This article was downloaded by: [Northwestern University] On: 03 February 2015, At: 20:25 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Cognitive Neuroscience Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/pcns20
Multiple attentional control settings influence late attentional selection but do not provide an early attentional filter a
a
Maha Adamo , Carson Pun & Susanne Ferber
a
a
University of Toronto , Toronto, Canada Published online: 18 Mar 2010.
To cite this article: Maha Adamo , Carson Pun & Susanne Ferber (2010) Multiple attentional control settings influence late attentional selection but do not provide an early attentional filter, Cognitive Neuroscience, 1:2, 102-110, DOI: 10.1080/17588921003646149 To link to this article: http://dx.doi.org/10.1080/17588921003646149
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
COGNITIVE NEUROSCIENCE, 2010, 1 (2), 102–110
Multiple attentional control settings influence late attentional selection but do not provide an early attentional filter
PCNS
Downloaded by [Northwestern University] at 20:25 03 February 2015
Multiple Attentional Control Sets
Maha Adamo, Carson Pun, and Susanne Ferber University of Toronto, Toronto, Canada
When one is responding to targets containing a specific feature, non-predictive peripheral cues that share this feature lead to faster responses to the target, while cues that do not contain the target feature effectively are ignored, providing evidence for the role of attentional control settings (ACSs) in the contingent capture hypothesis. It is unclear, however, at what stage of processing multiple ACSs are implemented. We took advantage of the excellent temporal resolution of electroencephalography to demonstrate that the maintenance of multiple ACSs influences later stages of attentional selection rather than providing an early attentional filter. N2pc analyses for cues and targets revealed a similar degree of spatial capture for any peripheral cue, regardless of control settings, with target P3s reflecting the application of the ACS color contingencies.
Keywords: Attentional control setting; Contingent capture; ERP.
INTRODUCTION In our typical environments, multiple sources of information compete for our attention at any given time, with more salient items automatically capturing our attention. Such stimulus-driven attentional orientation is fairly adaptive, as it biases attention toward significant stimuli, enhances perception, and ultimately leads to faster, more efficient behavioral responses (Posner & Petersen, 1990). Nonetheless, it is equally adaptive to apply top-down mediated goals to direct attention voluntarily, such that we shift the focus of our attention only to stimuli that meet our current goals. This idea underscores the contingent attentional capture hypothesis (Folk, Remington, & Johnston, 1992; Folk, Remington, & Wright, 1994): The task goal of detecting a target, defined by a particular stimulus feature, establishes an attentional control set (ACS) which in turn determines which stimuli will receive prioritized processing, namely only those that
contain the elementary target-defining feature. Thus stimulus-driven capture is contingent on top-down control settings, providing an interface between voluntary attentional allocation and automatic capture. Recently we explored the flexibility of ACSs by testing whether more than one goal can be maintained at a time (Adamo, Pun, Pratt, & Ferber, 2008). Participants were required to respond to targets defined by one critical feature on one side (e.g., green on the left) and only to targets with a different critical feature on the opposite side (e.g., blue on the right), thus generating two ACSs. Cues presented prior to targets were either congruent or incongruent with respect to the target’s spatial location and/or color. We found that fully congruent trials led to the fastest target reaction times (RTs), while fully incongruent trials resulted in the slowest RTs; critically, partially congruent trials (i.e., congruent spatial location but incongruent color, or vice versa) did not generate significantly different RTs from one another or from trials with no cue. We
Correspondence should be addressed to: Susanne Ferber, 100 St George Street, Room 4020, Toronto, Ontario, M5S3G3, Canada. E-mail: ferber@psych. utoronto.ca This project was funded by the Natural Sciences and Engineering Research Council, the Canadian Institutes for Health Research, and an Early Researcher Award to SF.
© 2010 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business www.psypress.com/cognitiveneuroscience DOI: 10.1080/17588921003646149
Downloaded by [Northwestern University] at 20:25 03 February 2015
MULTIPLE ATTENTIONAL CONTROL SETS
concluded that the two ACSs could be maintained simultaneously, with attentional capture generated only by cues that matched the ACS for the side of space on which they appeared. Despite our compelling behavioral evidence that the ability for top-down settings to influence bottomup processing is both flexible enough to account for multiple behavioral goals and specific enough to apply these goals over distinct regions, our RT-based findings do not provide unequivocal evidence that capture was only generated at each location by cues that fit the ACS for that side. That is, any cue might initiate spatial capture of attention, with rapid disengagement for cues that do not match an ACS (Theeuwes, Atchley, & Kramer, 2000). Alternatively, the role of the top-down settings may also be to enhance later attentional selection only following cues that match an ACS. Indeed, the latter alternative is more compelling in tasks where multiple ACSs must be maintained, as each ACS is defined by a particular color anchored to a specific location. That is, feature-based attention and spatial selection must operate in conjunction, which may be accomplished as two systems functioning in tandem or as one system modulating the other. While behavioral studies may lack sufficient sensitivity to tell these different accounts apart, electrophysiological studies provide a more refined answer. Therefore, the present study employed a similar paradigm to that previously reported (Adamo et al., 2008), with some modifications to make it suitable for event-related potential (ERP) analyses. Specifically, we were interested in two key ERP components: the N2pc and the P3. The N2pc reflects the degree to which stimuli engage the locus of attention (Brisson & Jolicoeur, 2008; Eimer, 1996; Luck & Hillyard, 1994). If the strongest application of the contingent capture hypothesis is correct, with capture only for cues that match an ACS, we would expect that only such cues would show a significant N2pc. If, on the other hand, any cue can lead to attentional capture, then any cue should show an N2pc. The ERPs for targets should further discriminate the level of processing at which attentional control settings operate, assessed both by target N2pc and by the second waveform of interest, the P3, whose amplitude and latency provide measures of late attentional selection and consolidation of targets into working memory (Kok, 2001; Picton, 1992; Polich & Kok, 1995). Target N2pcs should indicate the degree to which attention is engaged at the target location, which should vary as a function of the congruence of the preceding cue, and target P3s should reflect the availability of targets for conscious report, thus dictating behavioral response times.
103
METHODS Participants Thirty young adults were recruited from the University of Toronto and received bonus course credit and/or monetary compensation for their participation. Of these, datasets from 20 participants met selection criteria for inclusion in analyses (see Results). These participants (12 females, all right-handed; age range: 18–34 years, mean: 22.3 years) had normal or corrected-to-normal vision. Participants gave informed consent in accordance with the University of Toronto Ethics Review Board.
Apparatus The experiment was programmed and displayed using Presentation® software (Version 11.0, www.neurobs.com) running on a desktop PC, with a ViewSonic 21-inch monitor (1600 × 1200 resolution, 60 Hz refresh rate). The viewing distance was 57 cm, and participants made responses with their right hand on the spacebar of a standard keyboard. Continuous, unreferenced electroencephalography (EEG) was recorded at a sampling rate of 512 Hz using a BioSemi ActiveTwo system with 64 scalp electrodes in standard 10–20 placement and an additional electrode at each mastoid, at the outer canthus of each eye, and below each eye.
Stimuli and procedure The basic display, which was present throughout each trial, consisted of a central fixation cross (subtending 1° of visual angle) with two placeholder boxes centered at 5° eccentricity to the left and right (each subtending 2°). The fixation and placeholders were presented in pale gray against the black background. Cues consisted of a blue, green, or pale gray (i.e., neutral) border around one placeholder paired with the neutral border around the other placeholder (increasing each placeholder’s visual occlusion to 2.5°). Targets were blue or green squares (1.5°) that appeared unilaterally within either placeholder (see Figure 1). Each trial began with a blank screen for an intertrial interval (ITI) of 1000 ms, after which the appearance of the basic display signaled the onset of a trial. After 1000 ms presentation of the basic display, a cue appeared for 50 ms, followed by a 100 ms delay and then the appearance of a target in either placeholder for 100 ms (see Figure 1a). The trial ended and the ITI
Downloaded by [Northwestern University] at 20:25 03 February 2015
104
ADAMO, PUN, FERBER
fixated, to ignore cues when they appeared as they were non-predictive of the upcoming target, and to respond as quickly and as accurately as possible to targets depending on their color and spatial location. Half of the participants employed “blue-left” and “green-right” control sets: They responded only to blue targets appearing in the left placeholder and only to green targets appearing in the right placeholder, while ignoring green targets on the left and blue targets on the right. The other half of participants were given “green-left” and “blue-right” control sets. The experiment began with 10 practice trials with no cues to familiarize the participant with the response requirements, plus an additional 10 practice trials with cues, none of which were included in analyses. Trials were categorized according to the spatial (S) and color (C) congruence between colored cues and targets (see Figure 1b), yielding the following conditions: S+C+, where the colored cue was congruent with both the location and the color of the target; S+C–, where the colored cue appeared in the same location as the impending target but in the opposite color; S–C+, where the colored cue was the same color as the impending target but appeared at the opposite location; S–C–, where the colored cue appeared on the opposite side and in the opposite color relative to the impending target. Note that the fully congruent (S+C+) and fully incongruent (S–C–) conditions have colored cues that fit the ACS for the location at which they appear (“good” cues) while the partially congruent conditions (S+C– and S–C+) do not (“bad” cues). Finally, neutral trials had neutral cues on both sides and had no relevance for either ACS. Figure 1. (a) Temporal presentation of a single trial (S+C–) and (b) schematic representation of the four cue conditions, with cues and targets depicted together for simplicity (see text for details of all conditions and response requirements). These examples assume “blue-left” and “green–right” attentional control sets (ACSs). (c) Group mean response times to correctly detected targets, with error bars representing the standard error of the mean per condition. The leftmost bar represents the fully congruent condition (S+C+), the rightmost the fully incongruent condition (S–C–), and the middle bars the partially incongruent conditions (S+C– and S–C+).
was initiated as soon as a response was made, with the basic display remaining on the screen for 750 ms on trials in which no response was made. The same number of blue, green, and neutral cues appeared with equal probability at each placeholder. Likewise, the blue and green targets appeared with equal frequency within either placeholder. Cue color and location were fully crossed with target color and location, with 30 trials presented in each combination. Participants were instructed to keep their eyes centrally
ERP processing EEG datasets were converted into EEProbe format (version 3.3.118) using PolyRex (Kayser, 2003). Each individual dataset was filtered using a finite impulse response (FIR) filter high-passed at 0.5 Hz and lowpassed at 30 Hz. The continuous data were then re-referenced to the average of the two mastoids. Additionally, we computed lateral eye movements by re-referencing the right outer eye to the left outer eye. Rejection markers were then generated over all eye channels in addition to frontopolar channels FP1 and FP2. Each continuous dataset was averaged to generate individual ERPs: Cue epochs started 100 ms before and ended 500 ms after cue onset, while target epochs spanned from 100 ms before to 900 ms after target onset, with the 100 ms prior to stimulus onset used as a baseline for all analyses and the rejection window
MULTIPLE ATTENTIONAL CONTROL SETS
spanning both the cue and target epochs. Average waveforms for each individual were then imported to Matlab (version 7.0.4.352) for peak analyses using custom scripts.
RESULTS
Downloaded by [Northwestern University] at 20:25 03 February 2015
Exclusions We included only data from participants who showed a hit rate of greater than 90% and a false alarm rate of less than 15%. Additionally, in order to maintain an acceptable number of trials for ERP averages, we excluded participants whose datasets had fewer than 15 clean trials per condition per side. These criteria resulted in the exclusion of a total of 10 participants, leaving 20 acceptable datasets with average hit and false alarm rates of 98.6% and 3.9%, respectively, and an average of 24.4 artifact-free trials per condition per side contributing to final analyses.
Target RTs Because a relatively low false alarm rate was requisite to ensure that both ACSs were effectively maintained, likely resulting in floor effects for errors, only reaction times for correctly detected targets were analyzed. Anticipatory (RT < 150 ms) and extremely delayed (RT > 2000 ms) responses were excluded from analyses. Hits were collapsed across left and right targets for a single mean RT measure per condition for each individual (see Figure 1c for group performance), which were then submitted to a repeated measures ANOVA with two factors: spatial congruence (S+ vs. S–) and color congruence (C+ vs. C–). Confirming our behavioral study (Adamo et al., 2008), we found significant main effects of spatial congruence, F(1, 19) = 54.30, MSE = 389.23, p < .0001, and of color congruence, F(1, 19) = 88.82, MSE = 406.32, p < .0001, demonstrating that responses were fastest when the location and color of the cue and target were congruent (S+C+) and slowest when the location and color of the cue and target were incongruent (S–C–). The interaction between spatial and color congruence was also significant, F(1, 19) = 20.15, MSE = 205.71, p < .001, indicating that cues that fit an ACS were more effective at speeding processing for S+C+ trials than they were impeding processing for S–C– trials. Table 1 reports all planned pairwise t-tests (Bonferroni-corrected for a familywise error rate of 0.05), confirming that fully congruent and fully incongruent cues resulted in the fastest
105
TABLE 1 Pairwise comparisons between conditions
S+C+ S+C– S–C+ S–C–
S+C–
S–C+
S–C–
Neutral
8.85
7.54 1.70, p > .1
11.19 3.98 6.30
6.94 3.95 0.64, p > .5 5.96
All t-tests have df=19. Unless otherwise noted, the significance level is p < .005.
and slowest target RTs, respectively, and targets following the two types of partially congruent cue (S+C– and S–C+), which didn’t fit either ACS, did not differ significantly from one another.
Cue N2pc In order to examine the neural response to cues, we collapsed the ERPs for left- and right-sided cues that fit an ACS (“good” cues) or did not (“bad” cues), irrespective of the stimulus that followed (i.e., cue responses were collapsed across trials in which targets or non-targets followed at the same or the opposite location, effectively averaging out any activity related to this subsequent stimulus). We then calculated difference waves between contralateral and ipsilateral ERPs collapsed respectively over parietal (P7/8), parieto-occipital (PO7/8), and occipital (O1/2) channels (see Figure 2), and measured the negative deflection occurring between 200 and 300 ms after the onset of the cues. We submitted cue N2pc latencies and amplitudes to separate pairwise t-tests to compare the effects of cue condition (good vs. bad). While both good and bad cues generated an N2pc of significant amplitude, t(19)=5.40 and 5.94, respectively, both ps < .0001, there were no significant differences between cue conditions for either N2pc amplitude or latency, both t values < 1. Thus, any peripheral cue was associated with a shift of attention regardless of whether it fit an ACS.
Target N2pc We examined the degree of attentional engagement by correctly detected targets following the various cue conditions (see Figure 3) by computing the difference wave between target ERPs collapsed over contralateral and ipsilateral channels (same as above) and calculating target N2pc as the negative deflection between 200 and 300 ms. We submitted the peak amplitudes and latencies (see Figure 4) to separate
Downloaded by [Northwestern University] at 20:25 03 February 2015
106
ADAMO, PUN, FERBER
Figure 2. Grand-averaged waveforms for cues that (a) matched the ACS for the side on which they were presented (“Good” cues) and (b) did not match the ACS for the side on which they were presented (“Bad” cues), at contralateral and ipsilateral sites collapsed across parietal (P7/8), parieto-occipital (PO7/8) and occipital (O1/ 2) channels. The magnitude of the N2pc was not influenced by the nature of the cues.
repeated measures ANOVAs with factors of spatial congruence (S+ vs. S–) and color congruence (C+ vs.C–). Target N2pc latency (see Figure 4a) was affected by spatial congruence, F(1, 19) = 15.16, MSE = 565.44, p < .001, with overall faster peak times for targets following spatially congruent cues. The effect of color congruence was also significant, F(1, 19) = 7.03, MSE = 255.39, p < .02, such that peak times were slowed when the target color was congruent with that of the preceding cue. We found no interaction between spatial and color congruence, F(1, 19) = 0.41, MSE = 224.52, p > .5. With regard to target N2pc amplitude (see Figure 4b), we again found a significant effect of spatial congruence, F(1, 19) = 10.78, MSE = 1.35, p < .005, such that N2pc was significantly greater (i.e., larger negative amplitudes) for targets following spatially congruent cues, indicating a
Figure 3. Grand-averaged waveforms for correctly detected targets at parietal (Pz, P3/4, P7/8) and parieto-occipital (POz, PO3/4, PO7/8) sites in each cue condition, collapsed at the indicated sites.
greater degree of visual selection for the target when the cue had previously drawn attention to the target’s location, regardless of the color of the cue. Altogether,
107
Peak latency (ms)
320 300
400 380 360 340
460 440 420
500 480
210 200
230 220
250 240
270 260
290 280
300
Incongruent location (S-)
Congruent location (S+)
Incongruent location (S-)
Congruent color (C+) Incongruent color (C-) Neutral
(c) Target P3 latency
Congruent location (S+)
Congruent color (C+) Incongruent color (C-) Neutral
(a) Target N2pc latency
6 5
8 7
11 10 9
13 12
15 14
-5
-4
-3
-2
-1
0
1
2
Incongruent location (S-)
Congruent location (S+)
Incongruent location (S-)
Congruent color (C+) Incongruent color (C-) Neutral
(d) Target P3 amplitude
Congruent location (S+)
Congruent color (C+) Incongruent color (C-) Neutral
(b) Target N2pc amplitude
Downloaded by [Northwestern University] at 20:25 03 February 2015
Figure 4. (a, b) Target N2pc latencies and amplitudes. The target N2pc peaked earlier and had greater amplitude when the cue had accurately predicted the target’s location, regardless of the cue’s color. (c, d) Target P3 latencies and amplitudes for ERPs displayed in Figure 3, collapsed across all sites.
Peak latency (ms)
Peak amplitude (uV)
Peak amplitude (uV)
108
ADAMO, PUN, FERBER
these results indicate that target N2pc was faster and larger when the target’s location had just been cued, regardless of the cue’s color.
Downloaded by [Northwestern University] at 20:25 03 February 2015
Target P3 Finally, the target P3 analyses were intended to provide insights into late attentional processing and encoding into working memory. We analysed only trials in which the target was correctly detected, collapsing across left- and right-sided targets for each cue condition, as with the behavioral analysis. We computed P3 as the maximum peak within the window from 300 ms to 600 ms after target onset over parietal (Pz, P3/4, P7/8) and parieto-occipital (POz, PO3/4, PO7/8) sites, and submitted peak amplitude and peak latency to separate repeated measures ANOVAs with three factors: spatial congruence (S+ vs. S–), color congruence (C+ vs. C–), and electrode site (ipsilateral vs. midline vs. contralateral). The analysis of P3 peak amplitude (see Figure 4d) revealed a significant effect of color congruence, F(1, 19) = 14.15, MSE = 5.00, p < .002, with no effect of spatial congruence, F(1, 19) = 0.07, MSE = 9.86, p > .7, nor any interaction between the two, F(1, 19) = 0.00, MSE = 6.85, p > .9. We did, however, observe a main effect of site, F(2, 38) = 47.01, MSE = 4.56, p < .0001, with larger amplitudes overall at the midline site relative to lateral sites, in addition to an interaction between spatial congruence and site, F(2, 38) = 3.48, MSE = 0.68, p < .05, driven by higher ipsilateral amplitudes following any spatially congruent cue. The analysis of P3 peak latency (see Figure 4c) revealed a different pattern of results. We again found a significant main effect of color congruence, F(1, 19) = 14.92, MSE = 4928.50, p < .002. The effect of spatial congruence on P3 latency, however, failed to reach significance, F(1, 19) = 0.01, MSE = 5998.80, p < .9, due to the highly significant interaction with color congruence, F(1, 19) = 27.44, MSE = 2282.26, p < .0001. In other words, a cue that matched the ACS on the same side as the target sped the time to P3 peak, while a cue that did not fit the ACS on the target side slowed the P3 peak. This pattern held across all channels tested, with no effects of site on peak latency. In summary, target P3 latency primarily reflected the color congruence of cues that appeared on the same side as the target—that is, whether or not a spatially congruent cue matched the ACS for that side— while effects on target P3 amplitude were dominated by whether the preceding cue appeared in the same color as the target, regardless of whether that cue fit an ACS.
DISCUSSION The behavioral outcome of the current study replicated our previously published results: Participants were fastest to respond to targets that followed fully congruent (S+C+) cues and slowest for targets following fully incongruent (S–C–) cues, meaning that cues that matched an ACS had the greatest behavioral effect on subsequent targets. Partially congruent (S+C– and S–C+) cues, which did not match either ACS, did not lead to significantly different RTs from one another. While the observed behavioral profile suggests that only cues that exactly match an ACS generated attentional capture, it could be argued that all peripheral cues initiated capture but attention was rapidly disengaged when the cue did not match an ACS (Theeuwes et al., 2000). These very rapid attentional shifts may not be reflected in measures of reaction times; ERPs, however, should reveal their electrophysiological correlate. In fact, the cue ERPs show that both “good” and “bad” cues elicited comparable N2pc amplitudes. These findings are complemented by our target N2pc amplitude results. These were affected by spatial but not color congruence of the preceding cues, with greater visual selection of the target when its location had just been cued, regardless of whether or not that cue matched an ACS. In other words, even “bad” cues captured attention, allowing comparable attentional selection on the target side with no evidence of rapid disengagement. The rapid disengagement hypothesis would predict that spatial attention is reallocated within roughly 50 ms, which would result in an N2pc for the S+C– cue, as we observed, but followed by a diminished N2pc for the subsequent target at that location. Our results, however, indicate that targets following any spatially congruent cue elicited a larger N2pc, even when the preceding cue at that location did not fit the ACS for that location. While these results indicate rapid engagement of attention regardless of cue type, we found no evidence for rapid disengagement. How can this be reconciled with our behavioral finding for target RTs indicating that “good” cues confer some benefit on target processing when they appeared on the same side as the target? We found the electrophysiological signature of this behavioral advantage at a later processing stage: P3 peaked earliest for targets following fully congruent (S+C+) cues, and latest for targets following spatially congruent cues that were incongruent with respect to color (S+C–), with no differences in P3 latency for spatially incongruent (S–C+ and S–C–) cues. Target P3 amplitudes, on the other hand, reflected only the color congruence of the
Downloaded by [Northwestern University] at 20:25 03 February 2015
MULTIPLE ATTENTIONAL CONTROL SETS
cue and target, regardless of spatial congruence, suggesting that multiple mechanisms are at play in contingent attentional selection. Our task was designed to establish two attentional control settings, each anchored to a particular spatial location: Each rule generating an ACS was defined by a particular combination of color and location. The observed behavior demonstrates the successful implementation of both components of the rules. Nonetheless, our ERP findings suggest that this is achieved by a combination of space- and feature-based attentional selection operating in parallel until the point at which the target is consolidated into working memory, rather than an early, low-level, space-based mechanism that is tuned by color or vice versa. Feature-based attention seems to operate globally, such that looking for a particular target feature enhances neural representations for that feature across the visual cortex, irrespective of the particular location where stimuli containing that feature appear (Liu, Slotnick, Serences, & Yantis, 2003; Maunsell & Treue, 2006; Sàenz, Bura0as, & Boynton, 2003). In other words, attending to a particular feature generates a spatially non-selective template of that feature. We found that target P3 had larger peak amplitudes following cues that predicted the target color, regardless of the side of cue presentation, suggesting that the color-congruent cues activated the template against which the subsequent target was compared; when the target matched this activated template, P3 was larger and recognition memory for the target was enhanced (Kok, 2001). Target N2pcs, on the other hand, reflected spatial congruence irrespective of color and are much better aligned with a space-based account of attentional selection, as N2pc is elicited by stimuli that follow any spatially predictive, task-relevant information even prior to target selection (Eimer, 1998; Kiss, Van Velzen & Eimer, 2008). The combination of these underlying mechanisms resulted in the observed profile of RTs. That is, when a target follows a cue that has both spatial and color congruence, processing of that target benefits from enhanced spatial orientation, seen in the N2pc, and from matching an activated template, seen in the P3 amplitude (i.e., S+C+ trials): together these resulted in faster time to P3 peak, reflecting speeded consolidation and thus faster RT. When the cue allows for either spatial selection, as seen in a larger N2pc, but not feature-based selection (i.e., S+C– trials), or when the cue generates feature-based selection, seen in greater target P3 amplitude, but not spatial selection (i.e., S–C+ trials), the resulting RTs for targets are intermediate and indistinguishable from one another at the point of behavioral output. Finally, targets following
109
cues that do not promote either spatial orientation or feature-based selection (i.e., S–C– trials) show the slowest RTs, as target processing shows no enhancements in either N2pc or P3 amplitude. Target P3 latency, then, reflects the speed at which items are consolidated in working memory depending on these prior processes, allowing access to conscious report and response selection, which may be the point at which space-based and feature-based attention are integrated. Thus, this study confirms that multiple attentional control settings can be applied, but it also sheds light on the fundamental mechanisms by which top-down settings influence late attentional selection to produce the behavioral profile of contingent attentional capture. Manuscript received 13 October 2009 Manuscript accepted 8 January 2010 First published online 18 March 2010
REFERENCES Adamo, M., Pun, C., Pratt, J., & Ferber, S. (2008). Your divided attention, please! The maintenance of multiple attentional control sets over distinct regions in space. Cognition, 107(1), 295–303. Brisson, B., & Jolicoeur, P. (2008). Express attentional reengagement but delayed entry into consciousness following invalid spatial cues in visual search. PLoS ONE, 3(12), e3967. Eimer, M. (1996). The N2pc component as an indicator of attentional selectivity. Electroencephalography and Clinical Neuropsychology, 99, 225–234. Eimer, M. (1998). Mechanisms of visuospatial attention: Evidence from event-related brain potentials. Visual Cognition, 5(1/2), 257–286. Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary covert orienting is contingent on attentional control settings. Journal of Experimental Psychology: Human Perception & Performance, 18(4), 1030–1044. Folk, C. L., Remington, R. W., & Wright, J. H. (1994). The structure of attentional control: Contingent attentional capture by apparent motion, abrupt onset, and color. Journal of Experimental Psychology: Human Perception & Performance, 20(2), 317–329. Kayser, J. (2003). Polygraphic recording data exchange – PolyRex (http://psychophysiology.cpmc.columbia.edu/ PolyRex.htm). New York: New York State Psychiatric Institute. Kiss, M., Van Velzen, J., & Eimer, M. (2008). The N2pc component and its links to attention shifts and spatially selective visual processing. Psychophysiology, 45(2), 240–249. Kok, A. (2001). On the utility of P3 amplitude as a measure of processing capacity. Psychophysiology, 38, 557–577. Liu, T., Slotnick, S. D., Serences, J., & Yantis, S. (2003). Cortical mechanisms of feature-based attentional control. Cerebral Cortex, 13, 1334–1343. Luck, S. J., & Hillyard, S. A. (1994). Spatial filtering during visual search: Evidence from human electrophysiology.
110
ADAMO, PUN, FERBER
Downloaded by [Northwestern University] at 20:25 03 February 2015
Journal of Experimental Psychology: Human Perception & Performance, 20(5), 1000–1014. Maunsell, J.H. R, & Treue, S. (2006). Feature-based attention in visual cortex. Trends in Neurosciences 29(6), 317–322. Picton, T. (1992). The P300 wave of the human eventrelated potential. Journal of Clinical Neurophysiology, 9(4), 456–479. Polich, J., & Kok, A. (1995). Cognitive and biological determinants of P300: An integrative review. Biological Psychology, 41, 103–146.
Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain. Annual Reviews of Neuroscience, 13, 25–42. Sàenz, M., Bura0as, G. T., & Boynton, G. M. (2003). Global feature-based attention for motion and color. Vision Research, 43, 629–637. Theeuwes, J., Atchley, P., & Kramer, A. F. (2000). On the time course of top-down and bottom-up control of visual attention. In S. Monsell & J. Driver (Eds.), Control of cognitive processes: Attention and performance XVII (pp. 105–125). Cambridge, MA: MIT Press.
Cognitive Neuroscience Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/pcns20
Multiple attentional control settings influence late attentional selection but do not provide an early attentional filter a
a
Maha Adamo , Carson Pun & Susanne Ferber
a
a
University of Toronto , Toronto, Canada Published online: 18 Mar 2010.
To cite this article: Maha Adamo , Carson Pun & Susanne Ferber (2010) Multiple attentional control settings influence late attentional selection but do not provide an early attentional filter, Cognitive Neuroscience, 1:2, 102-110, DOI: 10.1080/17588921003646149 To link to this article: http://dx.doi.org/10.1080/17588921003646149
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
COGNITIVE NEUROSCIENCE, 2010, 1 (2), 102–110
Multiple attentional control settings influence late attentional selection but do not provide an early attentional filter
PCNS
Downloaded by [Northwestern University] at 20:25 03 February 2015
Multiple Attentional Control Sets
Maha Adamo, Carson Pun, and Susanne Ferber University of Toronto, Toronto, Canada
When one is responding to targets containing a specific feature, non-predictive peripheral cues that share this feature lead to faster responses to the target, while cues that do not contain the target feature effectively are ignored, providing evidence for the role of attentional control settings (ACSs) in the contingent capture hypothesis. It is unclear, however, at what stage of processing multiple ACSs are implemented. We took advantage of the excellent temporal resolution of electroencephalography to demonstrate that the maintenance of multiple ACSs influences later stages of attentional selection rather than providing an early attentional filter. N2pc analyses for cues and targets revealed a similar degree of spatial capture for any peripheral cue, regardless of control settings, with target P3s reflecting the application of the ACS color contingencies.
Keywords: Attentional control setting; Contingent capture; ERP.
INTRODUCTION In our typical environments, multiple sources of information compete for our attention at any given time, with more salient items automatically capturing our attention. Such stimulus-driven attentional orientation is fairly adaptive, as it biases attention toward significant stimuli, enhances perception, and ultimately leads to faster, more efficient behavioral responses (Posner & Petersen, 1990). Nonetheless, it is equally adaptive to apply top-down mediated goals to direct attention voluntarily, such that we shift the focus of our attention only to stimuli that meet our current goals. This idea underscores the contingent attentional capture hypothesis (Folk, Remington, & Johnston, 1992; Folk, Remington, & Wright, 1994): The task goal of detecting a target, defined by a particular stimulus feature, establishes an attentional control set (ACS) which in turn determines which stimuli will receive prioritized processing, namely only those that
contain the elementary target-defining feature. Thus stimulus-driven capture is contingent on top-down control settings, providing an interface between voluntary attentional allocation and automatic capture. Recently we explored the flexibility of ACSs by testing whether more than one goal can be maintained at a time (Adamo, Pun, Pratt, & Ferber, 2008). Participants were required to respond to targets defined by one critical feature on one side (e.g., green on the left) and only to targets with a different critical feature on the opposite side (e.g., blue on the right), thus generating two ACSs. Cues presented prior to targets were either congruent or incongruent with respect to the target’s spatial location and/or color. We found that fully congruent trials led to the fastest target reaction times (RTs), while fully incongruent trials resulted in the slowest RTs; critically, partially congruent trials (i.e., congruent spatial location but incongruent color, or vice versa) did not generate significantly different RTs from one another or from trials with no cue. We
Correspondence should be addressed to: Susanne Ferber, 100 St George Street, Room 4020, Toronto, Ontario, M5S3G3, Canada. E-mail: ferber@psych. utoronto.ca This project was funded by the Natural Sciences and Engineering Research Council, the Canadian Institutes for Health Research, and an Early Researcher Award to SF.
© 2010 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business www.psypress.com/cognitiveneuroscience DOI: 10.1080/17588921003646149
Downloaded by [Northwestern University] at 20:25 03 February 2015
MULTIPLE ATTENTIONAL CONTROL SETS
concluded that the two ACSs could be maintained simultaneously, with attentional capture generated only by cues that matched the ACS for the side of space on which they appeared. Despite our compelling behavioral evidence that the ability for top-down settings to influence bottomup processing is both flexible enough to account for multiple behavioral goals and specific enough to apply these goals over distinct regions, our RT-based findings do not provide unequivocal evidence that capture was only generated at each location by cues that fit the ACS for that side. That is, any cue might initiate spatial capture of attention, with rapid disengagement for cues that do not match an ACS (Theeuwes, Atchley, & Kramer, 2000). Alternatively, the role of the top-down settings may also be to enhance later attentional selection only following cues that match an ACS. Indeed, the latter alternative is more compelling in tasks where multiple ACSs must be maintained, as each ACS is defined by a particular color anchored to a specific location. That is, feature-based attention and spatial selection must operate in conjunction, which may be accomplished as two systems functioning in tandem or as one system modulating the other. While behavioral studies may lack sufficient sensitivity to tell these different accounts apart, electrophysiological studies provide a more refined answer. Therefore, the present study employed a similar paradigm to that previously reported (Adamo et al., 2008), with some modifications to make it suitable for event-related potential (ERP) analyses. Specifically, we were interested in two key ERP components: the N2pc and the P3. The N2pc reflects the degree to which stimuli engage the locus of attention (Brisson & Jolicoeur, 2008; Eimer, 1996; Luck & Hillyard, 1994). If the strongest application of the contingent capture hypothesis is correct, with capture only for cues that match an ACS, we would expect that only such cues would show a significant N2pc. If, on the other hand, any cue can lead to attentional capture, then any cue should show an N2pc. The ERPs for targets should further discriminate the level of processing at which attentional control settings operate, assessed both by target N2pc and by the second waveform of interest, the P3, whose amplitude and latency provide measures of late attentional selection and consolidation of targets into working memory (Kok, 2001; Picton, 1992; Polich & Kok, 1995). Target N2pcs should indicate the degree to which attention is engaged at the target location, which should vary as a function of the congruence of the preceding cue, and target P3s should reflect the availability of targets for conscious report, thus dictating behavioral response times.
103
METHODS Participants Thirty young adults were recruited from the University of Toronto and received bonus course credit and/or monetary compensation for their participation. Of these, datasets from 20 participants met selection criteria for inclusion in analyses (see Results). These participants (12 females, all right-handed; age range: 18–34 years, mean: 22.3 years) had normal or corrected-to-normal vision. Participants gave informed consent in accordance with the University of Toronto Ethics Review Board.
Apparatus The experiment was programmed and displayed using Presentation® software (Version 11.0, www.neurobs.com) running on a desktop PC, with a ViewSonic 21-inch monitor (1600 × 1200 resolution, 60 Hz refresh rate). The viewing distance was 57 cm, and participants made responses with their right hand on the spacebar of a standard keyboard. Continuous, unreferenced electroencephalography (EEG) was recorded at a sampling rate of 512 Hz using a BioSemi ActiveTwo system with 64 scalp electrodes in standard 10–20 placement and an additional electrode at each mastoid, at the outer canthus of each eye, and below each eye.
Stimuli and procedure The basic display, which was present throughout each trial, consisted of a central fixation cross (subtending 1° of visual angle) with two placeholder boxes centered at 5° eccentricity to the left and right (each subtending 2°). The fixation and placeholders were presented in pale gray against the black background. Cues consisted of a blue, green, or pale gray (i.e., neutral) border around one placeholder paired with the neutral border around the other placeholder (increasing each placeholder’s visual occlusion to 2.5°). Targets were blue or green squares (1.5°) that appeared unilaterally within either placeholder (see Figure 1). Each trial began with a blank screen for an intertrial interval (ITI) of 1000 ms, after which the appearance of the basic display signaled the onset of a trial. After 1000 ms presentation of the basic display, a cue appeared for 50 ms, followed by a 100 ms delay and then the appearance of a target in either placeholder for 100 ms (see Figure 1a). The trial ended and the ITI
Downloaded by [Northwestern University] at 20:25 03 February 2015
104
ADAMO, PUN, FERBER
fixated, to ignore cues when they appeared as they were non-predictive of the upcoming target, and to respond as quickly and as accurately as possible to targets depending on their color and spatial location. Half of the participants employed “blue-left” and “green-right” control sets: They responded only to blue targets appearing in the left placeholder and only to green targets appearing in the right placeholder, while ignoring green targets on the left and blue targets on the right. The other half of participants were given “green-left” and “blue-right” control sets. The experiment began with 10 practice trials with no cues to familiarize the participant with the response requirements, plus an additional 10 practice trials with cues, none of which were included in analyses. Trials were categorized according to the spatial (S) and color (C) congruence between colored cues and targets (see Figure 1b), yielding the following conditions: S+C+, where the colored cue was congruent with both the location and the color of the target; S+C–, where the colored cue appeared in the same location as the impending target but in the opposite color; S–C+, where the colored cue was the same color as the impending target but appeared at the opposite location; S–C–, where the colored cue appeared on the opposite side and in the opposite color relative to the impending target. Note that the fully congruent (S+C+) and fully incongruent (S–C–) conditions have colored cues that fit the ACS for the location at which they appear (“good” cues) while the partially congruent conditions (S+C– and S–C+) do not (“bad” cues). Finally, neutral trials had neutral cues on both sides and had no relevance for either ACS. Figure 1. (a) Temporal presentation of a single trial (S+C–) and (b) schematic representation of the four cue conditions, with cues and targets depicted together for simplicity (see text for details of all conditions and response requirements). These examples assume “blue-left” and “green–right” attentional control sets (ACSs). (c) Group mean response times to correctly detected targets, with error bars representing the standard error of the mean per condition. The leftmost bar represents the fully congruent condition (S+C+), the rightmost the fully incongruent condition (S–C–), and the middle bars the partially incongruent conditions (S+C– and S–C+).
was initiated as soon as a response was made, with the basic display remaining on the screen for 750 ms on trials in which no response was made. The same number of blue, green, and neutral cues appeared with equal probability at each placeholder. Likewise, the blue and green targets appeared with equal frequency within either placeholder. Cue color and location were fully crossed with target color and location, with 30 trials presented in each combination. Participants were instructed to keep their eyes centrally
ERP processing EEG datasets were converted into EEProbe format (version 3.3.118) using PolyRex (Kayser, 2003). Each individual dataset was filtered using a finite impulse response (FIR) filter high-passed at 0.5 Hz and lowpassed at 30 Hz. The continuous data were then re-referenced to the average of the two mastoids. Additionally, we computed lateral eye movements by re-referencing the right outer eye to the left outer eye. Rejection markers were then generated over all eye channels in addition to frontopolar channels FP1 and FP2. Each continuous dataset was averaged to generate individual ERPs: Cue epochs started 100 ms before and ended 500 ms after cue onset, while target epochs spanned from 100 ms before to 900 ms after target onset, with the 100 ms prior to stimulus onset used as a baseline for all analyses and the rejection window
MULTIPLE ATTENTIONAL CONTROL SETS
spanning both the cue and target epochs. Average waveforms for each individual were then imported to Matlab (version 7.0.4.352) for peak analyses using custom scripts.
RESULTS
Downloaded by [Northwestern University] at 20:25 03 February 2015
Exclusions We included only data from participants who showed a hit rate of greater than 90% and a false alarm rate of less than 15%. Additionally, in order to maintain an acceptable number of trials for ERP averages, we excluded participants whose datasets had fewer than 15 clean trials per condition per side. These criteria resulted in the exclusion of a total of 10 participants, leaving 20 acceptable datasets with average hit and false alarm rates of 98.6% and 3.9%, respectively, and an average of 24.4 artifact-free trials per condition per side contributing to final analyses.
Target RTs Because a relatively low false alarm rate was requisite to ensure that both ACSs were effectively maintained, likely resulting in floor effects for errors, only reaction times for correctly detected targets were analyzed. Anticipatory (RT < 150 ms) and extremely delayed (RT > 2000 ms) responses were excluded from analyses. Hits were collapsed across left and right targets for a single mean RT measure per condition for each individual (see Figure 1c for group performance), which were then submitted to a repeated measures ANOVA with two factors: spatial congruence (S+ vs. S–) and color congruence (C+ vs. C–). Confirming our behavioral study (Adamo et al., 2008), we found significant main effects of spatial congruence, F(1, 19) = 54.30, MSE = 389.23, p < .0001, and of color congruence, F(1, 19) = 88.82, MSE = 406.32, p < .0001, demonstrating that responses were fastest when the location and color of the cue and target were congruent (S+C+) and slowest when the location and color of the cue and target were incongruent (S–C–). The interaction between spatial and color congruence was also significant, F(1, 19) = 20.15, MSE = 205.71, p < .001, indicating that cues that fit an ACS were more effective at speeding processing for S+C+ trials than they were impeding processing for S–C– trials. Table 1 reports all planned pairwise t-tests (Bonferroni-corrected for a familywise error rate of 0.05), confirming that fully congruent and fully incongruent cues resulted in the fastest
105
TABLE 1 Pairwise comparisons between conditions
S+C+ S+C– S–C+ S–C–
S+C–
S–C+
S–C–
Neutral
8.85
7.54 1.70, p > .1
11.19 3.98 6.30
6.94 3.95 0.64, p > .5 5.96
All t-tests have df=19. Unless otherwise noted, the significance level is p < .005.
and slowest target RTs, respectively, and targets following the two types of partially congruent cue (S+C– and S–C+), which didn’t fit either ACS, did not differ significantly from one another.
Cue N2pc In order to examine the neural response to cues, we collapsed the ERPs for left- and right-sided cues that fit an ACS (“good” cues) or did not (“bad” cues), irrespective of the stimulus that followed (i.e., cue responses were collapsed across trials in which targets or non-targets followed at the same or the opposite location, effectively averaging out any activity related to this subsequent stimulus). We then calculated difference waves between contralateral and ipsilateral ERPs collapsed respectively over parietal (P7/8), parieto-occipital (PO7/8), and occipital (O1/2) channels (see Figure 2), and measured the negative deflection occurring between 200 and 300 ms after the onset of the cues. We submitted cue N2pc latencies and amplitudes to separate pairwise t-tests to compare the effects of cue condition (good vs. bad). While both good and bad cues generated an N2pc of significant amplitude, t(19)=5.40 and 5.94, respectively, both ps < .0001, there were no significant differences between cue conditions for either N2pc amplitude or latency, both t values < 1. Thus, any peripheral cue was associated with a shift of attention regardless of whether it fit an ACS.
Target N2pc We examined the degree of attentional engagement by correctly detected targets following the various cue conditions (see Figure 3) by computing the difference wave between target ERPs collapsed over contralateral and ipsilateral channels (same as above) and calculating target N2pc as the negative deflection between 200 and 300 ms. We submitted the peak amplitudes and latencies (see Figure 4) to separate
Downloaded by [Northwestern University] at 20:25 03 February 2015
106
ADAMO, PUN, FERBER
Figure 2. Grand-averaged waveforms for cues that (a) matched the ACS for the side on which they were presented (“Good” cues) and (b) did not match the ACS for the side on which they were presented (“Bad” cues), at contralateral and ipsilateral sites collapsed across parietal (P7/8), parieto-occipital (PO7/8) and occipital (O1/ 2) channels. The magnitude of the N2pc was not influenced by the nature of the cues.
repeated measures ANOVAs with factors of spatial congruence (S+ vs. S–) and color congruence (C+ vs.C–). Target N2pc latency (see Figure 4a) was affected by spatial congruence, F(1, 19) = 15.16, MSE = 565.44, p < .001, with overall faster peak times for targets following spatially congruent cues. The effect of color congruence was also significant, F(1, 19) = 7.03, MSE = 255.39, p < .02, such that peak times were slowed when the target color was congruent with that of the preceding cue. We found no interaction between spatial and color congruence, F(1, 19) = 0.41, MSE = 224.52, p > .5. With regard to target N2pc amplitude (see Figure 4b), we again found a significant effect of spatial congruence, F(1, 19) = 10.78, MSE = 1.35, p < .005, such that N2pc was significantly greater (i.e., larger negative amplitudes) for targets following spatially congruent cues, indicating a
Figure 3. Grand-averaged waveforms for correctly detected targets at parietal (Pz, P3/4, P7/8) and parieto-occipital (POz, PO3/4, PO7/8) sites in each cue condition, collapsed at the indicated sites.
greater degree of visual selection for the target when the cue had previously drawn attention to the target’s location, regardless of the color of the cue. Altogether,
107
Peak latency (ms)
320 300
400 380 360 340
460 440 420
500 480
210 200
230 220
250 240
270 260
290 280
300
Incongruent location (S-)
Congruent location (S+)
Incongruent location (S-)
Congruent color (C+) Incongruent color (C-) Neutral
(c) Target P3 latency
Congruent location (S+)
Congruent color (C+) Incongruent color (C-) Neutral
(a) Target N2pc latency
6 5
8 7
11 10 9
13 12
15 14
-5
-4
-3
-2
-1
0
1
2
Incongruent location (S-)
Congruent location (S+)
Incongruent location (S-)
Congruent color (C+) Incongruent color (C-) Neutral
(d) Target P3 amplitude
Congruent location (S+)
Congruent color (C+) Incongruent color (C-) Neutral
(b) Target N2pc amplitude
Downloaded by [Northwestern University] at 20:25 03 February 2015
Figure 4. (a, b) Target N2pc latencies and amplitudes. The target N2pc peaked earlier and had greater amplitude when the cue had accurately predicted the target’s location, regardless of the cue’s color. (c, d) Target P3 latencies and amplitudes for ERPs displayed in Figure 3, collapsed across all sites.
Peak latency (ms)
Peak amplitude (uV)
Peak amplitude (uV)
108
ADAMO, PUN, FERBER
these results indicate that target N2pc was faster and larger when the target’s location had just been cued, regardless of the cue’s color.
Downloaded by [Northwestern University] at 20:25 03 February 2015
Target P3 Finally, the target P3 analyses were intended to provide insights into late attentional processing and encoding into working memory. We analysed only trials in which the target was correctly detected, collapsing across left- and right-sided targets for each cue condition, as with the behavioral analysis. We computed P3 as the maximum peak within the window from 300 ms to 600 ms after target onset over parietal (Pz, P3/4, P7/8) and parieto-occipital (POz, PO3/4, PO7/8) sites, and submitted peak amplitude and peak latency to separate repeated measures ANOVAs with three factors: spatial congruence (S+ vs. S–), color congruence (C+ vs. C–), and electrode site (ipsilateral vs. midline vs. contralateral). The analysis of P3 peak amplitude (see Figure 4d) revealed a significant effect of color congruence, F(1, 19) = 14.15, MSE = 5.00, p < .002, with no effect of spatial congruence, F(1, 19) = 0.07, MSE = 9.86, p > .7, nor any interaction between the two, F(1, 19) = 0.00, MSE = 6.85, p > .9. We did, however, observe a main effect of site, F(2, 38) = 47.01, MSE = 4.56, p < .0001, with larger amplitudes overall at the midline site relative to lateral sites, in addition to an interaction between spatial congruence and site, F(2, 38) = 3.48, MSE = 0.68, p < .05, driven by higher ipsilateral amplitudes following any spatially congruent cue. The analysis of P3 peak latency (see Figure 4c) revealed a different pattern of results. We again found a significant main effect of color congruence, F(1, 19) = 14.92, MSE = 4928.50, p < .002. The effect of spatial congruence on P3 latency, however, failed to reach significance, F(1, 19) = 0.01, MSE = 5998.80, p < .9, due to the highly significant interaction with color congruence, F(1, 19) = 27.44, MSE = 2282.26, p < .0001. In other words, a cue that matched the ACS on the same side as the target sped the time to P3 peak, while a cue that did not fit the ACS on the target side slowed the P3 peak. This pattern held across all channels tested, with no effects of site on peak latency. In summary, target P3 latency primarily reflected the color congruence of cues that appeared on the same side as the target—that is, whether or not a spatially congruent cue matched the ACS for that side— while effects on target P3 amplitude were dominated by whether the preceding cue appeared in the same color as the target, regardless of whether that cue fit an ACS.
DISCUSSION The behavioral outcome of the current study replicated our previously published results: Participants were fastest to respond to targets that followed fully congruent (S+C+) cues and slowest for targets following fully incongruent (S–C–) cues, meaning that cues that matched an ACS had the greatest behavioral effect on subsequent targets. Partially congruent (S+C– and S–C+) cues, which did not match either ACS, did not lead to significantly different RTs from one another. While the observed behavioral profile suggests that only cues that exactly match an ACS generated attentional capture, it could be argued that all peripheral cues initiated capture but attention was rapidly disengaged when the cue did not match an ACS (Theeuwes et al., 2000). These very rapid attentional shifts may not be reflected in measures of reaction times; ERPs, however, should reveal their electrophysiological correlate. In fact, the cue ERPs show that both “good” and “bad” cues elicited comparable N2pc amplitudes. These findings are complemented by our target N2pc amplitude results. These were affected by spatial but not color congruence of the preceding cues, with greater visual selection of the target when its location had just been cued, regardless of whether or not that cue matched an ACS. In other words, even “bad” cues captured attention, allowing comparable attentional selection on the target side with no evidence of rapid disengagement. The rapid disengagement hypothesis would predict that spatial attention is reallocated within roughly 50 ms, which would result in an N2pc for the S+C– cue, as we observed, but followed by a diminished N2pc for the subsequent target at that location. Our results, however, indicate that targets following any spatially congruent cue elicited a larger N2pc, even when the preceding cue at that location did not fit the ACS for that location. While these results indicate rapid engagement of attention regardless of cue type, we found no evidence for rapid disengagement. How can this be reconciled with our behavioral finding for target RTs indicating that “good” cues confer some benefit on target processing when they appeared on the same side as the target? We found the electrophysiological signature of this behavioral advantage at a later processing stage: P3 peaked earliest for targets following fully congruent (S+C+) cues, and latest for targets following spatially congruent cues that were incongruent with respect to color (S+C–), with no differences in P3 latency for spatially incongruent (S–C+ and S–C–) cues. Target P3 amplitudes, on the other hand, reflected only the color congruence of the
Downloaded by [Northwestern University] at 20:25 03 February 2015
MULTIPLE ATTENTIONAL CONTROL SETS
cue and target, regardless of spatial congruence, suggesting that multiple mechanisms are at play in contingent attentional selection. Our task was designed to establish two attentional control settings, each anchored to a particular spatial location: Each rule generating an ACS was defined by a particular combination of color and location. The observed behavior demonstrates the successful implementation of both components of the rules. Nonetheless, our ERP findings suggest that this is achieved by a combination of space- and feature-based attentional selection operating in parallel until the point at which the target is consolidated into working memory, rather than an early, low-level, space-based mechanism that is tuned by color or vice versa. Feature-based attention seems to operate globally, such that looking for a particular target feature enhances neural representations for that feature across the visual cortex, irrespective of the particular location where stimuli containing that feature appear (Liu, Slotnick, Serences, & Yantis, 2003; Maunsell & Treue, 2006; Sàenz, Bura0as, & Boynton, 2003). In other words, attending to a particular feature generates a spatially non-selective template of that feature. We found that target P3 had larger peak amplitudes following cues that predicted the target color, regardless of the side of cue presentation, suggesting that the color-congruent cues activated the template against which the subsequent target was compared; when the target matched this activated template, P3 was larger and recognition memory for the target was enhanced (Kok, 2001). Target N2pcs, on the other hand, reflected spatial congruence irrespective of color and are much better aligned with a space-based account of attentional selection, as N2pc is elicited by stimuli that follow any spatially predictive, task-relevant information even prior to target selection (Eimer, 1998; Kiss, Van Velzen & Eimer, 2008). The combination of these underlying mechanisms resulted in the observed profile of RTs. That is, when a target follows a cue that has both spatial and color congruence, processing of that target benefits from enhanced spatial orientation, seen in the N2pc, and from matching an activated template, seen in the P3 amplitude (i.e., S+C+ trials): together these resulted in faster time to P3 peak, reflecting speeded consolidation and thus faster RT. When the cue allows for either spatial selection, as seen in a larger N2pc, but not feature-based selection (i.e., S+C– trials), or when the cue generates feature-based selection, seen in greater target P3 amplitude, but not spatial selection (i.e., S–C+ trials), the resulting RTs for targets are intermediate and indistinguishable from one another at the point of behavioral output. Finally, targets following
109
cues that do not promote either spatial orientation or feature-based selection (i.e., S–C– trials) show the slowest RTs, as target processing shows no enhancements in either N2pc or P3 amplitude. Target P3 latency, then, reflects the speed at which items are consolidated in working memory depending on these prior processes, allowing access to conscious report and response selection, which may be the point at which space-based and feature-based attention are integrated. Thus, this study confirms that multiple attentional control settings can be applied, but it also sheds light on the fundamental mechanisms by which top-down settings influence late attentional selection to produce the behavioral profile of contingent attentional capture. Manuscript received 13 October 2009 Manuscript accepted 8 January 2010 First published online 18 March 2010
REFERENCES Adamo, M., Pun, C., Pratt, J., & Ferber, S. (2008). Your divided attention, please! The maintenance of multiple attentional control sets over distinct regions in space. Cognition, 107(1), 295–303. Brisson, B., & Jolicoeur, P. (2008). Express attentional reengagement but delayed entry into consciousness following invalid spatial cues in visual search. PLoS ONE, 3(12), e3967. Eimer, M. (1996). The N2pc component as an indicator of attentional selectivity. Electroencephalography and Clinical Neuropsychology, 99, 225–234. Eimer, M. (1998). Mechanisms of visuospatial attention: Evidence from event-related brain potentials. Visual Cognition, 5(1/2), 257–286. Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary covert orienting is contingent on attentional control settings. Journal of Experimental Psychology: Human Perception & Performance, 18(4), 1030–1044. Folk, C. L., Remington, R. W., & Wright, J. H. (1994). The structure of attentional control: Contingent attentional capture by apparent motion, abrupt onset, and color. Journal of Experimental Psychology: Human Perception & Performance, 20(2), 317–329. Kayser, J. (2003). Polygraphic recording data exchange – PolyRex (http://psychophysiology.cpmc.columbia.edu/ PolyRex.htm). New York: New York State Psychiatric Institute. Kiss, M., Van Velzen, J., & Eimer, M. (2008). The N2pc component and its links to attention shifts and spatially selective visual processing. Psychophysiology, 45(2), 240–249. Kok, A. (2001). On the utility of P3 amplitude as a measure of processing capacity. Psychophysiology, 38, 557–577. Liu, T., Slotnick, S. D., Serences, J., & Yantis, S. (2003). Cortical mechanisms of feature-based attentional control. Cerebral Cortex, 13, 1334–1343. Luck, S. J., & Hillyard, S. A. (1994). Spatial filtering during visual search: Evidence from human electrophysiology.
110
ADAMO, PUN, FERBER
Downloaded by [Northwestern University] at 20:25 03 February 2015
Journal of Experimental Psychology: Human Perception & Performance, 20(5), 1000–1014. Maunsell, J.H. R, & Treue, S. (2006). Feature-based attention in visual cortex. Trends in Neurosciences 29(6), 317–322. Picton, T. (1992). The P300 wave of the human eventrelated potential. Journal of Clinical Neurophysiology, 9(4), 456–479. Polich, J., & Kok, A. (1995). Cognitive and biological determinants of P300: An integrative review. Biological Psychology, 41, 103–146.
Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain. Annual Reviews of Neuroscience, 13, 25–42. Sàenz, M., Bura0as, G. T., & Boynton, G. M. (2003). Global feature-based attention for motion and color. Vision Research, 43, 629–637. Theeuwes, J., Atchley, P., & Kramer, A. F. (2000). On the time course of top-down and bottom-up control of visual attention. In S. Monsell & J. Driver (Eds.), Control of cognitive processes: Attention and performance XVII (pp. 105–125). Cambridge, MA: MIT Press.
