Age-related changes in attentional tracking of multiple moving objects

Transcription

1 Perception Age-related changes in attentional tracking of multiple moving objects Robert Sekuler, Chris McLaughlin and Yuko Yotsumoto Volen Center for Complex Systems, Brandeis University In a multiple object tracking (MOT) task, young and older adults attentively tracked a subset of ten identical, randomly moving disks for several seconds, and then tried to identify those disks that had comprised the subset. Young adults who habitually played video games performed significantly better than those who did not. Compared to young subjects (mean age = 2 years) with whom they were matched for video game experience, older subjects (mean age = 75.3 years ) showed much reduced ability to track multiple moving objects, particularly with faster movement or longer tracking times. Control measurements with stationary disks show that the age-related decline in MOT was not caused by a general change in memory per se. To generate an item-wise performance measure, we examined older subjects proportion correct according to the serial order in which individual disks were identified. Correct identifications of target disks declined with the serial order in which targets were selected, suggesting that attentional tracking produced graded outcomes. Word count: 3470 Keywords: Attentional tracking, aging, serial order, attention, identification Selective attention amplifies the effective processing of visual information (Scholl, 2001; Connor, Egeth, & Yantis, 2004), enhancing the salience or perceived contrast of attended stimuli (Carrasco, Ling, & Read, 2004; Treue, 2004). Much is known about how rapid shifts in momentary attention are affected by normal aging (e.g., Nissen & Corkin, 1985; Greenwood, Parasuraman, & Haxby, 1993; Gottlob & Madden, 1999; Curran, Hills, Patterson, & Strauss, 2001), but less is known about how age affects attention that must be maintained steadily for several seconds. This distinction between phasic and tonic aspects of visual attention is an important one as the two seem to depend upon distinct neural circuits (Yantis & Serences, 2003). To examine potential age-related changes in sustained attention we adapted the multiple object tracking (MOT) paradigm (Pylyshyn & Storm, 1988). This task embodies key elements of sustained attention required for many activities, including driving (Anstey, Wood, Lord, & Walker, 2005; Owsley et al., 1998; Sekuler, Bennett, & Mamelak, 2000). In addition, this task shares neural substrates with other higher-order cognitive functions (Culham et al., 1998). A trial in a typical MOT experiment begins with a computer display of n randomly-positioned, stationary identical objects. During the trial s second phase, a subset of the n stationary objects are defined, by flashing or temporarily changing color, as the targets that must be tracked. During the We thank Arthur Wingfield for facilitating recruitment of older subjects. Supported by NIH grant MH trial s thirds phase, which lasts several seconds, the n identical objects move about randomly. Finally, when movement ceases, the subject tries to identify the n objects that were to have been tracked. Because visual attention is altered by habitual playing particular genres of video games (Green & Bavelier, 2003; Risenhuber, 2004), observed differences between young and older adults MOT performance could arise not only from known age-related changes in visual attention, but also from age-related differential experience with video games. To take account of this possibility, our design included three groups of subjects: one group of older adults and two groups of young adults. Of groups with young subjects, one comprised habitual video game players, and the other comprised agematched individuals who rarely if ever played video games. In addition to assessing video game playing s possible contribution to age-related differences in MOT, we measured MOT under conditions differing in difficulty, by varying disk speed, duration of the disks random movements, and by varying how many disks had to be tracked. Finally, because success in MOT requires sustained attention throughout the movement period, momentary lapses of attention can disrupt performance, as it does in many other tasks (Weissman, Roberts, Visscher, & Woldorff, 2006). In fact, during debriefing after an MOT session, subjects often report that they could tell when a lapse in attention caused tracking of an item to fail. We wondered, though, whether attentional tracking was actually as all-or-none as subjects claimed, or whether it might be graded in strength, as seems the case in many other situations. By modifying a standard measure of MOT to include a serial order analysis, we produced a simple, unobtrusive indicator of the strength with

2 2 SEKULER, MCLAUGHLIN & YOTSUMOTO which individual items were tracked. Our questions, then, were: (1) how might age alter MOT performance when varying demands were made on sustained visual attention, either by changing the time during which targets moved, or by varying their speed of movement? (2) could age-related differences in MOT be explained by differential, age-related video-game experience? and (3) is memory for individual tracked items all-or-none or graded in strength? Stimuli and Task Experiment 1 Stimuli for each trial consisted of ten solid black disks, each 1 visual angle in diameter. For 750 msec at the start of a trial, subjects saw ten stationary disks distributed randomly within a gray square region on a computer monitor. The disks initial, random locations were constrained so that no disk could lie nearer than 1 to any edge of the square grey region, nor could any disk lie nearer than 1 to any other disk. After 750 msec, that trial s randomly-selected, target disks were highlighted, turning from black to yellow for 1500 msec, and then returning to black. All ten black disks then began to move around within the window, at either 20 or 30 deg/sec. The number of targets, that is, disks that were to be tracked, varied randomly from trial to trial, with 2, 3, 4, and 5 targets being equally frequent. The duration of a trial s movement phase alternated randomly between 5 and 10 seconds. To avoid potential distractions from the abrupt onset of motion, as the disks began to move, their speed ramped up linearly over one second, from 0 to the trial s terminal speed, either 20 or 30 /sec. After the last of the subject s responses, feedback was given by displaying the number of correct identifications. Once terminal speed had been attained, all disks moved along linear paths at constant speed. To help each disk maintain its perceptual identity (Metzger, 1934; Pylyshyn, 2004), the disks were prevented from overlapping one another. A collision detection routine prevented the moving disks from hitting either the edge of the gray square field within which they moved or from hitting one another. On each frame of the motion sequence, the computer determined whether on the immediately succeeding frame, if disks continue moving along their current path, one or more disks would contact a bounding edge or another disk. In this way, the moving disks were kept from approaching closer than one disk diameter (1 ) to one another, or closer than 0.3 to an edge of the field. At the conclusion of the movement period, all disks stopped moving and the previously invisible mouse pointer was made visible once more, at the screen s center. Subjects were instructed to click directly on the disks originally defined as targets. Subjects were required to select disks one after another until they had selected the same number as had been highlighted as targets at the trial s start. This task was made easier by making a disk disappear once it had been selected. A disk was scored as selected if at the time of a mouse click, the cursor had been positioned inside the disk s area, or within 0.5 of a disk s circumference. In our analysis, the resulting proportion of correct selections was supplemented by information about the serial order in which disks were selected. We hypothesized that as subjects tried to identify the disks they had been tracking, subjects would select first disks for which their memory was stronger, proceeding seriatim through disks for which memory strength was less (Koriat & Goldsmith, 1996). We then used variation in proportion correct as a function of serial order of selection as an unobtrusive indicator of memory strength. After practice trials, each subject completed 20 repetitions of each condition, for a total of 320 trials. A subject s head was supported by a chin cup and forehead rest, 114 cm from the display. Subjects Twenty non-video-game-playing young adults (NVGP), sixteen video-game-playing young adults (VGP), and ten older adults participated in this study for monetary compensation. VGP were defined as individuals who have played an average of five hours per week of first-person, action/adventure video games over the past six months. These video games strongly influence visual attention as defined by a variety of experimental tasks (Green & Bavelier, 2003; Risenhuber, 2004). Of NVGP subjects, ten were male and ten were female; of the VGP group, ten were male, six female. The ten older subjects, who were evenly split between males and females, were drawn from a pool of independent living, active, healthy older adults. Previous studies with subjects from this pool, though not with the present subjects themselves, showed a striking preservation of visual memory, both for items and serial position (Sekuler, Kahana, McLaughlin, Golomb, & Wingfield, 2005; Sekuler, McLaughlin, Kahana, Wingfield, & Yotsumoto, in press). By selecting subjects from this pool we expected to reduce the likelihood that MOT results would be contaminated by general, age-related changes in memory per se. Each subject s visual acuity was assessed using a Snellen eye chart; contrast sensitivity was measured with a Lighthouse Letter Contrast Sensitivity Test (Arditi, 2005). Both sets of results are shown in Table 1, which also summarizes each group s demographic characteristics and video game playing habits. It should be noted that NVGP subjects, as well as older subjects who reported playing any games, did not play the sort of video games that alter visual attention, that is, first-person, action/adventure games. All of the older subjects rated their health and vision as average to excellent; all reported that their daily activities were hardly impaired or not impaired at all by their health. The ten older subjects had at least some college education, and six held graduate degrees. Several additional questionnaires were administered in order to assess the physical health and mental status of the older subjects. Performance on the Mini- Mental State Examination ranged from 28-30, with a mean = 29.2, SD =.08, which is at or above the population-based norm of 28 for year olds with college experience or a higher degree (Crum, Anthony, Bassett, & Folstein,

3 AGE-RELATED CHANGES IN MULTIPLE OBJECT TRACKING 3 Table 1: Subject Characteristics and Screening Results VGP NVGP Older Age Range Mean±SD 2±3.1 2± ±5.7 Visual Acuity Range 20/16-20/22 20/16-20/22 20/16-20/35 Mean±SD 20/18±1.8 20/17.6±1.6 20/25.2±5.7 Contrast Sensitivity Range Mean±SD 1.71± ± ±0.10 Video Game Range Hours per Week Mean±SD 11.6± ± 0.3± ). The older subjects also completed portions of the Visual Activities Questionnaire (Sloane, Ball, Owsley, Bruni, & Roenker, 1992), using a five-item Likert scale to describe the frequency with which they had difficulties with visual processing in everyday activities. On this scale, a response of 1 signified never had difficulty and a response of 5 signified always had difficulty. Subjects responses ranged from (mean = 2.60, SD = 0, age-appropriate norm = 2.96) for questions measuring acuity and spatial vision, from (mean = 2.52, SD = 2, norm = 2.37) for questions measuring visual search, and from (mean = 2.22, SD = 0.27, norm = 2.16) for questions related to speed of visual processing. So although the older subjects educational background may be atypical for their cohort, their performance on the Visual Activities Questionnaire was quite close to the norms for their cohort. Experiment 2 All subjects from Experiment 1 also served in a supplementary test meant to assess possible baseline memory differences among subjects. The procedure differed from Experiment 1 only in that the disks did not move after targets had been highlighted. Subjects had to remember the position either of one or five stationary target disks for 0, 5, or 10 sec. Subjects selected a target disk by clicking either within the disk or within 0.5 of its edge. Results and Discussion After a trial s movement phase, subjects were required to select disks one after another until they had selected the same number as had been highlighted as targets at the trial s start. Data were analyzed first in a customary way, without regard to the order in which particular disks were selected, and without correction for guessing (Hulleman, 2005). Effect of video game playing on MOT. On average, the VGP group consistently performed at a higher level than did the NVGP group, with overall means and standard errors of 0.96±0.013 and 0.93±0.007, respectively (t(34) = , p <.04, two-tailed test). Because video game playing was consistently associated with higher performance, and because our older subjects mainly eschewed video games, the analysis of age-related differences put the VGP subjects to the side, and concentrates on young subjects from NVGP, a group whose video game playing habits matched those of those of the older subjects. Figure 1 shows that performance declined with increases in the number of disks that had to be tracked, with the effect considerably more pronounced for older subjects (Panel C). An ANOVA confirmed the statistical significance of this main effect (F(3, 84) = , p < 0.01); additionally, older subjects were affected more than young subjects by an increase in the number of target disks (F(3,84) = 22.11, p < 0.01). Moreover, performance dropped as a trial s movement phase lengthened from 5 to 10 seconds (F(1,28) = , p < 0.01), with older subjects performance dropping relatively more than that of younger subjects (F(1, 28) = 11.15, p < 0.01). Finally, performance was better when disks moved more slowly, at 20 rather than 30 /sec, (F(1,28) = 87, p < 0.01). As with the other variables, increasing speed disk affected older subjects more than younger ones, (F(1, 28) = 6.75, p < 0.02). Benchmarking MOT performance Figure 1 reminds us that subjects ability to track multiple objects is limited. To evaluate the severity of those limits, we compared the results to theoretical expectations from a simple model of performance. Assume that of j disks that the subject attempts to track, the subject actually succeeds in tracking just k disks. Let the remaining j k choices that the subject is required to make be pure guesses, based on no information. If an MOT display contains N total objects, n=n-k is the pool of items from which the j k guesses are made (without replacement). Allowing i to stand for the number of guesses that are correct, we can compute E(x), the expected value of making x correct choices when only i items, i < x, are securely in memory, where, P i = E(x) = ( j k i j k k + ip i i=1 j )( ) n ( j k) j k i ( ) n j k (1)

4 4 SEKULER, MCLAUGHLIN & YOTSUMOTO Video Game Players Non-Video Game Players A B C Older Subjects _20 5_30 10_20 10_30 4 disks 3 disks 2 disks Trra acked 5_20 5_30 10_20 10_30 4 disks 3 disks 2 disks Trra acked 5_20 5_30 10_20 10_30 4 disks 3 disks 2 disks Trra acked Number of Disks in Target Set Number of Disks in Target Set Number of Disks in Target Set 0.5 Figure 1: Proportion correct identification varied with movement speed and number of target disks that were tracked. Open symbols represent movement at 20 /sec; closed symbols represent movement at 30 /sec. Red symbols are for movement duration of 5 sec; blue symbols are for movement duration of 10 sec. Panels A-C: Results for young video-game players, young non-video game players, and older subjects, respectively. Error bars represent ±1 SeM. The data points shown are not corrected for guessing. Dashed contours show predicted proportion correct assuming the subjects tracked 2, 3, or 4 disks perfectly, making information-free guesses for their remaining responses ( ) x and represents the number of combinations of x items y taken y at a time. We computed E(x) for the conditions in Experiment 1, assuming that the subject successfully tracked 2, 3, or 4 disks. The values of E(x) are shown by the dashed gray, iso-performance contours in each panel of Figure 1. A data point s location relative to a contour indicates whether observed performance exceeds the model s prediction. For example, as all four data points associated with n=5 in Figure 1A lie above the contour for 4 disks, the model suggests that the VGP subjects successfully tracked four of the five disks. In contrast, Figure 1C shows that older subjects successfully tracked at most three disks, and just two at the higher speed and longer duration. Memory control: Stationary disks. With just one stationary disk to hold in memory, every subject in both age groups achieved perfect performance, regardless of the retention interval, 0, 5 or 10 seconds. With five disks, although subjects in both groups continued to perform at a very high level, but did make some errors. With five disks, young and older subjects proportions of correct responses were 0.994±0.002 and 0.980±0.006, respectively. This small difference between groups was statistically significant, t(12) = 2.29, p <.04, but is insufficient in size to account for the substantial age-related differences in MOT (up to 15-20%) shown in Figure 1. It is important to note that neither group s memory for stationary disks varied as the retention interval increased from 0 to 5 to 10 seconds, p >.5. This means that decreases in older subjects performance observed when Experiment 1 s movement phase increased from 5 to 10 seconds probably did not come from a simple, time-dependent loss of items from working memory while subjects made their identifications. Instead, that decrease in performance most likely came from difficulty in continuing to track the positions of the moving objects. Serial order analysis of memory strength. Our examination of memory strength in MOT was stimulated by pilot subjects self-reports that they knew when an tracked item had fallen out of memory. Such claims are consistent with Kolers and Palef (1976) s suggestion that subjects can know directly that they do not know; they are consistent also with the notion that an item is either in memory or not, with no intermediate strengths. On each trial, our software forced subjects to identify the same number of disks as had been highlighted at the start of the trial. This mark all method (Hulleman, 2005) of data collection allowed us to examine the proportion correct identifications as function of the serial order of each choice, from first to last. In order not to bias their choices, subjects received no instructions about the order or speed with which they ought to identify target disks. We hypothesized that subjects successive selections of possible target disks would be influenced by their confidence that particular disks were in the original target set. In particular, we hypothesized that as they tried to identify the disks they had been tracking, subjects would first select disks for which they were most confident, and only later select disks for which they were less confident. Subjects successfully applying such a strategy would be more likely to err on later selections, misidentifying a non-target disk as having been among the disks to be tracked. An unbiased measure of performance that takes account of the order in which disks were selected requires a correction

5 AGE-RELATED CHANGES IN MULTIPLE OBJECT TRACKING 5 A Video Game Players B Non-Video Game Players C Older Subjects Targets 3 Targets 4 Targets 5 Targets 2 Targets 3 Targets 4 Targets 5 Targets 2 Targets 3 Targets 4 Targets 5 Targets Order of Disk Selection Order of Disk Selection Order of Disk Selection 0 Figure 2: Temporal order of identifications for young video game players (A), young non-video game players (B) and older subjects (C). Each panel contains the mean proportion of correct responses, corrected for guessing, averaged over speed of movement and duration of the trial s movement phase, using Eq. 2 to correct for guessing. The parameter of each family of curves is the number of target disks, 2-5. Note that data for young subjects are for non-video game players only. Error bars represent ±1 SeM for guessing that recognizes that (i) each successive identification altered the probability that one of the remaining nonselected disks could be identified correctly by chance alone, and that (ii) the probability of success by guessing depended also on the success of the previous identifications. Such a correction for i th selection can be obtained from the equation C i = n i 1 k=1 Pk T (i 1) where C i is the chance level for the i th selection, n is the number of target disks, Pk is the proportion correct of the preceding selections, and T is the total number of disks on the trial, both target and non-target. For our application, T =10. Note that this correction assumes that subjects sample without replacement, never selecting the same disk twice. This assumption was enforced by making a disk disappear following its selection, which kept the disk from being reselected. Note also that the plausibility of this correction requires that no feedback be given after each identification; otherwise, subjects could exploit such feedback to optimize succeeding response(s). The results, corrected for guessing on an item-by-item basis, are shown in Figures 2A and B, for young, NVGP subjects and older subjects, respectively. Data have been averaged over values of speed and duration of disk movement. Because the young subjects performed at such a high level overall, their serial order curves, shown in Figure 2A and B, are relatively flat, and therefore are fairly uninformative. In contrast, older subjects serial order results reveal characteristics of memory that might otherwise be obscured. To illustrate this point, look back to older subjects results shown in (2) Figure 1C. There, when five targets are being tracked, in even the most challenging conditions older subjects on average successfully track about 70% of the five disks. But Figure 2B casts the result into a different perspective, namely, that older subjects average proportion correct with five targets (in Figure 1B) is actually an aggregate of distinctly heterogeneous values. While trying to tracking five targets, older subjects first and second selections demonstrate a very high accuracy, in fact not noticeably below the corresponding values for younger subjects. So, older subjects had two of the five target disks firmly in memory. But when succeeding selections are considered, performance falls off sharply, suggesting that these later-selected targets were held in memory less firmly that the first two. On average, then, older subjects may appear to do fairly well tracking five targets, but this is misleading, the result of averaging over memories of markedly differing strength. In MOT, as in many other tasks, memory strength is graded. Absence of short-term practice effects. With practice on complex visual tasks, older as well as younger subjects sometimes exhibit rapid, long-lasting improvement in performance (e.g., Sekuler & Ball, 1986), including in tasks involving motion perception (Ball & Sekuler, 1986). To determine whether two hours intensive practice might have influenced our MOT results, we compared young and older subjects performance in each quarter of the experiment. A repeated measures ANOVA revealed no change in performance from the first to the last quarter of the experiment (F(1,28) = 1.81, p > 0.15), with both young and older subjects showing stable performance (p > ). Of course, we cannot rule out the possibility that additional practice would

6 6 SEKULER, MCLAUGHLIN & YOTSUMOTO have improved performance, particularly for the older subjects, whose initial performance was further from the upper limit of. Video game experience and age. While the present manuscript was being prepared, Trick, Perl, and Sethi (2005) independently reported an age-related difference in multiple object tracking. Although their general methods and subject populations resembled those we used, our experimental design took explicit account of a potentially-important variable that had been omitted by Trick et al. (2005): agerelated differences in experience playing video games. Comparisons of NVGP and VGP subjects showed that video game playing is associated with enhanced MOT performance. As a result, the substantial age-related differences in MOT reported here would have been even more exaggerated had we not evaluated and eliminated from the main analysis young subjects who frequently play action video games. Given the prevelance of video gaming among college students in North America (Jones, 2003) and elsewhere, and given game playing s demonstrated effects, not just on MOT but on other attention-demanding tasks (Green & Bavelier, 2003; Risenhuber, 2004), researchers interested in visual attention, including age-related changes in visual attention, should take notice of their subjects video game playing habits. References Anstey, K. J., Wood, J., Lord, S., & Walker, J. G. (2005). Cognitive, sensory and physical factors enabling driving safety in older adults. Clinical Psychological Review, 25, Arditi, A. (2005). Improving the design of the letter contrast sensitivity test. Investigative Ophthalmology & Visual Science. Ball, K., & Sekuler, R. (1986). Improving visual perception in older observers. Journal of Gerontology, 41, Carrasco, M., Ling, S., & Read, S. (2004). Attention alters appearance. Nature Neuroscience, 7, Connor, C. E., Egeth, H. E., & Yantis, S. (2004). Visual attention: bottom-up versus top-down. Current Biology, 14, R Crum, R. M., Anthony, J. C., Bassett, S. S., & Folstein, M. F. (1993). Population-based norms for the Mini-Mental Status Examination by age and education level. Journal of the American Medical Association, 269, 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, Curran, T., Hills, A., Patterson, M. B., & Strauss, M. E. (2001). Effects of aging on visuospatial attention: an ERP study. Neuropsychologia, 39, Gottlob, L. R., & Madden, D. J. (1999). Age similarities in the inertial properties of attention. Perception & Psychophysics, 61(4), Green, C. S., & Bavelier, D. (2003). Action video game modifies visual selective attention. Nature, 423(6939), Greenwood, P. M., Parasuraman, R., & Haxby, J. V. (1993). Changes in visuospatial attention over the adult lifespan. Neuropsychologia, 31(5), Hulleman, J. (2005). The mathematics of multiple object tracking: From proportions correct to number of objects tracked. Vision Research, 45, Jones, S. (2003). Let the games begin: Gaming technology and entertainment among college students (Report of the Internet & American Life Project). Washington, D.C.: The Pew Foundation. Kolers, P. A., & Palef, S. R. (1976). Knowing not. Memory & Cognition, 4(5), Koriat, A., & Goldsmith, M. (1996). Monitoring and control processes in the strategic regulation of memory accuracy. Psychological Review, 103, Metzger, W. (1934). Beobachtungen über phänomenale identität. Psychologische Forschung, 19, Nissen, M. J., & Corkin, S. (1985). Effectiveness of attentional cueing in older and younger adults. Journal of Gerontology, 40(2), Owsley, C., Ball, K., McGwin, G. J., Sloane, M. E., Roenker, D. L., White, M. F., & Overley, E. T. (1998). Visual processing impairment and risk of motor vehicle crash among older adults. Journal of the American Medical Association, 279, Pylyshyn, Z. W. (2004). Some puzzling findings in multiple object tracking: I. Tracking without keeping track of object identities. Visual Cognition, 11, Pylyshyn, Z. W., & Storm, R. W. (1988). Tracking multiple independent targets: evidence for a parallel tracking mechanism. Spatial Vision, 3(3), Risenhuber, M. (2004). An action video game modifies visual processing. Trends in Neuroscience, 27(2), Scholl, B. J. (2001). Objects and attention: the state of the art. Cognition, 80(1-2), Sekuler, A. B., Bennett, P. J., & Mamelak, M. (2000). Effects of aging on the useful field of view. Experimental Aging Research, 26, Sekuler, R., & Ball, K. (1986). Visual localization: age and practice. Journal of the Optical Society of America, A, 3(6), Sekuler, R., Kahana, M. J., McLaughlin, C., Golomb, J., & Wingfield, A. (2005). Preservation of episodic visual recognition memory in aging. Experimental Aging Research Aging, 31(1), Sekuler, R., McLaughlin, C., Kahana, M. J., Wingfield, A., & Yotsumoto, Y. (in press). Visual recognition and temporal order information are both well-preserved in aging. Psychology & Aging. Sloane, M. E., Ball, K., Owsley, C., Bruni, J. R., & Roenker, D. L. (1992). The visual activities questionnaire: Developing an instrument for assessing problems in everyday visual tasks. In Technical Digest, Noninvasive Assessment of the Visual System, Topical Meeting of the Optical Society of America (Vol. 1, p ). Washington, D. C.: Optical Society of America. Treue, S. (2004). Perceptual enhancement of contrast by attention. Trends in Cognitive Science, 8(10), Trick, L. M., Perl, T., & Sethi, N. (2005). Age-related differences in multiple-object tracking. Journal of Gerontology, B: Psychological Science, Social Science, 60(2), P Weissman, D. H., Roberts, K. C., Visscher, K. M., & Woldorff, M. G. (2006). The neural bases of momentary lapses in attention. Nature Neuroscience, 9(7), Yantis, S., & Serences, J. T. (2003). Cortical mechanisms of spacebased and object-based attentional control. Current Opinion in Neurobiology, 13(2),