Concurrent deployment of visual attention and response selection bottleneck in a dual-task: Electrophysiological and behavioural evidence

Abstract

Visual attention and response selection are limited in capacity. Here, we investigated whether visual attention requires the same bottleneck mechanism as response selection in a dual-task of the psychological refractory period (PRP) paradigm. The dual-task consisted of an auditory two-choice discrimination Task 1 and a conjunction search Task 2, which were presented at variable temporal intervals (stimulus onset asynchrony, SOA). In conjunction search, visual attention is required to select items and to bind their features resulting in a serial search process around the items in the search display (i.e., set size). We measured the reaction time of the visual search task (RT2) and the N2pc, an event-related potential (ERP), which reflects lateralized visual attention processes. If the response selection processes in Task 1 influence the visual attention processes in Task 2, N2pc latency and amplitude would be delayed and attenuated at short SOA compared to long SOA. The results, however, showed that latency and amplitude were independent of SOA, indicating that visual attention was concurrently deployed to response selection. Moreover, the RT2 analysis revealed an underadditive interaction of SOA and set size. We concluded that visual attention does not require the same bottleneck mechanism as response selection in dual-tasks.

Keywords

Electrophysiology N2pc Psychological refractory period Response selection bottleneck Visual attention Visual search

People experience difficulties when performing multiple tasks at the same time. The paradigm of the psychological refractory period (PRP) is well suited to examine these difficulties (Pashler, 1994; Welford, 1952). In the PRP paradigm, two choice reaction time (RT) tasks—Task 1 and Task 2—are presented with variable temporal intervals (stimulus onset asynchrony, SOA). The participants are asked to respond as fast and accurately as possible to both tasks with priority on Task 1. As is typical for the findings, RT of Task 2 (i.e., RT2) increases with decreasing SOA (i.e., PRP effect; Pashler, 1994; Schubert, 1999), but RT of Task 1 (i.e., RT1) is not affected by the SOA.

The central bottleneck model is an established theory that can account for the increase in RT2 when the SOA decreases (Pashler, 1994; Schubert, 1999, 2008; Welford, 1952; but see Logan & Gordon, 2001; Meyer & Kieras, 1997; Navon & Miller, 2002; Tombu & Jolicoeur, 2003). The model assumes that the processing of each task can be decomposed into a series of processing stages. The most distinct processing stages are perception, response selection, and motor response. In general, perception and motor response process concurrently to other processing stages, but the response selection stage is a central stage that processes for one task at a time, thus constituting a bottleneck. Accordingly, at short SOA, the response selection stage in Task 2 processes once the response selection stage in Task 1 has been finished. The waiting time for the response selection stage in Task 2—the so-called slack time—is prolonged and mirrored in increasing RT2. The present study is the first study to apply both behavioural and electrophysiological methods to investigate whether capacity-limited processes in visual attention (see below) operate during the slack time or whether they are subject to the same bottleneck mechanism as response selection in a PRP dual-task.

Capacity limitation in visual attention

Recently, we showed that visual attention is not subject to the same bottleneck mechanism as response selection in a dual-task consisting of an auditory two-choice discrimination Task 1 and a visual search Task 2 requiring feature conjunctions (Reimer, Strobach, Frensch, & Schubert, 2015). Before explaining the method that we applied to investigate this question, it is important to describe in what aspect visual attention deployment is considered as capacity limited in the conjunction search task that we used.

The capacity limitation in the visual attention processes of the conjunction search task can be referred to as the “feature binding problem” (Treisman & Gelade, 1980; Wolfe, 2012; Wolfe & Bennett, 1996; but see Di Lollo, 2012). A conjunction search task is suited to examine feature binding processes. In this search task, both the target and the distractors are composed of a conjunction of features—for example, colour and form. The target differs from the surrounding distractors in a unique combination of features—for example, a red–vertical target among red–horizontal and green–vertical distractors. The task is to correctly detect the presence versus absence of the target. Prominent visual search theories (feature integration theory, Treisman & Gelade, 1980; guided search model, Wolfe, 1994, 1998, 2007; Wolfe, Cave, & Franzel, 1989) claim that focal visual attention selects the items and binds the item features so that the item is correctly perceived as target or distractor. Because the cognitive system is limited with respect to the number of items it can process at the same time, feature binding cannot be done simultaneously for all items. Rather, the items are selectively attended in a serial search process. Accordingly, the search time increases with an increasing number of items, the set size. The increase in search time is called set size effect, which is assumed to reflect a limitation in the capacity of visual attention (Treisman & Gelade, 1980; Wolfe, 1994, 1998, 2007; Wolfe et al., 1989).

The locus-of-slack method

In our previous study (Reimer et al., 2015), we used the locus-of-slack method (Schweickert, 1978, 1980) to test whether visual attention is subject to the same bottleneck mechanism as response selection in a dual-task situation consisting of an auditory two-choice discrimination Task 1 and a conjunction search Task 2. We increased the set size from six to 18 items in the conjunction search task and modulated the SOA between both tasks to analyse the effects of these manipulations on RT2, the search time. As a result, we found an increase in search time (RT2) from the small to the large set size (i.e., set size effect) at long SOA, but the set size effect was significantly reduced at short SOA (see Figure 1A). The underadditive interaction of SOA and set size on RT2 pointed to an absorption of the visual search time into the slack time at short SOA (see also Schubert, Fischer, & Stelzel, 2008). According to the locus-of-slack method, this RT2 pattern is interpreted as evidence for the assumption that visual attention required for feature binding in Task 2 is concurrently deployed to the response selection processes in Task 1. Therefore, the processes leading to a longer search time in Task 2 operated during the slack time, so that the overall RT2 was not prolonged for the large compared to the small set size at short SOA.

By contrast, if visual attention had been subject to the same bottleneck mechanism as response selection in dual-tasks, the set size effect should have been additive with the effect of SOA (see Figure 1B). In that case visual attention deployment in Task 2 could have started only after the response selection in Task 1 would have been finished, and the set size effect would not have been absorbed into the slack time at short SOA. According to the locus-of-slack method, RT2 would have been equally prolonged for the large set size as for the small set size at short and at long SOA. Since this was not the case, the findings of our study are interpreted as evidence against the assumption that visual attention and response selection rely on a common capacity limitation (see also Pashler, 1989, 1991, using a different reasoning on error rates).

Electroencephalography (EEG) investigations on the interference between visual attention and response selection processes

In the present study, we extended the earlier findings and added an electrophysiological marker for visual attention processes—that is, the N2-posterior-contralateral (N2pc) component—to assess the validity of the main conclusion of our study (Reimer et al., 2015). Because the N2pc is assumed to reflect visual attention processes in a more direct way than RT measures, which cover the whole sequence from stimulus to response processing, a study using the N2pc should provide more evidence on the issue of whether visual attention proceeds concurrently to Task 1 response selection in a PRP dual-task situation.

The N2pc is a well-established marker of the focusing in visuospatial attention, specifically of the spatial selection of the target in visual search arrays (Eimer, 1996; Luck & Hillyard, 1994a, 1994b; Woodman & Luck, 1999, 2003). Magnetoencephalographic (MEG) recordings suggest that the N2pc is generated primarily in inferior occipitotemporal cortex and has probably an early parietal contribution (Hopf, Boelmans, Schoenfeld, Heinze, & Luck, 2002; Hopf et al., 2000). In particular, the electrophysiological marker is elicited as a negative-going deflection over parietal areas around 180–280 ms after the onset of the visual stimuli. Importantly, the stimuli are balanced—that is, target and distractors are presented equally to the ipsilateral and contralateral visual fields. The N2pc is more negative at electrodes contralateral to the side of the attended stimulus (i.e., the target) than at those that are ipsilateral. Accordingly, the N2pc difference wave is derived from subtracting the waveforms recorded ipsilateral to the attended stimulus from the waveforms recorded contralateral to the attended stimulus. When the stimuli are balanced, the subtraction procedure then eliminates any activity (i.e., pure sensory activity, any higher level cognitive activity equal for ipsilateral and contralateral targets) that is not attributable to the visual attention processes. Correspondingly, the N2pc amplitude is linked to the amount of visual attention allocated to the target in a visual search task, and the N2pc latency is linked to the speed of the attentional shift (Woodman & Luck, 1999, 2003).

The N2pc has already been reliably established in various manipulations of visual search tasks including manipulations of set size (Dowdall, Luczak, & Tata, 2012; Hickey, Di Lollo, & McDonald, 2008; Jolicoeur, Sessa, Dell'Acqua, & Robitaille, 2006; Kiss, Grubert, & Eimer, 2013; Mazza, Turatto, & Caramazza, 2009; Mazza, Turatto, Umiltà, & Eimer, 2007; Schubö, Wykowska, & Müller, 2007; Wolber & Wascher, 2003, 2005). Evidence suggests that the N2pc reflects the spatial selection of the target among distractors in visual search tasks (Eimer, 1996; Luck & Hillyard, 1994a, 1994b; Woodman & Luck, 1999). Support for this assumption comes from a spatial-cueing study in which the target required a conjunction of two features (Kiss et al., 2013). The results showed that visual attention was initially captured by all target-matching features, but remained exclusively at locations of candidate target items that shared all and not only some features with the target. Thus, the N2pc is well suited to reliably assess target selection, even when the target differs in more than one feature from the distractors.

Interestingly, the N2pc has already been used to investigate whether visual attention is subject to the same bottleneck mechanism as response selection in dual-task situations in studies of Brisson and Jolicoeur (2007a, 2007b) and of Lien, Croswaite, and Ruthruff (2011). In several conditions of these studies, the authors combined a visual search Task 2 with an auditory discrimination RT Task 1 at short and at long SOA and focused on the N2pc rather than on RT to measure visual attention deployment. Brisson and Jolicoeur reported that the N2pc amplitude was reduced at short SOA compared to long SOA (see also Lien et al., Experiments 3 and 4). The authors interpreted this finding as evidence for the assumption that the response selection processes in Task 1 interfered with the visual attention processes in Task 2 at short SOA. Therefore, less visual attention could be allocated to the target, which explained the observed reduction of the N2pc amplitude at short SOA compared to long SOA.

As these N2pc results were inconsistent with the results of our study (Reimer et al., 2015) that used RT2 analyses according to the locus-of-slack method, it should be considered that a number of differences between the studies could have caused the discrepancy. Most importantly, in the studies of Brisson and Jolicoeur (2007a, 2007b) as well as of Lien et al. (2011), the search display was masked after a brief exposure. Visual attention could only be deployed to the items during a critical period of time before the mask terminated any perceptual processing (Cameron, Tai, Eckstein, & Carrasco, 2004; Jolicoeur, 1999; Wolfe, 2007). Accordingly, the authors focused on fast spatial shifts of visual attention. On the other hand, we presented the search display until response. Accordingly, we focused on a serial search process around the search display (Treisman & Gelade, 1980; Wolfe, 1994, 2007).

In addition, the visual attention processes that were required to localize the target in the search tasks should be considered in more detail. In the studies of the authors, the target differed in colour from the distractors. Therefore, it is safe to assume that the target automatically captured attention. It seemed thus very likely that the automatic capture facilitated the visual attention shift to the target. On the other hand, in our search task, the target consisted of a conjunction of features. As explained above, visual attention was required to select the items and to bind the item features resulting in a serial search process (Treisman & Gelade, 1980; Wolfe, 1994, 2007). Finding the target in a serial search process is probably more visual attention demanding than finding the target via automatic capture of visual attention. We come back to this issue in the General Discussion section.

Taken together, in light of the findings of Brisson and Jolicoeur (2007a, 2007b) and of Lien et al. (2011, Experiments 3 and 4) it remains an open issue whether it is possible to reveal interference between visual attention in a serial search process (i.e., when the search display is not masked, but presented until response) and the response selection processes in Task 1 with electrophysiology compared to RT measures. Both methodologies measure temporal information, yet with different resolutions. Event-related potentials (ERPs) are online markers of particular task processes like visual attention, but information of all task processes is integrated in RT measures. Therefore, in the present study, we applied a visual search condition comparable to the condition in our previous study (Reimer et al., 2015) and focused on the N2pc in addition to RT to investigate whether visual attention (i.e., feature binding) in a target detection task is subject to the same bottleneck mechanism as response selection.

The present study

The present dual-task experiment consisted of an auditory two-choice discrimination Task 1 and a conjunction search Task 2 in which the target had to be detected among distractors. We presented visual search displays of a small and a large set size to manipulate the difficulty of feature binding (i.e., duration). In addition, we manipulated the SOA between Task 1 and Task 2 and examined whether the effect of the set size manipulation is absorbed into the slack time.

The hypotheses of the present study shown in Figure 1 were as follows. Concerning the behavioural measures, we analysed RT2 (i.e., visual search time) according to the locus-of-slack method (Schweickert, 1978, 1980). If visual attention is subject to the same bottleneck mechanism as response selection in dual-tasks, the effect of set size would be additive with the effect of SOA (Figure 1B). The set size effect would be similar at short and at long SOA, indicating that at short SOA, the visual attention processes in Task 2 (i.e., feature binding) could only start after the response selection processes in Task 1 would have been finished. This finding would provide evidence for the assumption that response selection and visual attention rely on a common capacity limitation. On the other hand, if visual attention is not subject to the same bottleneck mechanism as response selection in dual-tasks, the interaction of SOA and set size would be underadditive (Figure 1A). The set size effect at short SOA would be significantly reduced compared to the set size effect at long SOA, as visual search time would be absorbed into the slack time at short SOA. An underadditive interaction would indicate that the visual attention processes in Task 2 (i.e., feature binding) operate concurrently to the response selection processes in Task 1. This finding would provide evidence for the assumption that response selection and visual attention rely on distinct capacity limitations.

In addition to the RT analyses, we measured the N2pc (Eimer, 1996; Luck & Hillyard, 1994a, 1994b; Woodman & Luck, 1999, 2003). We locked the N2pc to the onset of the visual search display and analysed the peak latency, onset latency, and amplitude. As illustrated in Figure 1, the hypotheses were as follows. If the response selection processes in Task 1 delay the visual attention shift to the target in Task 2, the peak latency would be increased at short SOA compared to long SOA (Figure 1B). Likewise, if the response selection processes in Task 1 delay the onset of target selection, the onset latency would be increased at short SOA compared to long SOA (Figure 1B). In addition, if the amount of visual attention that could be allocated to the target depends on the response selection processes in Task 1, the amplitude would be attenuated at short SOA but not at long SOA (Figure 1B), because the amplitude is interpreted as reflecting the amount of allocated visual attention to the target (Eimer, 1996; Luck & Hillyard, 1994a, 1994b; Woodman & Luck, 1999, 2003). All in all, in case visual attention and response selection in Task 1 rely on a common capacity limitation, the N2pc parameters would change as a function of SOA (Figure 1B). However, in case they rely on distinct capacity limitations, the N2pc parameters would be independent of SOA (Figure 1A).

Furthermore, we presented the conjunction search task as single-task and measured the N2pc to compare visual attention performance in the dual-task and visual attention performance as single-task. The comparison allowed for investigating whether general dual-task demands (i.e., the demands that arise from the general need to perform an additional task; Jiang, Saxe, & Kanwisher, 2004; Schubert & Szameitat, 2003; Töllner, Strobach, Schubert, & Müller, 2012) affect visual attention deployment per se. If general dual-task demands influence visual attention deployment, the N2pc onset latency would be increased, and the amplitude would be attenuated in the dual-task compared to the single-task. If general dual-task demands do not influence visual attention deployment, the N2pc parameters would be independent of the task situation (i.e., dual- and single-tasks).

Experiment 1

Method

Participants

Fifteen participants (five men, 10 women, all right-handed) from the Humboldt-Universität zu Berlin with a mean age of 25.1 years (range = 18–33 years, SD = 4.5) took part in the experiment. We excluded two participants due to excessive eye movements and artefacts in the EEG recording. The data of the remaining 13 participants were used for the dual-task analysis. For the single-task conjunction search, we could only analyse the data of 10 of these 13 participants due to artefacts in the EEG recording.

Apparatus and stimuli

The experiment was programmed in Presentation (Version 14.9). The visual stimuli of the conjunction search Task 2 were presented on a 19″ CRT monitor at a resolution of 1280 × 1024 pixels with a refresh rate of 60 Hz at a viewing distance of 80 cm. We adapted the conjunction search task from a study of Wolber and Wascher (2003) (see also Treisman & Gelade, 1980). In this task, the participants had to detect the presence versus absence of a green T, the target, among red Ts and green Xs, the distractors, in both set sizes eight and 16 (green: CIE: x = 0.300, y = 0.600; luminance = 55.4 cd m^–²; red: CIE: x = 0.640, y = 0.329; luminance = 19.3 cd m^–²). The stimuli were presented on a black background (CIE: x = 0.313, y = 0.329; luminance = 0.2 cd m^–²). The number of Ts and Xs was equally distributed for every set size. As an example shown in Figure 2, in set size 8, there were one green T, three red Ts, and four green Xs when the target was present, and four red Ts and four green Xs when the target was absent. Each item subtended 0.25° × 0.48° on a 6.5° × 4.28° area. The central fixation point had a size of 0.05° × 0.05°.

We presented the stimuli in a lateralized format to measure the N2pc. As illustrated in Figure 3, the lateralization was provided by an underlying invisible matrix of potential target locations (see Wolber & Wascher, 2003). In target present trials, the target appeared in 45% of the trials in the dark grey area, in 36% in the light grey area and in 19% in the white area, but never on the vertical midline. The target and the distractors were equally distributed to the left and right side of fixation to avoid sensory artefacts. Concretely, the target replaced one of the distractors so that the same number of stimuli was presented to the left and right side of fixation. In target absent trials, the distractors were equally distributed to the left and right side of fixation.

Task 1 of the dual-task consisted of an auditory two-choice discrimination task. The participants should discriminate between two sine-wave tones of 350 Hz (78 dB) and 900 Hz (80 dB). The tones were presented for 100 ms via two speakers, which were placed behind the monitor.

Four external keys served as manual response keys. In the first half of each dual-task session, half of the participants responded with the left middle finger to the low tone, the left index finger to the high tone, the right index finger to target present, and the right middle finger to target absent. The assignment was switched finger-wise for the other half of the participants. They responded with the left middle finger to the high tone, the left index finger to the low tone, the right index finger to target absent, and the right middle finger to target present. In the second half of each dual-task session, left- and right-hand responses were changed—that is, the participants used their right hands for the tones and their left hands for the search task, using the same fingers as in the first half, respectively. In the single-task conjunction search session, the participants used their hand mapping from the dual-task sessions including a hand switch to guarantee similar response conditions for the conjunction search task.

Design and procedure

The experiment consisted of two dual-task sessions and a final single-task session. In the dual-task, we used a within-subjects design that consisted of orthogonally combining 3 SOAs (60 ms vs. 400 ms vs. 800 ms) × 2 set sizes (8 vs. 16) × 2 display types (target present vs. target absent) × 2 tones (350 Hz vs. 900 Hz). The 24 combinations were presented twice per block in random order, resulting in 48 trials per block. Each dual-task session consisted of 20 blocks. In these blocks, every combination was thus presented 160 times. In the single-task, we used a within-subjects design that consisted of orthogonally combining 2 set sizes (8 vs. 16) × 2 display types (target present vs. target absent). The four combinations were presented 12 times per block in random order. In 24 blocks, each combination was presented 288 times.

Each dual-task session started with 24 practice trials of the tone discrimination Task 1 and the conjunction search Task 2, respectively, followed by 24 trials of dual-task practice. There were also 24 practice trials of conjunction search in the single-task session. To get used to the hand switch after the first half of the experimental session, there were another 24 practice trials right after the switch for the new hand assignment both in the dual-task and in the single-task sessions. The additional practice consisted of dual-task trials in the dual-task sessions and of conjunction search trials in the single-task session. The participants were asked to respond as fast and accurately as possible to both Task 1 and Task 2 with priority on Task 1 in the dual-task session and as fast and accurately as possible to the conjunction search task in the single-task session. Additionally, the participants should keep their eyes on fixation.

The stimulus sequence of a trial in the dual-task is displayed in Figure 2. Each trial started with the fixation point shown at the centre of the screen. It was visible until the participants responded to the conjunction search task. In the dual-task, after 500 ms of fixation, the tone was played for 100 ms. Following the SOA of 60 ms, 400 ms, or 800 ms, the conjunction search display appeared until the participants responded. The maximum presentation time was 2500 ms. There was an intertrial interval (ITI) of 1000 ms showing a black screen after the response to Task 2. In the single-task session, the conjunction search display was presented until response after 500 ms of fixation. The maximum presentation time and the ITI were the same as those in the dual-task.

Electrophysiological recording

The experiment took place in a soundproof and electrically shielded EEG chamber. The EEG was continuously recorded at a sampling rate of 250 Hz from 56 silver/silver chloride (Ag/AgCl) electrodes mounted in an electrode-cap (ECI Inc.). The vertical and horizontal electrooculogram (EOG) was registered below the left eye and from the left and right outer canthi. The left and right mastoids served as reference for all electrodes. The electrode impedance was kept below 5 kΩ. Bandpass was set to 0.03–70 Hz.

Results

Behavioural data analysis

The RTs of Task 2 and Task 1 of the dual-task were separately analysed with repeated measures analysis of variance (ANOVA) with the three factors SOA, set size, and display type. The single-task RT was analysed with repeated measures ANOVA with the factors set size and display type. When assumptions of sphericity were violated, p values were adjusted using the Greenhouse–Geisser correction. Error trials included incorrectly committed and omitted responses. In the RT analyses, trials with errors in one or both tasks were excluded (4.7% in the dual-task and 4.6% in the single-task). Before submitting the error rates of Task 2 and Task 1 of the dual-task and of the single-task to the same ANOVAs as the RTs, they were transformed according to the formula 2 sin^–1√(error rate) (Kirk, 2013).

RT visual search Task 2

In the RT data of the conjunction search Task 2 that is presented in Figures 4A and 4B, we found increasing RT2 at short SOA (M = 1134 ms) compared to long SOA (M = 846 ms) indicating a PRP effect, F(2, 24) = 105.733, p < .001, $η_{p}^{2} = . 90$ . As is typical for conjunction search, RT2 increased with larger set size from 911 ms in set size 8 to 1019 ms in set size 16, F(1, 12) = 101.265, p < .001, $η_{p}^{2} = . 89$ . The participants responded faster when the target was present (M = 934 ms) than when it was absent (M = 1006 ms), F(1, 12) = 44.520, p < .001, $η_{p}^{2} = . 79$ . The factors set size and display type interacted significantly, F(1, 12) = 67.941, p < .001, $η_{p}^{2} = . 85$ , reflecting a higher increase from set size 8 to set size 16 for target absent (set size 8 vs. 16: M = 923 ms vs. 1089 ms) than for target present (set size 8 vs. 16: M = 898 ms vs. 969 ms).

Importantly, we found a significant underadditive SOA and set size interaction, F(2, 24) = 25.626, p < .001, $η_{p}^{2} = . 68$ . A planned comparison confirmed that the RT2 difference between set sizes 16 and 8 was significantly smaller at short SOA (M = 82 ms) than at long SOA (M = 143 ms), t(12) = 6.441, p < .001. Furthermore, we found a significant underadditive SOA and display type interaction, F(2, 24) = 4.272, p < .05, $η_{p}^{2} = . 26$ . The RT2 difference between target absent and target present was significantly smaller at short SOA (M = 58 ms) than at long SOA (M = 90 ms), t(12) = 2.308, p < .05. Both underadditive interactions revealed that the visual attention processes in Task 2 operated concurrently to the response selection processes in Task 1. No other main effects or interactions reached significance.

RT tone Task 1

RT1 of the auditory discrimination Task 1 is shown in Figure 4C. The participants responded faster in trials of the small than of the large set size (set size 8 vs. 16: M = 819 ms vs. 858 ms), F(1, 12) = 11.730, p < .01, $η_{p}^{2} = . 49$ , and when the target was present than when it was absent (target present vs. target absent: M = 832 ms vs. 846 ms), F(1, 12) = 6.397, p < .05, $η_{p}^{2} = . 35$ . The factors set size and display type interacted, F(1, 12) = 6.329, p < .05, $η_{p}^{2} = . 35$ , indicating that RT1 was exclusively slowed for set size 16 when the target was absent, but not for set size 8 (set size 16, target absent vs. target present: M = 872 ms vs. 845 ms; set size 8, target absent vs. target present: M = 819 ms vs. 819 ms). No other main effects or interactions reached significance.

Error visual search Task 2 and tone Task 1

Table 1 provides an overview of the error data for both Task 2 and Task 1. In the conjunction search Task 2, the participants made more errors for set size 16 (M = 6.4%) compared to set size 8 (M = 4.3%), F(1, 12) = 12.951, p < .01,

η_{p}^{2} = . 52

. They also committed more errors for target present (M = 6.9%) than for target absent (M = 4.1%), F(1, 12) = 15.220, p < .01,

η_{p}^{2} = . 56

. The factors set size and display type interacted, F(1, 12) = 21.528, p < .01,

η_{p}^{2} = . 64

, indicating a higher error rate for target present compared to target absent for set size 16 (target present vs. target absent: M = 8.8% vs. 4.1%) than for set size 8 (target present vs. target absent: M = 4.4% vs. 4.1%). In the auditory Task 1, the participants committed more errors at short SOA (M = 3.7%) than at long SOA (M = 1.7%), F(2, 24) = 15.734, p < .01,

η_{p}^{2} = . 57

. No other main effects or interactions reached significance.

Table 1.

Mean error percentages for Task 2, Task 1, and the single-task conjunction search.

Task condition	SOA (ms)	Visual search condition
		Target present		Target absent
		Set size 8	Set size 16	Set size 8	Set size 16
Dual-task Task 2	60	4.9	9.1	6.1	4.7
	400	4.3	9.1	3.5	4.0
	800	4.0	8.1	2.8	3.5
Dual-task Task 1	60	3.7	3.8	4.1	3.1
	400	1.9	1.9	2.2	2.5
	800	1.6	1.8	1.6	1.9
Single-task	–	1.3	2.8	3.6	10.6

Note: SOA = stimulus onset asynchrony; ms = millisecond. Mean error percentages are shown for the dual-task conjunction search Task 2, and for the auditory two-choice discrimination Task 1, as a function of SOA and Task 2 manipulation (set size, display type). Mean error percentages are also shown for the single-task conjunction search.

RT visual search single-task

The RT results of conjunction search performed as single-task are presented in Figure 5. The participants responded slower with increasing set size (set size 8 vs. 16: M = 675 ms vs. 853 ms), F(1, 9) = 26.999, p < .001, $η_{p}^{2} = . 75$ , and when the target was absent (M = 817 ms) than when it was present (M = 711 ms), F(1, 9) = 16.459, p < .01, $η_{p}^{2} = . 65$ . Furthermore, the RT increase from the small to the large set size was more pronounced for target absent (set size 8 vs. 16: M = 693 ms vs. 941 ms) than for target present (set size 8 vs. 16: M = 657 ms vs. 765 ms), F(1, 9) = 16.499, p < .01, $η_{p}^{2} = . 65$ .

Error visual search single-task

An overview of the error data of the single-task conjunction search is presented in Table 1. The participants made more errors for the large set size (M = 1.7%) than for the small set size (M = 0.6%), F(1, 9) = 33.000, p < .001, $η_{p}^{2} = . 77$ , and for target present (M = 1.8%) than for target absent (M = 0.5%), F(1, 9) = 23.115, p < .001, $η_{p}^{2} = . 72$ . The interaction of the factors display type and set size was significant, revealing that the pattern of a higher error rate for target present than for target absent was more pronounced for set size 16 (target present vs. target absent: M = 2.6% vs. 0.7%) than for set size 8 (target present vs. target absent: M = 0.9% vs. 0.3%), F(1, 9) = 18.413, p < .01, $η_{p}^{2} = . 67$ .

Electrophysiological analysis

Offline, the EEG data were re-referenced to common reference. We used Brain Electric Source Analysis (BESA 2.1, Megis Software GmbH) for eye blink artefact correction. In a subsequent step, we excluded deviations in amplitude larger than 80 μV. Continuous EEG was segmented into epochs of 3700 ms. For the N2pc analysis we segmented the EEG into epochs of 1000 ms relative to a 200-ms pre-stimulus interval. We performed a baseline correction on this pre-stimulus interval (−200 to 0 ms). Only correct trials and trials without artefacts were used for further analysis on an individual-channel basis before averaging the N2pc waveform. We computed the N2pc component as maximum amplitude during 180–380 ms post-stimulus by subtracting the ERPs measured at parieto-occipital electrodes (PO7/PO8) ipsilateral to the target location from contralateral ERPs.

We determined the N2pc parameters by using the jackknife approach suggested by Kiesel, Miller, Jolicoeur, and Brisson (2008). According to this approach, the N2pc parameters are identified in time courses averaged across participants (grand-averages). Averaging across participants reduces the noise, and the N2pc parameters can be reliably estimated. In more detail, the jackknife approach is a re-sampling technique that leads to the creation of a statistical distribution from grand-averages. Each of n participants is excluded from grand-averaging once. The resulting distribution of n grand-averages (each omitting a different participant) can then be used to calculate the N2pc parameters. Before we determined the N2pc parameters in the jackknife time courses, we smoothed them by applying a 15-Hz low-pass filter. We calculated the N2pc amplitude as maximum negative deflection during 180–380 ms post-stimulus. Accordingly, the peak latency was determined as the point in time when the amplitude reached its maximum activation. Further, we defined the N2pc onset as the point in time when the N2pc amplitude reached the criterion of −0.3 µV relative to the maximum activation. We calculated the amplitude, peak latency, and onset latency separately for each SOA and set size. For each parameter, we submitted the factors SOA and set size to repeated measures ANOVAs and adjusted the F-values and t-values according to these formulas F_adj = F/(n − 1)² and t_adj = t/(n − 1), respectively (n is the number of participants). Note that the N2pc could only be computed for target present trials.

Electrophysiology visual search Task 2

Figure 6 shows the N2pc effects of the conjunction search Task 2. Although the onset of the N2pc was 12 ms earlier at long SOA than at the other SOAs (SOA 60, 400, 800: M = 224 ms, 225 ms, 212 ms), this effect was not significant (F_adj < 1). There was no set size effect on the onset latency of the N2pc either (F_adj < 1; set size 8 vs. 16: M = 229 ms vs. 212 ms). Thus, the N2pc onset latency provided no evidence that the response selection processes in Task 1 delayed the onset of target selection in the visual search task, which is consistent with the findings of the RT2 analysis.

Interestingly, the peak latencies were not modulated by SOA (F_adj < 2), but the amplitude peaked significantly later for set size 16 (M = 351 ms) than for set size 8 (M = 336 ms), F_adj(1, 12) = 12.071, p < .05. The response selection processes in Task 1 did not delay target selection, but the set size effect on the peak latencies reflected that the attention shift to the target was delayed when the set size increased. It is important to note that the set size effect on the peak latencies was not modulated by SOA (F_adj < 2) suggesting that this was again not affected by concurrently processing response selection in Task 1.

Importantly, the N2pc amplitude was independent of SOA (F_adj < 1). The mean amplitudes were −1.60 μV, −1.74 μV, and −1.78 μV for the SOAs 60, 400, and 800, respectively. We found a set size effect on the N2pc amplitude, which was significantly reduced for the large compared to the small set size (set size 16 vs. 8: M = −1.362 μV vs. −2.06 μV), F_adj(1, 12) = 17.986, p < .05. Thus, the set size effect illustrated that the N2pc was sensitive to the amount of visual attention required for target selection. Still, there was no evidence that the response selection processes in Task 1 impaired the amount of visual attention that was allocated to the search task. No other effect or interaction reached significance.

Electrophysiology visual search single-task

The N2pc effects of conjunction search performed as single-task are shown in Figure 7. Although the onset latency was delayed for set size 16 (M = 197 ms) compared to set size 8 (M = 118 ms), the effect was not significant (t_adj < 2). The N2pc amplitude peaked later for the larger set size (M = 386 ms) than for the smaller set size (M = 338 ms), t_adj(9) = 2.480, p < .05. Correspondingly, there was a significant set size effect in the amplitude analysis, t_adj(9) = 3.913, p < .05. The amplitude was reduced for the larger set size (M = −1.699 μV) compared to the smaller set size (M = −2.539 μV). These findings stressed that the N2pc reflected attentional processes relevant for target selection.

Electrophysiology visual search Task 2 and single-task

We contrasted conjunction search performance in the dual-task with performance in the single-task to assess the influence of general dual-task demands on visual attention deployment. For that purpose, we compared the N2pc in the single-task and the N2pc at SOA 60 in the dual-task. We chose SOA 60, because this condition represented the task situation where the temporal overlap of Task 1 and Task 2 was maximal compared to the other two SOA conditions. Accordingly, the comparison between the N2pc in the single-task and at SOA 60 represented a comparison between the conditions that were least related with respect to cognitive processing (i.e., the condition of pure single-task processing vs. the condition of maximal dual-task processing). Therefore, this comparison was most informative to investigate the influence of general dual-task demands on visual attention deployment.

The amplitude peaked later in the single-task (M = 362 ms) than at SOA 60 in the dual-task (M = 338 ms), F_adj(1, 9) = 9.736, p < .05, and later for set size 16 (M = 367 ms) than for set size 8 (M = 333 ms), F_adj(1, 9) = 10.523, p < .05. Interestingly, the amplitude in the single-task did not differ significantly from the amplitude at SOA 60 in the dual-task, F_adj < 3. The amplitude was significantly reduced for set size 16 (M = −1.53 μV) compared to set size 8 (M = −2.22 μV), F_adj(1, 9) = 13.253, p < .05. Further, we did not find any statistical effects on the onset latency, all F_adjs < 5. Comparing conjunction search performance as single-task with performance at the shortest SOA in the dual-task revealed no effects of general dual-task demands on the N2pc onset latency and amplitude. Thus, general dual-task demands did not substantially impair the onset of the attention shift and the amount of visual attention allocated to the target. The later peak in the single-task (i.e., the peak latency effect) should not obscure these findings. Rather, it seemed plausible to assume that in the single-task it took more time to reach a numerically higher peak. No other main effects or interactions reached significance.

General Discussion

The aim of this study was to investigate whether capacity-limited processes in visual attention are subject to the same bottleneck mechanism as response selection in dual-tasks by applying the locus-of-slack method and by measuring the N2pc. For that purpose, we used a dual-task consisting of an auditory discrimination Task 1 and a conjunction search Task 2. ¹ We manipulated the SOA between both tasks and the set size in the conjunction search Task 2. Analysing conjunction search time according to the locus-of-slack method revealed that the visual attention processes operated during the slack time. This is consistent with the assumption that visual attention is not subject to the same bottleneck mechanism as response selection in dual-tasks. The analysis of the N2pc, an electrophysiological marker of visual attention deployment, coincided with the behavioural findings. Importantly, the N2pc parameters were independent of SOA. The response selection processes in Task 1 neither delayed the visual attention shift to the target nor attenuated the amount of visual attention that was allocated to the target. We found a set size effect on both the peak latency and the amplitude revealing that the N2pc was sensitive to target selection difficulty. Together, the behavioural and electrophysiological results suggested that visual attention is concurrently deployed to the response selection processes in an auditory two-choice discrimination task.

In addition, we measured the N2pc to contrast conjunction search performance as single-task with conjunction search performance at the shortest SOA in the dual-task. There was no evidence that the onset latency and the amplitude were modulated by general dual-task demands.

Relation to other electrophysiological studies

The results of the present study provided converging support and extended related findings of an ERP study of Lien et al. (2011, Experiments 1 and 2). In the first experiment of their study, the authors measured the N2pc to reveal whether spatial attention shifts in a visual search Task 2 are subject to the same bottleneck mechanism as response selection in dual-tasks. As the N2pc amplitude was not modulated by SOA, the authors concluded that the response selection processes of an auditory two-choice discrimination Task 1 did not impair the spatial attention shifts to the stimuli in the visual search Task 2. In our study, the N2pc amplitude was also independent of SOA. However, an important difference is that Lien et al. masked the visual search display, but we presented it until response. Accordingly, Lien et al. focused on the initial spatial attention shift to the target in the display. Our search task, however, allowed for examining the deployment of visual attention in a serial search process. The set size effect indicated that visual attention was limited in capacity, but this remained unclear for visual attention in Lien et al.'s task, as they did not manipulate the set size. Thus, our study design allowed for examining visual attention processing over time and revealed that though limited in capacity, visual attention is not subject to the same bottleneck mechanism as response selection in dual-tasks.

Interestingly, the results of the present study contradicted the findings of studies of Brisson and Jolicoeur (2007a, 2007b). Contrary to our results, the authors found an effect of SOA on the N2pc amplitude in their visual search tasks presented as Tasks 2 in the dual-tasks. Because the amplitude was attenuated at short SOA compared to long SOA, the authors concluded that visual attention interfered with the response selection processes in the auditory discrimination Task 1. It should be noted that the studies of Brisson and Jolicoeur and the present study differed in two major aspects. First, Brisson and Jolicoeur masked the visual search display, but we presented it until response. As outlined above, the masked search display enabled the authors to exclusively measure the initial spatial attention shift to the target in the display. The present study, however, investigated visual attention deployment in a serial search process. The set size effect showed that visual attention was limited in capacity, but this remained unclear for visual attention in the search tasks of the authors, as they did not manipulate the set size. Thus, theirs and our studies were designed to measure different aspects of visual attention deployment.

Second, the auditory Task 1 was a four-choice discrimination task in Brisson and Jolicoeur's (2007a, 2007b) studies, but the auditory Task 1 in the present study required a decision between two alternatives. Accordingly, the response selection demands of Task 1 were higher in their studies than in the present study. Lien et al. (2011) showed that the response selection difficulty in Task 1 is an important factor in the interplay of visual attention and response selection. The authors manipulated the response selection difficulty in Task 1 by pairing easy two-choice discrimination Tasks 1 (Experiments 1 and 2) and difficult four-choice discrimination Tasks 1 (Experiments 3 and 4) with a visual search Task 2, respectively. Their manipulation revealed that response selection and visual attention operate in parallel when response selection in Task 1 is easy (i.e., two-choice discrimination), but response selection and visual attention interfere when response selection in Task 1 is difficult (i.e., four-choice discrimination). Interestingly, the authors found this result pattern for both auditory and visual two-choice and four-choice discrimination Tasks 1, respectively, which were all combined with the same visual search Task 2. Thus, considering Lien et al.'s, Brisson and Jolicoeur's, and the present study together, easy response selection processes in Task 1 (i.e., two-choice discrimination) do not interfere with visual attention deployment in both non-masked (this study) and masked visual search Tasks 2 (Lien et al., 2011, Experiments 1 and 2), but difficult response selection processes in Task 1 (i.e., four-choice discrimination) interfere with visual attention deployment in masked visual search Tasks 2 (Brisson & Jolicoeur, 2007a, 2007b; Lien et al., 2011, Experiments 3 and 4). Further studies should investigate whether visual attention deployment in a non-masked search task—preferentially in a search task requiring a serial search process as the present conjunction search task—is subject to the same bottleneck mechanism as response selection in dual-tasks when Task 1 is more difficult than in the present study.

Finally, there is still another important difference between the studies. To our understanding, the authors used so-called compound tasks (Müller & Krummenacher, 2006). In more detail, a compound task consists of a target-related and a response-related visual attention process. In the studies of Brisson and Jolicoeur (2007a, 2007b), the search task required the localization of the target and the discrimination of the response relevant target feature. In their studies, the stimuli were squares with a gap on one side (i.e., left, right, up, or down), and the location of the gap of the target square that differed in colour from the distractor squares should be indicated. In the study of Lien et al. (2011), the search task required the localization and the identification of the target. In their study, the stimuli were Ts and Ls, and the identity of the target letter (i.e., T vs. L) that differed in colour from the distractor letters should be indicated. Since the target was probably localized via automatic capture of visual attention, the discrimination process of the response relevant target feature (Brisson & Jolicoeur, 2007a, 2007b) and the target identification process (Lien et al., 2011) could have been very visual attention demanding, which in turn could explain why the authors found interference when Task 1 required a four-choice discrimination. On the other hand, we used a target detection task, where the presence versus absence of the target should be detected. In other words, in the target detection task, there was no need to respond to a specific target feature, but only to the presence versus absence of the target. The difficulty of the search task (i.e., compound task vs. target detection) should be considered as a factor that could influence interference between visual attention and response selection.

Comparing visual attention deployment between dual- and single-tasks

Electrophysiology has one major advantage over RT measures. The temporal information of all task processes is integrated in one parameter in RT measures—that is, the resulting overall RT. Electrophysiology, however, allows for analysing the temporal course of individual processing stages. In the present study, we focused on the deployment of visual attention and its potential interference with general dual-task demands. Therefore, we measured the N2pc when conjunction search temporally overlapped with general dual-task demands (i.e., shortest SOA in the dual-task) and when it did not (i.e., single-task).

The N2pc analysis revealed no effect on the onset latency or the amplitude. Importantly, general dual-task demands do not delay the onset of target selection nor attenuate the amount of visual attention that is allocated to the target. However, the amplitude peaked later in the single-task compared to the shortest SOA of the dual-task. This finding could be explained with the size of the amplitudes. Because the amplitude was numerically larger in the single-task, it could be possible that it took more time for it to peak. Overall, the results provided evidence that visual attention deployment is to a large extent independent of general dual-task demands.

The set size effect on the N2pc

The debate on the set size effect on the N2pc and, in turn, the underlying attentional processes that the N2pc indexes has not been solved yet (Heitz, Cohen, Woodman, & Schall, 2010; Luck, 2005; Luck, Girelli, McDermott, & Ford, 1997; Luck & Hillyard, 1994b; Mazza et al., 2009; Wolber & Wascher, 2003, 2005). Two partially contradictory positions in this debate are worth mentioning. The set size effect that we found (i.e., decreasing amplitude with larger set size) is in line with the first presented work. Cohen, Heitz, Woodman, and Schall (2009) investigated the relationship between neural activity and the set size effect. They recorded activity from visually responsive neurons in the frontal eye field (FEF) of macaque monkeys during a visual search task in which the set size was increased from two to four to eight items. The authors showed that increasing the set size decreased the peak firing rate in the FEF. Considering that the FEF are reciprocally connected with parietal and occipito-temporal areas in which the N2pc is generated (Hopf, Boelmans, Schoenfeld, Luck, & Heinze, 2004; Hopf et al., 2000), it is most likely to assume a link between activity in the FEF and activity in the N2pc relevant areas. Accordingly, it is plausible that a lower firing rate in the FEF could lead to decreased orienting of covert attention that would be seen in a reduced N2pc amplitude (see also Gregoriou, Gotts, Zhou, & Desimone, 2009; Heitz et al., 2010; Moore & Armstrong, 2003).

On the other hand, the (electrophysiological) ambiguity resolution theory (ART) provides another approach to explain the set size effect on the N2pc. According to Luck and colleagues (Luck, 2005; Luck et al., 1997; Luck & Hillyard, 1994b), visual attention is needed to solve the competition between target and distractors by suppressing distractors. As described in ART, the N2pc is assumed to reflect the suppression of distractors during search for the target. Suppression or filtering of distractors is mirrored in the amplitude that depends on the number of distractors and their spatial proximity to the target location. A higher number of distractors (i.e., larger set size) increases interference and attentional demands for distractor suppression, and thus the N2pc amplitude. However, the set size effect in the present study cannot be explained within the framework of ART, because we found a reduced amplitude with larger set size. ART emphasizes the spatial proximity between target and distractors as one important influence on the set size effect, but we kept spatial proximity between target and distractors similar across all search conditions. This could be one reason why we did not find the set size effect pattern as proposed by ART.

Absorption of visual search time into the slack time

Although our behavioural results provided evidence that visual search time was absorbed into the slack time, a closer look at the data showed that the absorption was only partial, especially in the target absent condition (Figures 4A and 4B). To shed more light on this issue, we calculated the number of items that were processed during and after the slack time. Accordingly, we calculated the search time that was absorbed into the slack time by subtracting the set size effect at SOA 60 from the set size effect at SOA 800, as these set size effects indicate the search time that was not absorbed into the slack time. In a next step, we determined the search slopes (ms/item) for target present and target absent at SOA 800. The search slopes indicate the rate at which the items are processed. The slopes were 11.00 ms/item for target present and 24.80 ms/item for target absent. Finally, in order to calculate the number of items that were processed during the slack time, we divided the absorbed search time by the search slopes. We also added 8 items (i.e., the smallest set size) as the locus-of-slack method assumes that in case of absorption, the easiest condition is fully processed during the slack time. Thus, for target present, 11.98 items were processed during the slack time [(43.80/11) + 8], and for target absent, 11.22 items were processed during the slack time [(79.77/24.80) + 8]. Correspondingly, calculating the difference to 16 items, 4.02 items for target present and 4.78 items for target absent were processed after the slack time. All in all, the analysis showed that a large part of the visual search proceeded concurrently to the response selection processes in Task 1 and that only a small number of items were processed after the slack time, for both target present and target absent.

In line with this reasoning, it should be noted that a comparison between the results of the present study and other studies that applied the locus-of-slack method and found full absorption of the perceptual processes revealed an important difference in the experimental designs. We increased the number of stimuli without changing their physical appearance to test whether the relevant process in Task 2 requires the same bottleneck mechanism as response selection in dual-tasks. Other studies, however, kept the number of stimuli constant, but manipulated their physical appearance (i.e., Hendrich, Strobach, Buss, Müller, & Schubert, 2012; Jentzsch, Leuthold, & Ulrich, 2007; Johnston, McCann, & Remington, 1995; Maquestiaux, Hartley, & Bertsch, 2004; Pashler & Johnston, 1989; Tombu & Jolicoeur, 2005). The possibility should be considered that increasing the number of stimuli could prolong the relevant task process more than manipulating the physical appearance of the stimuli. Consequently, the increased number of stimuli could have prevented full absorption in our task situation.

Furthermore, Pashler (1991) conducted a study in which he used chronometric reasoning that is complementary to the present approach to disentangle possible interference between visual attention shifts and response selection. In Experiment 1, Task 1 required an auditory two-choice discrimination. In Task 2, visual attention was shifted from fixation to a probe. The probe was a short horizontal line that marked the target letter in a letter display. The task was to report the probed target letter. The search display was masked after a brief exposure, and the author focused on target report accuracy (see Strobach, Schütz, & Schubert, 2015, for a general discussion of RTs and accuracy in PRP dual-tasks). The author did not find an effect of SOA, indicating that from the statistical perspective, accuracy was similar at short and at long SOA. However, accuracy was numerically reduced for 1.9% at short compared to long SOA (mean percentage errors of 24.3% vs. 22.4%). Interestingly, in the following Experiment 2, Pashler investigated whether the magnitude of this accuracy reduction corresponds to the expected accuracy reduction if the visual attention shift is subject to the same mechanism as response selection. Therefore, in this experiment, Pashler examined the costs of delaying the visual attention shift by delaying the probe. Concretely, Task 1 and the SOA manipulation were omitted, but the interval between the display and the probe was varied. When the probe was delayed by 183 ms, the mean percentage errors to report the target increased by 30% compared to when the probe was delayed by only 17 ms. Pashler reasoned that when the probe was experimentally delayed, visual attention was shifted later to the probe, resulting in reduced target report accuracy of 30% compared to when the probe and the corresponding visual attention shift were not delayed. The result thus allowed for estimating the reduction in accuracy that could be expected for a delayed visual attention shift compared to a non-delayed visual attention shift. Pashler compared the result to the findings of Experiment 1 that was conducted to investigate if at short SOA performing the response selection in Task 1 delays the visual attention shift in Task 2. The author concluded that if the visual attention shift had been subject to the same bottleneck mechanism as response selection and, accordingly, had been postponed, the accuracy reduction at short SOA should have been much higher than the observed 1.9% compared to the accuracy at long SOA (i.e., rather in the range of 30% as suggested by Experiment 2). Overall, Pashler (see also Pashler, 1989) concluded that these findings are in line with the assumption that the visual attention shifts in Task 2 operate independently of the response selection processes in Task 1.

It is important to reconcile the author's reasoning with our quantification approach to reveal more about the magnitude of the visual attention effects. In our approach, the non-absorbed search time at short SOA indicated that the response selection processes in Task 1 affected the visual attention processing in Task 2 to some extent. As described above, we calculated the number of items that were processed during and after the slack time to understand in detail how many items were processed concurrently and sequentially to response selection. Similarly, in Pashler's (1991) study, the accuracy findings indicated that the response selection processes in Task 1 could have delayed the visual attention shifts to the probe, as there were slightly more errors at short than at long SOA. The author measured the percentage of errors when an experimentally later presented probe induced a delay of the visual attention shifts to the probed letter and compared this effect to the actual effect in the dual-task. As the actual effect was much smaller, Pashler reasoned that the response selection processes in Task 1 do not delay the visual attention shifts to the probed letter. Taken together, both approaches aim at better understanding the magnitude of the visual attention effects to show more precisely to what extent visual attention and response selection operate independently.

Conclusion

In the present study, we investigated whether the response selection processes of an auditory discrimination Task 1 affect concurrent visual attention deployment in a conjunction search Task 2. Both the behavioural and electrophysiological results indicated that simultaneously performing the auditory discrimination task does not interrupt the feature binding processes (see also Lien et al., 2011, Experiments 1 and 2; Pashler, 1989, 1991). However, our results were at odds with the findings of Brisson and Jolicoeur (2007a, 2007b) and parts of the findings of Lien et al. (2011, Experiments 3 and 4). To dissolve this discrepancy, it should be investigated whether the opposing results are based on the specific experimental set-up, in particular the visual search conditions (i.e., search display masked vs. search display presented until response) and/or the complexity of the visual attention processes (i.e., automatic attentional capture to localize the target and discrimination of the response relevant target feature in a compound task vs. feature binding processes in a target detection task).

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by a Deutsche Forschungsgemeinschaft grant [grant number Schu 1397/5-2], awarded to Torsten Schubert. Christina B. Reimer is supported by the Elsa-Neumann Scholarship.

Footnotes

1

As one reviewer pointed out, in the conjunction search task, there was an imbalance regarding the display distribution of the target colour green, which could have facilitated target selection. In more detail, there were 3 versus 2 green items (and 1 vs. 2 red items) for set size 8 as well as 5 versus 4 green items (and 3 vs. 4 red items) for set size 16 presented to the left and right of fixation (see ). However, this imbalance, which was caused by the methodological need to have an equal number of items on the left and right side of the display (including one target and the distractors), did not facilitate target selection in the present study. There was a large set size effect on visual search time and on the N2pc amplitude. Both set size effects indicate that the participants performed the visual search task by deploying visual attention serially around the items of the display and that they bound both stimulus features—that is, the colour and the form feature—during that process. Thus, there is evidence that the target-colour imbalance in the present search task did not obscure the serial character of target selection.

Acknowledgements

We thank Werner Sommer for giving us the opportunity to conduct the EEG experiment in his lab.

References

Brisson

, & Jolicoeur

(2007a). Electrophysiological evidence of central interference in the control of visuospatial attention. Psychonomic Bulletin & Review, 14(1), 126–132. doi:10.3758/BF03194039

Brisson

, & Jolicoeur

(2007b). A psychological refractory period in access to visual short-term memory and the deployment of visual? Spatial attention: Multitasking processing deficits revealed by event-related potentials. Psychophysiology, 44(2), 323–333. doi:10.1111/j.1469-8986.2007.00503.x

Cameron

E. L.

, Tai

J. C.

, Eckstein

M. P.

, & Carrasco

(2004). Signal detection theory applied to three visual search tasks — Identification, yes/no detection and localization. Spatial Vision, 17(4), 295–325. doi:10.1163/1568568041920212

Cohen

J. Y.

, Heitz

R. P.

, Woodman

G. F.

, & Schall

J. D.

(2009). Neural basis of the set-size effect in frontal eye field: Timing of attention during visual search. Journal of Neurophysiology, 101(4), 1699–1704. doi:10.1152/jn.00035.2009

Di Lollo

(2012). The feature-binding problem is an ill-posed problem. Trends in Cognitive Sciences, 16(6), 317–321. doi:10.1016/j.tics.2012.04.007

Dowdall

J. R.

, Luczak

, & Tata

M. S.

(2012). Temporal variability of the N2pc during efficient and inefficient visual search. Neuropsychologia, 50(10), 2442–2453. doi:10.1016/j.neuropsychologia.2012.06.015

Eimer

(1996). The N2pc component as an indicator of attentional selectivity. Electroencephalography and Clinical Neurophysiology, 99(3), 225–234. doi:10.1016/0013-4694(96)95711-9

Gregoriou

G. G.

, Gotts

S. J.

, Zhou

, & Desimone

(2009). High-frequency, long-range coupling between prefrontal and visual cortex during attention. Science, 324(5931), 1207–1210. doi:10.1126/science.1171402

Heitz

R. P.

, Cohen

J. Y.

, Woodman

G. F.

, & Schall

J. D.

(2010). Neural correlates of correct and errant attentional selection revealed through N2pc and frontal eye field activity. Journal of Neurophysiology, 104(5), 2433–2441. doi:10.1152/jn.00604.2010

10.

Hendrich

, Strobach

, Buss

, Müller

H. J.

, & Schubert

(2012). Temporal-order judgment of visual and auditory stimuli: Modulations in situations with and without stimulus discrimination. Frontiers in Integrative Neuroscience, 6, 63. doi:10.3389/fnint.2012.00063

11.

Hickey

, Di Lollo

, & McDonald

J. J.

(2008). Target and distractor processing in visual search: Decomposition of the N2pc. Visual Cognition, 16(1), 110–113. doi:10.1080/13506280701692097

12.

Hopf

J. M.

, Boelmans

, Schoenfeld

, Luck

S. J.

, & Heinze

H.-J.

(2004). Attention to features precedes attention to locations in visual search: Evidence from electromagnetic brain responses in humans. Journal of Neuroscience, 24(8), 1822–1832. doi:10.1523/JNEUROSCI.3564-03.2004

13.

Hopf

J. M.

, Boelmans

, Schoenfeld

A. M.

, Heinze

H.-J.

, & Luck

S. J.

(2002). How does attention attenuate target-distractor interference in vision? Evidence from magnetoencephalographic recordings. Cognitive Brain Research, 15(1), 17–29. doi:10.1016/S0926-6410(02)00213-6

14.

Hopf

J. M.

, Luck

S. J.

, Girelli

, Hagner

, Mangun

G. R.

, Scheich

, & Heinze

H.-J.

(2000). Neural sources of focused attention in visual search. Cerebral Cortex, 10(12), 1233–1241. doi:10.1093/cercor/10.12.1233

15.

Jentzsch

, Leuthold

, & Ulrich

(2007). Decomposing sources of response slowing in the PRP paradigm. Journal of Experimental Psychology: Human Perception and Performance, 33(3), 610–626. doi:10.1037/0096-1523.33.3.610

16.

Jiang

, Saxe

, & Kanwisher

(2004). Functional magnetic resonance imaging provides new constraints on theories of the psychological refractory period. Psychological Science, 15(6), 390–396. doi:10.1111/j.0956-7976.2004.00690.x

17.

Johnston

J. C.

, McCann

R. S.

, & Remington

R. W.

(1995). Chronometric evidence for two types of attention. Psychological Science, 6(6), 365–369. doi:10.1111/j.1467-9280.1995.tb00527.x

18.

Jolicoeur

(1999). Dual-task interference and visual encoding. Journal of Experimental Psychology: Human Perception and Performance, 25(3), 596–616. doi:10.1037/0096-1523.25.3.596

19.

Jolicoeur

, Sessa

, Dell'Acqua

, & Robitaille

(2006). On the control of visual spatial attention: Evidence from human electrophysiology. Psychological Research, 70(6), 414–424. doi:10.1007/s00426-005-0008-4

20.

Kiesel

, Miller

, Jolicoeur

, & Brisson

(2008). Measurement of ERP latency differences: A comparison of single-participant and jackknife-based scoring methods. Psychophysiology, 45(2), 250–274. doi:10.1111/j.1469-8986.2007.00618.x

21.

Kirk

R. E.

(2013). Experimental design: Procedures for the behavioral sciences (4th ed.). Thousand Oaks, CA: Sage.

22.

Kiss

, Grubert

, & Eimer

(2013). Top-down task sets for combined features: Behavioral and electrophysiological evidence for two stages in attentional object selection. Attention, Perception, & Psychophysics, 75(2), 216–228. doi:10.3758/s13414-012-0391-z

23.

Lien

M.-C.

, Croswaite

, & Ruthruff

(2011). Controlling spatial attention without central attentional resources: Evidence from event-related potentials. Visual Cognition, 19(1), 37–78. doi:10.1080/13506285.2010.491643

24.

Logan

G. D.

, & Gordon

R. D.

(2001). Executive control of visual attention in dual-task situations. Psychological Review, 108(2), 393–434. doi:10.1037/0033-295X.108.2.393

25.

Luck

S. J.

(2005). The operation of attention - Millisecond by millisecond - over the first half second. In Ogmen

, & Breitmeyer

B. G.

(Eds.), The first half second: The microgenesis and temporal dynamics of unconscious and conscious visual processes (pp. 187–206). Cambridge, MA: MIT Press.

26.

Luck

S. J.

, Girelli

, McDermott

M. T.

, & Ford

M. A.

(1997). Bridging the gap between monkey neurophysiology and human perception: An ambiguity resolution theory of visual selective attention. Cognitive Psychology, 33(1), 64–87. doi:10.1006/cogp.1997.0660

27.

Luck

S. J.

, & Hillyard

S. A.

(1994a). Electrophysiological correlates of feature analysis during visual search. Psychophysiology, 31(3), 291–308. doi:10.1111/j.1469-8986.1994.tb02218.x

28.

Luck

S. J.

, & Hillyard

S. A.

(1994b). Spatial filtering during visual search: Evidence from human electrophysiology. Journal of Experimental Psychology: Human Perception and Performance, 20(5), 1000–1014. doi:10.1037/0096-1523.20.5.1000

29.

Maquestiaux

, Hartley

A. A.

, & Bertsch

(2004). Can practice overcome age-related differences in the psychological refractory period effect? Psychology and Aging, 19(4), 649–667. doi:10.1037/0882-7974.19.4.649

30.

Mazza

, Turatto

, & Caramazza

(2009). Attention selection, distractor suppression and N2pc. Cortex, 45(7), 879–890. doi:10.1016/j.cortex.2008.10.009

31.

Mazza

, Turatto

, Umiltà

, & Eimer

(2007). Attentional selection and identification of visual objects are reflected by distinct electrophysiological responses. Experimental Brain Research, 181(3), 531–536. doi:10.1007/s00221-007-1002-4

32.

Meyer

D. E.

, & Kieras

D. E.

(1997). A computational theory of executive cognitive processes and multiple-task performance: Part I. Basic mechanisms. Psychological Review, 104(1), 3–65. doi:10.1037/0033-295X.104.1.3

33.

Moore

, & Armstrong

K. M.

(2003). Selective gating of visual signals by microstimulation of frontal cortex. Nature, 421, 370–373. doi:10.1038/nature01341

34.

Müller

H. J.

, & Krummenacher

(2006)). Visual search and selective attention. Visual Cognition, 14(4-8), 389–410. doi:10.1080/13506280500527676

35.

Navon

, & Miller

(2002). Queuing or sharing? A critical evaluation of the single-bottleneck notion. Cognitive Psychology, 44(3), 193–251. doi:10.1006/cogp.2001.0767

36.

Pashler

(1989). Dissociations and dependencies between speed and accuracy: Evidence for a two-component theory of divided attention in simple tasks. Cognitive Psychology, 21(4), 469–514. doi:10.1016/0010-0285(89)90016-9

37.

Pashler

(1991). Shifting visual attention and selecting motor responses: Distinct attentional mechanisms. Journal of Experimental Psychology: Human Perception and Performance, 17(4), 1023–1040. doi:10.1037/0096-1523.17.4.1023

38.

Pashler

(1994). Dual-task interference in simple tasks: Data and theory. Psychological Bulletin, 116(2), 220–244. doi:10.1037/0033-2909.116.2.220

39.

Pashler

, & Johnston

J. C.

(1989). Chronometric evidence for central postponement in temporally overlapping tasks. The Quarterly Journal of Experimental Psychology Section A, 41(1), 19–45. doi:10.1080/14640748908402351

40.

Reimer

C. B.

, Strobach

, Frensch

P. A.

, & Schubert

(2015). Are processing limitations of visual attention and response selection subject to the same bottleneck in dual-tasks? Attention, Perception, & Psychophysics, 77(4), 1052–1069. doi:10.3758/s13414-015-0874-9

41.

Schubert

(1999). Processing differences between simple and choice reactions affect bottleneck localization in overlapping tasks. Journal of Experimental Psychology: Human Perception and Performance, 25(2), 408–425. doi: 239143

42.

Schubert

(2008). The central attentional limitation and executive control. Frontiers in Bioscience, 13(13), 3569–3580. doi:10.2741/2950

43.

Schubert

, Fischer

, & Stelzel

(2008). Response activation in overlapping tasks and the response-selection bottleneck. Journal of Experimental Psychology: Human Perception and Performance, 34(2), 376–397. doi:10.1037/0096-1523.34.2.376

44.

Schubert

, & Szameitat

A. J.

(2003). Functional neuroanatomy of interference in overlapping dual tasks: An fMRI study. Cognitive Brain Research, 17(3), 733–746. doi:239254

45.

Schubö

, Wykowska

, & Müller

H. J.

(2007). Detecting pop-out targets in contexts of varying homogeneity: Investigating homogeneity coding with event-related brain potentials (ERPs). Brain Research, 1138, 136–147. doi:10.1016/j.brainres.2006.12.059

46.

Schweickert

(1978). A critical path generalization of the additive factor method: Analysis of a Stroop task. Journal of Mathematical Psychology, 18(2), 105–139. doi:10.1016/0022-2496(78)90059-7

47.

Schweickert

(1980). Critical-path scheduling of mental processes in a dual task. Science, 209(4457), 704–706. doi:10.1126/science.7394529

48.

Strobach

, Schütz

, & Schubert

(2015). On the importance of Task 1 and error performance measures in PRP dual-task studies. Frontiers in Psychology, 6, 403. doi:10.3389/fpsyg.2015.00403

49.

Tombu

, & Jolicoeur

(2003). A central capacity sharing model of dual-task performance. Journal of Experimental Psychology: Human Perception and Performance, 29(1), 3–18. doi:10.1037/0096-1523.29.1.3

50.

Tombu

, & Jolicoeur

(2005). Testing the predictions of the central capacity sharing model. Journal of Experimental Psychology: Human Perception and Performance, 31(4), 790–802. doi:10.1037/0096-1523.31.4.790

51.

Töllner

, Strobach

, Schubert

, & Müller

H. J.

(2012). The effect of task order predictability in audio-visual dual task performance: Just a central capacity limitation? Frontiers in Integrative Neuroscience, 6, 75. doi:10.3389/fnint.2012.00075

52.

Treisman

, & Gelade

(1980). A feature-integration theory of attention. Cognitive Psychology, 12(1), 97–136. doi:10.1016/0010-0285(80)90005-5

53.

Welford

A. T.

(1952). The psychological refractory period and the timing of high speed performance – A review and a theory. British Journal of Psychology, 43(1), 2–19. doi:10.1111/j.2044-8295.1952.tb00322.x

54.

Wolber

, & Wascher

(2003). Visual search strategies are indexed by event-related lateralizations of the EEG. Biological Psychology, 63(1), 79–100. doi:10.1016/S0301-0511(03)00028-0

55.

Wolber

, & Wascher

(2005). The posterior contralateral negativity as a temporal indicator of visuo-spatial processing. Journal of Psychophysiology, 19(3), 182–194. doi:10.1027/0269-8803.19.3.182

56.

Wolfe

J. M.

(1994). Guided Search 2.0 A revised model of visual search. Psychonomic Bulletin & Review, 1(2), 202–238. doi:10.3758/BF03200774

57.

Wolfe

J. M.

(1998). What can 1 million trials tell us about visual search? Psychological Science, 9(1), 33–39. doi:10.1111/1467-9280.00006

58.

Wolfe

J. M.

(2007). Guided search 4.0: Current progress with a model of visual search. In Gray

(Eds.), Integrated Models of Cognitive System (pp. 99–119). New York: Oxford.

59.

Wolfe

J. M.

(2012). The binding problem lives on: Comment on Di Lollo. Trends in Cognitive Sciences, 16(6), 307–308. doi:10.1016/j.tics.2012.04.013

60.

Wolfe

J. M.

, & Bennett

S. C.

(1996). Preattentive object files: Shapeless bundles of basic features. Vision Research, 37(1), 25–43. doi:10.1016/S0042-6989(96)00111-3

61.

Wolfe

J. M.

, Cave

K. R.

, & Franzel

S. L.

(1989). Guided search: An alternative to the feature integration model for visual search. Journal of Experimental Psychology: Human Perception and Performance, 15(3), 419–433. doi:10.1037/0096-1523.15.3.419

62.

Woodman

G. F.

, & Luck

S. J.

(1999). Electrophysiological measurement of rapid shifts of attention during visual search. Nature, 400, 867–869. doi:10.1038/23698

63.

Woodman

G. F.

, & Luck

S. J.

(2003). Serial deployment of attention during visual search. Journal of Experimental Psychology: Human Perception and Performance, 29(1), 121–138. doi:10.1037/0096-1523.29.1.121