Anticipatory attention: an event-related desynchronization approach

https://doi.org/10.1016/S0167-8760(01)00181-7Get rights and content

Abstract

This paper addresses the question of whether anticipatory attention — i.e. attention directed towards an upcoming stimulus in order to facilitate its processing — is realized at the neurophysiological level by a pre-stimulus desynchronization of the sensory cortex corresponding to the modality of the anticipated stimulus, reflecting the opening of a thalamocortical gate in the relevant sensory modality. It is argued that a technique called Event-Related Desynchronization (ERD) of rhythmic 10-Hz activity is well suited to study the thalamocortical processes that are thought to mediate anticipatory attention. In a series of experiments, ERD was computed on EEG and MEG data, recorded while subjects performed a time estimation task and were informed about the quality of their time estimation by stimuli providing Knowledge of Results (KR). The modality of the KR stimuli (auditory, visual, or somatosensory) was manipulated both within and between experiments. The results indicate to varying degrees that preceding the presentation of the KR stimuli, ERD is present over the sensory cortex, which corresponds to the modality of the KR stimulus. The general pattern of results supports the notion that a thalamocortical gating mechanism forms the neurophysiological basis of anticipatory attention. Furthermore, the results support the notion that Event-Related Potential (ERP) and ERD measures reflect fundamentally different neurophysiological processes.

Introduction

Most of the events that we experience everyday do not happen unexpectedly. This enables us to anticipate events that can be expected to happen in the near future. Anticipatory behavior plays a ubiquitous role in everyday life. As Requin et al. (1991, p. 360) state it, ‘at any moment a large part of the present activity of an organism is devoted to preparing for subsequent behavior’. Anticipatory behavior has two components, a motor component that ameliorates the processes at the output side of the organism, and a perceptual component that enhances the processes at the input side of the organism. The latter process, which can be termed anticipatory attention, is the focus of the present paper.

The neurophysiological model for anticipatory behavior described by Brunia (this issue), the so-called thalamocortical gating model, predicts that anticipatory attention manifests itself as a cortical activation preceding the presentation of stimuli, which is restricted to, or at least maximal in, the cortical area corresponding to the modality of the stimulus. This cortical activation would stem from an enhanced thalamocortical transfer in the relevant modality, and would serve the function of pre-setting the necessary physiological processes in the sensory cortex, in order to achieve a faster and/or more efficient processing of the upcoming sensory input.

Two EEG/MEG measures might be appropriate for revealing such cortical activation: negative slow potentials (or slow fields in case of MEG); and Event-Related Desynchronization. Negative slow potentials stem from a predominance of excitatory post-synaptic potentials at the apical dendrites of cortical pyramidal neurons, leading to a subthreshold depolarization of these cells. The concurrent increase in the readiness of cortical pyramidal neurons to fire might form the neurophysiological substrate of anticipatory attention. On the other hand, desynchronization of 10-Hz rhythms in the EEG/MEG over the human neocortex most probably reflects a thalamically driven disruption of synchronized activity in functionally related groups of cortical neurons (cf. Lopes da Silva, this issue; Steriade et al., 1990), which can be seen as a transition of cortical idling to cortical activity. Thus, desynchronization of 10-Hz rhythms in the EEG/MEG in sensory cortex preceding the presentation of a stimulus in the corresponding modality may be an expression of the thalamocortical gating mechanism that is proposed by Brunia and van Boxtel (this issue) to mediate anticipatory attention.

In the last two decades, quite some attention has been devoted at identifying slow negative potential shifts related to the anticipation of sensory input (cf. Böcker et al., this issue; Böcker and Van Boxtel, 1997). To this end, a paradigm has been developed in which subjects are asked to perform a time estimation task, and are subsequently informed about the quality of their time estimation by a stimulus providing Knowledge of Results (KR; Damen and Brunia, 1987). Preceding the KR stimulus, a negative slow potential can be recorded that has been termed the stimulus-preceding negativity (SPN). Note that any measure of brain activity that is a candidate for being a reflection of a purely perceptual anticipatory process as described in the thalamocortical gating model, would have to fulfil two criteria: it should be (1) independent of the type of information conveyed by the anticipated stimulus (that is, independent of whether the stimulus provides, for example, instructions on a subsequent task, or Knowledge of Results about a previously executed task); and (2) maximal over the sensory cortex corresponding to the modality of the stimulus. Subsequent research has been aimed at clarifying whether the SPN fulfils these criteria. Since detailed reviews have been presented elsewhere (e.g. Böcker et al., this issue; Böcker and Van Boxtel, 1997), we describe only the main results of these studies here. As far as the first criterion is concerned, the experimental evidence shows that the amplitude and lateralization of the SPN vary with the type of stimulus that is anticipated (cf. Böcker and Van Boxtel, 1997). With respect to modality differences in SPN topography, up to now only one study has addressed the question (Böcker et al., 1994). In this study, there was a striking similarity in scalp topography of the SPN for the two modalities: both had a frontotemporal maximum. Thus, the Böcker et al. study does not support the hypothesis that the SPN is maximal, or has a clearly distinguishable component, over the sensory cortex corresponding to the modality of the anticipated stimulus. However, recently Brunia and van Boxtel (unpublished data) demonstrated a small but significant difference in SPN scalp topography preceding auditory and visual KR stimuli, that went in the expected direction. Thus, although a small part of the SPN may be specific to the modality of the stimulus, the results do not support the hypothesis that the SPN is maximal over the sensory cortex corresponding to the modality of the anticipated stimulus.

It may be argued that analyzing the induced reactivity [e.g. by means of computing the Event-Related Desynchronization (ERD) and Event-Related Synchronization (ERS); cf. Pfurtscheller and Lopes da Silva, 1999a] of 10-Hz rhythms could possibly be a more fruitful approach in studying whether thalamocortical gating mechanisms are at the basis of anticipatory attention rather than studying slow potentials. Considering the processes that are thought to underlie the genesis of 10-Hz rhythms (cf. Lopes da Silva, this issue; Steriade et al., 1990), it seems reasonable to assume that ERD/ERS in the 10-Hz range can provide us with a window on the processes that operate in thalamocortical circuits. More specifically, ERD has been associated with an enhanced thalamocortical transfer (i.e. the opening of a sensory ‘gate’ to the cortex), while ERS has been associated with a disruption of such transfer (Lopes da Silva, 1991; see also Guillery et al., 1998). Therefore, it may be argued that if results obtained with ERD/ERS analyses fulfil the above mentioned two criteria for being a reflection of a purely perceptual anticipatory process, one would have a valuable empirical argument in favor of the thalamocortical gating model for anticipatory attention. Although mainly the SPN does not fulfil these criteria, there are good reasons to expect differences between ERP and ERD analyses in general.

One can distinguish between two types of event-related EEG/MEG activity: induced activity and evoked activity (e.g. Bullock, 1992, Pfurtscheller and Lopes da Silva, 1999b). This distinction is most easily seen when one considers rhythmic activity. An EEG rhythm (say, the occipital alpha rhythm) may be modulated by an event (say, opening of the eyes, which leads to an amplitude enhancement), while this event does not directly drive the amplitude enhancement (the eyes, or for that matter the eyelid movements, do not produce the electrical activity which is measured over occipital leads as the alpha rhythm). Thus, induced rhythms may be defined as ‘oscillations caused or modulated by stimuli or state changes that do not directly drive successive cycles of the rhythm’ (Bullock, 1992, p. 1). An instance of an evoked rhythm would be the steady-state response (SSR) evoked by 40-Hz stimulation (e.g. Rockstroh et al., 1996), in which each new stimulus directly drives each successive cycle of the SSR rhythm.

While evoked activity is by definition both time- and phase-locked to an event, induced activity is time-locked, but not necessarily phase-locked. For the present purpose, the important implication is that ERD/ERS analyses contain information (non-phase-locked, or induced EEG or MEG activity) that is lost when one computes a signal average as is done in the computation of ERPs, which leaves only the phase-locked (or evoked) part of the signal in the resulting average. From a signal analytic point of view, it is very well possible that relevant information is contained in the non-phase-locked part of the EEG/MEG. This means that a possible modality specificity of anticipatory cortical activation could have been overlooked because the analyses carried out so far on negative slow potentials have focused only on the phase-locked part of the EEG/MEG signal. ERD, with its focus on non-phase-locked activity, can be used to investigate this possibility.

The existence of modality-specific 10-Hz rhythms, alpha, mu and tau (see Bastiaansen et al., 1999a for a discussion of these rhythms), that react differentially to stimulation in one of the three main sensory modalities (visual, somatomotor and auditory, respectively) is promising, since their very existence opens the possibility that they independently may show a local desynchronization when a stimulus in the corresponding sensory modality is anticipated.

In sum, there are neurophysiological, signal-analytic and empirical arguments for the hypothesis that ERD might be a useful tool to investigate whether anticipatory attention for an upcoming stimulus, in one of the three main sensory modalities, is mediated by an enhanced thalamocortical transfer of sensory information to the cortical areas corresponding to the modality of the anticipated stimulus. Remarkably, however, although two studies have incidentally addressed the ERD related to anticipatory attention (Klimesch et al., 1992, Klimesch et al., 1998), systematic research into the relationship between ERD and anticipatory attention, in which stimulus modality is carefully manipulated, has not been performed. The present paper describes a series of experiments that have been performed recently in order to fill this gap.

Section snippets

Experimental paradigm

In all the experiments described in the present paper, anticipatory attention was induced in subjects by means of a time estimation task. In this task, subjects are asked to press a button 4 s after the presentation of an auditory warning signal. Two seconds after the button press they are informed about the quality of their time estimation by a stimulus providing Knowledge of Results (KR) in the following way: The KR stimulus (a minus sign, a horizontal bar or a plus sign) indicates whether

Methods

Experiment 1 (described in detail in Bastiaansen et al., 1999b) consisted of three conditions: voluntary movement; time estimation followed by an auditory KR; and time estimation followed by a visual KR. Ten subjects participated in the experiment. In the voluntary movement task, subjects were instructed to squeeze a response manipulandum at a slow, self-paced rate (6–10 times per min). The time estimation task has been described above. Each condition was repeated twice, once with a right-hand

Methods

In experiment 2 (described in detail in Bastiaansen et al., 2001a), five subjects performed the time estimation task in two separate sessions, which were at least 1 week apart. Auditory and visual KR stimuli were used in different conditions. In the first session, EEG was recorded from 29 electrodes; in the second session, MEG was recorded from 64 axial gradiometers. For the MEG recordings, we used a whole-cortex MEG system (CTF Systems Inc., Vancouver, Canada), in which the sensors are

Methods

In this experiment (described in detail in Bastiaansen et al., 2001b) 12 right-handed subjects, one man and 11 women, aged 18–29 (M=21, S.D.=3) performed a time estimation task while the EEG was measured from 27 electrodes. ERD was computed in the 8–10 Hz and the 10–12 Hz frequency bands. There were five conditions in the experiment: A voluntary movement (VM) task preceded the time estimation tasks, which were: KR auditory intact; KR auditory degraded; KR visual intact; and KR visual degraded.

Methods

In this experiment (described in detail in Bastiaansen, 2000), 13 healthy, right-handed subjects performed a time estimation task, in which visual and somatosensory KR stimuli were used. A voluntary movement task was included as a control condition. The data of two subjects were discarded from further analysis because of the presence of excessive eye movement artifacts, and the data of an additional subject were discarded from further analysis because of the absence of clear 10-Hz peaks in the

General discussion

Taken together, the results of the different experiments show one robust effect, that has been found in all the subjects of all the experiments: this effect is the occipital ERD preceding visual KR stimuli. Therefore, it can be concluded that anticipatory attention for a visual stimulus is accompanied by a pre-stimulus desynchronization of the visual cortex, which we interpret as a reflection of enhanced thalamocortical transfer of sensory information from the thalamus to the visual cortex. In

Acknowledgements

The research described in this paper was funded by the Cooperation Centre Tilburg and Eindhoven Universities (SOBU), grant 1995 AD. At the time of manuscript preparation, the author was funded by grant 400-56-384 of the Dutch Organization for Scientific Research (NWO). Thanks to Gert Pfurtscheller, Christa Neuper and Koen Böcker for valuable comments on an earlier version of the manuscript.

References (35)

  • G. Pfurtscheller et al.

    Event-related EEG–MEG synchronization and desynchronization: basic principles

    Clin. Neurophysiol.

    (1999)
  • M. Steriade et al.

    Basic mechanisms of cerebral rhythmic activities

    Electroencephalogr. Clin. Neurophysiol.

    (1990)
  • J. Tiihonen et al.

    Magnetoencephalographic 10 Hz rhythm from the human auditory cortex

    Neurosci. Lett.

    (1991)
  • Bastiaansen, M.C.M., 2000. Anticipatory attention: an Event-Related Desynchronization approach. Ph.D thesis, Tilburg...
  • M.C.M. Bastiaansen et al.

    ERD as an index of anticipatory behavior

  • M.C.M. Bastiaansen et al.

    Event-Related Desynchronization related to the anticipation of a stimulus providing Knowledge of Results

    Clin. Neurophysiol.

    (1999)
  • Bastiaansen, M.C.M., Böcker, K.B.E., Brunia, C.H.M, 2001b. ERD as an index of anticipatory attention? Effects of...
  • Cited by (0)

    View full text