Phase-coupling of theta–gamma EEG rhythms during short-term memory processing

https://doi.org/10.1016/S0167-8760(01)00199-4Get rights and content

Abstract

Because of the importance of oscillations as a general phenomenon of neuronal activity the use of EEG spectral analysis is among the most important approaches for studying human information processing. Usually, oscillations at different frequencies occur simultaneously during information processing. Thus, the question for synchronisation of different frequencies by phase coupling and its possible functional significance is of primary importance. An answer may be given by bispectral analysis. Estimation of the (cross-) bispectrum allows to identify synchronised frequencies and possibly, the existence of non-linear phase coupling of different oscillators. Previous studies have demonstrated the simultaneous occurrence of slow (4–7 Hz) and fast (20–30 Hz) oscillations at frontal and prefrontal electrode positions during memory processing. However, interrelations between these rhythms have not been investigated up to now. In order to test short-term memory, the Sternberg task with random figures and number words was carried out on 10 female subjects. During the task EEG was recorded. Power and bispectral analyses from frontal, prefrontal and frontopolar regions were performed off-line. Increased power was found in both the theta and the gamma bands. Strong phase-coupling between theta at Fz and gamma at F3 and at Fp1, respectively, was shown for memorising number words by means of cross-bicoherence. A possible reason for this is an amplitude modulation of gamma frequencies by slow oscillations. The correspondent coherence analysis between the envelope of gamma frequencies at Fp1 and the raw EEG at Fz supports this presumption. This finding is interpreted as an EEG aspect of the functional linking between the prefrontal areas and the G.cinguli (as part of the limbic system), which are both extremely important for memory functions.

Introduction

Oscillatory activity in different frequency bands of the EEG have long been known from various memory tasks. The role of oscillatory activity in the alpha and theta frequency bands has been explored in detail by Klimesch, 1996, Klimesch, 1999.

In recent years, more and more relations have been found between memory processes and both theta and gamma oscillations. Klimesch and co-workers (Klimesch, 1996, Klimesch et al., 1997) found encoding and retrieval processes to be reflected by task-related increases in theta power. In 1999, Klimesch demonstrated an increase in theta band power in response to memory demands comparable to the behaviour of hippocampal theta in animals. Miller (1991) in his studies on animals, considered this theta rhythm as an oscillatory component of the hippocampal EEG related to memory processes. In man, Gevins et al. (1997) observed increased frontal midline theta rhythm with increased memory load.

Recent physiological studies in cats and monkeys (Singer, 1993, Eckhorn et al., 1993) reported synchronisation of gamma activity of cortical neurons during the processing of visual stimuli. Singer (1993) found that high-frequency oscillations in the beta and gamma ranges, i.e. at frequencies of 15–30 and 30–60 Hz, respectively, occur spontaneously in both man and higher mammals (cats and monkeys) when the subjects are in a state of focused attention. He demonstrated the functional significance of gamma oscillations in single unit recordings. Nakamura et al. (1992) observed both low- and high-frequency oscillations associated with a recognition task in the temporal pole of macaca mulatta.

In human brains, Singer (1993), Joliot et al. (1994) and Llinas et al. (1998) demonstrated spontaneous oscillatory electrical activity at frequencies at approximately 40 Hz. Its resetting by sensory stimulation has been proposed to be related to cognitive processing and to the temporal binding of sensory stimuli. Tallon-Baudry and co-workers (Tallon-Baudry et al., 1997, Tallon-Baudry et al., 1998, Tallon-Baudry and Bertrand, 1999) found induced gamma activity during the delay of a visual short-term memory task in humans.

Beyond indications that activity from both bands are involved in memory processing, there is a growing body of evidence indicating co-occurrence and even co-operation of theta and gamma oscillations in memory tasks. Co-occurrences in the EEG were shown by Sarnthein et al. (1998), who found that working memory activity involves synchronisation between prefrontal and posterior association cortex by phase-locked low-frequency (4–7 Hz) activity. In addition, they discovered enhanced coherence in the gamma range (19–32 Hz) between prefrontal and posterior association cortex during retention intervals. From coherence studies during various cognitive tasks, Petsche and Etlinger (1998) concluded that the frontopolar cortical regions play a significant role in short-term memory.

Approaches to explore in more detail, the relationships between activity in the alpha/theta and gamma bands were based upon behavioural analyses referring to Sternberg's item-recognition task (see Sternberg, 1975). From this task it could be concluded that operation times for recognition — depending on item complexity — vary between 20 and 80 ms/item, i.e. cycles of processing which, expressed in frequencies, correspond to the gamma and beta bands. Cavanagh (1972) relating these epochs to retention performance, found a direct proportionality to reciprocal memory spans. This yields, on average, a period of some 250 ms which becomes segmented in memory scanning, a period duration corresponding to theta oscillations. On the basis of these results physiological interpretations were proposed by Geissler and co-workers (Geissler, 1991, Geissler, 1997, Geissler et al., 1999) and Lisman and co-workers (Lisman and Idiart, 1995, Jensen and Lisman, 1996, Jensen and Lisman, 1998). Whereas Geissler and co-workers centre upon relatively universal mechanisms of hierarchy formation among componential processes in perception and memory, the focus of Lisman and co-workers is the development of a specific multiplexing model explaining phase-coupled theta–gamma oscillations generated in the hippocampus.

Previous studies as mentioned above, showed that the location of gamma oscillations depends on the specificity of the cognitive task. Functional imaging techniques such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) are able to inform about the location of the activated cortex areas during memory tasks. Several recent neuroimaging studies suggest that prefrontal cortex is mainly involved in working memory.

A review of PET studies by Cabeza and Nyberg (1997) indicates the involvement of prefrontal cortex during information processing, among others: working memory; semantic memory retrieval; and episodic memory retrieval. A PET study by Tulving et al. (1994) demonstrates that left and right prefrontal lobes are part of an extensive neuronal network that subserves episodic memory. A preferential involvement of bilateral prefrontal cortex during episodic memory encoding and retrieval was also shown in a fMRI study by Iidaka et al. (1999). An activation of a common neural network for encoding was noticed regardless of the type of cognitive material presented in the left dorsal prefrontal cortex and the right cerebellum. In another fMRI study, Braver et al. (1997) showed that dorsolateral and left inferior regions of the prefrontal cortex exhibit a linear relationship between strength of activity and working memory load. Moreover, they identified additional brain regions showing a linear relationship with load, suggesting a distributed circuit in which the prefrontal cortex participates when working memory is involved. Petrides et al. (1993) observed in a PET study a strong bilateral activation within the mid-dorsolateral frontal cortex while maintaining numbers in working memory. In another PET study Swartz et al. (1991) examined abstract visual memory. They found that activations in the ventral premotor cortex and supramarginal and angular gyri were highly correlated with changes in the dorsolateral prefrontal cortex and also supported the role of the dorsal prefrontal region in non-spatial working memory in men.

Beside the general involvement of the prefrontal cortex in working memory the specificity of its different regions was investigated (Wilson et al., 1993, Goldman-Rakic, 1988, Goldman-Rakic, 1997, Paulesu et al., 1993, Shallice et al., 1994, Tulving et al., 1994, Iidaka et al., 1999, Smith et al., 1996). Other studies dealt with joint activations of cortex and hippocampus (Fernandez et al., 1999, Squire et al., 1992).

Up to now it is undecided whether phase-coupled theta-gamma oscillations do exist in the EEG of the prefrontal cortex, which would be associated to activities of neural networks during memory tasks. The bispectral analysis is a suitable methodical tool to investigate this question. The mathematical foundation of the definition and estimation of the spectral parameters of third order like (cross-) bispectrum, (cross-) biamplitude and (cross-) bicoherence is presented in detail in Nikias and Petropulu (1993). Additionally, our own methods of dynamic spectral analysis (Schack et al., 1999a, Schack et al., 1999b, Schack et al., 1999c, Möller et al., 1999) were used in order to examine temporal aspects. The definitions and estimation procedures of these methods will be given in Section 2.

Only recently several authors used the new tool of bispectral analysis for investigating the EEG. Jeffrey and Chamoun (1994) gave an introduction to this topic. By using this new tool, Schanze and Eckhorn (1997) found significant phase correlations between different frequencies in the visual cortex of the cat and monkey. Bullock et al. (1997) used bispectral analysis for the detection of short-term non-stationarity and non-linearity. Muthuswamy et al. (1999) detected phase coupling between two frequency components within the delta–theta frequency band of EEG bursts. Gajraj et al. (1998) applied the EEG bispectrum in order to distinguish the transition form unconsciousness to consciousness. Shils et al. (1996) studied the interactions between the electrocerebral activity resulting from stimulation of the left and right visual fields. He observed non-linear interactions between visual fields by means of bispectral analysis.

The main aim of this study is to test the general hypothesis of the existence of non-linear phase-coupled slow and fast rhythms within the frontal area during memory processing.

Section snippets

Experiment and subjects

Ten healthy, female right-handed adults (age=25–35 years) were performing the modified Sternberg task. Two different kinds of stimulus material served as items to be memorised: one-digit number words, zero, one, ..., nine (in German); and a set of 10 random figures. The figures were irregular rectangles of approximately the same area.

Sets with random set sizes 1, 2, 3 or 4 of words and random figures, respectively, were visually represented for 200 ms each followed by an interval of 1000 ms. A

Behavioural data

The reaction times for both stimuli — number words and random figures — show the well-known dependence on the memory load (see Fig. 2). In all cases reaction time increases with increasing set size.

Memorisation of unfamiliar random figures evidently requires more time than memorisation of number words. The comparison of the reaction times of the first and the second run shows an obvious learning effect. The significance of these results was proven by ANOVA with the factors modality, run and set

Discussion

The results in Section 3 manifest the existence of phase-coupling between theta- and gamma EEG activities during memory processing. Whereas the double peak of fast oscillations with a distance of approximately 5–7 Hz in bicoherence at Fp1, suggests only the presumption of possible phase-coupling (Section 3.3.1), the significantly increased cross-bicoherence between rhythms of 3–7 Hz at F3 and Fz, and 24–29 Hz at Fp1 clearly verifies the existence of theta–gamma phase-coupling (Section 3.3.2).

Acknowledgements

The authors would like to thank G. Lueer, U. Lass and D. Becker (Institute of Psychology, University of Goettingen) for their support in the design of the Sternberg task and their helpful discussions. This study was supported by the German Ministry of Research (DFG Scha 741/1-4 and Wi 1166/2-3).

References (59)

  • W. Singer

    Synchronisation of cortical activity and ist putative role in information processing and learning

    Annu. Rev. Physiol.

    (1993)
  • E. Tulving et al.

    Hemispheric encoding/retrieval asymmetry in episodic memory: positron emission tomography findings

    Proc. Natl. Acad. Sci. USA

    (1994)
  • J. Bredenkamp

    The relations between various invariance hypotheses in the psychology of memory

    Z. Exp. Angewandte Psychol

    (1993)
  • J. Bredenkamp et al.

    Experimental tests of a model connecting three invariance hypotheses on learning and memory processes

    Z. Psychol.

    (1998)
  • R. Cabeza et al.

    Imaging cognition: an empirical review of PET studies with normal subjects

    J. Cogn. Neurosci.

    (1997)
  • J.P. Cavanagh

    Relation between the immediate memory span and the memory search rate

    Psychol. Rev.

    (1972)
  • T.H. Bullock et al.

    Bicoherence of intracranial EEG in sleep, wakefulness and seizures

    Electroencephalogr. Clin. Neurophysiol.

    (1997)
  • R. Eckhorn et al.

    High frequencies (60–90 Hz) oscillations in primary visual cortex of awake monkey

    Neuroreport

    (1993)
  • G. Fernandez et al.

    Real-time tracking of memory formation in the human rhinal cortex and hippocampus

    Science

    (1999)
  • J.M. Fuster

    The Prefrontal Cortex, 2nd ed

    (1989)
  • R.J. Gajraj et al.

    Analysis of the EEG bispectrum, auditory evoked potentials and the EEG power spectrum during repeated transitions from consciousness to unconsciousness

    Br. J. Anaesth.

    (1998)
  • H.-G. Geissler

    The temporal architecture of central information processing: evidence for a tentative time–quantum model

    Psychol. Res.

    (1987)
  • H.-G. Geissler

    Zeitcodekonstanten-ein Bindeglied zwischen Psychologie und Physiologie bei der Erforschung kognitiver Prozesse? Hypothesen und Überlegungen zu Quantenstrukturen in der Alphaaktivität des Gehirns

    Z. Psychol.

    (1991)
  • H.-G. Geissler

    Is there a way from behavior to non-linear brain dynamics? On quantal periods in cognition and the place of alpha in brain resonances

    Int. J. Psychophysiol. (Special Issue)

    (1997)
  • A. Gevins et al.

    High-resolution EEG Mapping of cortical activation related to working memory: effects of task difficulty, type of processing, and practice

    Cereb. Cortex

    (1997)
  • P. Goldman-Rakic

    Topography of cognition: parallel distributed networks in primate association cortex

    Ann. Rev. Neurosci.

    (1988)
  • P.S. Goldman-Rakic et al.

    The circuitry of working memory revealed by anatomy and metabolic imaging

  • P. Goldman-Rakic

    Space and time in the mental universe

    Nature

    (1997)
  • G. Grieszbach et al.

    The dynamic description of a stochastic signal by its adaptive momentary power and momentary frequency estimation and its application to biological signals

    Med. Biol. Eng. Computing

    (1994)
  • Cited by (0)

    View full text