rendered far encounters.

The size of the display was 800 × 800 pixels (20.9 × 20.9° of visual angle). Objects had a radius of 15 pixels (0.4°). Trial duration was 10 sec. The displays were redrawn at a rate of 30 frames per second, and objects moved with a speed of 10 pixels per frame (0.27°). The objects changed directions at random time points or when colliding with the display perimeter. Reflection angles did not follow Newtonian physics. None of the objects overlapped. We generated a large pool of candidate stimuli with close and far encounters, from which we selected trials that were balanced on number of close encounters across load levels. In the trials with far encounters, the distance between targets and distractors was always larger than 80 pixels (2.1°; Figure 2D). In the trials with close encounters, we allowed the distance between targets and distractors to be as short as 15 pixels (0.4°; Figure 2D). Close encounters in those trials were thus defined as situations where a distractor traveled within 15 and 60 pixels from a target (0.4–1.6°) and kept a distance of less than 80 pixels for no more than 10 frames (333 msec). Thus, close encounters were brief to ensure that some close encounters would not last over several seconds. Each trial in the low interobject spacing set had exactly 13 close encounters, irrespective of load. In addition, it was ensured that each target had at least one close encounter with a distractor, but not more frequently than the 75th percentile of close encounter distributions in the pool of candidate trials. For the “far” encounters, we defined encounters as starting when a distractor approached a target closer than 100 pixels and ended when getting further away than 120 pixels. Objects did not get closer than 80 pixels. In selecting trials we ensured that the total number of long-range close encounters per trial would be less than the 75th percentile of such long-range close encounters in the pool of candidate trials.

In summary, we had a mixed design with factors load (three levels) and close encounters (two levels), making, in total, six different conditions.

Eye-tracking Study

Participants

Forty-one adults (28 women) were recruited through social media to participate in the study (mean age = 24.2 years, range = 19–40 years). They were compensated with a gift card equivalent to 100 Norwegian Kroner per hour, and each session took approximately 1 hr. The study was conducted according to institutional guidelines and was approved by the local ethics committee (institutional review board).

Experimental Procedure

The MOT task was written in code and delivered using custom software written in JavaScript. The participants read the instructions and performed a training session on a computer (10 trials). Afterwards, the full experiment was run in the lab. The total number of objects in the display was kept constant at 10. Each trial consisted of the presentation of the 10 objects in blue for 2 sec. Then, the target disks turned red for 2.5 sec before turning back to blue. In the passive viewing condition, no ball turned red. The disks started to move, and the tracking period lasted 10 sec (tracking trials) or 7 sec (passive viewing trials). At the end of the tracking period all objects' movement stopped. Participants had to click on all the disks that were originally selected to be tracked. Participants clicked with the left button if they had high confidence on their response and with the right button if they had low confidence. Response accuracy and time to start responding were recorded for each trial.

Pupillometry Analysis

Blinks were removed and interpolated before hampel and lowess filtering was applied. For each blink detected by the SMI software, we linearly interpolated from five samples before onset until five samples after offset. Trials were baseline-corrected using a baseline of 500 msec before the beginning of the tracking period. The pupillometry data were analyzed in the time region 2633 to 12,000 msec (determined by Monte Carlo analysis on load levels). To further explore whether the effect of close encounters would become evident by the end of the trial, we analyzed the pupil change in two intervals, one early (2633–4633 msec) and one late (10,000–12,000 msec) within the trial. Saccades were detected with the SMI software and extracted during the tracking interval (2000–12,000 msec).

fMRI Study

Participants

Forty adults (25 women) were recruited through social media to participate in the study (mean age = 25 years, range = 19–37 years). They were compensated with a gift card equivalent to 100 Norwegian Kroner per hour, and each session took approximately 2 hr.

Before starting testing, participants were asked to answer a brief questionnaire to assess whether they passed the inclusion criteria (no serious neurological or psychiatric illness, vision or language problems, etc.). Participants with an average proportion of accuracy of <.6 were removed since this level of performance approaches chance level. The study was conducted according to institutional guidelines and was approved by the local ethics committee.

Experimental Procedure

The MOT task included the same stimulus set as the eye-tracking study and was delivered using E-Prime 2. The participants read the instructions and performed a training session before the fMRI session (10 trials). Different from the eye-tracking study, in the fMRI study we used partial report. At the end of the tracking period, all objects' movement stopped, and one of the objects was highlighted with a yellow circle. The participant responded whether the probed object was one of the tracked objects. The response window was 2 sec, after which the trial ended and was followed by a 4-sec intertrial interval that consisted of a fixation cross. Response accuracy and RT were recorded for each trial. The probability that the probed object was one of the targets was 50% for all tracking conditions.

fMRI Acquisition and Analysis

Participants were scanned in a 3-T Philips MRI scanner at Rikshospitalet, Oslo. Each scanning session started with an anatomical scan (0.5 mm3). Four runs of functional images were acquired while participants performed the MOT task. One run consisted of 24 trials in two blocks of 12 trials. The trials with different levels of load and number of close encounters were semirandomized: Each block contained three trials of each load level in random order. A rest period of 20 sec always followed after each block of trials. Whole-brain functional images were acquired using a spin-echo EPI sequence sensitive to BOLD magnetic susceptibility (repetition time = 2208 msec, flip angle = 90°, number of slices: 42, voxel size = 3 mm3). Each functional scan lasted about 8 min, and in each scan, 234 volumes were collected. Finally, a neuromelanin-sensitive scan was acquired with the objective of further improving the localization of the LC. The slices for this scan were set to cover the brainstem (T1-TSE, repetition time = 600 msec, echo time = 14 msec, voxel dimension = 0.4 × 0.49 × 3 mm, flip angle = 90°, number of slices: 10). This scan lasted 12 min.

The stimuli were projected onto a screen positioned at the head end of the litter. Participants viewed the screen through a mirror placed on the head coil. Participants' response was given with the right hand through a joystick, with one button corresponding to “target” and another to “not a target” response.

The functional images of each participant were first visually inspected for anomalies and then submitted to a standard preprocessing pipeline using SPM 12 implemented on MATLAB (The MathWorks, Natick, MA). The data of one participant could not be used because of errors saving the images. Images were first corrected for time delays and realigned using six parameters of movement. The data were normalized to a standard template and smoothed (8 mm FWHM Gaussian kernel). For the analysis of the activity in the LC, a smaller Gaussian kernel was used (3 mm FWHM). Additional motion correction was performed by scrubbing the volumes with excessive movement using FSL functions.

Preprocessed images were submitted to a first-level analysis. Event-related activation was estimated with a general linear model. Stimulus presentation and tracking intervals of the different load and close encounters levels were modeled as separate events with a canonical hemodynamic response function. Images were high-pass filtered at 128 sec. Contrasts were generated for each tracking condition both collapsing across load levels and for each load level. Parameter estimates from each participant's general linear model were submitted to a second-level test. A full-factorial analysis was performed treating participants as a random factor and load and amount of close encounters as the fixed factors. The maps of Load 4 > Load 2 and close > far encounters were masked by the respective (load or level of close encounters) main effects. To obtain distinct areas related to load and close encounters, each of the former maps was masked by the other using thresholds of p = .05 and p = .001 for the exclusive mask and the contrast, respectively.

Given our interest in the brainstem, we separately assessed the activity in the VTA/SN and LC. For the VTA/SN, we performed a small volume correction in the contrasts of interest on a combined VTA/SN mask derived from probabilistic atlases of the structures (Murty et al., 2014). The VTA and SN masks partially overlap, and therefore, we combined them; the resulting mask was cropped to be spatially restricted to the brainstem. For the LC, given its small size and the variability of its location in the brainstem across individuals, we defined individual masks in the high-resolution T1-TSE images and used them for an ROI analysis. Specifically, the LC of each individual was delineated on the axial slices of the neuromelanin scans. The position of the nuclei was determined in the pons as the voxels of hyperintensity on either side of the fourth ventricle, following the procedure employed in previous studies (Krebs, Park, Bombeke, & Boehler, 2018). The T1-TSE scan of each individual was coregistered to the corresponding structural image, the structural was coregistered to the mean functional, and then the deformation field calculated during the normalization step was applied to the coregistered T1-TSE image. ROI analyses were performed on the individual masks of the LC. The data were extracted from the contrast images with rfxplot toolbox (Gläscher, 2009). The parameter estimates in the voxel of maximum effect size in each condition were extracted for the different load levels within the masks. Then, we performed repeated-measures (RM) ANOVA applying Load (levels: two, three, and four objects) and Encounters (far, close) as within-subject factors. For the comparison of groups, we included Group (high performers and low performers) as between-subject factor. For the comparison of task-related activity in the different groups in the VTA, a similar ROI analysis was applied on the VTA/SN combined mask.

Data Analysis: Behavior

Errors, RTs, confidence estimates, pupil change, number of saccades, and brainstem activity for each condition were compared with RM ANOVA, correcting for nonsphericity. Within-subject factors were Load (three levels) and Close encounters (two levels). Only correct trials were included in the RT analysis. For the analysis of the groups of performers, we divided the participants into high and low performers on the basis of a median split of total tracking accuracy. We performed RM ANOVA as before, adding Group as a between-subject factor. For this analysis, only the lowest and highest load levels were considered as to look at the largest effects. All effect sizes are reported as generalized eta squared (ηG2). The data were analyzed with IBM SPSS.

RESULTS

Eye-tracking Study

Behavior

Increasing load caused a moderate decrease in accuracy, F(1.92, 76.67) = 20.92, p < .001, ηG2 = .07 (Figure 3). Furthermore, close encounters also caused a decrease in accuracy, F(1, 40) = 510.56, p < .001, ηG2 = .54, and interacted weakly with load, F(1.68, 67.22) = 5.99, p = .006, ηG2 = .01. Confidence estimates were calculated for each trial as the percentage of responses that were given with the left mouse button (high confidence). An RM ANOVA on confidence estimates showed significant main effects of Load, F(1.39, 55.52) = 25.27, p < .0001, ηG2 = 7, and Close encounters, F(1, 40) = 50.64, p < .0001, ηG2 = .18, with lower confidence at higher difficulty in both effects. The interaction was not significant, F(1.69, 67.51) = 3.29, p = .051, ηG2 = .003.

Figure 3. 

Behavioral results of the eye-tracking study. Mean accuracy as a function of load for each level of close encounters.

Figure 3. 

Behavioral results of the eye-tracking study. Mean accuracy as a function of load for each level of close encounters.

Pupil Response

We analyzed the effects of Load and Close encounters on pupil dilation during tracking. We found a significant increase in pupil size with Load (main effect of Load: F(1.77, 70.86) = 57.6, p < .001, ηG2 = .04; Figure 4). Level of close encounters, F(1, 40) = 0.08, p = .78, and the interaction between Load and Level of close encounters, F(1.96, 78.25) = 1.26, p = .29, did not show significant effects.

Figure 4. 

Evoked pupil size during tracking for the different conditions. (A) Baseline-corrected change in pupil during tracking at the indicated conditions (far/close encounters). Shaded area indicates standard error. (B–D) Average pupil changes across the whole trial (A), an early interval (B), or a late interval (C) within a trial. Error bars indicate the SEM.

Figure 4. 

Evoked pupil size during tracking for the different conditions. (A) Baseline-corrected change in pupil during tracking at the indicated conditions (far/close encounters). Shaded area indicates standard error. (B–D) Average pupil changes across the whole trial (A), an early interval (B), or a late interval (C) within a trial. Error bars indicate the SEM.

When analyzing the pupil change early or late within the trial, the effect of Load was significant in both intervals (Early: F(1.93, 77.25) = 48.73, p < .001, ηG2 = .04; Late: F(1.83, 73.37) = 34.27, p < .001, ηG2 = .02; Figure 4), whereas the effect of Close encounters or the interaction was non-significant in both intervals (p > .28).

Saccade Frequencies

For this analysis, we removed two participants for having an unusual high number of saccades (more than an average of 25 saccades per trial). Both of the removed participants were classified as low performers. Number of saccades showed a significant effect of Load, F(1.59, 60.56) = 7.01, p = .004, ηG2 = .004, and Level of close encounters, F(1, 38) = 34.88, p < .001, ηG2 = .02. The interaction was weak but significant, F(2.00, 75.95) = 3.18, p = .047, ηG2 = .0006. On average, participants made 11.1 saccades in trials with close encounters and 9.2 saccades in trials with far encounters.

Individual Differences

We inspected the differences in behavior and ocular measures due to load and level of close encounters between high and low performers (Figure 5). The analysis of accuracy revealed a significant interaction between Group and Load, F(1.95, 75.94) = 3.85, p = .030, ηG2 = .03, and Group and Level of close encounters, F(1, 39) = 27.65, p < .001, ηG2 = .08, and three-way interaction, F(1.85, 72.2) = 10.69, p < .001, ηG2 = .04. In low level of close encounters, both groups had significantly lower accuracy with load (high performers: t(20) = 3.04, p = .006; low performers: t(19) = 9.56, p < .001). In high level of close encounters, high performers had lower accuracy with load, t(20) = 2.16, p = .043, whereas low performers showed no differences with load, t(19) = 1.15, p = .262.

Figure 5. 

Behavioral and pupillometry results for performance level groups. Mean accuracy (top) and mean pupil response (bottom) as a function of load for each of the levels of encounters and groups.

Figure 5. 

Behavioral and pupillometry results for performance level groups. Mean accuracy (top) and mean pupil response (bottom) as a function of load for each of the levels of encounters and groups.

The analysis of group differences on evoked pupil revealed a significant effect of Load, F(1.77, 69.18) = 56.24, p < .001, ηG2 = .04, and an interaction between performer Group, Load, and Difficulty, F(1.99, 77.45) = 5.17, p = .008, ηG2 = .003. In Load 4, low performers had significantly less pupil dilation in trials with close encounters compared with far encounters, t(19) = 3.31, p = .003, but not high performers (p = .31).

Next, we analyzed saccade frequencies between high and low performers. High performers made an average of 8.7 saccades per trial, whereas low performers made an average of 11.9 saccades per trial; this difference was however not statistically significant (t(25.9) = 1.5, p = .14).

Summary

In summary, load reduced accuracy even though the amount of spatial interference within the trials (close encounters) was equalized between the different load levels. Pupil size and saccades were sensitive to load, and saccades were also sensitive to the presence of close encounters within the trials. The individual differences showed that high performers were able to recruit effort, that is, increase pupil size, with load at both levels of spatial interference.

fMRI Study

Behavior

Increasing load caused a moderate decrease in accuracy, F(2, 76) = 11.98, p < .001, ηG2 = .08 (Figure 6). Furthermore, close encounters caused a large decrease in accuracy, F(1, 38) = 114.61, p < .001, ηG2 = .25, and did not interact with load, F(2, 76) = 0.40, p = .63. RT increased with higher load, F(2, 76) = 4.42, p = .015, ηG2 = .01, and close encounters, F(1, 38) = 25.85, p < .001, ηG2 = .02. There was no interaction between Load and Close encounters in RTs, F(2, 76)= 0.82, p = .42.

Figure 6. 

Behavioral results of fMRI study. Mean accuracy (top) and RT (bottom) as a function of load and for each level of close encounters. Error bars indicate the SEM.

Figure 6. 

Behavioral results of fMRI study. Mean accuracy (top) and RT (bottom) as a function of load and for each level of close encounters. Error bars indicate the SEM.

Whole-brain fMRI Analysis

A set of areas was associated with tracking as compared with passive viewing, comprising the bilateral inferior/middle occipital cortex; the bilateral inferior/superior parietal cortex; and bilateral middle frontal cortex, precentral area, and SMA. (Figure 7; Tables 1 and 2). This pattern replicates previous findings on MOT and fMRI (Culham et al., 1998).

Figure 7. 

Brain regions' activations and deactivations obtained from the indicated contrasts. L = left hemisphere; R = right hemisphere. The color bar indicates the obtained t value.

Figure 7. 

Brain regions' activations and deactivations obtained from the indicated contrasts. L = left hemisphere; R = right hemisphere. The color bar indicates the obtained t value.

Table 1. 
Summary of Brain Areas Presenting Hemodynamic Changes due to the Different Conditions: Tracking > Passive Viewing (Tracking), Load 4 > Load 2 (Load), and Close > Far Encounters
 TrackingLoadLevel of Close Encounters
Frontal 
Supplementary motor area   
Precentral gyrus   
Middle frontal gyrus (including frontal eye fields)   +(L) 
Middle frontal gyrus +(R) +(R) 
Inferior frontal gyrus   +(L)   
Medial frontal gyrus     
Frontal medial orbital   − − 
Anterior insula     
Posterior insula −     
  
Parietal 
Superior parietal lobule     
Inferior parietal lobule 
Middle/posterior cingulate cortex − −   
  
Temporal/Occipitial 
Fusiform   
Middle/superior temporal gyrus − −   
Precuneus     
Inferior occpitial gyrus     
Middle occpitial gyrus   
Lingual gyrus     
Calcarine − −   
  
Subcortical 
Thalamus (pulvinar)     
Midbrain     
Hippocampus   − − 
 TrackingLoadLevel of Close Encounters
Frontal 
Supplementary motor area   
Precentral gyrus   
Middle frontal gyrus (including frontal eye fields)   +(L) 
Middle frontal gyrus +(R) +(R) 
Inferior frontal gyrus   +(L)   
Medial frontal gyrus     
Frontal medial orbital   − − 
Anterior insula     
Posterior insula −     
  
Parietal 
Superior parietal lobule     
Inferior parietal lobule 
Middle/posterior cingulate cortex − −   
  
Temporal/Occipitial 
Fusiform   
Middle/superior temporal gyrus − −   
Precuneus     
Inferior occpitial gyrus     
Middle occpitial gyrus   
Lingual gyrus     
Calcarine − −   
  
Subcortical 
Thalamus (pulvinar)     
Midbrain     
Hippocampus   − − 

Positive activations are marked with a plus sign (+), and negative activations are marked with a minus sign (−). If activations were lateralized, they are indicated as L = left or R = right. For precise coordinates, please refer to Tables 2 and 3.

Table 2. 
Track-PVAreaBAClusterCoordinatesPeak pT
Hemispherep (FWE Corr)Sizexyz
RH Middle occipital gyrus BA 37 <.01 16399 46 −68 <.001 19.38 
LH Inferior occipital gyrus BA 19     −44 −76 <.001 18.17 
RH Precuneus BA 7/31     36 40 54 <.001 15.07 
LH Superior parietal lobule BA 7     −24 −60 58 <.001 13.64 
RH Inferior parietal lobule BA 40     42 −30 42 <.001 11.00 
RH Middle frontal gyrus BA 6 <.01 6445 24 −6 52 <.001 17.67 
LH Middle frontal gyrus BA 6     −24 −6 60 <.001 13.93 
RH Precentral gyrus BA 6     56 32 <.001 9.63 
RH Supplementary motor area BA 6     52 <.001 8.42 
LH Precentral gyrus BA 6     −40 −8 54 <.001 7.75 
RH Midbrain   <.01 1253 −18 −6 <.001 7.56 
RH Thalamus Pulvinar .161 131 14 −24 14 <.001 5.95 
Track-PVAreaBAClusterCoordinatesPeak pT
Hemispherep (FWE Corr)Sizexyz
RH Middle occipital gyrus BA 37 <.01 16399 46 −68 <.001 19.38 
LH Inferior occipital gyrus BA 19     −44 −76 <.001 18.17 
RH Precuneus BA 7/31     36 40 54 <.001 15.07 
LH Superior parietal lobule BA 7     −24 −60 58 <.001 13.64 
RH Inferior parietal lobule BA 40     42 −30 42 <.001 11.00 
RH Middle frontal gyrus BA 6 <.01 6445 24 −6 52 <.001 17.67 
LH Middle frontal gyrus BA 6     −24 −6 60 <.001 13.93 
RH Precentral gyrus BA 6     56 32 <.001 9.63 
RH Supplementary motor area BA 6     52 <.001 8.42 
LH Precentral gyrus BA 6     −40 −8 54 <.001 7.75 
RH Midbrain   <.01 1253 −18 −6 <.001 7.56 
RH Thalamus Pulvinar .161 131 14 −24 14 <.001 5.95 

RH = right hemisphere; LH = left hemisphere.

Load evoked parametrically higher activation in bilateral inferior parietal area and left inferior frontal (Figure 7; Tables 1 and 3). Close encounters evoked higher activity in middle occipital areas, as well as bilateral inferior parietal and bilateral middle frontal gyrus/precentral area including FEFs, and insula (Figure 7; Tables 1 and 3). The difference between these contrasts revealed that close encounters presented additional activity in bilateral inferior parietal and medial frontal regions, insula, and primary visual areas (lingual/calcarine) as compared with load (Figure 8; Table 3).

Table 3. 
Brain Coordinates of Activation at the Indicated Contrasts
HemisphereAreaBACluster pCoordinatesPeak pT
FWE CorrSizexyz
Load 4-2 
LH Inferior parietal lobule BA 40 .008 836 −42 −36 40 <.001 4.96 
RH Inferior parietal lobule BA 40 .003 1020 40 −44 46 <.001 5.50 
RH Fusiform gyrus BA 37 .518 85 46 −56 −12 <.001 3.95 
LH Inferior frontal gyrus BA 9 .786 23 −52 10 30 <.001 3.49 
RH Middle frontal gyrus BA 46 .807 19 46 40 18 0.001 3.26 
  
Close–Far Encounters 
LH Inferior parietal lobule BA 40 .000 4802 −44 −30 42 <.001 6.41 
RH Inferior parietal lobule BA 40     44 −36 50 <.001 4.56 
LH Middle occipital gyrus BA 19 .000 4673 −34 −78 18 <.001 3.41 
RH Middle occipital gyrus BA 19     32 −86 −2 <.001 3.40 
RH Lingual gyrus BA 18     10 −70 <.001 4.92 
LH Lingual gyrus BA 18     −12 −74 <.001 4.44 
RH Medial frontal gyrus BA 32 .034 525 16 46 <.001 5.02 
RH Middle frontal gyrus BA 9 .151 270 40 40 24 <.001 4.12 
RH Precentral gyrus BA 6 .003 1196 44 28 <.001 4.87 
RH Insula       42 18 −4 <.001 5.26 
LH Precentral gyrus BA 6 .055 437 −46 34 <.001 4.94 
LH Insula BA 13 .212 217 −36 14 <.001 4.43 
LH Middle frontal gyrus BA 6 .448 106 −28 54 <.001 3.96 
RH Middle frontal gyrus BA 6 .717 37 22 64 <.001 3.36 
  
Effect of Close Encounters Masked by Load 
RH Middle occipital gyrus BA 18 .000 3548 32 −86 −2 <.001 7.74 
LH Middle occipital gyrus BA 18     −22 −90 −2 <.001 6.20 
LH Fusiform gyrus BA 37     −46 −68 −8 <.001 4.58 
LH Medial frontal gyrus BA 32 .100 336 12 48 <.001 4.72 
RH Cingulate gyrus BA 32     18 42 <.001 3.82 
RH Insula BA 13 .026 578 40 20 −8 <.001 4.96 
LH Insula BA 13 .610 61 −36 16 <.001 4.10 
LH Postcentral gyrus BA 2 .050 455 −52 −26 40 <.001 5.63 
LH Inferior parietal lobule BA 40 .514 86 −40 38 58 <.001 4.62 
RH Inferior parietal lobule BA 40 .086 361 56 −32 42 <.001 4.35 
RH Supramarginal gyrus BA 40     48 −42 32 <.001 3.46 
LH Precuneus BA 7 .088 357 −20 −68 48 <.001 4.09 
RH Precuneus BA 7     12 −66 44 <.001 3.79 
LH Precentral gyrus BA 6 .703 40 −46 32 <.001 3.94 
LH Inferior frontal gyrus BA 9 .848 56 −40 10 30 <.001 3.65 
RH Middle frontal gyrus BA 9 .766 27 36 32 32 <.001 3.61 
HemisphereAreaBACluster pCoordinatesPeak pT
FWE CorrSizexyz
Load 4-2 
LH Inferior parietal lobule BA 40 .008 836 −42 −36 40 <.001 4.96 
RH Inferior parietal lobule BA 40 .003 1020 40 −44 46 <.001 5.50 
RH Fusiform gyrus BA 37 .518 85 46 −56 −12 <.001 3.95 
LH Inferior frontal gyrus BA 9 .786 23 −52 10 30 <.001 3.49 
RH Middle frontal gyrus BA 46 .807 19 46 40 18 0.001 3.26 
  
Close–Far Encounters 
LH Inferior parietal lobule BA 40 .000 4802 −44 −30 42 <.001 6.41 
RH Inferior parietal lobule BA 40     44 −36 50 <.001 4.56 
LH Middle occipital gyrus BA 19 .000 4673 −34 −78 18 <.001 3.41 
RH Middle occipital gyrus BA 19     32 −86 −2 <.001 3.40 
RH Lingual gyrus BA 18     10 −70 <.001 4.92 
LH Lingual gyrus BA 18     −12 −74 <.001 4.44 
RH Medial frontal gyrus BA 32 .034 525 16 46 <.001 5.02 
RH Middle frontal gyrus BA 9 .151 270 40 40 24 <.001 4.12 
RH Precentral gyrus BA 6 .003 1196 44 28 <.001 4.87 
RH Insula       42 18 −4 <.001 5.26 
LH Precentral gyrus BA 6 .055 437 −46 34 <.001 4.94 
LH Insula BA 13 .212 217 −36 14 <.001 4.43 
LH Middle frontal gyrus BA 6 .448 106 −28 54 <.001 3.96 
RH Middle frontal gyrus BA 6 .717 37 22 64 <.001 3.36 
  
Effect of Close Encounters Masked by Load 
RH Middle occipital gyrus BA 18 .000 3548 32 −86 −2 <.001 7.74 
LH Middle occipital gyrus BA 18     −22 −90 −2 <.001 6.20 
LH Fusiform gyrus BA 37     −46 −68 −8 <.001 4.58 
LH Medial frontal gyrus BA 32 .100 336 12 48 <.001 4.72 
RH Cingulate gyrus BA 32     18 42 <.001 3.82 
RH Insula BA 13 .026 578 40 20 −8 <.001 4.96 
LH Insula BA 13 .610 61 −36 16 <.001 4.10 
LH Postcentral gyrus BA 2 .050 455 −52 −26 40 <.001 5.63 
LH Inferior parietal lobule BA 40 .514 86 −40 38 58 <.001 4.62 
RH Inferior parietal lobule BA 40 .086 361 56 −32 42 <.001 4.35 
RH Supramarginal gyrus BA 40     48 −42 32 <.001 3.46 
LH Precuneus BA 7 .088 357 −20 −68 48 <.001 4.09 
RH Precuneus BA 7     12 −66 44 <.001 3.79 
LH Precentral gyrus BA 6 .703 40 −46 32 <.001 3.94 
LH Inferior frontal gyrus BA 9 .848 56 −40 10 30 <.001 3.65 
RH Middle frontal gyrus BA 9 .766 27 36 32 32 <.001 3.61 

RH = right hemisphere; LH = left hemisphere.

Figure 8. 

Effect of close encounters masked by load.

Figure 8. 

Effect of close encounters masked by load.

Brainstem Nuclei

We inspected the effect of increasing task demands on the activity within individually defined LC masks. We found a significant effect of Load, F(2, 74) = 3.21, p = .046, ηG2 = .02 (Figure 9). We found no effect of Close encounters, F(1, 37) = 0.59, p = .44, or interaction, F(2, 74) = 3.21, p = .066.

Figure 9. 

Brainstem activity evoked by the different conditions. Top: LC; bottom: VTA/SN.

Figure 9. 

Brainstem activity evoked by the different conditions. Top: LC; bottom: VTA/SN.

Moreover, Load had no significant effect on activity in the VTA/SN nuclei, F(2, 74) = 1.71, p = .194 (Figure 9) but trials with close encounters evoked a larger activity than trials with far encounters, F(1, 37) = 4.96, p = .032, ηG2 = .02. We found no interaction, F(2, 74) = 0.375, p = .656.

Individual Differences

High and low performers showed different levels of accuracy for the different conditions. In addition to the main effects of Load and Level of close encounters (F(1, 36) = 28.51, p < .001 ηG2 = .11 and F(1, 36) = 88.14, p < .001, ηG2 = .27, respectively; Figure 10), we found a significant interaction between Performance level and Load, F(1, 36) = 9.20, p = .004, ηG2 = .04. Low performers had lower accuracy in Load 4 as compared with Load 2, T(19) = 5.50, p < .001, whereas the effect of Load in high performers did not reach significance, T(17) = 1.83, p = .084. No group effects were observed for RT.

Figure 10. 

Behavioral and brainstem effects on high and low performers. Mean accuracy (top), LC (middle), and VTA/SN (bottom) in the close encounters conditions for the lowest and highest load split by high performers versus low performers.

Figure 10. 

Behavioral and brainstem effects on high and low performers. Mean accuracy (top), LC (middle), and VTA/SN (bottom) in the close encounters conditions for the lowest and highest load split by high performers versus low performers.

In the LC, we found a significant three-way interaction of Load × Performance level × Level of close encounters, F(1, 36) = 4.44, p = .042, ηG2 = .02 (Figure 10). When looking separately at the different level of close encounters, we observed a significant Load × Group interaction, F(1,36) = 5.55, p = .024, ηG2 = .06, in close but not in far encounters. In trials with close encounters, high performers had higher activity in Load 4 as compared with Load 2, T(17) = 2.51, p = .022, but not low performers (p > .2 for all contrasts).

In the VTA/SN, we found no main effect or interaction effects (Figure 10). High performers presented larger activity than low performers, although the results did not reach significance, F(1, 36) = 3.17, p = .083.

Summary

Similar to the eye-tracking experiment, load reduced accuracy even though the amount of spatial interference within the trials (number of close encounters) was equalized across load levels. We found overlap in the regions that were active due to load and spatial interference. Additional areas, including VTA/SN, seemed to be involved in processing high number of close encounters. High performers were able to recruit effort, that is, increase LC, with load and under high level of spatial interference.

DISCUSSION

Based on the convergent findings obtained from our two independent experiments, we identified different neural mechanisms dealing with load and close encounters that may reflect different sources of capacity limitations. This conclusion is based on the observation that, even when the number of close encounters was kept equal, accuracy decreased and RT increased with load level (number of targets). Increasing demands due to load elicited pupil dilations and LC activity throughout the tracking period, greater activation in the frontoparietal attention network, and more frequent saccades. In addition to the load effect, high number of close encounters reduced accuracy, engaged the VTA/SN, and further implicated the attentional network and number of saccades. Furthermore, low performers were challenged by object numbers even at low level of perceptual challenge, whereas high performers appeared able to successfully resolve stimulus-driven conflict, accompanied by higher pupil and LC activity during trials with close encounters. Taken together, the above results suggest that the brain can cope with load and instances of close encounters in a highly specific and adaptive manner.

Task Manipulation and Behavioral Effects of Load and Level of Close Encounters

Different factors are known to affect performance in MOT: number of objects or cognitive workload (both targets and nontargets), number of close encounters, speed, and hemifield (Meyerhoff et al., 2017; Scimeca & Franconeri, 2015; Franconeri et al., 2013). Some studies have assessed the effect of interobject spacing on accuracy, either at different or at a fixed level of load (Feria, 2013; Bae & Flombaum, 2012; Iordanescu et al., 2009; Franconeri et al., 2008; Shim et al., 2008; Alvarez & Franconeri, 2007). Others have assessed the relation between load and speed (Srivastava & Vul, 2016; Franconeri et al., 2010) or between speed and object spacing (Vul et al., 2009; Alvarez & Franconeri, 2007). In the current study, we systematically varied the number of targets at two levels of amount of close encounters. This allowed us to assess the source of capacity limitations associated to the two factors.

We found a linear decrease in accuracy with load at both levels of object spacing, in which the number of close encounters was equalized between load levels. This suggests that, in addition to the frequency of close encounters, the number of objects posed a demand on the participants, and therefore, those factors seem to be additive (Figure 1B). This agrees with previous work suggesting that recruitment of higher order processes allows for dynamic allocation of attention to crowded situations (Srivastava & Vul, 2016; Iordanescu et al., 2009; Pylyshyn, 2004; Intriligator & Cavanagh, 2001). Higher order mechanisms likely involve visual selection and sustained visual attention, which may be specifically engaged during target assignment and during the tracking period, respectively (Meyerhoff et al., 2017). It has been shown that cognitive processes, such as update in working memory, are engaged at the assignment of objects (Ma & Flombaum, 2013; Allen et al., 2006) and during tracking (Eayrs & Lavie, 2018; Oksama & Hyönä, 2004), consistent with computational simulations (Srivastava & Vul, 2016; Vul et al., 2009). Similarly, Vul et al. (2009) argued that “while many results in MOT arise as consequences of the information available for the computational task, the speed-number tradeoff seems to be the result of a flexibly-allocated resource such as memory or attention.” Possibly, cognitive processes may help to circumvent the limitations of the visual attentional processes and vice versa (Oksama & Hyönä, 2004). In line with this, a series of EEG studies have shown that different ERP signatures were related to the demands exerted by target number and the demands of the tracking itself (Drew et al., 2013; Drew, McCollough, Horowitz, & Vogel, 2009; Drew & Vogel, 2008). For example, a parietocentral negativity was related to increasing target number during the assignment period, a contralateral delay activity was related to load during the tracking period, and early negative and positive components were related to following targets rather than distractors during the tracking period. These results argue for a highly coordinated and dynamic process developing during the tracking to keep target information in working memory, resolve confusion and remain engaged.

Pupil Size Increases with Load but Not with Close Encounters

We replicated the finding that pupil size is sensitive to load (Alnæs et al., 2014; Wright et al., 2013). One possibility is that pupil size reflects preparatory effort mechanisms recruited in the beginning of the trial and sustained during the whole trial. This would agree with the proposal that pupil dilation reflects a readiness to expend resources (Bruya & Tang, 2018; van der Wel & van Steenbergen, 2018). Another possibility is that pupil dilation is associated with the effort needed to perform the tracking itself throughout the duration of the trial, perhaps similar to the contralateral delay activity (Drew et al., 2009, 2013; Drew & Vogel, 2008). Such an interpretation would correspond with work in rhesus monkeys showing that LC neurons were activated both before and during task performance to energize behavior to face challenges (Varazzani et al., 2015). Importantly, we found that pupil size was not sensitive to our manipulation of object spacing. This is consistent with load and close encounters relying on different physiological mechanisms. We also note that, to successfully perform the task, effort due to the load manipulation has to be sustained over the tracking period, whereas the close encounters are transient events. Because there are multiple such close encounters, one could expect that the effect of them would accumulate to such an extent that they would influence the average pupil size over the entire tracking period, but it is also possible that these effects would wash out in the average. However, our analysis of pupil size in early and late intervals within trials suggests that the LC was not associated to the perceptual challenge imposed by close encounters.

The number of saccades increased with load and close encounters, although to a greater extent with close encounters. This result may correspond to the increase in activity that we observed in the FEFs in high versus low object spacing. FEFs are involved in the control of eye movement (Vernet, Quentin, Chanes, Mitsumasu, & Valero-Cabré, 2014) and may have been implicated in directing the eyes toward highly conflicting events (i.e., close encounters) or the execution of rescue saccades, consistent with previous results on MOT and eye tracking (Vater et al., 2017; Zelinsky & Todor, 2010). Similarly, saccades were associated to identity–location binding during multiple-identity tracking (Oksama & Hyönä, 2016). These results argue for specific ocular signatures related to the different sources of challenge during dynamic attention.

Cortical Attentional Networks Related to Load and Close Encounters

We found that a frontoparietal network including inferior parietal and dorsal frontal areas, as well as occipital areas, were recruited during tracking. Load was further related to increased activity in the bilateral inferior parietal and inferior frontal areas. These activations closely match the ones reported by previous MOT studies (Alnæs et al., 2014; Jahn et al., 2012; Shim et al., 2009; Tomasi et al., 2004; Culham et al., 1998, 2001; Jovicich et al., 2001). The frontoparietal network is generally activated during demanding visuospatial tasks of various kinds (Hugdahl, Raichle, Mitra, & Specht, 2015; Duncan, 2010; Corbetta & Shulman, 2002), suggesting that its components cooperate to represent voluntary and stimulus-driven attentional priority (Scolari, Seidl-Rathkopf, & Kastner, 2015; Ptak, 2012; Serences & Yantis, 2006). As part of this network, the inferior parietal cortex is suggested to index the location of selected or prioritized objects (Serences & Yantis, 2006) and its level of activity depends on capacity limitations in visual STM (Mitchell & Cusack, 2007). The inferior parietal cortices interact with superior parietal and frontal areas (Howe, Horowitz, Morocz, Wolfe, & Livingstone, 2009) and constitute the so-called dorsal attention network (Vossel, Geng, & Fink, 2014; Fox, Corbetta, Snyder, Vincent, & Raichle, 2006; Corbetta & Shulman, 2002), whose involvement is consistent with top–down processes engaged during the task to guide the tracking of the target objects.

Close encounters were associated with strong activity in bilateral inferior parietal, dorsal–frontal, including FEF, and occipital regions. Strong activation in the medial superior parietal lobe (including precuneus) has been suggested to reflect voluntary shifts of attention between perceptual entities (Serences & Yantis, 2006; Serences, Liu, & Yantis, 2005; Serences, Shomstein, et al., 2005; Yantis et al., 2002). In addition, insula and medial frontal areas were engaged. The insula and medial frontal areas are considered part of the ventral attention network (Vossel et al., 2014; Corbetta, Patel, & Shulman, 2008), whose function is to detect salient events and assist attention orienting. During tracking, events such as close encounters may engage the ventral attention network to flexibly allocate attention where it is most needed. Therefore, an increased activity in close versus far encounters may be related to coping with the increasing on-line demands imposed by that manipulation.

The FEFs were found to be activated during high versus low number of close encounters. In a seminal study on MOT with fMRI, Culham et al. (1998) found that the activity in FEF was related to tracking but was not modulated by load. Here, we corroborate this finding, and in addition, we propose that the FEF activity during tracking is related to the events of close encounters, which were not controlled for in the Culham et al. study. Therefore, its involvement in the task would be the resolution of moment-to-moment confusion, rather than barely “suppressing eye movements” (Culham et al., 1998).

There has been some discussion on whether the challenges posed by object motions are handled by low-level visual processing or higher order cognitive functions (Cavanagh, Battelli, & Holcombe, 2014). Our results support the latter by indicating that anatomical areas associated with higher order cognitive functions are recruited during close encounters. These areas may also support the suggested flexible allocation of attentional resources between spatial locations requiring different levels of attention (Alvarez & Franconeri, 2005).

We cannot exclude the possibility that the network of cortical areas resolving load and close encounters were the same and that differences in the demand levels of our paradigm caused that the effects of close encounters in brain activity to be stronger than the effect of load. It is not possible to evaluate whether Load 4, compared with Load 2, imposed as high demands as close versus far encounters, as indicated by the different effect sizes of those two factors on accuracy. Indeed, previous studies have shown larger brain effects of load when including Load 5 (e.g., Alnæs et al., 2014). However, previous studies have shown specificity in the factors accounting for the activation of some areas during MOT (Nummenmaa, Oksama, Glerean, & Hyönä, 2016; Merkel, Hopf, Heinze, & Schoenfeld, 2015; Atmaca et al., 2013; Howe et al., 2009). Here, we observed large activations in attentional and visual areas due to close encounters as compared with load. In light of the tasks to which these areas have been linked, we conclude that they reflect the need of rapidly increasing attentional resolution to resolve confusing events around the targets.

Brainstem Neuromodulatory Systems

Neuromodulators play an active role on attention by adjusting responsivity of neurons to optimize processing of inputs, a concept known as “gain” (Aston-Jones & Cohen, 2005; Servan-Schreiber, Printz, & Cohen, 1990). In the context of the MOT, we found that two nuclei, LC (noradrenergic) and VTA/SN (dopaminergic), presented different profiles for the different task manipulations. LC increased activity with load, whereas VTA/SN showed increased activity with close encounters. The involvement of LC agrees with the findings from a previous study, which employed an atlas-based mask of the LC to extract the task-related activity (Alnæs et al., 2014). The LC-NE system has been studied in the context of mental effort, and its function has been posited to be to increase the gain of the neural circuits relating co-occurring events. This result may be explained by the recruitment of LC by a top–down frontal system engaged in the preparation of load-related effort. We further found that LC activity was not significantly related to close encounters. In contrast, the VTA/SN presented higher activity in trials with high as compared with low spatial interference but showed no effect of load. Such a double dissociation between brainstem systems is (to the best of our knowledge) novel within human imaging studies in the context of an attentional task with varying levels of task demands. Interestingly, the VTA/SN and LC share reciprocal connections (Weinshenker & Schroeder, 2007), and therefore, they are expected to be tightly related. Within theories of cognitive effort, LC activity may be recruited after the computation of cost at the initiation of the trial, when the upcoming load is known. A limited but growing literature suggests a role of DA on spatial attention. For example, direct infusion of dopaminergic agents in the FEFs influence saccades and visual processing in macaques (Noudoost & Moore, 2011), and DA has potent modulatory effects on prefrontal activation to spatial working memory tasks (Arnsten, 1997; Sawaguchi & Goldman-Rakic, 1991). VTA/SN activity may be related to working memory processes that enable propagation of state estimates of the position of the targets through time. In other words, one mechanism (VTA/SN) may depend on events occurring in spatiotopic regions, whereas the other (LC) may be essentially nonspatial and more related to temporal relations between events and the recruitment of resources (effort or arousal). Although coherent with the roles associated with these neuromodulatory systems, the ventral attentional network (which we found to be involved as a function of amount of close encounters) has been proposed to recruit LC after detection of salient events (Corbetta et al., 2008); however, because we did not observe that LC was related to changes in spatial interference, its role may mainly relate to nonspatial aspects of the attention network. In summary, activity in the brainstem nuclei appear to be linked to specific demands imposed by the tracking of multiple objects.

Individual Differences

Our analysis of individual differences showed that, across studies, load and close encounters exerted different effects on participants' accuracy. With far encounters, low performers had a larger decrease in accuracy as a function of load than high performers, and this effect was observed in both studies. This result suggests that the processes of load and close encounters are separate in terms of compromising participants' performance, as has already been proposed (Oksama & Hyönä, 2004; Intriligator & Cavanagh, 2001). The pupillometry results showed pupil dilation with load in both levels of close encounters in high performers. Similarly, the brainstem analysis indicated that, in addition to being recruited with increasing load, the LC was also, to some extent, recruited with close encounters but only in high performing participants. Engagement of this neuromodulatory system increases neural gain and enhances the representation of visual stimuli. In a similar line, previous findings indicate that better tracking performance correlates with the ability of sustaining attention and updating object representations in working memory (Drew & Vogel, 2008). Our results render the assessment of recruitment of neuromodulatory systems as promising to explain individual differences in capacity.

Relation with Theories on Limitations in Dynamic Visual Attention

Theoretically, two competing models have been proposed to account for the perceptual and cognitive limitations observed during a task requiring dynamic attention, such as MOT (Franconeri et al., 2013; Alvarez & Cavanagh, 2004). One is the coding of individual stimuli into cortical spatial maps that may inhibit each other due to close perceptual proximity of the stimuli since an inhibitory surround of one target may interfere with the enhancement of other nearby targets (“cortical real-estate” account). Another is the coding of stimuli in separate working memory slots (“resource” account). Although the first predicts that load and close encounters engage the same substrates in primary visual circuits, the second would predict that different networks deal with load and close encounters (i.e., objects are represented as informational chunks and additional mechanisms that deal with spatial interference due to close encounters are recruited as needed). Although our results do support the cortical real-state account, given that close encounters influence accuracy and cortical as well as some of the brainstem activations, this view meets difficulty in explaining effects on accuracy, cortical and brainstem activation, and pupil dilations when the number of close encounters is held constant. Moreover, we observed activations in LC and pupil responses related to load and not to close encounters. Hence, a general “resource” account as originally proposed by Kahneman (1973), although it may be unsatisfactory for a number of reasons (Franconeri et al., 2013), remains a candidate explanation for our results. This agrees with empirical results using object tracking with circular trajectories and varying degrees of separation (Holcombe, Chen, & Howe, 2014). We conclude that target number imposes a constraint in the ability to track multiple objects, loading on some sort of distributed attentional resource as suggested by a multifocal model (Cavanagh & Alvarez, 2005) and that the events of close encounters load on some other attentional resource that can be dynamically engaged to resolve spatial interference (Franconeri et al., 2010). However, we note that the above two views have mainly focused on the perceptual level. Previous studies suggest that not only perceptual but also executive processing is crucial for resolving MOT, not only during tracking but also at the moment of target selection (Eayrs & Lavie, 2018; Lapierre et al., 2017; Ma & Flombaum, 2013; Vul et al., 2009; Tombu & Seiffert, 2008; Franconeri, Alvarez, & Enns, 2007; Allen et al., 2006). Participants with high capacity may have benefited from the ability of managing both types of resources in a timely manner.

MOT shares many characteristics with other visuospatial tasks that rely on perceptual as well as executive processes (i.e., visual STM; Fougnie & Marois, 2006; Marois & Ivanoff, 2005; subitizing: Chesney & Haladjian, 2011; selection of objects: Xu & Chun, 2009), and consequently, its study may inform how limitations in the other processes occur.

Study Limitations and Future Directions

For the present work, we only had two levels of close encounters (i.e., spatial interference). Future experiments with more levels of minimum target–distractor object spacing in addition to load would allow a more complete picture of the relation between the two factors. Although we found no interaction between them, adding object spacing levels may present a more complex picture. Another remaining question is what would be the effect of informing participants at the start of the trial whether there will be high or low close encounters in the display at the same time as telling the number of targets. A prediction would be that, if preparatory activity related to load would be engaged to prepare for higher close encounters, the accuracy patterns of load and close encounters would be dependent (Figure 1C). For the purpose of this study, we focused on the role of attention in resolving spatial interference during tracking and therefore did not study brain and brainstem activity during visual selection (i.e., activity during the target assignment phase). Studies focusing on this stage may confirm previous results suggesting that executive functions are involved in this stage and influence tracking, and extend our results regarding the involvement of brain and brainstem regions in the encoding of targets.

The study of nuclei in the brainstem conveys limitations due to their small size. Some spillover from nearby regions may be expected. However, to minimize this problem, we acquired high-resolution images of the LC, used a mask of the VTA/SN, and used a small smoothing kernel.

Conclusion

Our results are inconsistent with the view that load can be merely explained by instances of spatial interference and support the view that, in the MOT task, distinct neural basis deal with the increasing demands posed by load when perceptual factors are controlled for. In particular, brainstem neuromodulatory nuclei play a key role. Both load and close encounters compromised accuracy during MOT and recruited an overlapping set of regions within the dorsal attention network. Additional areas within the ventral attention network were recruited during high number of close encounters. The ventral and dorsal attentional systems collaborate to solve the challenges of current goals, likely engaging top–down control, as well as spontaneously arising demands, engaging bottom–up processes (Vossel et al., 2014; Corbetta et al., 2008). LC and pupil dilation signaled effort driven by increasing load and, in high performers, with higher close encounters. VTA/SN differed from LC in that it was related to close encounters. The MOT task, together with physiological measures, may offer a unique opportunity to interrogate how cortical and neuromodulatory systems are adaptively recruited when cognition is challenged.

Reprint requests should be sent to Veronica Mäki-Marttunen or Thomas Hagen, Postboks 1094 Blindern 0317 Oslo, Norway, or via e-mail: makimarttunen.veronica@gmail.com, thomas.hagen@psykologi.uio.no.

REFERENCES

REFERENCES
Allen
,
R.
,
McGeorge
,
P.
,
Pearson
,
D. G.
, &
Milne
,
A.
(
2006
).
Multiple-target tracking: A role for working memory?
Quarterly Journal of Experimental Psychology
,
59
,
1101
1116
.
Alnæs
,
D.
,
Sneve
,
M. H.
,
Espeseth
,
T.
,
Endestad
,
T.
,
van de Pavert
,
S. H.
, &
Laeng
,
B.
(
2014
).
Pupil size signals mental effort deployed during multiple object tracking and predicts brain activity in the dorsal attention network and the locus coeruleus
.
Journal of Vision
,
14
,
1
.
Alvarez
,
G. A.
, &
Cavanagh
,
P.
(
2004
).
The capacity of visual short-term memory is set both by visual information load and by number of objects
.
Psychological Science
,
15
,
106
111
.
Alvarez
,
G. A.
, &
Franconeri
,
S. L.
(
2005
).
How many objects can you track? Evidence for a flexible tracking resource
.
Journal of Vision
,
5
,
641
.
Alvarez
,
G. A.
, &
Franconeri
,
S. L.
(
2007
).
How many objects can you track?: Evidence for a resource-limited attentive tracking mechanism
.
Journal of Vision
,
7
,
14
.
Arnsten
,
A. F.
(
1997
).
Catecholamine regulation of the prefrontal cortex
.
Journal of Psychopharmacology
,
11
,
151
162
.
Aston-Jones
,
G.
, &
Cohen
,
J. D.
(
2005
).
An integrative theory of locus coeruleus-norepinephrine function: Adaptive gain and optimal performance
.
Annual Review of Neuroscience
,
28
,
403
450
.
Atmaca
,
S.
,
Stadler
,
W.
,
Keitel
,
A.
,
Ott
,
D. V.
,
Lepsien
,
J.
, &
Prinz
,
W.
(
2013
).
Prediction processes during multiple object tracking (MOT): Involvement of dorsal and ventral premotor cortices
.
Brain and Behavior
,
3
,
683
700
.
Bae
,
G. Y.
, &
Flombaum
,
J. I.
(
2012
).
Close encounters of the distracting kind: Identifying the cause of visual tracking errors
.
Attention, Perception, & Psychophysics
,
74
,
703
715
.
Bruya
,
B.
, &
Tang
,
Y. Y.
(
2018
).
Is attention really effort? Revisiting Daniel Kahneman's influential 1973 book attention and effort
.
Frontiers in Psychology
,
9
,
1133
.
Cavanagh
,
P.
, &
Alvarez
,
G. A.
(
2005
).
Tracking multiple targets with multifocal attention
.
Trends in Cognitive Sciences
,
9
,
349
354
.
Cavanagh
,
P.
,
Battelli
,
L.
, &
Holcombe
,
A. O.
(
2014
).
Dynamic attention
. In
A. C. (Kia)
Nobre
&
S.
Kastner
(Eds.),
The Oxford handbook of attention
(pp.
503
).
Oxford
:
Oxford University Press
.
Chesney
,
D. L.
, &
Haladjian
,
H. H.
(
2011
).
Evidence for a shared mechanism used in multiple-object tracking and subitizing
.
Attention, Perception, & Psychophysics
,
73
,
2457
2480
.
Corbetta
,
M.
,
Patel
,
G.
, &
Shulman
,
G. L.
(
2008
).
The reorienting system of the human brain: From environment to theory of mind
.
Neuron
,
58
,
306
324
.
Corbetta
,
M.
, &
Shulman
,
G. L.
(
2002
).
Control of goal-directed and stimulus-driven attention in the brain
.
Nature Reviews Neuroscience
,
3
,
201
215
.
Culham
,
J. C.
,
Brandt
,
S. A.
,
Cavanagh
,
P.
,
Kanwisher
,
N. G.
,
Dale
,
A. M.
, &
Tootell
,
R. B.
(
1998
).
Cortical fMRI activation produced by attentive tracking of moving targets
.
Journal of Neurophysiology
,
80
,
2657
2670
.
Culham
,
J. C.
,
Cavanagh
,
P.
, &
Kanwisher
,
N. G.
(
2001
).
Attention response functions: Characterizing brain areas using fMRI activation during parametric variations of attentional load
.
Neuron
,
32
,
737
745
.
Drew
,
T.
,
Horowitz
,
T. S.
, &
Vogel
,
E. K.
(
2013
).
Swapping or dropping? Electrophysiological measures of difficulty during multiple object tracking
.
Cognition
,
126
,
213
223
.
Drew
,
T.
,
McCollough
,
A. W.
,
Horowitz
,
T. S.
, &
Vogel
,
E. K.
(
2009
).
Attentional enhancement during multiple-object tracking
.
Psychonomic Bulletin & Review
,
16
,
411
417
.
Drew
,
T.
, &
Vogel
,
E. K.
(
2008
).
Neural measures of individual differences in selecting and tracking multiple moving objects
.
Journal of Neuroscience
,
28
,
4183
4191
.
Duncan
,
J.
(
2010
).
The multiple-demand (MD) system of the primate brain: Mental programs for intelligent behaviour
.
Trends in Cognitive Sciences
,
14
,
172
179
.
Eayrs
,
J.
, &
Lavie
,
N.
(
2018
).
Establishing individual differences in perceptual capacity
.
Journal of Experimental Psychology: Human Perception and Performance
,
44
,
1240
1257
.
Feria
,
C. S.
(
2013
).
Speed has an effect on multiple-object tracking independently of the number of close encounters between targets and distractors
.
Attention, Perception, & Psychophysics
,
75
,
53
67
.
Fougnie
,
D.
, &
Marois
,
R.
(
2006
).
Distinct capacity limits for attention and working memory: Evidence from attentive tracking and visual working memory paradigms
.
Psychological Science
,
17
,
526
534
.
Fox
,
M. D.
,
Corbetta
,
M.
,
Snyder
,
A. Z.
,
Vincent
,
J. L.
, &
Raichle
,
M. E.
(
2006
).
Spontaneous neuronal activity distinguishes human dorsal and ventral attention systems
.
Proceedings of the National Academy of Sciences, U.S.A.
,
103
,
10046
10051
.
Franconeri
,
S. L.
,
Alvarez
,
G. A.
, &
Cavanagh
,
P.
(
2013
).
Flexible cognitive resources: Competitive content maps for attention and memory
.
Trends in Cognitive Sciences
,
17
,
134
141
.
Franconeri
,
S. L.
,
Alvarez
,
G. A.
, &
Enns
,
J. T.
(
2007
).
How many locations can be selected at once?
Journal of Experimental Psychology: Human Perception and Performance
,
33
,
1003
1012
.
Franconeri
,
S. L.
,
Jonathan
,
S. V.
, &
Scimeca
,
J. M.
(
2010
).
Tracking multiple objects is limited only by object spacing, not by speed, time, or capacity
.
Psychological Science
,
21
,
920
925
.
Franconeri
,
S. L.
,
Lin
,
J. Y.
,
Pylyshyn
,
Z. W.
,
Fisher
,
B.
, &
Enns
,
J. T.
(
2008
).
Evidence against a speed limit in multiple-object tracking
.
Psychonomic Bulletin & Review
,
15
,
802
808
.
Gläscher
,
J.
(
2009
).
Visualization of group inference data in functional neuroimaging
.
Neuroinformatics
,
7
,
73
82
.
He
,
S.
,
Cavanagh
,
P.
, &
Intriligator
,
J.
(
1996
).
Attentional resolution and the locus of visual awareness
.
Nature
,
383
,
334
337
.
Holcombe
,
A. O.
,
Chen
,
W. Y.
, &
Howe
,
P. D.
(
2014
).
Object tracking: Absence of long-range spatial interference supports resource theories
.
Journal of Vision
,
14
,
1
.
Howe
,
P. D.
,
Horowitz
,
T. S.
,
Morocz
,
I. A.
,
Wolfe
,
J.
, &
Livingstone
,
M. S.
(
2009
).
Using fMRI to distinguish components of the multiple object tracking task
.
Journal of Vision
,
9
,
10
.
Hugdahl
,
K.
,
Raichle
,
M. E.
,
Mitra
,
A.
, &
Specht
,
K.
(
2015
).
On the existence of a generalized non-specific task-dependent network
.
Frontiers in Human Neuroscience
,
9
,
430
.
Husain
,
M.
, &
Roiser
,
J. P.
(
2018
).
Neuroscience of apathy and anhedonia: A transdiagnostic approach
.
Nature Reviews Neuroscience
,
19
,
470
484
.
Intriligator
,
J.
, &
Cavanagh
,
P.
(
2001
).
The spatial resolution of visual attention
.
Cognitive Psychology
,
43
,
171
216
.
Iordanescu
,
L.
,
Grabowecky
,
M.
, &
Suzuki
,
S.
(
2009
).
Demand-based dynamic distribution of attention and monitoring of velocities during multiple-object tracking
.
Journal of Vision
,
9
,
1
.
Jahn
,
G.
,
Wendt
,
J.
,
Lotze
,
M.
,
Papenmeier
,
F.
, &
Huff
,
M.
(
2012
).
Brain activation during spatial updating and attentive tracking of moving targets
.
Brain and Cognition
,
78
,
105
113
.
Joshi
,
S.
,
Li
,
Y.
,
Kalwani
,
R. M.
, &
Gold
,
J. I.
(
2016
).
Relationships between pupil diameter and neuronal activity in the locus coeruleus, colliculi, and cingulate cortex
.
Neuron
,
89
,
221
234
.
Jovicich
,
J.
,
Peters
,
R. J.
,
Koch
,
C.
,
Braun
,
J.
,
Chang
,
L.
, &
Ernst
,
T.
(
2001
).
Brain areas specific for attentional load in a motion-tracking task
.
Journal of Cognitive Neuroscience
,
13
,
1048
1058
.
Kahneman
,
D.
(
1973
).
Attention and effort
(
Vol. 1063
).
Englewood Cliffs, NJ
:
Prentice Hall
.
Krebs
,
R. M.
,
Park
,
H. R. P.
,
Bombeke
,
K.
, &
Boehler
,
C. N.
(
2018
).
Modulation of locus coeruleus activity by novel oddball stimuli
.
Brain Imaging and Behavior
,
12
,
577
584
.
Lapierre
,
M. D.
,
Cropper
,
S. J.
, &
Howe
,
P. D. L.
(
2017
).
Shared processing in multiple object tracking and visual working memory in the absence of response order and task order confounds
.
PLoS One
,
12
,
e0175736
.
Ma
,
Z.
, &
Flombaum
,
J. I.
(
2013
).
Off to a bad start: Uncertainty about the number of targets at the onset of multiple object tracking
.
Journal of Experimental Psychology: Human Perception and Performance
,
39
,
1421
1432
.
Marois
,
R.
, &
Ivanoff
,
J.
(
2005
).
Capacity limits of information processing in the brain
.
Trends in Cognitive Sciences
,
9
,
296
305
.
Merkel
,
C.
,
Hopf
,
J. M.
,
Heinze
,
H. J.
, &
Schoenfeld
,
M. A.
(
2015
).
Neural correlates of multiple object tracking strategies
.
Neuroimage
,
118
,
63
73
.
Meyerhoff
,
H. S.
,
Papenmeier
,
F.
, &
Huff
,
M.
(
2017
).
Studying visual attention using the multiple object tracking paradigm: A tutorial review
.
Attention, Perception, & Psychophysics
,
79
,
1255
1274
.
Meyerhoff
,
H. S.
,
Papenmeier
,
F.
,
Jahn
,
G.
, &
Huff
,
M.
(
2016
).
Not FLEXible enough: Exploring the temporal dynamics of attentional reallocations with the multiple object tracking paradigm
.
Journal of Experimental Psychology: Human Perception and Performance
,
42
,
776
787
.
Meyerhoff
,
H. S.
,
Schwan
,
S.
, &
Huff
,
M.
(
2018
).
Oculomotion mediates attentional guidance toward temporarily close objects
.
Visual Cognition
,
26
,
166
178
.
Mitchell
,
D. J.
, &
Cusack
,
R.
(
2007
).
Flexible, capacity-limited activity of posterior parietal cortex in perceptual as well as visual short-term memory tasks
.
Cerebral Cortex
,
18
,
1788
1798
.
Murty
,
V. P.
,
Shermohammed
,
M.
,
Smith
,
D. V.
,
Carter
,
R. M.
,
Huettel
,
S. A.
, &
Adcock
,
R. A.
(
2014
).
Resting state networks distinguish human ventral tegmental area from substantia nigra
.
Neuroimage
,
100
,
580
589
.
Noudoost
,
B.
, &
Moore
,
T.
(
2011
).
The role of neuromodulators in selective attention
.
Trends in Cognitive Sciences
,
15
,
585
591
.
Nummenmaa
,
L.
,
Oksama
,
L.
,
Glerean
,
E.
, &
Hyönä
,
J.
(
2016
).
Cortical circuit for binding object identity and location during multiple-object tracking
.
Cerebral Cortex
,
27
,
162
172
.
Oksama
,
L.
, &
Hyönä
,
J.
(
2004
).
Is multiple object tracking carried out automatically by an early vision mechanism independent of higher-order cognition? An individual difference approach
.
Visual Cognition
,
11
,
631
671
.
Oksama
,
L.
, &
Hyönä
,
J.
(
2016
).
Position tracking and identity tracking are separate systems: Evidence from eye movements
.
Cognition
,
146
,
393
409
.
Pessiglione
,
M.
,
Le Bouc
,
R.
, &
Vinckier
,
F.
(
2018
).
When decisions talk: Computational phenotyping of motivation disorders
.
Current Opinion in Behavioral Sciences
,
22
,
50
58
.
Petersen
,
S. E.
, &
Posner
,
M. I.
(
2012
).
The attention system of the human brain: 20 Years after
.
Annual Review of Neuroscience
,
35
,
73
89
.
Ptak
,
R.
(
2012
).
The frontoparietal attention network of the human brain: Action, saliency, and a priority map of the environment
.
Neuroscientist
,
18
,
502
515
.
Pylyshyn
,
Z.
(
2004
).
Some puzzling findings in multiple object tracking: I. Tracking without keeping track of object identities
.
Visual Cognition
,
11
,
801
822
.
Pylyshyn
,
Z. W.
, &
Storm
,
R. W.
(
1988
).
Tracking multiple independent targets: Evidence for a parallel tracking mechanism
.
Spatial Vision
,
3
,
179
197
.
Sawaguchi
,
T.
, &
Goldman-Rakic
,
P. S.
(
1991
).
D1 dopamine receptors in prefrontal cortex: Involvement in working memory
.
Science
,
251
,
947
950
.
Scimeca
,
J. M.
, &
Franconeri
,
S. L.
(
2015
).
Selecting and tracking multiple objects
.
Wiley Interdisciplinary Reviews: Cognitive Science
,
6
,
109
118
.
Scolari
,
M.
,
Seidl-Rathkopf
,
K. N.
, &
Kastner
,
S.
(
2015
).
Functions of the human frontoparietal attention network: Evidence from neuroimaging
.
Current Opinion in Behavioral Sciences
,
1
,
32
39
.
Serences
,
J. T.
,
Liu
,
T.
, &
Yantis
,
S.
(
2005
).
Parietal mechanisms of attentional control: Locations, features, and objects
. In
I.
Laurent
,
G.
Rees
, &
J. K.
Tsotsos
(Eds.),
Neurobiology of attention
(pp.
35
41
).
Philadelphia, PA
:
Elsevier
.
Serences
,
J. T.
,
Shomstein
,
S.
,
Leber
,
A. B.
,
Golay
,
X.
,
Egeth
,
H. E.
, &
Yantis
,
S.
(
2005
).
Coordination of voluntary and stimulus-driven attentional control in human cortex
.
Psychological Science
,
16
,
114
122
.
Serences
,
J. T.
, &
Yantis
,
S.
(
2006
).
Spatially selective representations of voluntary and stimulus-driven attentional priority in human occipital, parietal, and frontal cortex
.
Cerebral Cortex
,
17
,
284
293
.
Servan-Schreiber
,
D.
,
Printz
,
H.
, &
Cohen
,
J. D.
(
1990
).
A network model of catecholamine effects: Gain, signal-to-noise ratio, and behavior
.
Science
,
249
,
892
895
.
Shim
,
W. M.
,
Alvarez
,
G. A.
, &
Jiang
,
Y. V.
(
2008
).
Spatial separation between targets constrains maintenance of attention on multiple objects
.
Psychonomic Bulletin & Review
,
15
,
390
397
.
Shim
,
W. M.
,
Alvarez
,
G. A.
,
Vickery
,
T. J.
, &
Jiang
,
Y. V.
(
2009
).
The number of attentional foci and their precision are dissociated in the posterior parietal cortex
.
Cerebral Cortex
,
20
,
1341
1349
.
Srivastava
,
N.
, &
Vul
,
E.
(
2016
).
Attention modulates spatial precision in multiple-object tracking
.
Topics in Cognitive Science
,
8
,
335
348
.
Störmer
,
V. S.
,
Li
,
S. C.
,
Heekeren
,
H. R.
, &
Lindenberger
,
U.
(
2013
).
Normal aging delays and compromises early multifocal visual attention during object tracking
.
Journal of Cognitive Neuroscience
,
25
,
188
202
.
Strobach
,
T.
,
Wendt
,
M.
, &
Janczyk
,
M.
(
2018
).
Multitasking: Executive functioning in dual-task and task switching situations
.
Frontiers in Psychology
,
9
,
108
.
Thiele
,
A.
, &
Bellgrove
,
M. A.
(
2018
).
Neuromodulation of attention
.
Neuron
,
97
,
769
785
.
Tomasi
,
D.
,
Ernst
,
T.
,
Caparelli
,
E. C.
, &
Chang
,
L.
(
2004
).
Practice-induced changes of brain function during visual attention: A parametric fMRI study at 4 Tesla
.
Neuroimage
,
23
,
1414
1421
.
Tombu
,
M.
, &
Seiffert
,
A. E.
(
2008
).
Attentional costs in multiple-object tracking
.
Cognition
,
108
,
1
25
.
Unsworth
,
N.
, &
Robison
,
M. K.
(
2015
).
Individual differences in the allocation of attention to items in working memory: Evidence from pupillometry
.
Psychonomic Bulletin & Review
,
22
,
757
765
.
van der Wel
,
P.
, &
van Steenbergen
,
H.
(
2018
).
Pupil dilation as an index of effort in cognitive control tasks: A review
.
Psychonomic Bulletin & Review
,
25
,
2005
2015
.
Varazzani
,
C.
,
San-Galli
,
A.
,
Gilardeau
,
S.
, &
Bouret
,
S.
(
2015
).
Noradrenaline and dopamine neurons in the reward/effort trade-off: A direct electrophysiological comparison in behaving monkeys
.
Journal of Neuroscience
,
35
,
7866
7877
.
Vater
,
C.
,
Kredel
,
R.
, &
Hossner
,
E. J.
(
2017
).
Disentangling vision and attention in multiple-object tracking: How crowding and collisions affect gaze anchoring and dual-task performance
.
Journal of Vision
,
17
,
21
.
Vernet
,
M.
,
Quentin
,
R.
,
Chanes
,
L.
,
Mitsumasu
,
A.
, &
Valero-Cabré
,
A.
(
2014
).
Frontal eye field, where art thou? Anatomy, function, and non-invasive manipulation of frontal regions involved in eye movements and associated cognitive operations
.
Frontiers in Integrative Neuroscience
,
8
,
66
.
Vossel
,
S.
,
Geng
,
J. J.
, &
Fink
,
G. R.
(
2014
).
Dorsal and ventral attention systems: Distinct neural circuits but collaborative roles
.
Neuroscientist
,
20
,
150
159
.
Vul
,
E.
,
Alvarez
,
G.
,
Tenenbaum
,
J. B.
, &
Black
,
M. J.
(
2009
).
Explaining human multiple object tracking as resource-constrained approximate inference in a dynamic probabilistic model
.
Paper presented at the Advances in Neural Information Processing Systems
.
Weinshenker
,
D.
, &
Schroeder
,
J. P.
(
2007
).
There and back again: A tale of norepinephrine and drug addiction
.
Neuropsychopharmacology
,
32
,
1433
1451
.
Westbrook
,
A.
, &
Braver
,
T. S.
(
2015
).
Cognitive effort: A neuroeconomic approach
.
Cognitive, Affective, & Behavioral Neuroscience
,
15
,
395
415
.
Westbrook
,
A.
, &
Braver
,
T. S.
(
2016
).
Dopamine does double duty in motivating cognitive effort
.
Neuron
,
89
,
695
710
.
Williges
,
R. C.
, &
Wierwille
,
W. W.
(
1979
).
Behavioral measures of aircrew mental workload
.
Human Factors
,
21
,
549
574
.
Wright
,
T. J.
,
Boot
,
W. R.
, &
Morgan
,
C. S.
(
2013
).
Pupillary response predicts multiple object tracking load, error rate, and conscientiousness, but not inattentional blindness
.
Acta Psychologica
,
144
,
6
11
.
Xu
,
Y.
, &
Chun
,
M. M.
(
2009
).
Selecting and perceiving multiple visual objects
.
Trends in Cognitive Sciences
,
13
,
167
174
.
Yantis
,
S.
,
Schwarzbach
,
J.
,
Serences
,
J. T.
,
Carlson
,
R. L.
,
Steinmetz
,
M. A.
,
Pekar
,
J. J.
, et al
(
2002
).
Transient neural activity in human parietal cortex during spatial attention shifts
.
Nature Neuroscience
,
5
,
995
1002
.
Zelinsky
,
G. J.
, &
Todor
,
A.
(
2010
).
The role of “rescue saccades” in tracking objects through occlusions
.
Journal of Vision
,
10
,
29
.

Author notes

*

These authors contributed equally to the paper.