One way of dealing with the problem of scientific evidence for dream theories would be to restrict the search for data to experience shared by analyst and patient, or at which analyst and patient are both present. Such occasions might be all those on which the patient said he had had a dream, or all those on which there appear to be events taking place, e.g. the patient sits up and looks around in a dazed way; the analyst, identifying himself with the patient, feels that the experience the patient is having would be more understandable if the patient were asleep and dreaming.