MMN elicitation during natural sleep to violations of an auditory pattern

Abstract

The mismatch negativity (MMN) ERP component is generally considered to reflect the outcome of a pre-conscious change detection mechanism. The manipulation of active task demands has typically demonstrated that the MMN operates relatively independently of inferred attention. It remains a possibility, however, that subjects are capable of covertly sampling, or “eavesdropping” on, the irrelevant auditory stimuli, even during the most demanding of diversion tasks. The presence of the MMN in an unconscious state, such as natural sleep, provides strong evidence that its operations take place at a pre-conscious level. There exists consistent evidence that the MMN can be elicited at least during REM sleep, but these MMNs were typically elicited using oddball paradigms in which the new physical properties of deviants may trigger fresh afferent activation. The current sleep study employed a standard pattern in which two pure tones alternated (e.g. ABABABAB). Deviants were repetitions (e.g., ABABBBAB), and therefore physically identical to the preceding standard. In different conditions, the tones of the pattern were separated by either 1 or 6 semitones. A clear MMN was elicited in the waking state in the 6 semitone condition. The MMN was also elicited in the 6 semitone condition during REM sleep. No MMN was apparent in REM sleep in the 1 semitone condition. The MMN was not apparent in either the 6 or 1 semitone condition during NREM sleep. These results confirm the operation of the MMN in REM sleep, and support the view that the MMN operates at a pre-conscious level of processing.

Introduction

The ability to detect changes in the environment that occur outside the realm of active focus is crucial for survival. Näätänen (1990) proposed a model of attention and automaticity in auditory processing in which two passive routes of processing have the potential to interrupt the limited-capacity central executive. This interruption allows the output of passive analysis to become available for further detailed analysis in the perceptual and cognitive systems, presumably requiring some form of conscious awareness. One route, associated with the N1 ERP component, detects obtrusive changes in the transient energy of stimuli, such as their onsets and offsets. A second route, associated with the mismatch negativity (MMN) ERP component, detects deviations from an otherwise homogeneous train of stimuli. The present study examines this second route to awareness.

The MMN exhibits a fronto-central maximum scalp distribution, peaks 100–250 ms following the onset of the deviant stimulus, and inverts in polarity over the mastoids when a nose reference is used (Sams et al., 1985a). Fundamental to the Näätänen model is the claim that the detection of change, signaled by the MMN, occurs whether the subject attends to the auditory stimuli or not. In other words, Näätänen's original model assumes that MMN generation occurs at a pre-attentive (or pre-conscious) level.

A number of studies have tested this assertion by examining the susceptibility of the MMN to manipulations of attention and task demands (for a recent review, see Sussman, 2007). Many intramodal studies (attend to one auditory channel, ignore another) have demonstrated a reduced MMN to deviant stimuli occurring in the unattended auditory channel, compared to when the same channel was attended (e.g. Woldorff et al., 1991, Woldorff et al., 1998, Trejo et al., 1995, Alain and Woods, 1997, Arnott and Alain, 2002, Alain and Izenberg, 2003). One criticism of intramodal studies of attention and the MMN is that overt attention to the stimuli in one channel elicits another negativity, the N2b, in addition to the MMN (Näätänen et al., 1993). The N2b peaks about 250–300 ms after stimulus onset, and shares a similar fronto-central distribution with the MMN. The N2b does not, however, invert at the mastoids. Some manipulations yield an N2b that appears as a distinct peak (e.g. Sams et al., 1984, Sams et al., 1985b), but the differences between the N2b and the MMN are often quite subtle. Several studies have nevertheless demonstrated attenuation of the MMN using procedures that minimized the possibility of measuring an overlapping N2b (Alain and Alain and Woods, 1997, Woldorff et al., 1998, Alain and Izenberg, 2003).

The N2b confound can also be overcome by requiring subjects to engage in a task in a different modality (usually visual) while ignoring the concurrently presented auditory stimuli¹. In these intermodal paradigms, the difficulty of the visual task is often manipulated with the assumption that an easy visual task requires fewer attentional resources for its successful completion than a difficult one. Thus, during an easy visual task, more resources would presumably be available to sample (or “eavesdrop” on) the to-be-ignored auditory channel, compared to a difficult task. Most studies have shown the MMN to be unaffected by the visual task (Alho et al., 1994, Dittmann-Balcar et al., 1999, Dyson et al., 2005, Harmony et al., 2000, Kathmann et al., 1999, Muller-Gass et al., 2005, Muller-Gass et al., 2006, Muller-Gass et al., 2007, Otten et al., 2000, Sams et al., 1984, Sams et al., 1985b, Sculthorpe et al., 2008), though not all have (Kramer et al., 1995, Yucel et al., 2005, Restuccia et al., 2005, Zhang et al., 2006).

Muller-Gass et al. (2006) questioned the underlying assumption of intermodal studies that subjects cannot sample the auditory channel during a difficult visual task. They thus employed both a focused condition, in which subjects actively performed a visual–perceptual task and ignored the auditory stimuli, and a divided attention condition, in which subjects were instructed to divide their attention and perform the visual task while also responding to targets in the auditory oddball sequence. Subjects were capable of successfully dividing their attention between the two tasks. Performance on the visual task declined slightly during the divided compared to the focused attention condition. Importantly, their ability to detect auditory targets was not modulated by visual task difficulty. This suggests that it is possible for subjects to covertly sample the auditory channel while performing a visual task, even if that task is very difficult. This might explain why many intermodal studies have shown the MMN to be unaffected by visual task difficulty.

Since both intramodal and intermodal studies run the risk that subjects might be covertly monitoring stimuli in a “to-be-ignored” channel, the best test for theories that claim certain aspects of processing are carried out independently of conscious awareness may be whether this processing can be demonstrated in an unconscious state. A number of researchers have therefore examined the effects of natural sleep on the MMN. Sleep is associated with methodological problems that are different from those in the waking state. Sleep is not uniform. It consists of a series of stages, including non-REM (NREM) stages 1–4, reflecting the depth of sleep², and REM sleep. The amplitude of the background EEG is much higher during sleep than in the waking state, particularly during N3 (often > 200 μV). This can make the MMN (often < 1 μV) difficult to observe (Sabri and Campbell, 2002).

Most studies have failed to demonstrate an MMN in stage 2 of NREM sleep (Loewy et al., 1996, Nashida et al., 2000, Nielsen-Bohlman et al., 1991, Nittono et al., 2001, Paavilainen et al., 1987, Sallinen et al., 1997; Winter et al., 1995). Those that have done so used very large deviants (Sabri et al., 2000, Sabri et al., 2003, Sabri and Campbell, 2005, Sallinen et al., 1994, Ruby et al., 2008). There is some evidence of an MMN in stage 3 (Ruby et al., 2008), but it cannot be elicited in stage 4 sleep. Unlike NREM, the MMN is very consistently reported in REM sleep, often with an attenuated amplitude compared to the waking state (Atienza and Atienza and Cantero, 2001, Atienza et al., 1997, Atienza et al., 2000, Loewy et al., 1996, Nashida et al., 2000, Ruby et al., 2008, Sabri and Campbell, 2005).

There exist, however, methodological problems with much of the existing sleep MMN literature. These problems are primarily a consequence of the so-called “oddball” paradigm. In the oddball paradigm, the subject is presented with a sequence of discrete, identical standard auditory stimuli. At rare and unpredictable times, a physical feature is changed to form a “deviant” stimulus that elicits the MMN. A change in any physical feature of the standard stimulus can elicit a MMN, including its tonal frequency (Näätänen et al., 1978, Sams et al., 1985b), intensity (Näätänen et al., 1987), duration (Näätänen et al., 1989), or spatial location (Paavilainen et al., 1989).

Every investigation of the MMN in sleep has used oddball paradigms, usually with frequency deviants (Atienza and Atienza and Cantero, 2001, Atienza et al., 1997, Atienza et al., 2000, Campbell et al., 1992, Loewy et al., 1996, Nashida et al., 2000, Nielsen-Bohlman et al., 1991, Nittono et al., 2001, Paavilainen et al., 1987, Sabri and Campbell, 2005, Sabri et al., 2000, Sabri et al., 2003, Sallinen et al., 1994, Sallinen et al., 1997; Winter et al., 1995). The frequency oddball paradigm may not elicit a pure MMN, particularly when the extent of deviance is large (Jacobsen and Schröger, 2001). Auditory afferent processing occurs in a tonotopic fashion, in which populations of sensory neurons are maximally responsive to specific tonal frequencies. In a frequency oddball paradigm, afferent activation by the often-presented standard pitch rapidly diminishes because neurons in the responsive population enter their refractory period. Deviants that are very different from the standards activate a different neuronal population. Since these deviants are rarely presented, the neuronal population responsive to the deviant pitch remains “fresh” and produces a large response when activated.

The N1 ERP component is much affected by the refractoriness of afferent neuronal populations, and its amplitude increases as the rate of stimulation presentation is slowed (Näätänen and Picton, 1987). The use of a frequency oddball paradigm therefore elicits a deviance-related negativity (DRN) that is composed of the superimposition of both the MMN and the N1. Since the scalp-recorded N1 can overlap both spatially and temporally with the MMN, the separation of its contribution to the resultant DRN is often very difficult. The problem of fresh afferent activation remains a concern during sleep. N1 is reduced to baseline level during NREM sleep, but returns to 25–50% of its waking amplitude during REM sleep (Colrain and Campbell, 2007).

Few sleep studies have examined the MMN using paradigms that avoid N1 enhancement. Oddball sequences in which the deviants are created by decreasing stimulus intensity appear to be free of N1 enhancement (Jacobsen et al., 2003), but no MMN has been demonstrated to such stimuli during sleep (Loewy et al., 2000, Macdonald et al., 2008). Oddball sequences using deviants that have a shorter duration than the standard also appear to elicit a pure MMN (Jacobsen and Schröger, 2003). Duration decrement deviants have been shown to elicit the MMN in all stages of sleep, except stage 4 (Ruby et al., 2008). The same group has even reported MMNs to duration decrement deviants in comatose patients (Fischer et al., 1999). Thus, a definitive MMN has only been demonstrated in sleep using duration decrement deviants, and it is not known whether an MMN can be elicited in sleep by deviance along other stimulus dimensions.

Some sleep studies using frequency oddball paradigms attempted to minimize the risk of N1 overlap by using a small deviant (Loewy et al., 1996), or by using a deviant that represented a change to a complex standard stimulus that could only be detected after training (Atienza and Cantero, 2001)³. Both of these studies demonstrated the MMN during REM sleep, but the possibility of N1 enhancement cannot be discounted.

The MMN is not restricted to the oddball paradigm. It can also be elicited by a violation of a complex regularity (Näätänen et al., 2001). In these paradigms, no physically identical standard stimuli exist. What becomes “standard” is a more psychological rule, or regularity, that governs the relationships among physically different standard stimuli. Deviants in these paradigms are rule violations. Nordby et al. (1988) developed a simple rule violation paradigm using two alternating tones. In this paradigm, two tones of different pitch, A and B, consistently alternate in a standard pattern (ABABABAB…) that is occasionally broken by a deviant repetition (ABABAAB… or ABABBAB…). The deviant repetition elicits an MMN because it violates rules that have been extrapolated from the standard sequence based on the temporal organization of its two constituent tonal frequencies (Horváth et al., 2001). Since deviant stimuli in this paradigm are physically identical to the standard tones that precede them, the MMN cannot be attributed fresh afferent activation.

The MMN elicited by violations of such two-tone alternating patterns generally exhibits the same morphology as that elicited in the oddball paradigm. It possesses a peak latency in the 100–250 ms range, a maximum amplitude at fronto-central sites, and inverts at the mastoid when a nose reference is used (e.g., Alain et al., 1994, Takegata et al., 2005, Sculthorpe et al., 2008, Sculthorpe et al., 2009). Mastoid inversion for the MMN to violations of an alternating pattern does not, however, appear to be as robust as that elicited by simple feature deviants in an oddball paradigm. While an intermodal study of attention demonstrated a mastoid inversion even during a very difficult visual task (Sculthorpe et al., 2008), an intramodal dichotic listening study demonstrated a small MMN that did not invert in polarity at the mastoids in response to pattern violations in the unattended ear (Alain and Woods, 1997).

The present study will employ a similar two-tone alternating pattern to test for the presence of an MMN in natural sleep. The demonstration of an MMN in sleep to violations of an alternating pattern would not only provide convincing evidence of a true MMN to frequency-related deviance at the pre-conscious level, but would also indicate that the auditory system is capable of detecting more abstract rule violations even in an unconscious sleeping state.

Some studies employing the frequency oddball in sleep have also examined the effect of extent of deviance (Loewy et al., 1996, Sabri et al., 2000, Sabri et al., 2003, Sabri and Campbell, 2005). Under these conditions, the MMN to large deviants was typically better preserved in sleep. The current study therefore employs two conditions, in which the constituent tones of the alternating pattern are separated by either 1 or 6 semitones. Unlike in frequency oddball paradigms, this increased tonal separation does not represent an increased “extent of deviance”, as tonal frequency is not the only dimension upon which deviance is established. Like the frequency oddball, however, MMN amplitude does appear to increase with tonal separation, probably due to increased perceptibility of the pattern (Alain et al., 1994). In the waking state, it was expected that the MMN in the 6 semitone condition would be larger than that in the 1 semitone condition. These patterned auditory sequences were also presented during a single night of natural sleep.