relationship between doctor and patient. Certain groups are especially at risk in this respect, for example, involuntary patients and mentally abnormal offenders in prison. ii) Problems of decision and action: Our capacities for decision and action may be adversely affected by a wide variety of psychological disorders: for example, obsessive-compulsive disorders often involve a generalised inability to make decisions; mood disorders may involve pathological motivations (such as the depressed patient who agrees to take part in a research project in the hope that it will kill him); and in psychotic disorders there may be profound impairments of insight (as with the deluded patient who believes that a research project is really an initiation ceremony offering access to supernatural powers). The corresponding difficulties in relation to issues raised by the information element in consent include: i) How much information is appropriate: In ordinary clinical practice full disclosure of all the risks and benefits of a proposed procedure is generally not required for consent to be valid. In the UK, for example, a doctor has to reveal only as much information as would be revealed by any responsible and skilled body of his or her peers (the so-called 'prudent doctor' test). With research procedures, however, there is a strong presumption that more complete disclosure is necessary. It is recognised that full understanding of a research project - amounting as it would to a lengthy and to many patients incomprehensible seminar - is generally impractical. Nonetheless, in the Royal College of Physicians' guidelines, adequate information is taken to include 'any benefits and hazards' (1 2), and where simplification is necessary this should 'not have the effect of understating any risks or glossing over inconvenience or discomfort' (13). Achieving this degree of information, however, as a basis for acopyright. on November 22, 2021 by guest. Protected by http://jme.bmj.com/ J Med Ethics: first published as 10.1136/jme.19.2.85 on 1 June 1993. Downloaded from K WM Fulford and KHowse 87 balanced decision about participation in research, presents many difficulties. These have been widely discussed in relation to the problem of consent to inclusion in randomised trials (14,15). But in some areas of psychiatric and psychological research a degree of active deception is sometimes required (14). Full disclosure even of the purposes of the research can sometimes so alter the subject's responses as to invalidate the results: false feedback experimental designs, for example, are not uncommon in psychological research; and some of these may even involve paradoxical intent, in which the therapist suggests certain responses or courses of action with the intention of provoking their contraries (16). The standard proposed by the Royal College of Psychiatrists is less comprehensive than that suggested by the Royal College of Physicians (17). After noting that it is not possible to give detailed guidelines covering every case, it suggests that 'important' risks (rather than 'any', as in the Royal College of Physicians' guidelines) should be made clear. What is important is in turn to be defined by reference to an ethics of research committee which should 'apply commonsense to decide what level of risk would be likely to affect a reasonable person's decision'. Further, where incomplete disclosure of information is proposed, the research should not be carried out where it is considered that on a basis of full information, the patients concerned would be expected to withhold their consent. What this amounts to, then, is a 'reasonable person' rather than, as in ordinary clinical practice, a 'reasonable doctor' test of adequacy of information. However, while providing sound general principles, the reference to commonsense raises many of the difficulties involved in achieving adequate understanding in this context. In psychiatric research patients are by definition, in one or other respect, and to a lesser or greater degree, unreasonable. And as Taylor has emphasised (1 8), while there are many excellent theoretical discussions of the nature of competence in the literature (19), the common assumption that