SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. References

Aim  Event-related potentials (ERPs) obtained when focused attention is kept away from the stimulus (unnoticed stimulation) are possibly linked to automatic mismatch-detection mechanisms, and could be a useful tool to investigate sensory discrimination ability. By considering the high impact of impaired somatosensory integration on many neurological disturbances in children, we aimed to verify whether mismatch-related responses to somatosensory stimulation could be obtained in healthy children.

Method  Eleven healthy participants (age range 6–11y, mean 8y 2mo, SD 1y 7mo; seven males, four females) underwent ‘oddball’ electrical stimulation of the right hand (80% frequent stimuli delivered to the thumb, 20% deviant stimuli delivered to the fifth finger). Data were compared with those obtained when the frequent stimuli to the thumb were omitted (‘standard-omitted’ protocol). ERPs were recorded at frontal, central, and parietal scalp locations. Children’s overt attention was engaged by a demanding video game.

Results  In the oddball protocol, deviant stimulation elicited a left central negativity at about 160ms latency, followed by a left frontal negative response at about 220ms latency. Standard-omitted traces showed only a left parietal negative response spreading to right parietal regions.

Interpretation  Mismatch-related somatosensory responses can be reliably obtained in children, providing that appropriate technical contrivances are used. In clinical use, the frontal components, which are present only during the oddball protocol, could be a reliable and unequivocal neurophysiological marker of the automatic mismatch-detection mechanism.

List of abbreviations
ERP

Event-related potentials

MMN

Mismatch negativity

N140

Parieto-central negative response at about 160ms latency

P3a

Frontal positive response at about 300ms latency

In recent years, increasing attention has been paid to the study of event-related potentials (ERPs) elicited when focused attention is kept away from the stimulus (unnoticed stimulation). In particular, for acoustic stimulation, earlier studies have clarified that two main types of ERPs can be elicited by unnoticed, deviant stimuli interspersed between regular and frequent stimuli, depending on their physiological characteristics.1 Deviant stimuli whose physiological features are different from regular ones, but which are not intrusive enough to cause an overt attention shift, usually elicit a frontal negative response in the 120 to 180ms latency range, labelled mismatch negativity (MMN).1,2 By contrast, intrusive deviant stimuli (i.e. stimuli that are able to catch the attention) elicit a frontal positive response at about 300ms latency (labelled P3a3), preceded by a negative response in the 120 to 180ms latency range (labelled N2b4). MMN and P3a often coexist in the same recording; however, MMN can also be elicited in the absence of a P3a response and when the stimulus does not catch the attention of the participant. In particular, earlier reports strongly indicate that MMN can also be elicited in comatose patients5 or when the stimulus is fully unnoticed,6 although focused attention can partially modify the final response.7 Assuming that the preattentive comparison between the memory track of the regular stimulus and the physiological features of the incoming one plays a crucial role in MMN elicitation, its recording can, in theory, provide useful information about a person’s ability to automatically discriminate unnoticed, slightly different, sensory stimuli.

So far, MMN responses have been studied more in auditory than in somatosensory modalities. In fact, because of the physiological characteristics of any somatosensory stimulus, it is difficult to conceive that a deviant somatosensory stimulation could not catch the overt attention of the participant. Nevertheless, somatosensory MMN-like responses have been reliably obtained in adults.8–13 In particular, it has been demonstrated recently that somatosensory MMN-like responses are often abnormal in patients with unilateral cerebellar lesions,13 suggesting that the discrimination of slightly different somatosensory stimuli is impaired in these patients. This finding seems particularly relevant given that, when patients with cerebellar lesions focused their attention on the stimulus during a clinical sensory test, they did not show significant somatosensory deficits. Therefore, MMN recording should be particularly promising in the assessment of disturbances characterized by a reduced somatosensory discrimination. Looking at potential clinical uses of somatosensory MMN in the paediatric neurological field, somatosensory discrimination impairment has been claimed to play a role in some developmental disorders, such as developmental coordination disorder,14 autism,15 and dyslexia.16

Because no data are yet available about somatosensory mismatch-related responses in children, we aimed to verify whether reliable mismatch-related responses could be obtained in a population of healthy children, whether the characteristics of the MMN-related responses in children were similar to those reported in adults, and whether their recording in children required specific technical contrivances. We therefore recorded scalp responses in children following electrical stimulation of the fifth right finger interspersed among frequent electrical stimulation of the right thumb (‘oddball’ protocol). Predictably, the technical procedure that we previously used in adults13 could not be automatically applied to young participants, because of their high distractibility by the deviant stimulus, as well as by electrical stimulation in general; the ultimate procedure was therefore adjusted after a number of preliminary experiments, described below. We also recorded responses elicited by the deviant stimulus alone, omitting the frequent stimulus (‘standard-omitted’ protocol). A similar procedure, already described in earlier reports8,9,13 is specifically addressed to ensure that differences between ERPs obtained after frequent and deviant stimulation during the oddball protocol are due to mechanisms related to the mismatch detection, rather than to the lower deviant stimulation rate, because it is well known that the stimulation rate is critical in affecting ERPs.17

Method

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. References

Preliminary experiments

Initially, we applied the same protocol as in our previous study in adults.13 In the first experiments, a clear P3a response was invariably found over frontal regions, suggesting that the deviant electrical stimulation could have caught the participants’ attention (Fig. 1). Looking at the potential clinical uses of this technique, we considered that any attention shift towards the deviant stimulus should be avoided. Therefore, we modified both stimulus intensity and primary task, to keep the participants’ overt attention away from the deviant stimulus. Stimulus intensity was kept just above the sensory threshold; this allowed poorly defined and low-voltage traces. By contrast, the increase in the attentional load of the primary task (initially consisting of book reading, and then substituted by an engaging video game) was paralleled by the clearest definition of ERPs, with no contamination from P3a responses.

image

Figure 1.  Participant 1, oddball protocol, subtraction traces (deviant – frequent) in two different conditions: less demanding primary task (reading a book, red traces) and demanding primary task (video game, black traces). During the less demanding primary task, a positive response, designated P3a (black arrowheads) is evident over frontal regions, suggesting a possible distraction toward the deviant stimulation. The same traces also show two extranegativities preceding the P3a, peaking at about 250ms on frontal leads (red arrowhead) and at about 200ms on parietal leads (grey arrowheads), roughly similar to those obtained during the very demanding primary task, but different in amplitude, latency, and distribution; mechanisms other than the preattentive mismatch detection may be contributing.

Download figure to PowerPoint

Final technical procedure

After these preliminary experiments, we decided to engage the participants’ overt attention by a demanding video game and to keep the stimulation intensity at more than twice the sensory threshold (mean 4mAmp, range 3.5–4.5mAmp), because such an intensity never produced a clear P3a response indicating an overt distraction toward the deviant stimulus.

We therefore managed the final experiments as follows: electrical stimulation was delivered via ring electrodes placed on the finger (stimulating electrode proximal, placed above the proximal phalanx; anode placed above the distal phalanx). The frequent and deviant stimuli were delivered to the first and fifth fingers of the right hand respectively (80% frequent stimuli; 20% deviant stimuli). Two successive runs of 500 stimuli were delivered with an interstimulus interval of 1000ms. Traces from each run were superimposed to check reproducibility and were then averaged. An electro-oculogram was recorded using an electrode placed on the lateral cantus of the right eye. To limit ocular artefacts as much as possible, the screen (24cm width) was kept 75cm from the participants’ eyes. Recording electrodes were placed on 15 regularly spaced scalp locations (10–20 system): F7, F3, Fz, F4, F8, T3, C3, Cz, C4, T4, T5, P3, Pz, P4, and T6. A reference electrode was placed on the nose, but before the analysis traces were re-referenced offline using an average reference including all 15 scalp electrodes. Electrode impedance was kept below 3000Ω. The signal was filtered with a bandpass of 1 to 60Hz; trials exceeding 40μV were automatically edited out from the averaging. Signals were further digitally filtered offline with a bandpass of 2 to 35Hz. Baseline stabilization was ensured by subtracting the mean voltage during the 50ms prestimulus period from the signal.

The above procedure was repeated in all participants with the frequent stimulation to the right thumb omitted (standard-omitted protocol). The characteristics of the deviant stimulus (intensity, stimulus rate, stimulated finger) were kept unchanged. To avoid any habituation effect, the two procedures were separated by about 10 minutes of relaxation, during which the child freely read a book.

According to the final procedure we studied, 11 participants; their age and sex, as well as their participation in the preliminary phases of the experiment, are described in Table I. Participants were recruited from the families of the technical staff of our hospital (UDGEE La Nostra Famiglia, Polo Friuli-Venezia Giulia, Udine, Italy). The procedure was approved by the local ethics committee IRCCS La Nostra Famiglia, Sede Centrale ethics committee. Caregivers of the participants gave their informed consent to the research and to the publication of the results.

Table I.   Participants’ age, sex, and involvement in preliminary and final study procedures
ParticipantAge (y)Sex1st test2nd testFinal procedureStandard omitted
  1. Mean age: 8y 2mo (SD 1y 7mo). 1st test: less demanding primary task (reading a novel); 2nd test: less demanding primary task (reading a novel) + low-stimulation intensity (sensory threshold); final procedure: demanding primary task (video game) + high stimulation intensity (twice sensory threshold). +, participant took part in designated test; −, participant did not take part in designated test.

 111Female++++
 2 9Male++++
 3 7Male++++
 410Male++++
 510Female+++
 6 7Female+++
 7 7Male++
 8 7Male++
 9 8Male++
10 8Male++
11 6Female++

Analysis of traces

Components analyzed

The statistical analysis was preceded by a visual inspection of traces, which revealed some differences between conditions, as follows.

During the standard-omitted protocol (Fig. 2), the following components were recognizable in the 250ms after stimulus onset: P40, N60, P100, and a further negative component widely distributed over parieto-central regions, which we labelled N140.18–20

image

Figure 2.  Grand-average of all 11 participants. Black lines show traces issued during standard-omitted stimulation. Dashed lines show subtraction traces (deviant – frequent stimulation) issued during oddball stimulation. The left frontal negative response at about 220ms (black arrowheads) is evident only on oddball subtraction traces, but it is virtually absent on standard-omitted traces. Both oddball subtraction and standard-omitted traces show a negative response on parietal regions (grey arrowheads).

Download figure to PowerPoint

During the oddball protocol, on the deviant traces (Fig. 3), after the P40, N60, and P100 components, two major components (whose polarity depended on scalp location) were respectively identifiable in the 120 to 180ms and 180 to 250ms latency ranges. The 120 to 180ms latency was recognizable as a negativity over left centro-parietal regions, whereas the 180 to 250ms latency was identifiable as a negativity over left frontal regions.

image

Figure 3.  Oddball protocol, grand-average of all 11 participants. Black lines show subtraction traces (deviant – frequent stimulation). Dashed lines show deviant stimulation traces. Dotted lines show frequent stimulation traces. On F4, C4, and P4 leads, a negativity at about 100ms latency is evident, which should represent the negative counterpart of the contralateral P100 response on C3 and P3 locations (asterisks). A large negative response at about 220ms (black arrowheads) is evident on left and mid-frontal regions after deviant, but not after frequent stimulation. A negative response on left centro-parietal regions (grey arrowheads), although present in both deviant and frequent stimulation traces, is larger after deviant stimulation, reaching its maximum on P3.

Download figure to PowerPoint

During the oddball protocol, the same ERPs were elicited on the frequent stimulation as on the deviant stimulation (see Fig. 3), but ERPs following the P100 wave were generally poorly represented and of low amplitude.

In summary, visual inspection suggested that deviant stimulation during the oddball protocol elicited two successive extra negativities in the 120 to 180ms and 180 to 250ms latency ranges, whereas the standard-omitted procedure elicited only a parieto-central response in the 120 to 180ms latency range. Therefore, the statistical analysis was expressly designed to verify whether there were significant differences in these two latency windows between (1) standard and deviant stimulation in the oddball protocol and (2) deviant stimulation in the oddball and standard-omitted protocols. Because the analysis of components occurring in the first 120ms latency was substantially outside the main aims of the study, the statistical analysis of early responses was not performed. To compare differences between conditions and between scalp recording sites, statistical analysis of ERP amplitudes was performed by two-way analysis of variance (ANOVA). The threshold for statistical significance was p<0.05.

Amplitude measurements

Before measuring amplitudes, peaks were automatically identified for each recording site as the point of maximal amplitude in the 120 to 180ms and 180 to 250ms latency ranges. To avoid single-point amplitude measurement, for each identified peak we measured the mean amplitude in a window ranging from 10ms before to 10ms after the identified latency peak.

Statistical analysis

For amplitude analysis, we compared results obtained after frequent stimulation with those obtained after deviant stimulation during the oddball protocol. As stated above, measures were performed for two successive latency windows, i.e. 120 to 180ms and 180 to 250ms, because two main components had been visually identified in these latency ranges. Amplitudes were evaluated using two-way ANOVAs [condition (deviant − frequent] × electrode location). When significance was reached, post-hoc analysis was performed using Scheffe’s tests. We preferred to use such a conservative test,21 rather than pairwise comparisons followed by Bonferroni’s correction, because the latter approach can, in theory, inflate type II statistical errors.22 We then compared results from deviant stimulation during the oddball protocol with those obtained after deviant stimulation during the standard-omitted protocol. Measures were performed for the two successive latency windows, 120 to 180ms and 180 to 250ms. Amplitudes were evaluated using two-way ANOVAs [condition (deviant during oddball protocol − deviant during standard-omitted protocol) × electrode location]. When significance was reached, post-hoc analysis was performed using Scheffe’s tests.

Results

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. References

Oddball protocol

After the visual inspection of grand-average traces (see Fig. 3), two main components were evident: a negative component peaking at about 160ms on left parieto-central regions (contralateral to the stimulated side) and a further negative component peaking at about 220ms on left frontal regions (contralateral to the stimulated side). Such findings were confirmed by distribution maps (Fig. 4), also showing that the former parietal component had a positive counterpart on right frontal regions, whereas the latter frontal component showed a positive counterpart on parieto-temporal right regions. The statistical evaluation of amplitudes in the first latency window showed significant differences between conditions (frequent vs deviant stimulation, two-way ANOVA: F(1,13)=8.2, p=0.004), confirming that ERPs after deviant stimulation showed higher amplitudes than those obtained after frequent stimulation. A significant difference was also found between electrode locations (frequent vs deviant stimulation, two-way ANOVA: F(1,13)=6.61, p<0.001). Post-hoc analysis showed a significant difference between P3 and all right fronto-central recordings (Scheffe’s test p<0.05). By contrast, no significant interaction effects were found (frequent vs deviant stimulation, two-way ANOVA: F(1,13)=1.45, p=0.13).

image

Figure 4.  Grand-average of all 11 participants. Top: spline distribution maps issued from difference traces in the oddball protocol. Middle: spline distribution maps issued from the standard-omitted protocol; maps (negativity in blue, positivity in red) are shown at three different latencies. Bottom: traces issued from the standard-omitted protocol (black lines) and from the oddball protocol (deviant – frequent difference traces, dashed lines), corresponding in (A), (B), and (C) to the distribution maps. At 155ms and 165ms (A and B respectively), maps show a clear negative response on left parietal regions, spreading to right parietal regions during the standard-omitted protocol. At about 230ms (C), the map shows a clear negative response on left frontal regions during the oddball protocol, whereas no negative responses are obtained during the standard-omitted protocol.

Download figure to PowerPoint

The statistical evaluation of amplitudes in the second latency window showed a significant difference between conditions (frequent vs deviant stimulation, two-way ANOVA: F(1,13)=17.5, p<0.001), confirming that higher responses were obtained after deviant stimulation, but not between electrode locations (two-way ANOVA: F(1,13)=1.32, p=0.19). Significant interaction effects were found (frequent vs deviant stimulation, two-way ANOVA: F(1,13)=1.87, p=0.03), suggesting a different distribution of scalp responses between frequent and deviant stimulation. Post-hoc analysis showed a significant difference between F3 and all right parieto-temporal recordings (Scheffe’s test: p<0.05). The F3 prevalence was not just a group statistical effect, because a negative left frontal response was elicited by deviant stimulation in all participants.

Standard-omitted protocol

The visual inspection of grand-average traces allowed us to recognize an evident response at about 160ms latency (see Fig. 2), widely distributed over parieto-central regions, with a positive counterpart on frontal regions (see Fig. 4). To simplify the exposition of data, we decided to label this wave N140, although its latency and distribution slightly differ from those previously described as N140.18–20

By comparing the 120 to 180ms parietal component of deviant traces in the oddball protocol with the N140 wave issued from the standard-omitted protocol, statistical analysis of amplitudes showed a difference among electrodes, but did not reveal any significant interaction effect, which could have suggested a different scalp distribution of these two waves (two-way ANOVA, condition × electrodes: condition, F(1,13)=3.46, p=0.06; electrodes, F(1,13)=8.43, p<0.001; interaction, F(1,13)=0.54, p=0.89).

We compared ERP amplitudes in the second latency window (180–250ms) after deviant stimulation during the oddball and standard-omitted protocols. Predictably, given the lack of clear frontal responses in the standard-omitted protocol (see Fig. 4), statistical analysis revealed a significant interaction effect and a significant difference between electrodes (two-way ANOVA, condition × electrodes: condition, F(1,13)=1.14, p=0.28; electrodes, F(1,13)=2.95, p<0.004; interaction, F(1,13)=2.49, p=0.003). As stated above, a negative left frontal response was invariably elicited by deviant stimulation during the oddball protocol in all participants.

Discussion

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. References

The results of this study confirm that mismatch-related responses can be reliably obtained in children. In a mismatch-detection condition (oddball protocol), traces issued from deviant stimulation clearly show a significant difference from those issued from frequent stimulation. Such a difference was mainly due to the constant presence of two successive components in deviant traces, the former peaking at about 160ms and the latter at about 220ms. The first component is recorded as a negativity in parieto-central regions contralateral to the stimulus, whereas the second one is recorded as a negativity in frontal regions contralateral to the stimulus. Moreover, traces issued from the standard-omitted protocol did not show any reliable left frontal component, thus confirming that this wave is unequivocally linked to a mismatch-detection mechanism. By contrast, despite some slight differences in scalp distribution, statistical analysis was not able to demonstrate any significant difference between the central negativity in the oddball protocol and the parietal negativity in the standard-omitted protocol. This could, however, be due to the small sample and to the low number of recording electrodes, which make it unlikely that statistical significance could be reached by means of the classical interaction approach. A definitive answer to the question whether the centro-parietal components in the oddball and standard-omitted protocols are generated by different mechanisms might be provided by a clear-cut dissociation in pathological conditions (see, for instance, cerebellar patients in Restuccia et al.13), but this is obviously impossible in our sample of healthy children.

Comparison with earlier data

As far as auditory MMN is concerned, Näätänen and Michie2 hypothesized that MMN-generating mechanisms require two successive steps: mismatch detection in the temporal lobes (sensory-specific for the auditory modality) and covert attention switching toward the unnoticed stimulus, involving frontal areas. Further studies largely confirmed that auditory MMN is built from temporal as well as frontal components.23,24 Thus, our present data seem to fit well with this general mechanism, as they demonstrate two main components of the somatosensory MMN, the former located in centro-parietal regions (sensory-specific for the somatosensory modality), and the latter in frontal areas contralateral to the stimulus.

Surprisingly, earlier studies in adults did not always show the coexistence of these two components. Frontal responses have been described by Kekoni et al.8 and by Kida et al.10 Shinozaki et al.9 also found a frontal extranegativity when the interstimulus interval was 1000ms as in our present study. Akatsuka et al.12 during an automatic 2-point discrimination protocol, found that parietal (SII) areas were mainly involved in building a mismatch-related somatosensory response, but, in a successive functional magnetic resonance imaging study, they found that not only parietal but also frontal regions are activated during a 2-point discrimination task.25 In our earlier study in adults13 we were not able to find frontal components, whereas mismatch responses to somatosensory stimulation were mainly localized on parieto-occipital scalp regions.

The simplest explanation for this discrepancy is related to the recording technique. In the earlier study, we used a reference electrode located over the nose; this recording technique can allow a reduction in amplitude of frontal responses, which are picked up not only from active leads, but also from the nose reference electrode, leading to an obvious decrease of the voltage difference between active and reference leads. However, another possible explanation could be related to the different characteristics of the primary task. Although the precise mechanism of this phenomenon has not been fully elucidated, many earlier findings suggest that frontal generators are mainly active when attentional resources are allocated toward a very demanding primary task,22 or when the deviance detection is difficult: for example, recent neuroimaging studies found that the magnetic resonance signal change in the right frontal cortex was larger for smaller deviants.26,27 It is, therefore, conceivable that frontal components are enhanced when stimuli are difficult to discriminate (e.g. vibratory stimuli8) or when attentional resources are allocated toward a very demanding primary task, as in the present study. However, whatever the reason the frontal response has been so variably recorded in previous adult studies, our present data demonstrate that a frontal response can be reliably recorded in children, and that this ERP is unequivocally related to mismatch-detection mechanisms, because it is fully lacking in control (standard-omitted) protocols.

Practical implications

From a practical point of view, our data demonstrate that somatosensory MMN responses can be reliably recorded in children, providing that appropriate contrivances are used. The use of a very demanding primary task seems to be mandatory, for two reasons. First, traces recorded during a less demanding primary task were characterized by a clear P3a component. In theory, this might suggest an overt attention shift toward the deviant stimulation and, in general, toward the stimulated hand; looking at some possible clinical uses of the somatosensory MMN, in particular the assessment of disturbances of the automatic somatosensory discrimination, this phenomenon should be avoided. Second, the use of a demanding primary task seems to allow an enhancement of the frontal MMN component, which is unequivocally related to a mismatch-detection mechanism.

References

  1. Top of page
  2. Abstract
  3. Method
  4. Results
  5. Discussion
  6. References