In previous studies we suggested that liberal acceptance (LA) represents a fundamental cognitive bias in schizophrenia and may explain why patients are more willing to accept weak response alternatives and display overconfidence in incorrect responses. The aim of the present study was to test a central assumption of the LA account: false alarms in schizophrenia should be particularly increased when the distractor–target resemblance is weak relative to a control group.

Sixty-eight schizophrenia patients were compared to 25 healthy controls on a visual memory task. At encoding, participants studied eight complex displays, each consisting of a unique pairing of four stimulus attributes: symbol, shape, position and colour. At recognition, studied items were presented along with distractors that resembled the targets to varying degrees (i.e. the match between distractors and targets ranged from one to three attributes). Participants were required to make old/new judgements graded for confidence.

The hypotheses were confirmed: false recognition was increased for patients compared to controls for weakly and moderately related distractors only, whereas strong lure items induced similar levels of false recognition for both groups. In accordance with prior research, patients displayed a significantly reduced confidence gap and enhanced knowledge corruption compared to controls. Finally, higher neuroleptic dosage was related to a decreased number of high-confident ratings.

These data assert that LA is a core mechanism contributing to both enhanced acceptance of weakly supported response alternatives and metamemory deficits, and this may be linked to the emergence of positive symptomatology.