This issue contains two papers which concern the occurrence of hallucinations in Parkinson's disease. This is a matter of great importance in the management of the disease. Hallucinations and related phenomena are of wider interest because they may provide clues to the brain mechanisms involved in their production.

Almost 65 years have passed since the publication of the now classic work of Wolff and Curran1: Nature of delerium and allied states. Their study reported that “no evidence was found that there was any specific relationship between a particular noxious agent and the form and content of the accompanying psychobiologic disturbances.” These findings suggest that a range of events may provoke a reaction in the functioning of the brain but with a common clinical presentation. The occurrence of hallucinations in Parkinson's disease raises similar issues.

Barnes and David (this issue, pp 727–733) present detailed accounts of the phenomenology of hallucinations in Parkinson's disease and relate their occurrence to features of the disease which are associated with a higher risk of the experience—namely, greater age and duration of disease, cognitive impairment, depression of mood, and sleep disturbance.2 They draw comparison with the hallucinations experienced by patients with visual impairment and comment on the marked similarity in the form and content of the experiences. They suggest that a common “physiological substrate” may underlie the experience in the two settings.

In the study reported by Holroyd et al (this issue, pp 734–738) more than a quarter of a consecutive series of patients with Parkinson's disease were found to have experienced visual hallucinations in the previous week.3 There was a clear association between the occurrence of visual hallucinations and visual impairment as well as with impaired cognitive function, depression of mood, and severity of disease. It is notable that no association was found between the use, dosage, and duration of administration of levodopa or of other antiparkinsonian drugs. Unsurprisingly, in view of previous reports, few patients showed evidence of a paranoid illness, of which hallucinations might form a part.

Research on Parkinson's disease is influenced by the sample of subjects studied.Because the onset of the disease is usually insidious the identification of patients is difficult and estimation of the duration of illness imprecise.4 Barnes and Anthony recognise that their sample is difficult to define and is unlikely to be typical of patients with Parkinson's disease. Holroydet al took consecutive subjects attending a specialised clinic for the first time and their sample has the advantages of an “inception cohort” where patients enter a study by the same route and at roughly the same stage of the disease.

Holroyd et al use standardised criteria for the diagnosis of Parkinson's disease and for the exclusion of other illnesses but their criteria for the recognition of hallucinations lack detail. Barnes and Anthony use more rigorous criteria for the identification and categorisation of the phenomena found.

Somewhat paradoxically, as research in Parkinson's disease has progressed, it is the diagnosis of the disease itself which has become problematic.The central category of idiopathic Parkinson's disease is diminishing as similar but distinct syndromes are separated off.4-6 Furthermore, the boundaries between neurodegenerative diseases are becoming less distinct.6This aspect is not really addressed in either paper but is central to the matter stressed by Barnes and Anthony, and touched upon by Holroydet al, that hallucinations may be a non-specific response to a range of circumstances in conditions which predispose to their occurrence.

Both papers are a welcome addition to the research in this difficult field. Relating psychopathology to brain mechanisms is essential if progress is to be made in this area.