Physiological Foundations of Quantitative EEG Analysis - Quantitative EEG Analysis Methods and Clinical Applications - page 18

scales. Here the term phase locking is used to indicate “synchronization” with arbi-
trary phase lag. Several approaches to estimate phase locking are discussed in
Chapter 4.
One measure of phase locking between any pair of signals is coherence , a corre-
lation coefficient (squared) expressed as a function of frequency band. For example,
while performing mental calculations subjects often exhibit increased EEG coher-
ence in the theta (near 5 Hz) and upper alpha (near 10 Hz) bands, whereas the same
data may show decreased coherence in the lower alpha band (near 8 Hz) in most
electrode pairs [2, 14]. Coherence of steady-state visually evoked potentials indi-
cates that mental tasks consistently involve increased 13-Hz coherence in select elec-
trode pairs but decreased coherence in other pairs [29, 30]. Binocular rivalry
experiments using steady-state magnetic field recordings show that conscious per-
ception of a stimulus flicker is reliably associated with increased cross hemispheric
coherence at 7 Hz [31]. These data are consistent with the formation of large-scale
cell assemblies (e.g., cortical dipole layers) at select frequencies with center-to-center
cortical separations of roughly 5 to 20 cm.
These cognitive studies emphasize relatively low frequencies ( 15 Hz) because

of the very low signal-to-noise ratio (SNR) associated with higher frequencies in
scalp recordings. By contrast, intracranial studies in lower mammals often empha-
size gamma-band phase locking (

<

40 Hz). It is, however, difficult (if not impossible)
to record gamma-band spontaneous EEG data from the scalp that are not substan-
tially contaminated by muscle and other artifact. Brain signals above about 15 to 20
Hz have very low scalp amplitudes, often lower than muscle activity in the same gen-
eral frequency range. By contrast, much higher frequency potentials may be
recorded with intracranial electrodes. For example, human studies using subdural
electrodes have found increased electrocorticogram (ECoG) power in the 80- to
150-Hz range over auditory and prefrontal cortex when epilepsy patients attended
to external stimuli [32].
Past emphasis on the 40-Hz gamma band may have been based partly on two
unproven assumptions: (1) Frequency bands in which robust correlations between
cognition/behavior and phase locking occur are similar in humans and lower mam-
mals. (2) Physiologically interesting phase locking is mainly confined to frequencies
near 40 Hz. Nunez and Srinivasan [2] have challenged these assumptions, suggest-
ing that cognition may easily produce concurrent signatures in multiple frequency
bands that may differ in humans and lower mammals. For example, observations of
concurrent theta and alpha coherence effects (in three distinct bands) do not pre-
clude additional concurrent effects in multiple bands well above 15 Hz occurring at
spatial scales too small to be recorded from the scalp. The most robust effects are
likely to be observed in frequency bands that most closely “match” the spatial scale
of the recording, as determined by electrode size and distance from sources in
intracranial recordings (refer to Figure 1.4).

∼

1.12 “Simple” Theories of Cortical Dynamics

Human EEG recordings have been carried out for the past 80 years, but the physio-
logical bases for EEG and P ( r , t ) dynamic behavior remain mostly obscure. We