Conflict monitoring in multi-sensory flanker tasks: Effects of cross-modal distractors on the N2 component

Highlights

• Effects of cross-modal distractors on conflict monitoring were examined using ERP.

• Cross-modal distractors enhanced the effect of congruency on RT and N2 amplitude.

• N2-RT correlation in the visual task supports “response conflict” account of N2.

• N2-RT correlation was generally weakened in the audiovisual task.

• Exploratory data analysis was conducted on the NoGo congruency effect.

Abstract

The N2 component is a well-known neural correlate of conflict monitoring (CM), being more negative in the presence of conflicting information in visual conflict tasks. However, whether to-be-ignored auditory distractors can introduce additional conflict remains unknown. In the present work, subjects performed a visual (V) and audiovisual (AV) version of a Go/NoGo flanker task, and responded only if the target arrow pointed toward a pre-specified direction (e.g., left). In the AV task, in which to-be-ignored auditory distractors that were semantically associated with the flankers were concurrently presented, the congruency effect on both RT and N2 amplitude was enhanced, confirming that additional conflict can be brought about by cross-modal distractors at both behavioural and neural levels. Consistent with the hypothesis that N2 amplitude reflects response conflict in visual conflict tasks, within-subject correlation between N2 amplitude and RT was significant in the Go conditions for the V task (congruent/incongruent). However, for the AV task, the correlation was significant only in the congruent condition. These findings suggest that while the cross-modal conflict is registered by the CM process, only part of this conflict could effectively induce response conflict.

Introduction

Successful goal-oriented behaviour depends on the functioning of cross-modal selective attention, which enables us to focus our attention on one modality, and filter out irrelevant/conflicting information in unattended modalities. Previous works have extensively studied how conflict processing operates in uni-sensory conflict tasks, e.g., Eriksen's flanker task [1]. The conflict monitoring (CM) theory [[2], [3]] is especially influential in elucidating the brain network that subserves conflict processing. It hypothesizes that the anterior cingulate cortex (ACC) plays a key role to detect conflicts between competing response options (see also [4]). Since its original proposal, the theory has been extended to other aspects of conflict processing, including conflict adaptation [5] and error detection [6]. In event-related potential (ERP) research, the N2 component—a frontocentral component that peaks about 200–350 ms post-stimulus [7]—is known to be an index of conflict monitoring [8]. The N2 amplitude is larger (i.e., more negative) in the presence of response conflict [[9], [10]]. For example, in Go/NoGo tasks, the N2 amplitude is larger for rare NoGo trials because subjects need to overcome a response tendency induced by frequent Go trials [[11], [12]]. Similarly, in flanker tasks, it is larger for incongruent than congruent trials, because the incorrect response option is activated involuntarily by the flankers in incongruent trials [[12], [13]].

To date, there has been little work on whether the CM process is sensitive to cross-modal distractors presented in a to-be-ignored modality. While it is highly plausible that the presence of such cross-modal distractors in a visual (V) conflict task would introduce additional interference, results of behavioural studies have been mixed [[14], [15], [16]]. For example, Elliott et al. [14] compared the congruency effect in three audiovisual (AV) Stroop paradigms with the conventional Stroop task. In these AV paradigms, auditory distractors that either match or mismatch the visual stimuli in colour were simultaneously presented. Compared to the conventional task, the congruency effects were all significantly reduced. The authors suggested that such reduction might be due to effects of dilution, i.e., the reduction in interference in the presence of a neutral distractor. However, using a similar stimulus design as Elliott et al. [14], a later work by Francis et al. [15] demonstrated significant enhancement in the RT congruency effect, regardless of whether the distracting verbal dimension was embedded within the coloured word target or presented in its vicinity as flankers.

To our knowledge, only one study has examined a similar issue [17]. In this ERP study, subjects made Go responses to the visual word DRÜCK (“press” in German). The NoGo trials were split into three conditions—the visual word STOPP (“stop”) was presented concurrently with the spoken word STOPP (“congruent”) or DRÜCK (“incongruent”), or in isolation. Despite the effectiveness of cross-modal selective attention, the result revealed a congruency effect on N2 amplitude, confirming that cross-modal distractors can bring about response conflict. However, whether cross-modal distractors remain effective in inducing conflict in the presence of distractors in the attended modality remains unknown.

Another unaddressed issue concerns the functional role of N2 component in AV conflict tasks. In V conflict tasks, a number of works found a within-subject correlation between N2 amplitude and RT [[6], [18]]—N2 amplitude was larger (i.e., more negative) for trials with slower than faster RT. This provided solid evidence that N2 amplitude primarily reflects the degree of response conflict in V conflict tasks. However, it is known that the ACC is not only sensitive to response conflict but also sensory conflicts [4]—both within-modal [19] and cross-modal [20]. It is plausible that if N2 amplitude in an AV task is a mixture of sensory and response conflicts, the N2–RT correlation may be weakened when compared with V tasks.

The present EEG study employed a flanker paradigm, which has been central in the development of the CM theory, to test whether to-be-ignored cross-modal distractors introduce additional conflict in a visual conflict task. Specifically, we modified the Go/NoGo flanker task used in a recent study (see Table 1 of [21], “inhibition block”), and created a visual (V) and audiovisual (AV) version of the task. The task of Groom and Cragg [21] was unique in that subjects are only required to respond with the dominant hand. In examining the correlation between N2 amplitude and RT with Go trials, this has the advantage of avoiding complications that arise due to the slower responses by the non-dominant hand.

In the V task, subjects were presented with spatial arrays, each comprising a target symbol (left or right arrows) in the middle and flankers that were either the same as (congruent, C) or different from (incongruent, I) the target symbol. They were instructed to ignore the flankers, and respond to the target arrow at the central fixation point only if the arrow pointed toward a pre-specified direction (left/right). This task comprised four critical conditions: for both Go and NoGo trials, the two types of trials (C/I) were presented with equal probability. Neutral trials were also included as fillers, to ensure compatibility with the bulk of the previous works [e.g.,13]. Go trials occurred twice as often as NoGo trials (probability: Go, 67%; NoGo, 33%). Similarly, the AV task comprised the same four conditions, except that a word (e.g., left, presented in subjects’ native language) that always matched the direction of the flanker was presented aurally and simultaneously with the spatial array.

Comparing the two tasks, we hypothesized that the congruency effect (I minus C) on RT would be significantly greater for AV than V task. Correspondingly, the congruency effect on N2 amplitude would also be significantly larger (more negative) in the AV task. Neutral trials were excluded in testing these two a priori hypotheses in order to increase the statistical power. In addition to these main hypotheses, further analyses were conducted to clarify the functional role played by the N2 component, with respect to whether the N2 amplitude reflects response conflict even in an audiovisual scenario. If the additional conflict introduced by to-be-ignored auditory distractors primarily occurs at a response level in the AV task, the within-subject N2 amplitude–RT correlation should be similar or even strengthened when compared with that in the V task. However, if it instead represents a mixture of sensory and response conflicts, the same correlation should be weakened in the AV task. In addition, it remains unknown whether the congruency effect for NoGo conditions is comparable to that for Go conditions. Exploratory analysis focusing on NoGo conditions was therefore conducted to clarify this point.