Electrophysiological Studies of Mind

Intracellular neural currents carry inhibition and excitation from the synapses where neurons communicate to where they decide whether to fire. Those same currents give rise directly to the magnetic fields measured by the magnetoencephalogram (MEG); the equal but opposite extracellular return currents instantaneously cause potential differences between locations on the scalp, which are measured by the electroencephalogram (EEG) (Figure 1). The EEG and MEG thus directly and instantaneously reflect the computational currents used by the information-processing mechanisms of the human brain.

Figure 1 (A) Charges entering at a synapse flow inside the apical dendrite of a pyramidal cell. This current (narrow arrow inside apical dendrite) generates the MEG. The extracellular return current (thick, darker arrow) results in voltage differences at distant sites, which is the EEG. (B) MEG is measured with superconducting quantum interference devices (SQUIDs, marked with square boxes) that are kept near absolute zero in liquid helium. EEG is measured with ordinary voltage amplifiers.

Consequently, the EEG and MEG have a systematic relationship to different aspects of mental activity, including higher mental functions such as attention, orienting, memory, language, face perception, action, error correction, reanalysis, and closure. Typically, these relationships are revealed by averaging the EEG or MEG with respect to repeated cognitive events, yielding event-related potentials (ERP) and event-related fields (ERF), or generically, evoked responses (ER). Across thousands of studies, it has been observed that ERs can be described as a series of positive and negative peaks with characteristic latencies, scalp topographies, and cognitive correlates. These regularly occurring clusters of mental and physical characteristics have been reified as named components. This approach provides a vocabulary for communicating discoveries in EEG and MEG and thus bootstraps a physiological lexicon of the mind.

Because they measure directly and instantaneously the currents that perform information processing in the human brain, EEG and MEG have a temporal and physiological accuracy that hemodynamic measures, such as positron emission tomography (PET) and functional magnetic resonance imaging (fMRI), lack. However, EEG and MEG have other problems. First, it is impossible to localize with certainty from extracranial measures where they are generated in the brain. This not only renders problematic the interpretation of ER with respect to neural systems; it also deprives components of an unambiguous defining characteristic and thus makes component-based analysis somewhat circular. Second, most of the information-processing currents cannot be recorded at a distance because of spatiotemporal cancellation. Action potentials are not recorded at all, nor are currents from interneurons. EEG and MEG arise mainly from relatively slow and synchronized currents in the apical dendrites of pyramidal cells. Third, the polarity of the EEG or MEG signal does not tell us whether the underlying neural activity is excitatory or inhibitory. To some degree, these difficulties can be ameliorated by combining information from PET or fMRI with that of EEG and MEG, within the context of individual brain anatomy provided by structural MRI. In certain clinical situations, it is possible to record potentials directly within the human brain, yielding very high spatial, temporal, and physiological accuracy; such measures can be used to infer the location [Page 274]and physiological meaning of EEG and MEG components recorded in healthy subjects.

...