The article (Reference Rao and KeshavanRao 2006) that Clark & Crossfield refer to in their e-Letter does not in fact justify their assumption that art does not indicate mental state. Rao & Keshavan resorted to a chi-squared test to analyse their results, finding that untrained lay people were much less able than psychiatrists to infer mental illness in the tragic paintings of Gauguin, Van Gogh, Munch and Rothko. The article then concedes that psychiatrists are not so sure-footed when dealing with the ordinary works of individuals not found in museums.

Rao & Keshavan's article belongs to a series in the American Journal of Psychiatry, ‘Images in psychiatry’, in which the overall implication is, in fact, that art does mirror the often turbulent mental state within. One article in the series introduces the American Visionary Art Museum in Baltimore, which shows the works of self-taught artists, many of whom have mental illnesses. The authors entitle their article ‘Art as a portal into the minds of those with mental illness’ (Reference Fujimoto and Douglas-FujimotoFujimoto 2008). Another focuses on Caravaggio, with the grim painting of David holding, in place of the head of Goliath, Caravaggio's own severed head. The author comments that this reflects Caravaggio's ‘insight into … his psychopathology – pitilessly self-destructive and ultimately fatal masochism’ (Reference BuckleyBuckley 2008). Caravaggio was indeed a suicidal Goliath.

According to Dubuffet, all art requires instability, rather than moderation and reason, at its core. Art is the pursuit of the abnormal. Caravaggio – a homosexual, brawler and murderer hunted down by authority across Italy – was the typical artist as doomed rebellious outsider.

The association between art and mental illness is an old one. We could not cope without the insight of Clark & Crossfield that art is a dialogue, but they are denying the history and essence of art in divorcing the mental state from what is depicted and displayed. Art seeks the heart of shadows that is in us all.