The effect of visual task difficulty and attentional direction on the detection of acoustic change as indexed by the Mismatch Negativity

Abstract

Näätänen's model of auditory processing purports that attention does not affect the MMN. The present study investigates this claim through two different manipulations. First, the effect of visual task difficulty on the passively elicited MMN is assessed. Second, the MMNs elicited by stimuli under attended and ignored conditions are compared. In Experiment 1, subjects were presented with mixed sequences of equiprobable auditory and visual stimuli. The auditory stimuli consisted of standard (80 dB SPL 1000 Hz), frequency deviant (1050 Hz), and intensity deviant (70 dB SPL) tone pips. In a first instance, subjects were instructed to ignore the auditory stimulation and engage in an easy and difficult visual discrimination task (focused condition). Subsequently, they were asked to attend to both modalities and detect visual and auditory deviant stimuli (divided condition). The results indicate that the passively elicited MMN to frequency and intensity deviants did not significantly vary with visual task difficulty, in spite of the fact that the easy and difficult tasks showed a wide variation in performance. The manipulation of the attentional direction (focused vs. divided conditions) did result in a significant effect on the MMN elicited by the intensity, but not frequency, deviant. The intensity MMN was larger at frontal sites when subjects' attention was directed to both modalities as compared to only the visual modality. The attentional effect on the MMN to the intensity deviants only may be due to the specific deviant feature or the poorer perceptual discriminability of this deviant from the standard. Experiment 2 was designed to address this issue. The methods of Experiment 2 were identical to those of Experiment 1 with the exception that the intensity deviant (60 dB SPL) was made to be more perceptible than the frequency deviant (1016 Hz) when compared to the standard stimulus (80 dB SPL 1000 Hz). The results of Experiment 2 also demonstrated that the passively elicited MMN was not affected by large variations in visual task difficulty; this provides convincing evidence that the MMN is independent of visual task demands. Similarly to Experiment 1, the direction of attention again had a significant effect on the MMN. In Experiment 2, however, the frequency MMN (and not the intensity MMN) was larger at frontal sites during divided attention compared to focused visual attention. The most parsimonious explanation of these results is that attention enhances the discriminability of the deviant from the standard background stimulation. As such, small acoustic changes would benefit from attention whereas the discriminability of larger changes may not be significantly enhanced.

Introduction

The mismatch negativity (MMN), a component of the auditory event-related potential (ERP), is evoked by a rare change in an otherwise homogeneous stream of auditory stimuli. It has been elicited by changes in a variety of static and temporal stimulus features (Näätänen and Winkler, 1999). In Näätänen's model of auditory processing (Näätänen, 1990), the generator mechanism of the MMN is associated with the functions of a brief duration auditory sensory memory. According to this model, a feature detector system forwards information about the physical properties of the incoming stimulus to a sensory memory store where feature and temporal integration takes place. The resulting stimulus representation is strengthened with repeated presentations of the same auditory stimulus (i.e., standard). However, if a subsequent incoming stimulus is detected as different (i.e., deviant) from the well-formed stimulus trace, the MMN generator mechanism is activated. It is hypothesized that the automatic change-detector sends an interrupt signal to higher executive mechanisms in order to redirect attention to potentially significant changes in the environment (Escera et al., 1998).

Central to Näätänen's model is the claim that the physical features of auditory stimuli are fully analyzed and encoded whether the channel in which the stimuli occur is attended or not. That is, the quality and quantity of sensory information entering the MMN generator mechanism are not modulated by attention. Näätänen proposes that the purpose of attention is to increase the likelihood that a stimulus will gain access to a limited-capacity system where it will be further evaluated. The additional processing afforded to attended stimuli is nevertheless performed by a different system operating in parallel with the obligatory sensory analysis.

Early studies of the MMN indicated that it could indeed be elicited by small changes in auditory stimulation when the subject's attention was directed to another cognitive activity in either auditory or visual modalities (Näätänen et al., 1978, Sams et al., 1984). Further support for the attentional insensitivity of the MMN came from experiments reporting comparable MMNs elicited by attended and unattended stimuli (Näätänen et al., 1978, Näätänen et al., 1980). These studies were criticized, however, for employing conditions that did not require a strong selective focusing of attention. As such, subjects may have diverted resources to the processing of the auditory stimuli even when they were presented in the to-be-ignored channel. In order to facilitate early selective attention, stimuli must be presented at very rapid rates, the subject's task must be difficult, and the attended and ignored channels must be easily distinguished (Woldorff and Hillyard, 1990).

More recent studies, in particular those employing intramodal auditory selective attention tasks requiring a strong attentional focus, indicated that attention can affect the morphology of the MMN (Alain and Woods, 1997, Arnott and Alain, 2002, Näätänen et al., 1993, Szymanski et al., 1999, Trejo et al., 1995, Woldorff et al., 1991, Woldorff et al., 1998). These studies reported that the MMN elicited by a deviant stimulus delivered in the unattended channel appeared to be smaller than that elicited by the same deviant when it was part of the attended sequence. The ERP elicited by a deviant stimulus presented in the attended channel, however, is characterized by additional components, notably another negative deflection, the N2b. The N2b may overlap both spatially and temporally with the MMN, resulting in a composite negative deflection. A contribution of the N2b component to the attentional effect on the MMN cannot therefore be excluded in these studies.

The problem of the N2b overlap is generally overcome in most MMN studies by instructing the subject to engage in a visual diversion task. The extent to which attentional focus is required by the visual diversion task is often considered incidental. The primary aim of the present study was to determine if the attentional requirements imposed by visual tasks do, in fact, modulate the passively elicited MMN. The few studies that considered this question generally concluded that the demands of the visual task have little effect on the MMN (Alho et al., 1992, Dittmann-Balcar et al., 1999, Harmony et al., 2000, Kathmann et al., 1999, Muller-Gass et al., 2005, Otten et al., 2000; see, however, Kramer et al., 1995, Yucel et al., 2005). In these studies, two visual diversion tasks have typically been employed. One task is made to be more difficult than the other and thus presumed to demand a greater focusing of attention. The underlying assumption in these studies is that the strong demands of the difficult visual task reduce the likelihood that the subject will either rapidly switch or allocate resources to the processing of the task-irrelevant auditory stimuli. It is assumed that during the easy task, subjects have available resources to switch to and process the (irrelevant) auditory stimuli. On this basis, it has generally been expected that, if the MMN were modulated by attention, a much-reduced MMN would be elicited during the difficult relative to easy visual task condition.

The absence of a visual task difficulty effect on the MMN in most studies may be because the experimental conditions needed to demonstrate an effect of visual task demands were not optimized. As described in the following sub-section, the effect of visual attention on the MMN is more rigorously tested in the present study, by maximizing the likelihood of finding an effect of task demands. The assumption at the basis of these studies, that fewer attentional resources are available for auditory processing during a difficult than easy visual task, is also examined in the present study by assessing the effect of visual task difficulty on concomitant auditory processing (see Section 1.1.2).

The effects of attention are often only apparent under highly specific conditions. In previous MMN studies, the failure to find effects of visual task demands may be explained by methodological issues (slow stimulus presentation rate; lack of variation in visual task difficulty; choice of deviant). Several of these studies employed a slow auditory (Dittmann-Balcar et al., 1999) or visual (Dittmann-Balcar et al., 1999, Harmony et al., 2000, Kathmann et al., 1999, Otten et al., 2000) stimulus presentation rate. When visual stimuli are presented slowly, there is sufficient time between presentations to switch between the attended-visual and apparently ignored-auditory channels (Schwent et al., 1976). Furthermore, when the unattended auditory stimuli are presented at slow rates, these stimuli are likely to cause involuntary attention shifts toward the to-be-ignored channel (Hansen and Hillyard, 1984). Slow rates of stimulus presentation are hence not conducive to the selective focusing of attention.

The manipulation of task demands in previous studies has not resulted in a wide range of task performance. For example, in the Kathmann et al. and Otten et al. studies, both “easy” and “difficult” tasks were easily completed, hit rates ranging from 0.83 to 0.95 in the difficult condition. In contrast, in the Alho et al. study, “easy” and “difficult” tasks were difficult to complete, both tasks having hit rates below 0.80. A wider range of task difficulty may be required in order to elicit statistically measurable differences in the MMN. Other studies, such as those by Harmony et al. and Muller-Gass et al., employed visual diversion tasks that presumed to vary in difficulty; task difficulty was, however, not quantified and, therefore, the difference between easy and more difficult tasks could not be assessed.

Finally, previous studies have almost exclusively employed small (Alho et al., 1992, Harmony et al., 2000, Kathmann et al., 1999) or large (Alho et al., 1992, Otten et al., 2000) frequency deviants. MMNs elicited by large deviants have generally not been affected by attentional manipulations (Woods et al., 1992). Further, the MMN to small intensity deviants has been reported to be more susceptible to the effects of attention than the MMN to small frequency deviants (Näätänen et al., 1993). The intensity deviant has not been employed in studies of visual task difficulty and MMN (see, however, Muller-Gass et al., 2005).

The present study optimized the experimental conditions necessary for finding an effect of visual task difficulty on the MMN. Subjects were instructed to perform a visual discrimination task while ignoring the auditory stimuli (this will be referred to as the focused attention condition). In one condition, this task was very easy, with hit rates close to 1.0. In another condition, the discriminability of the visual target was adjusted for each subject in order to obtain hit rates that were close to 0.5. Therefore, in contrast to previous studies, this experiment was designed to assure a large separation in terms of performance between the “easy” and “difficult” tasks. Furthermore, auditory and visual stimuli were presented at rapid and unpredictable rates. Finally, the auditory stimulus sequence included both small frequency and intensity deviants.

Critically, this study specifically tests the often-made assumption that as the demands of the visual task increase, the available resources for covert processing of the task-irrelevant auditory stimuli will decrease. Divided attention paradigms are often used to determine the extent to which processing of different tasks actually compete for the same resources (Pashler, 1994). In the second part of this study, subjects were again asked to perform, in separate conditions, an easy or difficult visual discrimination task. In addition, they were instructed to concurrently discriminate the rare intensity and frequency stimulus changes presented in the auditory channel (this will be referred to as the divided attention condition). It was expected that if processing resources were to a large extent shared by the auditory and visual tasks, then auditory performance measures and auditory ERPs would be sensitive to the manipulation of visual task difficulty.

A second and related question also addressed in this study concerned the effects of the direction of attention on the MMN. This issue was previously investigated in studies employing intermodal auditory–visual selective attention tasks. These studies showed that, under some circumstances, the MMN appeared to be larger when subjects attended to the MMN-eliciting auditory stimuli compared to when they were engaged in a visual task such as reading (Alain and Woods, 1997, Schröger, 1995) a visual discrimination task (Alho et al., 1992, Woods et al., 1992) or a video game (Gomes et al., 2000, Kramer et al., 1995, Woods et al., 1992). This attentional effect emerged primarily when the MMN-eliciting deviants closely resembled the standard stimuli (Alho et al., 1992, Woods et al., 1992), such as those employed in the present study. As mentioned previously, the larger MMN reported in the attended condition may, however, in part be explained by an overlapping attention-dependent N2b component. Subtle differences between the characteristics of the MMN and N2b may be useful in disentangling their respective contribution to the composite negativity. The MMN often has maximal amplitude at fronto-central scalp sites, inverts in polarity below the Sylvian fissure, and occurs prior to 200 ms following stimulus onset. The N2b is typically largest at central sites, remains negative-going at sites below the Sylvian fissure, and reaches peak amplitude after 250 ms (Alho et al., 1986). In the present study, attention was directed exclusively to the visual task during the focused visual condition, whereas subjects attended to both the visual and auditory channels during the divided attention condition. The effects of the direction of attention were evaluated by comparing the MMNs obtained when the auditory stimuli were ignored (focused condition) with that obtained when the auditory stimuli were attended (divided condition).