I’ve used to on neural spike data before. If your responses are not too “messy”, you can estimate the part of the response that was due to a stimulus instead of just noise, pretty well.