Psychotic phenomena are far from uncommon in Alzheimer's disease; delusions are present in up to three quarters of patients, hallucinations in up to a half. They appear as the disease progresses into its middle phase and are associated with cognitive decline. A better understanding of the psychopathogenetic mechanisms through which psychotic experience is mediated would have implications that would extend well beyond Alzheimer's disease; in this respect it could provide a neurological model for the so called functional psychoses. Some studies have used functional neuroimaging to identify those parts of the brain implicated in the psychotic process. The findings have been inconsistent. The paper by Mega et al in this issue (pp167–171)1 describes a methodology that represents a distinct advance. The neuropsychiatric inventory2, a care giver based instrument, was employed to control for concomitant behaviours other than those that are psychotically driven. Ten behavioural variables were rated, only two of which, delusions and hallucinations, are psychotic. A close match was achieved for non-psychotic behaviours as well as for sex, cognitive performance, and education, but only at a cost: the initial pool of 210 patients was reduced to psychotic and non-psychotic groups of 10 patients each.

The groups so formed underwent single photon emission computed tomography (SPECT) neuroimaging. In the psychotic group perfusion was significantly lower in 11 regions, most of which were clustered in the prefrontal cortex bilaterally, the left striatum, and the left parietal cortex. Reductions in regional perfusion in the temporoparietal cortex are characteristic of Alzheimer's disease3: reductions in prefrontal cortical perfusion would not necessarily have been predicted, so inevitably these findings provoke speculation concerning the role of executive function (or rather dysfunction) in the appearance of psychotic phenomena. These are not necessarily generated in the prefrontal striatal areas (for example, there is some evidence to implicate the middle temporal cortex in the production of hallucinations4) but Mega et alare right to emphasise the critical significance of executive failure: apsychotic hallucination is not simply an isolated perceptual aberration; it owes its distinctive psychotic quality to the failure of insight, influential judgement, and self monitoring capacity that are characteristic of executive dysfunction.

Psychotic symptoms are not usually present during the early stages of Alzheimer's disease. Does their emergence signify the advance of Alzheimer's pathology into the prefrontal regions? Future studies that seek to replicate these findings should include measures of executive function in their study design.