It is well known that many adults with aphasia demonstrate concomitant deficits in higher-level cognitive functions, including attention, executive function, and short-term and working memory. This has led to two premises: (a) the domain-specific hypothesis, in which aphasia is associated with additional cognitive deficits only to the extent that these are dependent upon language; and (b) the domain-general hypothesis, in which aphasia is associated with nonlinguistic cognitive impairments as a consequence of either overlapping anatomy or widespread cortical changes post-insult.

The purpose of this research was to disentangle these competing hypotheses with regards to working memory (WM) in adults with aphasia. Like other categories of cognitive impairment in this patient group, past research has identified but failed to elucidate WM impairments in aphasic language processing. Toward this end, 15 adults with left-hemisphere damage and aphasia (LHD) and 12 non-brain-damaged controls (NBD) completed a parametric WM task with systematic variation of psycholinguistic complexity (high-frequency, low-frequency, or non-nameable stimuli) and WM load (0-, 1-, and 2-back). Data were analyzed with respect to the differential impact of these variables within and across subjects and groups.

Whereas expected effects of word frequency were elicited in stimulus confrontation naming, LHD subjects were affected only minimally by frequency manipulations during the n-back task. Instead, these subjects demonstrated a significant performance decrement relative to controls with increasing WM load. Moreover, aphasia severity was moderately correlated with WM for non-nameable (i.e., more difficult) but not nameable stimuli. At the theoretical level, these results support a resource-based processing model in aphasia; at the neurobiological level, these findings are consistent with the proposition of widespread cortical connectivity changes irrespective of type or location of brain damage.

A secondary purpose of this study was to investigate the reliability of LHD performance on the n-back task, given the known performance variability associated with aphasia and the general dearth of reliability data for higher-level tasks. Results demonstrated that the n-back task is a reliable WM indicator over time for this population.