Data obtained from electroencephalograms (EEGs) can be characterized in several different ways and each provides unique insight into the physiological meaning contained within. The most common way to view the data is in the time-domain. This describes how the value of the signal changes in time and is the most familiar form of EEG representation. It allows for analyses of the change in signal just before or after an event (see Event Related Potential). The second important interpretation of the data is in the frequency-domain. Electrical activity in the brain, like any signal, can be described as the sum of regularly oscillating sine and cosine functions of various frequencies. Determining which frequencies are prevalent in a signal, a method called spectral decomposition, is important especially since it is known that oscillating patterns of neuronal activity at various frequencies are biologically relevant.
Time-Frequency analysis attempts to merge these two interpretations of EEG data. The goal is to understand not only the overall frequency content of the signal but to determine when in time different frequency contributions are greatest. This hybrid approach also allows for the determination of the particular phase of a certain frequency at any given time. Recent work has shown that there is significant, coherent activity ("phase locking") in the gamma band (36-50 Hz) between 20 and 60 milliseconds after an auditory event. This characteristic is dramatically reduced in patients with schizophrenia.
1) Roach BJ, Mathalon DH (2008): Event-related EEG time-frequency analysis: An overview of measures and an analysis of early gamma band phase locking in schizophrenia. Schizophrenia Bulletin. 34(5), 1-20 [pdf]