time-lock and phase-lock signal

• Time-lock
  A time-locked signal is so called because it repeats itself over the time usually after a trigger event. Many physiological signals are time-locked: when a stimuls is given, after some time you will get the response, f.i in terms of a voltage pulse. Since the signal will repeat in the same way after each stimulus, sending a number of N stimuli and summing all the N responses it is possible to increase the S/N ratio. In fact the amplitude of the pulse after the sum of the response to N stimuli will be N times the single pulse amplitude; the noise, instead since it is uncorrelated by definition will increase as sqrt(N).
  An EEG response is time-locked simply if it manifests the same pattern at roughly the same time on each trial after the stimulus onset (or whatever is the time=0 event). It doesn’t matter whether the EEG pattern is in power, phase, cross-frequency-coupling, or anything else. In contrast, non-time-locked activity means that the change in EEG signal occurs after a variable time on different trials. Whether you can measure non-time-locked activity with time-frequency-based analyses depends on how non-time-locked it is (that is, how variable the response is over trials).
• phase-lock
  the EEG signal needs to have the same phase angle at the onset of each trial.
  Phase-locked is a more stringent feature. A signal is phase-locked only if it takes the same phase angle on each trial. An EEG response has to be strongly time-locked in order to be phase-locked. Non-phase-locked, on the other hand, simply means that the time=0 event doesn’t affect the phase characteristics of the signal. Imagine that the stimulus increases the amplitude of an ongoing theta oscillation. Because the theta phase will be random on different trials, this response will be non-phase-locked, and will tend towards zero in time-domain averaging (the ERP).

The ERP reflects the phase-locked and time-locked part of the EEG signal