A patient, AB, is reported who showed clear signs of neglect but no extinction (N+ E-). Several hypotheses proposed to account for this dissociation were put to the test. The postulated association between motor neglect and extinction did not hold good, nor did the possibility that the N+ E- dissociation may be traced back to the difference in test requirements and therefore observed only in patients with object-centred neglect. Likewise, manipulating the physical features of the stimuli (relative size, exposure time, presentation synchrony) did not elicit extinction. However, when the task demands were modified by asking the patient to perform a further spatial analysis of the stimuli, rather than simply detect them, extinction emerged. Since AB performed well on several neglect tasks requiring parallel processing, while failing all tasks calling for serial processing, the hypothesis is put forward that AB's N+ E- dissociation could be interpreted within the parallel/serial distinction framework.