Multi-unit Extracellular Recording
The phrase "multiunit recording" has been used for the measurement of neuronal activity at a variety of scales, encompassing both averaged measurements of the activity of many thousands (sometimes millions) of neurons, as well as measurement of the individual action potentials from a handful, perhaps a hundred, of cells. In all cases, the underlying motivation is that only by observing the activity of many neurons in concert can we hope to understand how collective computation in the brain occurs.
The average measurements may be electrical (current or field), magnetic, optical (using voltage-sensitive dyes) or indirect (measuring metabolic parameters by NMR or "intrinsic signal" optical imaging). Thus far, all the individual-cell measurements, at least in intact animals, have been electrical, and this is likely to remain true for the short-term future.
Extracellular Recording
Electrical signals from individual cells are acquired by inserting a conductive microelectrode near the cell membrane, which then provides a source or sink for the trans-membrane currents associated with an action potential. Each spike appears as a sterotyped waveform on the voltage trace recorded from the microelectrode. Neural tissue in the brain is dense enough that a probe inserted at random into the gray matter will lie close to many cell membranes, and will couple to currents accross all of them.
Extracellular single-unit recording has traditionally proceeded by manoeuvering the electrode tip until it lies close to a single membrane and the recorded signal is dominated by that one source (a procedure referred to as "isolation"). This positioning needs to be precise on a scale of microns, and in alert non-paralyzed animals, can only be maintained for a limited period.
One approach to multiple single-unit recording, then, would be to introduce many electrodes and position each to isolate a single cell. This turns out to be impractical, at least in awake animals, beyond a small number of electrodes. One issue is that it takes time to isolate cells, and once isolation is achieved it can only be held for so long. Thus, by the time one has isolated a cell on the tenth electrode there is a risk that the isolation on the first electrode has been lost. This problem is made worse when one tries to record from cells that are close to each other, in general a sensible goal. As each electrode is manoeuvered, it drags the adjacent brain tissue with it, disturbing the isolation of cells on nearby electrodes.
Thus, scalable recording techniques must rely on algorithmic approaches, rather than physical positioning, to sort out signals from the various membranes.
Tetrodes
The problem of recognizing action potentials from different pieces of
membrane is made much simpler by use of a tetrode. A tetrode
is a bundle of four individually insulated fine wire electrodes, whose
tips lie closer together than their respective spheres of
sensitivity. An exact analogy can be drawn between neural recordings
obtained from such a device and quadraphonic musical recordings. A
neuron that lies in the overlap of two or more of these spheres is
detected by two or more electrodes. Assymetries, whether in electrode
construction or relative geometry, insure that different electrodes
record this neural signal through slightly different filters. The
comparison of signal across channels allows the disambiguation of
signals that appear identical to a single electrode.
The tetrode was introduced by Recce and O'Keefe in 1987 for use in rat hippocampus where the cells are so close packed that conventional isolation is very difficult. Since then, other groups have used it successfully in the same preparation. More recently,