ReviewWhat does phase information of oscillatory brain activity tell us about cognitive processes?
Introduction
For 80 years by now EEG has been used to record electrical activity from the human brain (Berger, 1929). It is a popular method to acquire neural signals in a non-invasive manner. Electrodes are placed on the scalp and the so recorded electrical fields are then amplified by a factor of approximately 1000.
The signal that is acquired by the EEG is comparable to the local field potential in cortex, but on a much larger spatial scale. This means that the sum activity of many millions of neurons generates the EEG. However, the recorded activity comes not from action potentials of cortical neurons but rather their dendritic activity (excitatory and inhibitory post-synaptic potentials EPSP/IPSP). So, what we see in the human EEG is the synchronous excitatory and/or inhibitory input into a large population of nerve cells. With EEG it is possible to get a glimpse of neural activity from the whole cortex. This makes EEG a very potent tool to study the interaction between brain areas and different cortical networks. Although spatially very imprecise (spatial resolution of scalp EEG is in the range of several centimeters), EEG provides excellent temporal resolution in the range of milliseconds. This is a big advantage over other modern neuroimaging tools such as functional magnetic resonance imaging (fMRI) or positron emission tomography (PET), as EEG does not rely on the hemodynamic response but records neural activity in real time. This provides also the possibility to analyze oscillatory brain activity, which will be a main focus in this review.
Already Berger (1929) recognized that the electric activity of the human brain exhibits certain rhythmicity. He was the first who reported high amplitude oscillations around 10 Hz during a resting condition in which the subject had his/her eyes closed. He termed this activity ‘alpha rhythm’. When the subject opened his/her eyes this 10 Hz alpha activity vanished and much faster rhythmic activity with lower amplitudes became dominant. He called this pattern ‘beta rhythm’. Later, brain oscillations that are not easily visible in the healthy human (awake) resting EEG were labeled: ‘Delta’ refers to a frequency range between 0 and 4 Hz, ‘theta’ represents a rhythm between 4 and 8 Hz and ‘gamma’ oscillations describe activity above 30 Hz.
Any oscillation can be described by various parameters. These are: (i) the oscillation's frequency, (ii) its amplitude and (iii) its instantaneous phase. In Fig. 1 two examples of periodic signals are shown. The upper one has slower frequency (f = 1000/200 = 5 Hz) than the second one (f = 1000/100 = 10 Hz). Moreover, the second signal shows smaller amplitude than the upper one. In the upper panel the instantaneous phase angle of this cosine wave is given for three time points. In the EEG, all these parameters can bear important information. It will be discussed later how these parameters of oscillatory brain activity contribute to a better understanding of the human mind.
Given the idea that brain circuits of different size show different resonance properties one should expect that brain rhythms of different frequency can help dissociating specific brain networks (Von Stein and Sarnthein, 2000). This is underpinned by the fact that the classical EEG brain rhythms show different neural generators and also different functionality, as will be demonstrated in the following part.
Oscillations between 0 and 4 Hz are classically termed delta. Following Steriade (1999) oscillations below 4 Hz are generated by neocortical and thalamo-cortical networks. In terms of its functions in the brain, delta is important for large-scale cortical integration (Bruns and Eckhorn, 2004) and for attentional and syntactic language processes (Devrim et al., 1999, Schürmann et al., 2001, Roehm et al., 2004).
Theta oscillations can be found in the human cortex and the hippocampus (e.g., Kahana et al., 2001). A whole network of theta pacemakers is discussed (for an overview see O’Keefe and Nadel, 1978, Steriade, 1999) including the medial and lateral septum, hypothalamus, the hippocampus, the reticular-formation and further brain-stem structures. There is also the hypothesis that it is similarly generated as alpha oscillations, namely by thalamic nuclei (Hughes et al., 2004) and thalamo-cortical loops (Talk et al., 1999). Theta oscillations seem to be important for a variety of cognitive functions. For instance, in rats hippocampal theta, and its phase in particular, codes locations in space by influencing the temporal firing pattern of place cells (for reviews see O’Keefe and Nadel, 1978, Redish, 1999). Kahana et al. (1999) provided evidence that dominant theta activity can also be found in the human hippocampus. And it was shown that hippocampal and cortical theta activity or rhinal-hippocampal interplay was associated with virtual navigation (Kahana et al., 1999), declarative memory processes (Fell et al., 2003), successful memory encoding (Sederberg et al., 2003, Klimesch et al., 1996), the amount of information held in memory (Mecklinger et al., 1992, Tesche and Karhu, 2000, Klimesch et al., 1999, Jensen and Tesche, 2002) and episodic memory processing (e.g., Klimesch et al., 2001a, Klimesch et al., 2001b).
Inhibitory thalamic interconnection and thalamo-cortical feedback-loops are discussed to result in oscillatory activity between 8 and 13 Hz as are cortico-cortical networks (Lopes da Silva et al., 1980, Steriade, 1999, Nunez, 2000, Nunez et al., 2001). The functional relevance of these so-called alpha oscillations is very widespread. There is strong evidence that alpha amplitudes are related to the level of cortical activation. A strong alpha activity is associated with cortical and behavioral deactivation or inhibition (e.g., Klimesch et al., 1999, Klimesch et al., 2007a, Ray and Cole, 1985, Cooper et al., 2006, Hummel et al., 2002, Thut et al., 2006, Rihs et al., 2007, Worden et al., 2000, Jensen et al., 2002). But it is also involved in highly specific perceptual (Ergenoglu et al., 2004, Hanslmayr et al., 2005, Thut et al., 2006), attentional (Von Stein et al., 2000, Worden et al., 2000, Sauseng et al., 2005c, Thut et al., 2006, Rihs et al., 2007) and memory processes (for reviews see Klimesch, 1997, Klimesch, 1999, Klimesch et al., 2005).
It has been emphasized that beta oscillations are cortically generated (e.g., Salenius and Hari, 2003), due to their local strictness. This is in contrast to Gross et al. (2004) who demonstrated widespread cortical beta networks in humans. From a functional perspective beta oscillations have mainly been associated with motor activity. During movements primary motor cortices exhibit a pronounced decrease of beta amplitudes whereas there occurs a strong beta power rebound when movements are stopped (see Neuper and Pfurtscheller, 2001). But beta has also been suggested to play an important role during attention (Bekisz and Wróbel, 2003, Gross et al., 2004, Wróbel et al., 2007) or higher cognitive functions (Razumnikova, 2004).
There is large agreement that gamma oscillations (30–80 Hz) are cortically generated (see Steriade, 1999). It is emphasized that gamma oscillations arise from intrinsic membrane properties of interneurons or from neocortical excitatory-inhibitory circuits (Llinás et al., 1991, Gray et al., 1990). Since the pioneering work on the cat visual cortex by Gray, Singer and co-workers in the 1980 s gamma oscillations are well investigated and associated with visual awareness (Gray et al., 1989; for review see Engel and Singer, 2001). Synchronization phenomena of this brain rhythm were related to binding of information. More recently, effects at human gamma frequency were also reported for the encoding, retention and retrieval of information independent of sensory modality (e.g., Tallon-Baudry and Bertrand, 1999, Herrmann et al., 2004, Sederberg et al., 2003, Kaiser and Lutzenberger, 2005, Kaiser et al., 2006, Leiberg et al., 2006; for review see Kahana, 2006). Although it has also been discussed that gamma binds large-scale brain networks (Rodriguez et al., 1999; for a review see Varela et al., 2001) this view has been challenged more recently. Instead it is more likely that gamma oscillation reflect strictly local activity (Von Stein and Sarnthein, 2000, Bruns and Eckhorn, 2004). However, it should be mentioned that recording gamma activity in the human EEG is difficult due to the very small amplitude of gamma oscillations and the similarity in terms of its frequency characteristics with electrical muscle activity that is accidentally also recorded by the EEG.
For most of the above described frequency bands reactivity in response to cognitive processes was reported in terms of amplitude modulations. Slow frequency bands (delta and theta) as well as very fast oscillations (gamma) tend to increase in amplitude during cognitive effort whereas alpha and beta rhythm usually show an amplitude reduction during active cognitive processing (see Basar et al., 2001, Basar-Eroglu et al., 1996 for reviews). Measures for the reactivity of frequency bands were developed to describe event-related amplitude in- and decrease (e.g., Pfurtscheller and Aranibar, 1977, Salmelin and Hari, 1994). Using these methods it was shown that amplitude variations very selectively indicated cortical activation/deactivation in various sensory and cognitive modalities (see, e.g., Klimesch, 1999, Neuper and Pfurtscheller, 2001, Neuper et al., 2006, Hummel and Gerloff, 2006). These event-related or task-related changes have a temporal preciseness of a few hundred milliseconds (Woertz et al., 2004). Compared to the hemodynamic response this is fast. However, neuronal communication in the brain relies on dramatically faster processes (in the range of only a few milliseconds; see, i.e. Buzsaki and Draguhn, 2004, Dragoi and Buzsaki, 2006, Siapas et al., 2005). Therefore, EEG amplitude seems not to be very informative about fast changes in task-related neural processing but rather reflects sustained activation and deactivation patterns of larger cortical patches. It has been reported that the instantaneous phase of (primate) brain oscillations is associated with particular neuronal firing patterns and high temporal preciseness of neural activity (Buzsaki and Draguhn, 2004, Hirase et al., 1999, Harris et al., 2002). It was also shown that the instantaneous phase of hippocampal theta activity (in the rat) directly influences the temporal firing pattern of place cells CA1 neurons (for reviews see O’Keefe and Nadel, 1978, Redish, 1999). Thus, also in the human brain, phase information from EEG might tell us much more about the neural activity related to cognitive processes per se than do amplitude estimates. However, compared to the large body of research regarding amplitude modulations in the EEG, human studies in cognitive neuroscience using phase information from cortical oscillatory activity still are underrepresented. As outlined in the following, recently, new methods measuring different aspects of phase synchronization in the human brain have been applied to the EEG to more precisely describe how memory and attentional processes are neuronally implemented.
Section snippets
Phase synchronization in the brain
Neurons in the human brain are to a very large extent interconnected. Therefore, one can see the whole brain as a huge network consisting of millions of sub-networks ranging from micro-level to large-scale connections (see Varela et al., 2001 for review). It is widely accepted that information is stored in neural networks and human behavior arises from extremely complex communication between neurons in these networks and also between separate networks or assemblies (see Fuster, 1997, Varela et
Phase synchronization and behavior: merely correlative or causal?
Many of the above referred studies suggest that phenomena of phase synchronization are the underlying neural mechanisms of certain cognitive processes. Formally, it is not clear whether the association between these electrophysiological mechanisms and cognitive functions is of correlative or causal nature. However, there are some psychopharmacological and some clinical studies that provide evidence for the causal nature of phase synchronization as neural substrates of cognitive behavior. For
General conclusions: do brain oscillatory phase measures bear any new information about cognitive processes?
As outlined in the previous section, we find phenomena of phase synchronization in the human EEG (i) across brain sites, (ii) across frequencies and (iii) towards onset of external events. Each of these EEG responses yields important information that is not contained in any other measure of brain activity. As discussed above, the advantage of phase measures in the human EEG is its high temporal preciseness. Phase synchronization seems more tangible to measure neural communication on different
Acknowledgments
PS was supported by Grant BI 195/51-1 of the Deutsche Forschungsgemeinschaft (DFG).
References (158)
- et al.
Fronto-parietal coupling of brain rhythms in mild cognitive impairment: a multicentric EEG study
Brain Res. Bull.
(2006) - et al.
Preferred EEG brain states at stimulus onset in a fixed interstimulus interval auditory oddball task, and their effects on ERP components
Int. J. Psychophysiol.
(2003) - et al.
Gamma, alpha, delta, and theta oscillations govern cognitive processes
Int. J. Psychophysiol.
(2001) - et al.
Gamma-band responses in the brain: a short review of psychophysiological correlates and functional significance
Int. J. Psychophysiol.
(1996) - et al.
Pre-stimulus spectral EEG patterns and the visual evoked response
Electroencephalogr. Clin. Neurophysiol.
(1991) - et al.
Task-related coupling from high-to low-frequency signals among visual cortical areas in human subdural recordings
Int. J. Psychophysiol.
(2004) - et al.
Modelling event-related responses in the brain
Neuroimage
(2005) - et al.
Gamma amplitudes are coupled to theta phase in human EEG during visual perception
Int. J. Psychophysiol.
(2007) - et al.
A model for P300 generation based on responses to near-threshold visual stimuli
Cogn. Brain Res.
(1999) - et al.
EEG alpha power and intelligence
Intelligence
(2002)