Two classes of electrical activity in the central nervous system have been known for a long time: spikes with synaptic potentials and "slow" fluctuations (components mainly below ca. 100 Hz). Their relations to each other are still little known and an unfortunate schism persists in mutual disparagement by investigators who chiefly study one class or the other. The news I wish to highlight is that this schism is waning and this essay will be outmoded as more workers study both. I focus here on the class of slow potentials which in certain respects is the more neglected. This class should extend down into the less-known "infraslow" domain (power mainly below 0.1 Hz) - omnipresent, higher in amplitude and clearly significant functionally.
One view of our knowledge underlines the probability of many kinds of cellular sources, including, but not limited to neuron somata, axons and their terminals, dendrites and postsynaptic sites, neuroglia of more than one kind, ependyma and possibly blood vessels (including streaming potentials at the walls). Corresponding diversity of kinds of activity includes local, graded potentials of more than one kind, generator potentials of central sensory neurons, pacemaker potentials of more than one kind, such as quantal miniatures and extremely regular relaxation oscillations, spikes, after-potentials, classical synaptic potentials and subthreshold hyper- and depolarizations of longer duration with opposite membrane impedance change from the classical ones - and others. The intercellular fields of all these sources (more numerous than the sum of the cells and their synapses) add as vector sums, are dynamically interdependent and influence each other nonlinearly.
The long-standing question, whether brain waves could be
causal or only epiphenomena should now be recast to read:
Which of these kinds of activity can influence which others and
in what ways? Pacemaker potentials, for example, must in the generalcase be influenced by the
vector sum of local field potentials (LFPs) that alter transmembrane
voltage gradients differently at dendritic arbors, somata and
axon hillocks, at axon terminals and collateral branch points.
Particularly important must be the multiple mechanisms of synchrony
among populations of cells, both in firing and in subthreshold
oscillations: these are most probably modulated by LFPs as well
as chemical modulators.
The results to be expected from this view are observed: a great
variety of manifestations of assembly activity, recorded from
populations of different microstructure, connectivity, and cellular
"personality" properties (e.g. facilitation, accomodation,
iterativeness, burstiness, gap junction coupling, modulation by
any of scores of substances). Even an extensive knowledge of single
cell dynamics cannot, in general, predict the recorded microEEG
seen by extracellular microelectrodes and wideband amplifiers.
Likewise, even an extensive knowledge of such micro-fields, with
multiple electrodes, cannot, in general, predict the macro-fields
such as the scalp- recorded EEG. The three-dimensional spatial
and temporal structure of the microEEG is probably information-rich
about the vastly more complex activity of the distributed sources.
EEGs recorded from the scalp of humans show activity in
all spectral frequencies between <0.1 and >200 Hz, with
an amplitude maximum below 15 Hz and a steep amplitude decline
above ca. 15 Hz. Amplitude is so small above ca. 50 Hz that these
components are seldom studied outside the skull. Activity is generally
rather similar at scalp loci four centimeters apart; coherence
is quite significant out to some decimeters. Underlying strong
local differentiation, fluctuations in coherence, with place as
well as time, tend to occur in parallel at all frequencies, belying
the Fourier assumption of independent oscillators.
EEGs recorded intracranially with macroelectrodes are similar
but somewhat larger and decline in coherence, on average, to insignificant
levels within 10 mm. Microelectrodes see much more variable activity,
often with no appreciable coherence even at a fraction of a millimeter.
They show neuronal spikes in some loci and these can be phase-related
to some slow wave or not.
Rhythms or peaks in the power spectrum may become obvious
in certain brain states or after adequate stimuli, for example
delta, theta, alpha, beta or gamma waves at ca. 2, 5, 10, 20,
or 40 Hz, respectively. Rarely do two or more of these rhythms,
other than harmonics, occur together. A large literature has established
correlations between the presence, or scalp distribution, of an
oscillation and the behavioral or clinical state or the onset
of an adequate stimulus or expectation. It is probable that during
most of our waking hours, no rhythm is present in the scalp or
subdural EEG - defining "rhythm" (in an abbreviated
form, without discussion in this place) as more than just a component
of a wide-band spectrum, such as as the spectrum seen by band-
passing white noise.
A rhythm, for my purposes should be a distinct oscillation
that maintains a periodicity within arbitrary, reasonable
limits (e.g. ±25% of the mean) for more than a few cycles.
Even when a clear rhythm is seen, wide-band energy is also present
in EEGs and the resultant may not be equivalent to a stochastic
time series. It can have a subtle temporal structure for example
a nonlinear, quadratic phase coupling among the frequency components
(bicoherence), and in addition a spatial fine or gross structure.
Remarkable new analytical tools have come to hand for revealing
temporal and spatial structure, such as "bispectrum",
"independent component analysis", "mutual information"
and the "additive periodogram", in addition to the battery
of more classical methods.
Responses evoked by stimuli or expectations can be slow
wave transients with a characteristic form or a succession of
peaks or a more or less time-locked oscillation in addition to
nerve impulse spikes. These "event-related potentials"
and "induced rhythms" share many of the foregoing features
of ongoing activity and are typically dependent on stimulus quality,
intensity and repetition interval, locus in the brain and a number
of modulating states.
Compound, local fields can be viewed heuristically as phenomena
presented by nature that deserve adequate description and analysis
at their own level, with the expectation of eventual clarification
of the relations with other levels, such as single cell spiking.
Long before such relations are well understood, empirical knowledge
of the slow field potentials will undoubtedly reveal - and has
already turned up - emergent phenomena, such as kindling, spindles,
long range coherence, bicoherence and shifts in power spectra,
in mutual information and in measures of chaos. Some of the emergents
are useful, as in EEG diagnostic signs; others are the first neural
and objective physical signs of cognitive processes such as expectation,
attention, recognition and subtle mental discriminations; and
some are the basic data for studying synchrony, microstructure
and dynamical cooperativity..
In spite of over 70 years of research, major descriptive, as well
as analytical and mechanistic questions remain. We do not
know what differences in the EEG correlate with the vastly increased
complexity of the cerebral mantle, comparing mammals with amphibians
and fish, for example. Still less can we explain the differences
between EEGs as well as evoked potentials from different parts
of the brain, even in the best studied mammals. Insects, crustaceans,
gastropods and annelids all show similar ongoing, unstimulated
electrical activity in their higher centers that looks drastically
different from activity of any vertebrate (or cephalopod!) higher
center, but we have no understanding of the basis of the more
conspicuous spikes and relatively much weaker slow waves from
the invertebrates. A number of insects and gastropods can, under
certain conditions, show repeatable slow (<1 to >40 Hz)
rhythms and wide- band activity but dominated by spikey background.
We have reason to expect wide variation in the degree and distribution
of synchrony but lack quantitative data; this important variable
can cause or enhance slow waves and hence LFPs. Such "induced
rhythms" have attracted a great deal of recent interest,
yet we have little understanding of what they do, especially in
the majority of cases, where they have not been studied with adequate
stimuli. These and other problems offer opportunities for discovery
of new principles of system dynamics and organization.
References
Basar, E. 1999 "Brain Function and Oscillations" 2 vols., Springer, Berlin
Bullock TH 1997 Signals
and signs in the nervous system: the dynamic anatomy of electrical
activity is probably information-rich. Proc Natl Acad Sci, USA
94:1-6