Elsevier

Biological Psychology

Volume 68, Issue 3, March 2005, Pages 309-329
Biological Psychology

Bursts of occipital theta and alpha amplitude preceding alternation and repetition trials in a task-switching experiment

https://doi.org/10.1016/j.biopsycho.2004.06.004Get rights and content

Abstract

The instantaneous amplitude of the theta and alpha bands of the electroencephalogram (EEG) was studied during preparation periods in a task-switching experiment. Subjects had to switch between tasks in which they were to respond to either the visual or the auditory component of the stimulus. 11–13 Hz occipital amplitude increased prior to auditory, relative to visual repetition trials. The effect was transient, ending well before presentation of the stimulus that was being prepared for. Alternation trials were preceded by an increase in occipital theta-band activity, relative to repetition trials, for the visual task. This effect was also transient. The effects suggest tentative hypotheses for the function of transient bursts of alpha- and theta-band oscillations and indicate the possibility of a psychophysiological resolution of theoretical questions concerning the origin of switch costs.

Introduction

Before a task can be performed, the brain must somehow implement the correct task set: the set of all stimulus–response mappings necessary to perform that task (Monsell, 2003). For any task involving stimuli, which are not always associated with the same responses, its task set must have a transient, context-sensitive implementation. However, while the brain has the ability to switch tasks, the replacement of one task set by another is associated with switch costs (Jersild, 1927): increased average response times on trials following a task switch, relative to trials requiring the same task as the previous trial. Subsequent experiments (e.g. Allport et al., 1994) have shown that switch costs diminish as the time available to replace task sets increases, but never completely disappear. The persistent difference between switch and repetition trials is called the residual switch cost.

According to the task-set inertia (TSI) hypothesis (Allport et al., 1994), switch costs are due to a conflict between task sets that share common features. While the influence of a previous task set decays over time, a full stimulus–response cycle seems to be necessary and sufficient for its complete replacement (Rogers and Monsell, 1995, Gilbert and Shallice, 2002). Rogers and Monsell (1995) propose that task-set reconfiguration (TSR), the required binding of stimuli to responses, does partially occur prior to stimulus presentation. The benefit of increasing preparation time derives from the opportunity to complete more of TSR. Residual switch costs are explained by an inability to completely prepare for an upcoming task: the actual stimulus is necessary to complete TSR, although not to initiate it. The TSI and TSR hypotheses do not seem to be incompatible; residual switch costs might be multiply determined, by both previous task interference and an inability to completely prepare. Such a combined explanation is of course less parsimonious than either separate hypothesis, and this added complexity of a combination has yet to be justified (Gilbert and Shallice, 2002).

The failure-to-engage hypothesis (de Jong, 2000) is an explanation of residual switch costs that is based on reaction time distribution rather than average reaction time. The residual switch costs found when subjects are provided with the opportunity to prepare for a switch in task prior to stimulus onset have been found to have a distribution that consists of a mixture of reaction times drawn from two other distributions: those when subjects were already prepared and those when subjects had not had a chance to prepare (de Jong et al., 1999, de Jong, 2000, de Jong, 2001, Nieuwenhuis and Monsell, 2002). Residual switch costs may not, therefore, be due to an effect present on every trial when a switch must be performed, but to the proportions of prepared and unprepared trials. According to the failure-to-engage hypothesis, the subset of prepared trials is fully prepared, while the subset of unprepared trials follows from a failure to engage preparatory capabilities that would have been sufficient to switch tasks.

A binary mixture of reaction time distributions changes what must be explained about residual switch costs. Hypotheses must explain why trials are divided into a fast and a slow subset, instead of why a single average reaction time is slower. In the failure-to-engage hypothesis, a task set is either recalled or not, and the theoretical questions reduce to the way memory recall is implemented in general, and how the memory and recall of task sets are implemented in particular. How are task sets stored? How are they instantiated? How does the brain remember what to recall?

Theta-band (5–7 Hz) oscillations may be involved in one way to switch tasks, namely the reactivation of the memory of a task set. Theta oscillations are involved in a number of cognitive functions (Kahana et al., 2001). The phase of theta oscillations modulates long-term potentiation; hence, synaptic plasticity is partially determined by theta oscillations (Huerta and Lisman, 1993, Hölscher et al., 1997). The phase of theta oscillations in the hippocampus of rats has been shown to code for location (O’Keele and Recce, 1993, Skaggs et al., 1996); for a review of the relationship of the hippocampus to memory, see Nader (2003). The power of theta oscillations measured by the electroencephalogram (EEG) increases with working memory load (e.g. Gevins et al., 1997, Krause et al., 2000). As might be expected due to its role in synaptic plasticity, EEG theta power also increases when information is encoded to (Klimesch et al., 1994) or retrieved from (Klimesch et al., 2001) long-term memory. Hence, the theta band is of interest in this context because, first, preparation is in a sense the recall of the memory of what task to perform and, second, remembering what to recall is a situation involving working-with-memory (WWM; Moscovitch, 1995).

WWM refers to three operations of the prefrontal cortex on information. First, the organisation of information before it is encoded to memory; second, the strategic search for cues to activate associated information in memory; and third, the relation of information retrieved from memory to current goals. WWM has a dependent relation with automatic, associative memory. Before memory can be worked with, its content must have been determined by associative memory at a previous time—that is, the function of working with memory does not include memory itself. Further, the prefrontal organisation that could be said, incautiously, to work on memory must itself be a kind of memory, which can be activated by the right conditions (including, e.g., emotional information), to avoid introducing a homunculus. The neuronal implementation of the “working” of WWM will therefore always involve an interplay between organizational memories in prefrontal cortex and the activity in and relations between neuronal areas that represent a specific kind of information, e.g. occipital areas for visual information. In the case of task-switching, the memory being worked with is the task set: stimuli, responses and the relations between them.

WWM as defined above is a broad concept, able to cover various specific functions that could be characterized as kinds of WWM. For example, Miller and Cohen (2001) argue that the function of PFC is to manipulate the competition between activity in other cortical areas, and Mayr (2002) showed how flexible behaviour is achieved by inhibiting action rules, which could be seen as WWM described at smaller and larger scales, respectively. The PFC also contains functional divisions, as in for instance Hoshi and Tanji (2003), who present evidence, from an intracortical study using monkeys, for a ventral-to-dorsal gradient in the involvement of dorsolateral PFC cells in information retrieval to motor planning. WWM may provide a framework for explaining how these cells are involved in these functions. FMRI studies have also shown specificity within the PFC. For example, Sohn and Carlson (2000) separated endogenous and exogenous components of task switching by manipulating foreknowledge of the task to be performed. Endogenous preparation, where foreknowledge was used during the preparation interval, was related to higher activation in inferior lateral prefrontal cortex, as well as superior posterior parietal cortex. Exogenous adjustment, occurring after stimulus presentation on trials without foreknowledge, was associated with different areas, namely superior prefrontal cortex and posterior parietal cortex.

If WWM involves the manipulation of sensory memory distributed over the cortex, effects specific to the areas being worked with would be expected to be found. In a study by Kirk and Mackay (2003) linguistic and spatial task demands resulted in left and right hemisphere theta maxima, respectively, providing an indication that theta topography may be in part determined by the relevance of specific cortices. Task-region compatibility of effects on theta amplitude would provide some support for the suggestion that theta oscillations may occur as a cortical expression of hippocampal activity (e.g. Klimesch, 1996). If the hippocampus is involved in memory recall and memory exists as activity distributed over the cortex, then what the behaviour of the hippocampus, which involves theta oscillations, achieves, may be the induction of cortical behaviour related to those theta oscillations. For instance, activating the memory of a visual stimulus would then not only involve increased firing rate, but perhaps also a theta-band oscillation over visual cortex. The present study provides data that may be relevant for this hypothetical relationship, for visual information and the occipital cortex.

Another way to switch tasks is to leave a set of stimulus–response mappings intact, but to block a subset of either stimuli or responses so that the remaining paths form a task set. This could be considered as a kind of inhibition of action rules (Mayr, 2002); note that it is not the episodic traces of specific stimuli that are proposed here to be inhibited, but subsets of stimuli. In the present study, the subsets were separated by modality, being either visual or auditory. Alpha-band (8–12 Hz) oscillations may be involved in such a function, as they are related either to simply the absence of processing (Pfurtscheller et al., 1996, Berger, 1930) or to the active and selective inhibition of external sensory influences (Cooper et al., 2003). Bastiaansen et al. (2001) found anticipatory event-related desynchronization (i.e. a drop in amplitude) in the alpha-band when visual feedback was expected, which was explained in terms of a thalamo-cortical gating mechanism (Guillery et al., 1998, Brunia, 1993, Skinner and Yingling, 1977). Temporal desynchronization due to auditory preparation was found only using magnetoencephalogram (MEG) recordings; the EEG does not seem able to register such activity. A general hypothesis on the effect of alpha oscillations, that fits with either of the ideas mentioned above, is based on a consequence of large-scale synchrony, namely that enforced synchronous firing diminishes the parallel processing that a large number of neurons would otherwise be able to perform (Nunez, 2000). A simple cause of synchronous oscillations in the alpha band is rhythmic inhibition due to excitatory-to-inhibitory back to excitatory feedback combined with long-lasting (NMDA) post-synaptic dynamics (Wang, 1999). Such an interaction, due to the periods of uniform inhibition, also seems likely to cause a decrease in the information a neuronal population undergoing synchrony could transmit or transform.

These theories and results suggest hypotheses concerning effects in the theta and alpha frequency bands related to the implementation of task sets. This is not to say that other frequency bands, for instance the gamma band (Tallon-Baudry and Bertrand, 1999), might not be of interest, or even be intimately related to the same processes associated with the theta and alpha bands. The present study, however, is focused on the theta and alpha frequency bands because of their known relevance for memory and inhibition, respectively, and the possible importance of these processes in understanding task switching.

In the present study, subjects were confronted with ambiguous stimuli: a letter and a tone, presented simultaneously. Subjects had to perform one of two tasks using these stimuli, either responding to the letter or to the tone (see Section 2). The tasks were otherwise similar, so effectively preparing for tasks should be related to changing which modality will be responded to. The period of interest is the interval during which subjects could prepare to respond in the correct way to the upcoming stimulus. In this interval, effects related to the next kind of trial—visual versus auditory, alternation versus repetition—could be related to the recall of a task set.

The final focus of the present study is the occipital region. Although both the visual and auditory modality were relevant in the task, the occipital region was chosen due to the sensitivity of the EEG to activity in visual cortex, as opposed to auditory cortex (see above). This selection of frequency bands, period and location allows the following hypotheses to be formulated.

First, preparing to switch to the visual task is expected to involve increased occipital theta amplitude in the preparation interval. This expectation follows from the relation of theta activity to memory and the hypothesis that switching to a new task can be considered to be the recall of the memory of the task set. By letting subjects switch between a visual and auditory task, effects associated with the recall of a visual task might be found, and such effects can be contrasted with the recall of a non-visual task.

Second, effects on occipital alpha amplitude might be expected in this interval, in the case that switching in this task is achieved by inhibiting irrelevant cortices, as opposed to recalling memories involving relevant cortices. Such inhibition would be expected to cause increased occipital alpha amplitude—associated with increased inhibition—when preparing for the auditory task, in which situation information in the visual cortex is irrelevant.

Section snippets

Subjects

Subjects were eight paid volunteers (age 18–25 years), with normal or corrected-to-normal vision. Four subjects were male. All subjects were right-handed.

Task

A version of a Rogers and Monsell (1995) task-switching task, implemented in MEL (Psychology Software Tools, 1995), was used. In this task, visual stimuli were presented for 200 ms in one of four cells in a 2 × 2 matrix. Auditory stimuli, which were either a high (400 Hz) or a low (1200 Hz) tone, were presented simultaneously and for the same

Behavior

Fig. 1 shows the reaction time data. An important effect is the diminishing switch costs, which indicates that subjects were preparing for the upcoming task during preparation intervals.

Repeated measures MANOVA was performed on the reaction time data using the factors congruence, modality (visual versus auditory task), trial type (alternation versus repetition trial) and preparation interval.

Congruent stimuli were associated with faster reaction times than incongruent stimuli (F (1, 7) = 11.8, P

Discussion

Two experimental effects on instantaneous amplitude were found using the FDR procedure. First, an increase in occipital theta amplitude was found for alternation versus repetition trials, but only for the visual task. The theta-amplitude time course preceding alternation trials is initially similar for both modalities and no alternation by modality interaction was found over the whole period selected by the FDR procedure. However, the visual and auditory alternation time courses diverge around 1

Acknowledgements

The authors would like to thank Prof. Dr. R. Ridderinkhof and two anonymous reviewers for their comments and ideas for improving this paper.

References (49)

  • W. Klimesch

    Memory processes, brain oscillations and EEG synchronization

    International Journal of Psychophysiology

    (1996)
  • W. Klimesch

    EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis

    Brain Research Reviews

    (1999)
  • W. Klimesch et al.

    Theta synchronization during episodic retrieval: neural correlates of conscious awareness

    Cognitive Brain Research

    (2001)
  • C.M. Krause et al.

    The effects of memory load on event-related EEG desynchronization and synchronization

    Clinical Neurophysiology

    (2000)
  • S. Monsell

    Task switching

    Trends in Cognitive Sciences

    (2003)
  • K. Nader

    Memory traces unbound

    Trends in Neurosciences

    (2003)
  • G. Pfurtscheller et al.

    Event-related synchonization (ERS) in the alpha band—an electrophysiological correlate of cortical idling: a review

    International Journal of Psychophysiology

    (1996)
  • W. Singer

    Time as coding space?

    Current Opinion in Neurobiology

    (1999)
  • C. Tallon-Baudry et al.

    Oscillatory gamma activity in humans and its role in object representation

    Trends in Cognitive Sciences

    (1999)
  • T. Tsujimoto et al.

    Prefrontal theta oscillations associated with hand movements triggered by warning and imperative stimuli in the monkey

    Neuroscience Letters

    (2003)
  • A.A. Wijers et al.

    Visual search and spatial attention: ERPs in focused and divided attention conditions

    Biological Psychology

    (1987)
  • D.A. Allport et al.

    Shifting intentional set: exploring the dynamic control of tasks

    Attention and Performance XV: Conscious and Nonconscious Information Processing

    (1994)
  • Y. Benjamini et al.

    Controlling the false discovery rate: a practical and powerful approach to multiple testing

    Journal of the Royal Statistical Society B

    (1995)
  • H. Berger

    On the electroencephalogram of man

    Journal fur Psychologie und Neurologie

    (1930)
  • Cited by (34)

    • Detrimental effects of a high-dose alcohol intoxication on sequential cognitive flexibility are attenuated by practice

      2019, Progress in Neuro-Psychopharmacology and Biological Psychiatry
      Citation Excerpt :

      Stimulus-related theta is however more often observed over posterior regions (Freunberger et al., 2007; Gladwin and de Jong, 2005; Sauseng et al., 2006) and thought to reflect more “proactive” aspects of cognitive control (Cooper et al., 2017). Posterior theta has been found at parietal and/or occipital electrodes during task switching (Freunberger et al., 2007; Gladwin and de Jong, 2005) and is supposed to reflect top-down regulation processes in recalling the task set memory, as its power is enhanced when the previous irrelevant visual task must be reactivated (Freunberger et al., 2007; Gladwin and de Jong, 2005). Likewise, posterior theta has been reported to increase when cognitive content needs to be suppressed (Depue et al., 2013).

    View all citing articles on Scopus
    1

    Tel.: +31 50 3636463.

    View full text