Abstract

Even when our attention is dedicated to an important task, background processes monitor the environment for significant events. The mismatch negativity (MMN) event-related potential is thought to reflect such a monitoring process. Nevertheless, there is continuing debate concerning the susceptibility of the MMN to attentional manipulation. We investigated the trial-by-trial relationship between brain activity related to change detection, reflected in the MMN, and visual psychophysical performance—while varying task difficulty. We find that auditory change detection is indeed “automatic” in that MMN remains robust despite increasing (visual) task load. However, the MMN amplitude and latency are susceptible to both visual load and to momentary attentional fluctuations as reflected in success or failure to identify a following visual target. We conclude that background central auditory processing is sensitive to the demands of a visual task, and fluctuates based on moment-to-moment allocation of attentional resources to the visual task.

INTRODUCTION

When attention is focused on a specific task, or subset of the sensory input, survival demands that other input be continuously monitored for unexpected events so that attention may be turned toward salient novel events. The auditory modality is well suited for such surveillance because it acquires information simultaneously from the entire surrounding space. A signature of the process of background monitoring is the mismatch negativity (MMN) event-related potential (ERP). The MMN is evoked upon an acoustic change during a sequence of sounds, which the subject is asked to ignore (Näätänen, 1990; Näätänen, Gaillard, & Mantysalo, 1978). Thus, it is generally considered a “preattentive” response. If the MMN truly reflects the process of detecting changes outside the focus of attention, then it should not depend on available attentional resources. Congruently, it was found to occur despite performance of attention-demanding tasks (Alho, Woods, & Algazi, 1994; Näätänen, 1990). However, findings regarding attentional modulation of MMN are inconsistent (Sussman, Winkler, & Wang, 2003; Woldorff, Hillyard, Gallen, Hampson, & Bloom, 1998; Näätänen, Paavilainen, Tiitinen, Jiang, & Alho, 1993; Woldorff & Hillyard, 1991). Here, we revisit this question from a cross-modal perspective, examining the effect of visual attentional load and moment-to-moment attentional fluctuations on auditory change detection.

In the prototypical MMN paradigm, subjects ignore the sounds and read a book or watch a movie. However, the fact that the MMN is elicited under these conditions cannot be taken as sufficient evidence for attention independence because these tasks do not require highly focused attention (Woldorff & Hillyard, 1991). Previous studies investigating the cross-modal effect of visual task load on the MMN found contradictory evidence. Dittmann-Balcar, Thienel, and Schall (1999) used visual and auditory discrimination tasks of various loads. The MMN decreased significantly for the most demanding auditory task but was unaffected by visual task difficulty, when deviant sounds were task irrelevant. Other studies similarly did not find a significant effect of visual load on the MMN (Muller-Gass, Stelmack, & Campbell, 2006; Dyson, Alain, & He, 2005; Sussman, Bregman, Wang, & Khan, 2005; Takegata et al., 2005; Muller, Achenbach, Oades, Bender, & Schall, 2002; Harmony et al., 2000; Otten, Alain, & Picton, 2000; Kathmann, Frodl-Bauch, & Hegerl, 1999; Alho et al., 1994; Alho, Woods, Algazi, & Näätänen, 1992). However, all of these studies used discrete stimulus presentations, leaving open the possibility that subjects shifted their attention to the auditory stimuli between visual stimulations. Recent studies limited participant attentional shifts between modalities by using visual tasks that required continuous vigilance. Conflicting results were found, with some studies reporting MMN decrements during more difficult visual tracking tasks (Yucel, Petty, McCarthy, & Belger, 2005a, 2005b; Singhal, Doerfling, & Fowler, 2002; Kramer, Trejo, & Humphrey, 1995), and others reporting an increase of MMN amplitude with increasing load (Zhang, Chen, Yuan, Zhang, & He, 2006; Muller-Gass, Stelmack, & Campbell, 2005; Restuccia, Della Marca, Marra, Rubino, & Valeriani, 2005).

Here, we explored the effects of visual attentional load on the MMN using a controlled load manipulation in a rapid-serial-visual-presentation (RSVP) paradigm. Critically, the paradigm allowed us to examine the effect of trial-by-trial attentional fluctuations on the mismatch response.

METHODS

Participants

Data of 13 participants were included in the analysis (age = 19–28 years, mean = 23.7 years; 6 women); data of three additional participants were discarded due to excessive eye movement and blink artifacts. The experiment conformed with institutional ethics regulations; participants provided informed consent and received course credit or remuneration for participation; all reported normal hearing and normal or corrected-to-normal vision.

Stimuli and Procedure

The experiment was run in a dimly lit, noise-attenuated, double-walled acoustic booth (Eckel C-26). Participants sat in an armchair 100 cm from a 17-in. CRT screen (resolution 1024 × 768 pixels, refresh rate 100 Hz). We presented a rapid series of 1° visual angle Latin capital letters (A–Z) and numerals (1–9) in 141-pt Times New Roman font. The stimuli were centered in a white square (200 × 200 pixels, RGB intensity values 255, 255, 255) at the center of the otherwise medium gray screen (RGB 128, 128, 128). Stimulus presentation frequency was constant with a stimulus onset asynchrony (SOA) of 90 msec (11.1 Hz). Stimuli were presented in blocks with fixed visual load. Visual load was manipulated by varying both contrast and stimulus duration. For the low load condition, letters and numerals differed in contrast level: Numeral targets were presented with a 100% contrast (black, RGB = 0, 0, 0), whereas letter distractors appeared at 5% contrast (RGB = 243, 243, 243); both were presented with a duration of 90 msec and no interstimulus interval (ISI). For the high load condition, all stimuli were presented at 5% contrast (243, 243, 243), for 20 msec, followed by a 70-msec blank (white) square ISI. These parameters were set according to the results of a pilot study for which data are not shown. Stimulus contrast was measured with a United Detector Technology type 61 Optometer.

High and low load blocks alternated, and the sequence began with either a high or low block in a counterbalanced fashion. In total, each participant performed six blocks of trials.

A single block consisted of 72 RSVP trials. Trial length varied, containing 28–62 stimuli. All but two stimuli in the sequence were letters. The other two were numerals, and served as targets, designated as T1 and T2. T1 appeared with equal probabilities in odd-numbered Positions 17–51 (i.e., with steps of 2 positions), and T2 appeared following a lag of one, three, or seven stimuli from T1. This design is akin to the one used in “attentional blink” paradigms (Hein, Alink, Kleinschmidt, & Müller, 2009; Chun & Potter, 1995; Shapiro, Raymond, & Arnell, 1994; Raymond, Shapiro, & Arnell, 1992). All 26 letters and 9 numerals were used across trials, in pseudorandom order. The same numeral target was not used twice successively. Six stimulus lists, specifying the order of letters and numerals and conforming to the above constraints, were formed. List assignment to load conditions was counterbalanced across participants. T1–T2 lags of 1, 3, or 7 were equally represented in each list in pseudorandom order.

The trial ended 11 stimuli after T1. Each trial was followed by a sequence of 27–29 “fixation” stimuli (medium gray; RGB 128, 128, 128; symbols *, #, $, or + presented for 90 msec each with no ISI). A similar sequence preceded the first trial as well. Participants were asked to report the identity of the numeral targets by typing them in the presented order on a keyboard number pad during the fixation period following the trial.

Auditory stimuli were presented throughout the experiment via a loudspeaker immediately under the monitor, synchronized with every fourth visual stimulus. Stimuli consisted of 75-msec complex tones (10 msec rise and fall time) with fundamental frequency and first three harmonics with relative amplitudes of 100%, 75%, 50%, and 25%. A series of 85% standard tones (fundamental frequency = 600 Hz) and 15% pitch deviants (fundamental 540 Hz) were presented at 70 dB with an SOA of 360 msec. The order of standards and deviants was random with the constraint that each deviant was preceded by at least three standards. Only ERP responses to deviants and standards that appeared before T1 were analyzed for the purposes of this study. At this time, the subject's search efforts were presumably well engaged, yet the ERPs were not confounded with the response to the visual targets. We therefore ensured that each trial sequence contained a deviant sound at Position 12, 13, or 14 from trial onset, that is, before T1 (which could appear no sooner than Position 17 in the list). All together, there were 282 pre-T1 deviants in our design per load condition (high and low).

The inclusion of the second target served to increase general load and for future comparison with ongoing experiments examining post-T2 effects. Issues of deviant probability, counterbalancing, and experiment length precluded examination of both pre- and post-T1 effects in the same study. Participants were instructed to disregard the sounds and concentrate on the visual task. Figure 1 presents a schematic representation of one trial in the low and one in the high visual load condition.

Figure 1. 

Trial design. Two target digits (T1 and T2) were embedded in an 11.1-Hz RSVP of Latin capital letters. Tones were presented simultaneously with every fourth visual stimulus. Subjects reported the targets' identities during the presentation of the nonletter, nondigit symbols at the end of each trial (only 2 of which are shown). Only mismatch responses for deviant and standard sounds appearing before T1 were further analyzed for the purposes of this study. Visual load was manipulated by decreasing target contrast level and decreasing target and distractor presentation time.

Figure 1. 

Trial design. Two target digits (T1 and T2) were embedded in an 11.1-Hz RSVP of Latin capital letters. Tones were presented simultaneously with every fourth visual stimulus. Subjects reported the targets' identities during the presentation of the nonletter, nondigit symbols at the end of each trial (only 2 of which are shown). Only mismatch responses for deviant and standard sounds appearing before T1 were further analyzed for the purposes of this study. Visual load was manipulated by decreasing target contrast level and decreasing target and distractor presentation time.

EEG Recording and Analysis

The EEG was recorded from 64 scalp sites using a Biosemi Active 2 system (Biosemi, Amsterdam, The Netherlands). Electrodes were embedded in an elastic electrode cap according to the extended 10–20 system. Two additional electrodes were placed on the mastoid processes. Eye movements and blinks were recorded by two horizontal EOG channels placed at the outer canthi of the two eyes and two vertical EOG channels above and below the right eye. EEG was recorded with 24-bit resolution at a sampling rate of 256 Hz, low-pass filtered at 67 Hz to prevent high-frequency aliasing, amplified × 15. Off-line, data were referenced to the nose and digitally high-pass filtered with a cutoff of 0.5 Hz. Prior to segmentation, epochs containing blinks, eye movements, excessive muscle activity, or recording artifacts were rejected following a supervised semiautomatic predefined procedure modified to suit individual variation in the EOG artifact amplitude. EEG segments of 500 msec (including 100 msec prestimulus), time-locked to auditory stimulus onsets, were extracted from the continuous recordings and averaged separately for each participant, electrode, visual load condition, and auditory stimulus (standard or deviant). About 30% of segments were excluded based on the above artifact rejection procedure, similarly for both load conditions. MMN was computed at each channel by subtracting the average pre-T1 standard response from the average pre-T1 deviant response (Schröger, 1998). The difference waves were filtered with a band pass of 1–12 Hz (Sinkkonen & Tervaniemi, 2000).

Data Reduction

MMN peak latencies were measured at electrode Fz on the difference waveforms of each subject, within a window of 80–220 msec from tone onset. The corresponding MMN amplitudes were measured in each individual, channel, and condition by averaging 14 data points around the time of the individual MMN peak.

RESULTS

Psychophysical Results

All individuals performed well above chance level, that is, above 11% correct, in both load conditions. Nevertheless, psychophysical results confirmed that the RSVP task became more difficult when the target's contrast level was decreased and presentation duration was shortened. A one-way repeated measure ANOVA with the fraction of correct T1 identification as dependent variable confirmed that task performance significantly decreased from low load (90.9 ± 2.4%)1 to high load (73.8 ± 3.5%) [F(1, 12) = 17.72, p < .001]. Note that we report T1 performance rather than T2 performance, as is customary for attentional blink studies, because we are interested here in the attentional load prior to T1, when the MMN was measured (see Methods).

To determine whether fatigue effects play an important role in producing T1 errors, we analyzed the distribution of incorrect responses across the experiment. This analysis could only be applied to high load trials, where a considerable number of incorrect trials were available. A repeated measures two-way ANOVA of the number of incorrect responses by block and by thirds of a block for the 13 participants revealed no significant difference in the number of incorrect responses between the three different high load blocks [mean ± SE, 1st: 18.23 ± 2.74; 2nd: 17.54 ± 2.4; 3rd: 19.62 ± 2.97; F(2, 12) = 1.32, p = .29], no difference between error rates for the beginning, middle, and end of each block [F(2, 12) = 0.65, p = .53], and no interaction between these effects [F(2, 12) = 0.58, p = .68]. These results rule out fatigue as a substantial source of T1 errors.

Checking for other possible causes of errors, we found no difference in number of pre-T1 deviants in correct versus incorrect trials (corresponding average ± SE: 1.32 ± 0.01, 1.29 ± 0.02, paired t test: p = .30). We also found no difference in lag between the last deviant and target in correct versus incorrect trials (corresponding average ± SE: 989 ± 1 msec, 993 ± 3 msec, paired t test: p = .30). We found that trial length, despite its large variability, had a tendency of an effect on performance (correct trials: 34 ± 0.14 stimuli; incorrect trials: 33.0 ± 0.47; p = .06), and it was in the direction opposite to that expected (i.e., that longer trials would lead to performance decrease).

ERP Results

The potentials evoked by standard and deviant tones in the two load conditions are shown in Figure 2A. There was a clear difference between the potentials recorded for the standard and the deviant tones in both conditions. Note, however, that these potentials also reflect the response to the simultaneously presented visual distractors. The visual response is filtered out when computing the difference wave, that is, the MMN. We therefore further discuss only the subtraction outcome to prevent confounding the results with visual response effects.

Figure 2. 

MMN effects. (A) ERP waveforms at frontal electrodes F3, Fz, and F4 (referenced to the nose) for standard (solid line) and deviant (dashed line) tones, for low (top row) and high (lower row) load conditions. Subtraction of these waveforms (solid from dashed) gives (B) the MMN for the two load conditions (black vs. gray lines). The separate standard and deviant ERPs contain a confounding visual response to the simultaneous visual (distractor) stimulus, which is removed by the subtraction producing the MMN. Note that the MMN was curtailed with higher visual load so that the latency is briefer and the amplitude decreased with higher visual load. These effects are shown quantitatively in (C), where we plot MMN mean latencies and amplitudes at Fz for the two load conditions (measured in a ±28-msec window around individual MMN peaks). Note that error bars represent standard errors of the mean across subjects, whereas comparisons were within subjects. (D) MMN topography for low versus high load demonstrating that the classical MMN topography was observed under both load conditions with a gradual polarity reversal from the frontal electrodes to the mastoid sites. Topographies were rendered using the topoplot function of the EEGLAB Matlab toolbox. M1,M2 = left and right mastoids.

Figure 2. 

MMN effects. (A) ERP waveforms at frontal electrodes F3, Fz, and F4 (referenced to the nose) for standard (solid line) and deviant (dashed line) tones, for low (top row) and high (lower row) load conditions. Subtraction of these waveforms (solid from dashed) gives (B) the MMN for the two load conditions (black vs. gray lines). The separate standard and deviant ERPs contain a confounding visual response to the simultaneous visual (distractor) stimulus, which is removed by the subtraction producing the MMN. Note that the MMN was curtailed with higher visual load so that the latency is briefer and the amplitude decreased with higher visual load. These effects are shown quantitatively in (C), where we plot MMN mean latencies and amplitudes at Fz for the two load conditions (measured in a ±28-msec window around individual MMN peaks). Note that error bars represent standard errors of the mean across subjects, whereas comparisons were within subjects. (D) MMN topography for low versus high load demonstrating that the classical MMN topography was observed under both load conditions with a gradual polarity reversal from the frontal electrodes to the mastoid sites. Topographies were rendered using the topoplot function of the EEGLAB Matlab toolbox. M1,M2 = left and right mastoids.

Figure 2B shows the average MMN waveforms for the two conditions at frontal electrodes F3, Fz, and F4. Figure 2D demonstrates the classical MMN topography elicited in both conditions with the gradual polarity reversal from frontal sites to the mastoid electrodes. The MMN amplitudes, as measured around the individual MMN peaks elicited at the frontal electrode Fz, were marginally reduced with the high load compared to the low load condition [Figure 2B, middle and Figure 2C; one-tailed paired t test: t(12) = 1.96, p < .05].2 Furthermore, the peak MMN latency was decreased when the load increased from low to high [one-tailed paired t test: t(12) = 2.1, p < .05]. These effects seem to derive largely from the reduced MMN width with high load (see Figure 2B, middle). Therefore, these results suggest that visual attention load affects the MMN response.

In the high load condition, there was a significant number of trials in which subjects did not correctly identify T1. We hypothesized that in the correct trials (hits), subjects were more likely to have strongly focused their attention on the visual task, as requested, whereas in incorrect trials (misses), their attention may have drifted, resulting in an error. Hence, any visual attention effect over the MMN should be maximal for correct trials and minimal for incorrect trials. We therefore analyzed the results for correct and incorrect trials separately in the high load condition. In the low load condition, errors were very rare (and were caused mainly by early T1 responses, before culmination of the distractor sequence, which went unrecorded). Thus, in this condition, we do not compare correct and incorrect trials, but look only at the results for correct trials.

Taking into account only correct trials from both conditions, a significant reduction of MMN peak amplitude was found when the visual load was increased from low to high [paired t test: t(12) = −2.61, p < .01; Figure 3A, B], as seen in the average waveforms of Figure 3A (compare full and dashed black lines) and the bar graphs of Figure 3B, bottom (compare left bars). This effect is also clearly evident on a subject-by-subject basis. Each dot in Figure 3C shows the MMN amplitudes of a single subject for correct T1 responses in the two load conditions. Note that most dots lie above the central diagonal, indicating smaller MMN amplitude for the high load condition than for the low load condition.

Figure 3. 

Moment-by-moment visual-task attention effect on the MMN. (A) MMN difference waveforms at electrode Fz, referenced to the nose, separating trials with correct versus incorrect T1 identification in the high load condition. Note that in the low load condition, performance was near perfect so only one waveform is shown. (B) MMN individual peak latencies and amplitudes measured ±28 msec around these peaks. Error bars represent standard error of the mean across subjects. (C) Single-subject MMN amplitudes for trials with correct T1 identification for low versus high load conditions. Note that for most subjects, the amplitudes are lower for the high load condition. (D) Single-subject MMN amplitudes comparing correct and incorrect trials within the high load condition. Note larger amplitudes of incorrect trials, despite similar stimulation.

Figure 3. 

Moment-by-moment visual-task attention effect on the MMN. (A) MMN difference waveforms at electrode Fz, referenced to the nose, separating trials with correct versus incorrect T1 identification in the high load condition. Note that in the low load condition, performance was near perfect so only one waveform is shown. (B) MMN individual peak latencies and amplitudes measured ±28 msec around these peaks. Error bars represent standard error of the mean across subjects. (C) Single-subject MMN amplitudes for trials with correct T1 identification for low versus high load conditions. Note that for most subjects, the amplitudes are lower for the high load condition. (D) Single-subject MMN amplitudes comparing correct and incorrect trials within the high load condition. Note larger amplitudes of incorrect trials, despite similar stimulation.

Moreover, within the high load condition, the MMN was significantly smaller for correct compared to incorrect trials [paired t test: t(12) = 3.79, p < .001], as seen in the waveforms of Figure 3A (compare dashed to gray line)3 and the bar graph of Figure 3B, bottom (compare right bars). Note that there were too few incorrect trials in the low load condition for comparable analysis. This “correct–incorrect” effect is prominent also on a subject by subject basis. As demonstrated in Figure 3D, almost all individual data points reflect a reduced MMN amplitude for the correct versus incorrect trials. The fact that stimulation was identical for correct and incorrect trials suggests that, indeed, attention and not perceptual parameters per se were responsible for this MMN effect.

The correct high load trials had a significantly shorter MMN latency than the low load trials [paired t test: t(12) = −2.35, p < .02] and tended to be shorter than the incorrect high load trials, as seen in Figure 3B (top): In two subjects MMN was abolished in the high correct trials, but was elicited for the low load condition, and out of the remaining 11 subjects, 8 displayed this tendency. Thus, for latency, it seems that when more attention was needed for the visual task (i.e., for the high load vs. low load condition), and more attention was actually paid to this task (i.e., for correct vs. incorrect visual task performance), the MMN peaked earlier.

DISCUSSION

We explored the effects of visual load in a task requiring continuous allocation of attention to a stream of visual stimuli, minimizing the possibility of occasional shifts of attention toward the simultaneous auditory stream. The MMN remained robust even in the most attention-demanding condition (except for 2 subjects; see Figure 3C). Nevertheless, this signal was not immune to the attentional demands of the visual task. When visual task difficulty was increased, not only did performance on this task decrease, but the MMN amplitude also decreased. Although this decrease was quite modest when combining data for all conditions, it is noticeably present on a subject-by-subject basis when comparing only data for trials where subjects correctly identified the first visual target, presumably allocating the required attention to the visual task. We introduced a novel analysis, comparing MMN amplitude for correct versus incorrect visual performance trials. Presence of a cross-modal auditory effect due to attention allocation to the visual task was further corroborated by the finding that MMN amplitude depended greatly on actual performance of the visual task—on a trial-by-trial basis. This dependence presumably reflects fluctuation in the degree of attention allocation to the visual task (leading to correct or incorrect response), even when the perceptual load of the visual task is kept constant.

Effects of Different Load Types on the MMN

Previous attempts to examine the effect of visual attention load on MMN yielded conflicting results, as reviewed in Table 1. Whereas most studies failed to find any effect (Table 1: 1–13a [here and in the following, numbers refer to numbered citations in Table 1]), some studies, as ours, reported reduced MMN with increased visual load (13b–16), and others even found an increase in MMN amplitude (17–19). Perhaps the source of these different results lies in the different types of load used in these studies and in the manner in which load was manipulated. This conjecture is outlined in the following paragraphs, which review the studies listed in Table 1.

Table 1. 

Studies of Visual Attention Load Effects on the MMN



Visual Task
Auditory MMN Effect
Taska
Load Control Manipulation
ISI/Duration
MMN (S = Standard; D = Deviant)
Effect Found
Alho et al., 1992  Detect visual deviant in mixed auditory/visual sequence. Target size ISI = 200–400 msec between any 2 stimuli Frequency: S = 1000 Hz; D = 1050/1500 Hz, plus broadband masking noise No visual load effect 
Duration = 50 msec 
Alho et al., 1994  detect occasional visual stimuli; (ignore secondary lateralized auditory stream) Compare to detect secondary auditory stream deviant tones (ignore visual) mean ISI = 6 sec Frequency: S = 1000 Hz; D = 1050/1300 Hz; ISI = 200–400 msec No visual load effect 
Dittmann-Balcar et al., 1999  Visual discrimination task Target/Nontarget size ratio ISI = 1.1–1.5 sec Duration: S = 100 msec; D = 50 msec; ISI = 1 sec No visual load effect 
Frequency: S = 1000 Hz; D = 940/970/985 Hz 
Kathmann et al., 1999  Detect orientation changes (horizontal to vertical); continuously presented grating Control group had no task. ISI 5–20 sec Frequency: S = 800 Hz; D = 880 Hz No visual load effect 
vertical orientation duration 500 msec 
Duration: S = 50 msec; D = 100 msec 
SOA = 500 msec 
Harmony et al., 2000  Reorder 5 simultaneously presented visual letters to form word Control: repeat a previously presented letter 5 times ISI ≥ 12 sec; self-paced trials; Duration = 1 sec Frequency: S = 1000 Hz; D = 1050 Hz No visual load effect 
ISI = 550–700 msec 
Otten et al., 2000  1- or 0-back task: report if digit is <, >5; 1- or 0-back depending on color Control: 0-back task; report if current digit is <, >5, regardless of color ISI 1300–1800 msec Frequency: S = 1000 Hz; D = 1250 Hz (D probability = .25) No visual load effect 
Duration 100 msec 
ISI = 400 or 900 msec 
Muller et al., 2002  Visual discrimination task (red vs. blue circles) Ignore visual and detect frequency deviants. ISI = 1.1 sec Frequency: S = 1000 Hz; D = 500 Hz No visual load effect 
Duration: S = 80 msec; D = 40 msec 
ISI = 1 sec 
Sussman et al., 2005  Detect letter repetition in temporal series Watch a video ISI = 1.4 sec 3 interleaved streams; Frequency of high tone No visual load effect (statistical comparison not reported) 
Duration = 150 msec S = 2489 Hz; D = 2637 Hz 
plus medium/low tone repeats 
SOA = 90 msec 
Dyson et al., 2005  1- or 0-back depending on current digit color; report digit < or > 5 0-back always; report current digit < or > 5, regardless of color ISI = 1.4–1.9 sec Temporal order No visual load effect 
S = 1189–841 Hz 
Duration = 100 msec D = 1189–1189–841 Hz or 841–841–1189 Hz groupings 
ISI = 400–600 msec 
10 Takegata et al., 2005  Detect 3-back location repetition 1-back task ISI = 1995 msec S & D were different combinations of pitch [e.g., C5 (523 Hz) or E5 (659 Hz)] and timber (violin-like or piano-like); SOA = 400 or 800 msec No visual load effect 
Duration = 200 msec 
11 Muller-Gass et al., 2006  Detect visual deviant in mixed visual–auditory sequence Target shape/color difference large or small SOA between any two stimuli ≥ 250 msec. Duration = 50 msec Frequency: S = 1000 Hz; D = 1050 Hz No visual load effect 
Intensity: S = 80 dB; D = 70 dB 
12 Wei, Chan, & Luo, 2002  Delayed response visual discrimination (press R or L key for visual standard or deviant, following go signal) Ignore visual: delayed response auditory discriminate standard/deviant ISI between any 2 stimuli = 250–700 msec. 0–2 auditory stimuli per visual stimulus Frequency: S = 800 Hz; D = 1000 Hz Shift in MMN distribution for attend visual vs. attend auditory, perhaps due to N2b contamination 
13 Singhal et al., 2002  Simulated landing of aircraft: Low vs. High turbulence level Continuous task Frequency: S = 1000 Hz; D = 1500 Hz 
a: dual task with dichotic listening a: No visual load effect 
b: ignore auditory stream b: MMN decreased with visual load 
ISI = 200–600 msec 
14 Kramer et al., 1995  Simulated radar-monitoring task (experts)—detect & identify aircraft Target density Continuous task, Duration: 3 min Frequency: S = 1500 Hz; D = 1000/2000 Hz MMN decreased with visual load 
ISI = 700 msec 
15 Yucel et al., 2005a  Visuomotor tracking task Manipulate joystick control dynamics Continuous tracking for at least 3.4 min Frequency: S = 600 Hz; D = 780 Hz MMN decreased with visual load 
ISI = 1.3 sec 
16 Yucel et al., 2005b  Visuomotor tracking task manipulate joystick control dynamics Continuous tracking for 3 min Frequency: S = 600 Hz; D = 700 Hz MMN decreased with visual load 
ISI = 1.3 sec 
17 Muller-Gass et al., 2005  Read experimenter-selected book; told would be questioned on text Controls: No questioning and/or self-selected text; or no reading at all Reading self-paced Frequency: S = 1000 Hz; D = 1050 Hz Frequency MMN increased only for experimenter text with query vs. self-selected text without query. 
Intensity: S = 80 dB; D = 70 dB No intensity MMN load effect 
ISI = 600 msec 
18 Restuccia et al., 2005  Multiple Feature Target Cancellation (MFTC) task Read a book Self-paced Frequency: S = 800 Hz; D = 500 Hz MMN increased with visual load 
ISI = 1 sec 
19 Zhang et al., 2006  Visuospatial tracking task Number of monitored targets (1, 3, 5 balls) Continuous tracking; 21 sec trials Frequency: S = 1000 Hz; D = 1500 Hz MMN increased with visual load 
ISI = 550 msec 


Visual Task
Auditory MMN Effect
Taska
Load Control Manipulation
ISI/Duration
MMN (S = Standard; D = Deviant)
Effect Found
Alho et al., 1992  Detect visual deviant in mixed auditory/visual sequence. Target size ISI = 200–400 msec between any 2 stimuli Frequency: S = 1000 Hz; D = 1050/1500 Hz, plus broadband masking noise No visual load effect 
Duration = 50 msec 
Alho et al., 1994  detect occasional visual stimuli; (ignore secondary lateralized auditory stream) Compare to detect secondary auditory stream deviant tones (ignore visual) mean ISI = 6 sec Frequency: S = 1000 Hz; D = 1050/1300 Hz; ISI = 200–400 msec No visual load effect 
Dittmann-Balcar et al., 1999  Visual discrimination task Target/Nontarget size ratio ISI = 1.1–1.5 sec Duration: S = 100 msec; D = 50 msec; ISI = 1 sec No visual load effect 
Frequency: S = 1000 Hz; D = 940/970/985 Hz 
Kathmann et al., 1999  Detect orientation changes (horizontal to vertical); continuously presented grating Control group had no task. ISI 5–20 sec Frequency: S = 800 Hz; D = 880 Hz No visual load effect 
vertical orientation duration 500 msec 
Duration: S = 50 msec; D = 100 msec 
SOA = 500 msec 
Harmony et al., 2000  Reorder 5 simultaneously presented visual letters to form word Control: repeat a previously presented letter 5 times ISI ≥ 12 sec; self-paced trials; Duration = 1 sec Frequency: S = 1000 Hz; D = 1050 Hz No visual load effect 
ISI = 550–700 msec 
Otten et al., 2000  1- or 0-back task: report if digit is <, >5; 1- or 0-back depending on color Control: 0-back task; report if current digit is <, >5, regardless of color ISI 1300–1800 msec Frequency: S = 1000 Hz; D = 1250 Hz (D probability = .25) No visual load effect 
Duration 100 msec 
ISI = 400 or 900 msec 
Muller et al., 2002  Visual discrimination task (red vs. blue circles) Ignore visual and detect frequency deviants. ISI = 1.1 sec Frequency: S = 1000 Hz; D = 500 Hz No visual load effect 
Duration: S = 80 msec; D = 40 msec 
ISI = 1 sec 
Sussman et al., 2005  Detect letter repetition in temporal series Watch a video ISI = 1.4 sec 3 interleaved streams; Frequency of high tone No visual load effect (statistical comparison not reported) 
Duration = 150 msec S = 2489 Hz; D = 2637 Hz 
plus medium/low tone repeats 
SOA = 90 msec 
Dyson et al., 2005  1- or 0-back depending on current digit color; report digit < or > 5 0-back always; report current digit < or > 5, regardless of color ISI = 1.4–1.9 sec Temporal order No visual load effect 
S = 1189–841 Hz 
Duration = 100 msec D = 1189–1189–841 Hz or 841–841–1189 Hz groupings 
ISI = 400–600 msec 
10 Takegata et al., 2005  Detect 3-back location repetition 1-back task ISI = 1995 msec S & D were different combinations of pitch [e.g., C5 (523 Hz) or E5 (659 Hz)] and timber (violin-like or piano-like); SOA = 400 or 800 msec No visual load effect 
Duration = 200 msec 
11 Muller-Gass et al., 2006  Detect visual deviant in mixed visual–auditory sequence Target shape/color difference large or small SOA between any two stimuli ≥ 250 msec. Duration = 50 msec Frequency: S = 1000 Hz; D = 1050 Hz No visual load effect 
Intensity: S = 80 dB; D = 70 dB 
12 Wei, Chan, & Luo, 2002  Delayed response visual discrimination (press R or L key for visual standard or deviant, following go signal) Ignore visual: delayed response auditory discriminate standard/deviant ISI between any 2 stimuli = 250–700 msec. 0–2 auditory stimuli per visual stimulus Frequency: S = 800 Hz; D = 1000 Hz Shift in MMN distribution for attend visual vs. attend auditory, perhaps due to N2b contamination 
13 Singhal et al., 2002  Simulated landing of aircraft: Low vs. High turbulence level Continuous task Frequency: S = 1000 Hz; D = 1500 Hz 
a: dual task with dichotic listening a: No visual load effect 
b: ignore auditory stream b: MMN decreased with visual load 
ISI = 200–600 msec 
14 Kramer et al., 1995  Simulated radar-monitoring task (experts)—detect & identify aircraft Target density Continuous task, Duration: 3 min Frequency: S = 1500 Hz; D = 1000/2000 Hz MMN decreased with visual load 
ISI = 700 msec 
15 Yucel et al., 2005a  Visuomotor tracking task Manipulate joystick control dynamics Continuous tracking for at least 3.4 min Frequency: S = 600 Hz; D = 780 Hz MMN decreased with visual load 
ISI = 1.3 sec 
16 Yucel et al., 2005b  Visuomotor tracking task manipulate joystick control dynamics Continuous tracking for 3 min Frequency: S = 600 Hz; D = 700 Hz MMN decreased with visual load 
ISI = 1.3 sec 
17 Muller-Gass et al., 2005  Read experimenter-selected book; told would be questioned on text Controls: No questioning and/or self-selected text; or no reading at all Reading self-paced Frequency: S = 1000 Hz; D = 1050 Hz Frequency MMN increased only for experimenter text with query vs. self-selected text without query. 
Intensity: S = 80 dB; D = 70 dB No intensity MMN load effect 
ISI = 600 msec 
18 Restuccia et al., 2005  Multiple Feature Target Cancellation (MFTC) task Read a book Self-paced Frequency: S = 800 Hz; D = 500 Hz MMN increased with visual load 
ISI = 1 sec 
19 Zhang et al., 2006  Visuospatial tracking task Number of monitored targets (1, 3, 5 balls) Continuous tracking; 21 sec trials Frequency: S = 1000 Hz; D = 1500 Hz MMN increased with visual load 
ISI = 550 msec 

aUnless otherwise mentioned, subjects were asked to ignore auditory stream.

Studies that found that greater attention to a visual task led to a reduced MMN required subjects to constantly monitor the visual stream (13–16)—as was the case in the present study. A potential caveat in the interpretation of these previous studies is that a rather large standard–deviant difference was used (on the order of 30–50% in frequency). With such large difference, the deviant-related response may conflate a true MMN and a larger N1 for nonadapted rare deviants (Schröger, 1998). Thus, the attention effect may be at least partially related to attentional sensitivity of the overlapping the N1 component (Woods, Alho, & Algazi, 1993). The much smaller deviation magnitude (10%) used in the present study is expected to reduce the contribution of N1-related effect, although it can clearly not preclude adaptation effects. Future studies should use designs which directly address adaptive and nonadaptive mechanisms (e.g., Jacobsen & Schröger, 2001) to dissociate the specific effects of attention load on these two complementary mechanisms of sensory memory.

In contrast to studies that found a visual attentional effect, a major characteristic of studies that did not find such an effect was that visual stimuli were presented with nonnegligible interstimulus gaps, in which auditory stimuli were presented (1–11). That is, auditory stimuli were presented at moments when there was no competing visual stimulus. For example, a recent study used mixed sequences of auditory and visual stimuli and subjects detected easy or hard visual targets (11). However, visual and auditory stimuli never overlapped and the minimum SOA between stimuli was 250 msec. Perhaps the reason that no visual load effect was found on the auditory MMN was that attention had shifted to the auditory modality between the occasional visual stimulus presentations.

The third variant, found in a few recent studies, was that MMN amplitude to unattended pitch deviants actually increased with increasing load, rather than decreased (17–19). These studies, too, required subjects to attend continuously to the visual task (unlike 1–11). Participants were engaged in a Multiple Features Target Cancellation task compared with a passive reading condition (18) or were engaged in a visuospatial tracking task, where the load was a function of the number of visual targets maintained and tracked (19). Zhang et al. (2006, 19) explain the unexpected MMN increase by suggesting that, under high cognitive load, resources are not available to suppress processing of irrelevant (auditory) information, resulting in a more robust mismatch response.4

The type of load may, in fact, be critical for the direction of the effect. When load is related to working memory, increased load tends to increase the auditory MMN signal. In contrast, increasing more perceptual loads leads to a decreased MMN response. Load was manipulated in Zhang et al.'s (2006, 19) study by varying the number of moving dots monitored (i.e., kept in working memory). In the Multiple Features Target Cancellation task (18), subjects detected targets that matched a combination of features, maintained in working memory. Subjects in the study of Muller-Gass et al. (2005, 17) expected questions regarding their reading material, again probably taxing working memory. These manipulations led to an increase of MMN amplitude relative to less demanding tasks. Thus, increasing working memory load seems to increase MMN amplitude (provided that stimulation is continuous; see 6, 9–10, who found no working memory load effect for noncontinuous stimulation). In contrast, in the current study and in others that found a decreased MMN with increasing load (Table 1: 13–16), load was manipulated by changing perceptual difficulty, requiring more attention to be allocated to perceptual processes: Subjects monitored radar screens and detected targets among distractors of different density and type (14); they “landed” a simulated aircraft under different turbulence conditions (13); or they manually tracked a target using a joystick, albeit with varying forces applied to make the task more difficult (15, 16). In the latter two tasks, the need to calculate the difference between the expected position of the target and its perceived position arguably increased perceptual load. The increased perceptual load may reduce the resources available for monitoring the acoustic environment, leading to MMN reduction.

Thus, we suggest that when the visual task load involves perceptual processes, these may compete on a cross-modal basis with early auditory perception, leading to the reduction in MMN amplitude (see also Kislyuk, Möttönen, & Sams, 2008). Two mechanisms may contribute to this reduction: First, shared resources may be depleted by the visual task, reducing the amount of resources available to auditory processing. Second, the increase in visual perceptual load may lead to active suppression of the auditory signal, perhaps via a central executive, also implicated in the working memory system (Baddeley, 1996). Consistent with the latter, when the visual task load involves a working memory executive function, the overloaded executive may find it hard to suppress processing of background auditory signals, leading to enhanced MMN signals.

Effect of Trial-by-Trial Attentional Fluctuations

The overall effect of visual load on MMN amplitude and latency was quite modest in our experiment. The mismatch responses were elicited by deviants occurring in the period prior to the presentation of T1. During that period, subjects were supposed to attend to the visual stream, searching for the target. A possible reason for the load effect being so small is that we averaged across trials in which subjects paid more or less attention to the visual task, due to ongoing fluctuations of sustained attention. We reasoned that these fluctuations should be reflected in subjects' target identification accuracy, and thus, trials with correct responses could be considered as reflecting, on average, more attention paid to the visual task than trials with incorrect responses. Indeed, separating the MMNs on the basis of visual performance revealed that within visual load, MMNs were significantly smaller when visual-task performance was correct compared with incorrect (Figure 3A, B, D). Moreover, examining the MMN only for correctly performed trials, during which participants are more likely to have allocated their attention to the visual task, showed a more prominent effect of increasing visual load on MMN amplitude than when all trials were included. It is possible that a major determinant of the amplitude reduction and shortened peak latency in correct high load trials is the earlier curtailment of the signal (see Figure 3A). In this case, it would seem that visual attentional load leads to a shortened auditory change detection process. It is as if the system is deciding to “get back to the important task at hand,” namely, the visual signal.

We suggest that the amount of attention devoted by subjects to the visual task fluctuates from trial to trial, and that with less attention devoted to the visual task, marked by reduced performance, more resources are available for the auditory stream. This finding is consistent with those of a recent fMRI study in which subjects were engaged in a visual identification task (Weissman, Roberts, Visscher, & Woldorff, 2006). Slower reaction times were associated with reduced activity in right dorsolateral prefrontal and anterior cingulate regions prior to the stimulus, suggesting a momentarily diminishing top–down attention control (i.e., an attentional lapse). The stimulus-evoked BOLD responses in visual sensory areas were concomitantly diminished, and most importantly, activity related to distractor interference was enhanced in these trials. Presumably, when attentional control is fully functioning, the competition between task-relevant and task-irrelevant streams is biased toward the former, whereas the latter may be suppressed. During lapses of attention, the biasing signal is reduced and suppression of the irrelevant stream diminishes. This results in improved processing of irrelevant stimuli (Weissman et al., 2006).

An alternative interpretation could be that the mismatch response has some random (or at least currently unexplained) variability and that occasional larger mismatch responses attract more attention, and thus, disrupt subsequent visual performance (reflected here as impaired T1 detection). Indeed, Escera, Corral, and Yago (2002) found that deviants occurring 300 msec prior to a visual target hinder visual task performance. However, Escera et al. (personal communication, 10 January 2007) actually found that this effect was limited to the first 400 msec after the deviant. In the current study, in more than 96% of the trials, deviant–target gap exceeded this limit (and 84% exceeded double this limit). Moreover, we found that the distance between the deviant and the target did not affect the error rate. Thus, the suggestion that MMN amplitude affected visual performance seems unlikely.

Conclusion

We conclude that auditory mismatch detection is maintained even under conditions demanding very high visual attention (Muller-Gass et al., 2006; Sussman et al., 2005; Fischer, Morlet, & Giard, 2000; Dittmann-Balcar et al., 1999; Kathmann et al., 1999; Näätänen & Winkler, 1999; Näätänen, 1990; Näätänen et al., 1978), yet the MMN response is not completely encapsulated from the degree of attention deployed by a concomitant visual task. The most direct interpretation of these results is that attentional resources are at least partly shared between modalities. Alternatively, use of attentional resources by a visual task could lead to an active suppression of auditory signals, even without assuming shared limited resources. Presumably, this alternative process would depend on a central cross-modal executive control.

One of the functions of the deviance detection system is detecting and alerting to the presence of a salient and, perhaps, important external event, outside the current focus of attention (Näätänen, 1990). The robustness of this system under highly demanding tasks suggests that this system is, indeed, automatic in the sense that it acts by default, and may not require focused attention to operate. However, it may be advantageous that even such a system will have some flexibility, so that its interfering effect may be reduced when the primary task becomes highly challenging and requires all available resources. By pushing the system to its limits and examining its ongoing dynamics, the present study reveals this interaction between the onstage processes involved in performing an attended task and the backstage monitoring operations, within a cross-modal context.

Acknowledgments

We thank Istvan Winkler for helpful discussion. This study was supported by grants from the Israel Science Foundation of the Israel Academy of Sciences and the US–Israel Binational Science Foundation (S. H.) and grant 9-2004-5 from the National Institute of Psychobiology in Israel founded by the Charles E. Smith family (L. Y. D.).

Reprint requests should be sent to Dr. Leon Deouell, Department of Psychology, The Hebrew University of Jerusalem, Jerusalem 91905, Israel, or via e-mail: Leon.Deouell@huji.ac.il.

Notes

1. 

This may be an underestimate of the real accuracy, as some subjects responded very rapidly in the low load condition with some responses occurring before the permitted time (i.e., before the fixation symbols), and this was considered an error by the experimental program. High load condition responses were slower and never too early.

2. 

Whereas the statistics were computed based on individual subjects' peak latencies, the difference in amplitude across conditions is smeared in the grand-average waveforms which do not take into account the intersubject latency jitter.

3. 

The MMN for high load in incorrect trials seems to differ from the other waveforms in Figure 3A already prior to 80 msec following stimulus onset, but this difference was found insignificant [paired t test: correct high load vs. incorrect high load t(12) = −0.99, p = .34; correct high load vs. low load t(12) = −0.86, p = .4].

4. 

Note, however, that according to Lavie's (2005) Load Theory, increasing item number is related to perceptual rather than cognitive load and should favor early selection, that is, increased inhibition of auditory change detection. Lavie and De Fockert's (2003) results would also predict that changes in stimulus contrast, as used in the present study, would affect task difficulty, not perceptual load. Space limitations preclude further discussion of these issues.

REFERENCES

Alho
,
K.
,
Woods
,
D. L.
, &
Algazi
,
A.
(
1994
).
Processing of auditory-stimuli during auditory and visual-attention as revealed by event-related potentials.
Psychophysiology
,
31
,
469
479
.
Alho
,
K.
,
Woods
,
D. L.
,
Algazi
,
A.
, &
Näätänen
,
R.
(
1992
).
Intermodal selective attention: 2. Effects of attentional load on processing of auditory and visual-stimuli in central space.
Electroencephalography and Clinical Neurophysiology
,
82
,
356
368
.
Baddeley
,
A.
(
1996
).
Exploring the central executive.
Quarterly Journal of Experimental Psychology
,
49
,
5
28
.
Chun
,
M. M.
, &
Potter
,
M. C.
(
1995
).
A 2-stage model for multiple-target detection in rapid serial visual presentation.
Journal of Experimental Psychology: Human Perception and Performance
,
21
,
109
127
.
Dittmann-Balcar
,
A.
,
Thienel
,
R.
, &
Schall
,
U.
(
1999
).
Attention-dependent allocation of auditory processing resources as measured by mismatch negativity.
NeuroReport
,
10
,
3749
3753
.
Dyson
,
B. J.
,
Alain
,
C.
, &
He
,
Y.
(
2005
).
Effects of visual attentional load on low-level auditory scene analysis.
Cognitive Affective & Behavioral Neuroscience
,
5
,
319
338
.
Escera
,
C.
,
Corral
,
M. J.
, &
Yago
,
E.
(
2002
).
An electrophysiological and behavioral investigation of involuntary attention towards auditory frequency, duration and intensity changes.
Cognitive Brain Research
,
14
,
325
332
.
Fischer
,
C.
,
Morlet
,
D.
, &
Giard
,
M. H.
(
2000
).
Mismatch negativity and N100 in comatose patients.
Audiology and Neuro-Otology
,
5
,
192
197
.
Harmony
,
T.
,
Bernal
,
J.
,
Fernandez
,
T.
,
Silva-Pereyra
,
J.
,
Fernandez-Bouzas
,
A.
,
Marosi
,
E.
,
et al
(
2000
).
Primary task demands modulate P3a amplitude.
Cognitive Brain Research
,
9
,
53
60
.
Hein
,
G.
,
Alink
,
A.
,
Kleinschmidt
,
A.
, &
Müller
,
N. G.
(
2009
).
The attentional blink modulates activity in the early visual cortex.
Journal of Cognitive Neuroscience
,
21
,
197
206
.
Jacobsen
,
T.
, &
Schröger
,
E.
(
2001
).
Is there pre-attentive memory-based comparison of pitch?
Psychophysiology
,
38
,
723
727
.
Kathmann
,
N.
,
Frodl-Bauch
,
T.
, &
Hegerl
,
U.
(
1999
).
Stability of the mismatch negativity under different stimulus and attention conditions.
Clinical Neurophysiology
,
110
,
317
323
.
Kislyuk
,
D. S.
,
Möttönen
,
R.
, &
Sams
,
M.
(
2008
).
Visual processing affects the neural basis of auditory discrimination.
Journal of Cognitive Neuroscience
,
20
,
2175
2184
.
Kramer
,
A. F.
,
Trejo
,
L. J.
, &
Humphrey
,
D.
(
1995
).
Assessment of mental workload with task-irrelevant auditory probes.
Biological Psychology
,
40
,
83
100
.
Lavie
,
N.
(
2005
).
Distracted and confused?: Selective attention under load.
Trends in Cognitive Sciences
,
9
,
75
82
.
Lavie
,
N.
, &
De Fockert
,
J. W.
(
2003
).
Contrasting effects of sensory limits and capacity limits in visual selective attention.
Perception & Psychophysics
,
65
,
202
212
.
Muller
,
B. W.
,
Achenbach
,
C.
,
Oades
,
R. D.
,
Bender
,
S.
, &
Schall
,
U.
(
2002
).
Modulation of mismatch negativity by stimulus deviance and modality of attention.
NeuroReport
,
13
,
1317
1320
.
Muller-Gass
,
A.
,
Stelmack
,
R. M.
, &
Campbell
,
K. B.
(
2005
).
“… and were instructed to read a self-selected book while ignoring the auditory stimuli”: The effects of task demands on the mismatch negativity.
Clinical Neurophysiology
,
116
,
2142
2152
.
Muller-Gass
,
A.
,
Stelmack
,
R. M.
, &
Campbell
,
K. B.
(
2006
).
The effect of visual task difficulty and attentional direction on the detection of acoustic change as indexed by the Mismatch Negativity.
Brain Research
,
1078
,
112
130
.
Näätänen
,
R.
(
1990
).
The role of attention in auditory information-processing as revealed by event-related potentials and other brain measures of cognitive function.
Behavioral and Brain Sciences
,
13
,
201
232
.
Näätänen
,
R.
,
Gaillard
,
A. W. K.
, &
Mantysalo
,
S.
(
1978
).
Early selective-attention effect on evoked-potential reinterpreted.
Acta Psychologica
,
42
,
313
329
.
Näätänen
,
R.
,
Paavilainen
,
P.
,
Tiitinen
,
H.
,
Jiang
,
D.
, &
Alho
,
K.
(
1993
).
Attention and mismatch negativity.
Psychophysiology
,
30
,
436
450
.
Näätänen
,
R.
, &
Winkler
,
I.
(
1999
).
The concept of auditory stimulus representation in cognitive neuroscience.
Psychological Bulletin
,
125
,
826
859
.
Otten
,
L. J.
,
Alain
,
C.
, &
Picton
,
T. W.
(
2000
).
Effects of visual attentional load on auditory processing.
NeuroReport
,
11
,
875
880
.
Raymond
,
J. E.
,
Shapiro
,
K. L.
, &
Arnell
,
K. M.
(
1992
).
Temporary suppression of visual processing in an RSVP task: An attentional blink.
Journal of Experimental Psychology: Human Perception and Performance
,
18
,
849
860
.
Restuccia
,
D.
,
Della Marca
,
G.
,
Marra
,
C.
,
Rubino
,
M.
, &
Valeriani
,
M.
(
2005
).
Attentional load of the primary task influences the frontal but not the temporal generators of mismatch negativity.
Cognitive Brain Research
,
25
,
891
899
.
Schröger
,
E.
(
1998
).
Measurement and interpretation of the mismatch negativity.
Behavior Research Methods, Instruments, & Computers
,
30
,
131
145
.
Shapiro
,
K. L.
,
Raymond
,
J. E.
, &
Arnell
,
K. M.
(
1994
).
Attention to visual-pattern information produces the attentional blink in rapid serial visual presentation.
Journal of Experimental Psychology: Human Perception and Performance
,
20
,
357
371
.
Singhal
,
A.
,
Doerfling
,
P.
, &
Fowler
,
B.
(
2002
).
Effects of a dual task on the N100–P200 complex and the early and late Nd attention waveforms.
Psychophysiology
,
39
,
236
245
.
Sinkkonen
,
J.
, &
Tervaniemi
,
M.
(
2000
).
Towards optimal recording and analysis of the mismatch negativity.
Audiology and Neuro-Otology
,
5
,
235
246
.
Sussman
,
E.
,
Winkler
,
I.
, &
Wang
,
W. J.
(
2003
).
MMN and attention: Competition for deviance detection.
Psychophysiology
,
40
,
430
435
.
Sussman
,
E. S.
,
Bregman
,
A. S.
,
Wang
,
W. J.
, &
Khan
,
F. J.
(
2005
).
Attentional modulation of electrophysiological activity in auditory cortex for unattended sounds within multistream auditory environments.
Cognitive Affective & Behavioral Neuroscience
,
5
,
93
110
.
Takegata
,
R.
,
Brattico
,
E.
,
Tervaniemi
,
M.
,
Varyagina
,
O.
,
Näätänen
,
R.
, &
Winkler
,
I.
(
2005
).
Preattentive representation of feature conjunctions for concurrent spatially distributed auditory objects.
Cognitive Brain Research
,
25
,
169
179
.
Wei
,
J. H.
,
Chan
,
T. C.
, &
Luo
,
Y. J.
(
2002
).
A modified oddball paradigm “cross-modal delayed response” and the research on mismatch negativity.
Brain Research Bulletin
,
57
,
221
230
.
Weissman
,
D. H.
,
Roberts
,
K. C.
,
Visscher
,
K. M.
, &
Woldorff
,
M. G.
(
2006
).
The neural bases of momentary lapses in attention.
Nature Neuroscience
,
9
,
971
978
.
Woldorff
,
M. G.
, &
Hillyard
,
S. A.
(
1991
).
Modulation of early auditory processing during selective listening to rapidly presented tones.
Electroencephalography and Clinical Neurophysiology
,
79
,
170
191
.
Woldorff
,
M. G.
,
Hillyard
,
S. A.
,
Gallen
,
C. C.
,
Hampson
,
S. R.
, &
Bloom
,
F. E.
(
1998
).
Magnetoencephalographic recordings demonstrate attentional modulation of mismatch-related neural activity in human auditory cortex.
Psychophysiology
,
35
,
283
292
.
Woods
,
D. L.
,
Alho
,
K.
, &
Algazi
,
A.
(
1993
).
Intermodal selective attention—Evidence for processing in tonotopic auditory fields.
Psychophysiology
,
30
,
287
295
.
Yucel
,
G.
,
Petty
,
C.
,
McCarthy
,
G.
, &
Belger
,
A.
(
2005a
).
Graded visual attention modulates brain responses evoked by task-irrelevant auditory pitch changes.
Journal of Cognitive Neuroscience
,
17
,
1819
1828
.
Yucel
,
G.
,
Petty
,
C.
,
McCarthy
,
G.
, &
Belger
,
A.
(
2005b
).
Visual task complexity modulates the brain's response to unattended auditory novelty.
NeuroReport
,
16
,
1031
1036
.
Zhang
,
P.
,
Chen
,
X. C.
,
Yuan
,
P.
,
Zhang
,
D. R.
, &
He
,
S.
(
2006
).
The effect of visuospatial attentional load on the processing of irrelevant acoustic distractors.
Neuroimage
,
33
,
715
772
.