Marshall and Halligan’s (1988) pioneering study of unconscious processing in visuospatial neglect is one of the most influential neuropsychological single case studies of the last 40 years. Here we report a pre-registered conceptual replication of this study in a large group of patients. Fifty-four stroke survivors (21 with neglect), unselected for lesion location, completed a computerised and extended variation of the Burning House Task. Patients were asked to report whether pairs of pictures were the same or different, and then asked to indicate which image they preferred. On critical trials, one image was normal (intact), and the other had a lateralised addition of either fire (burning) or shading (shaded). In pre-registered analyses, one patient reliably preferred the intact images despite reporting the two to be identical. This replicates Marshall & Halligan’s main finding, except that our patient did not have neglect. In exploratory analyses, with adjusted criteria, we identified five additional patients with this pattern, only one of whom showed signs of neglect. All six patients showed similar preferences for intact over burning and shaded pictures, suggesting that the preference was not due to unconscious processing of semantic content (‘fire’). Overall, the results suggest that the preference bias preference reported by Marshall & Halligan is neither common in neglect nor exclusive to neglect, and may not be driven by semantic processing of the meaning of fire.

Visual neglect

Stroke

Pre-attentive processing

Neuropsychology

Cognitive impairment

© 2025 The Author(s). Published by Elsevier Ltd.