How does EMG measure facial activity?
EMG can involve needle electrodes inserted precisely into individual muscle fibers, or as in Duchenne's experiments (image below shows stimulation), it can use surface electrodes pasted on the skin of the face. Needle EMG is more accurate about what muscle is being monitored than surface EMG, but much less pleasant, and surface EMG is used in virtually all psychophysical human research.
Raw EMG is the straight reading of the potential difference detected at the electrode. The graph of raw EMG is wildly spikey and not readily interpretable. Integrated EMG is a running average of the raw EMG. It looks smoother when graphed and is usually used as data in the analysis.
The advantages of EMG include that it is relatively inexpensive and is performed by a machine. It yields a lot of data that is continuous and scalar, increasing its apparent credibility. It can detect more subtle muscular activity than visual measurement, and is the only useful approach when movement is not visible.
The disadvantages include that it is very intrusive and may alter natural expression. Surface EMG has channel crosstalk and the number of muscles it can work with is limited by how many electrodes can be attached to the face. The quantity of validity work with EMG is not as great as with FACS. As with facial measurement, the costs of technical training is high.
Some work has been done to assure that EMG and visual facial measurement are measuring the same underlying phenomena.