| Goldman-Rakic offers a conceptualization of working memory (WM) that emphasizes how prefrontal (PFC) and posterior cortical areas are linked in reverbatory circuits that keep information on-line. Goldman-Rakic argues that these WM circuits are modality-specific, with dorsal PFC and parietal activation for spatial processing, and ventral PFC and inferior temporal activation when non-spatial or pattern information is being processed. Cohen et al. (1997) demonstrated fMRI activation of the dorsolateral PFC as a function of load in an n-back WM task. We employed a similar task to test the modality-specific WM model, with a spatial condition (position being the matched criterion for letter stimuli presented either 1, 2 or 3 items earlier) and a verbal condition (specific letter to be matched). Electrophysiological measures were employed, including ERP and EEG spectral analysis, as well as EEG coherence, which can provide a functional measure of corticocortical communication. EEG was recorded from 9 frontal and 9 parietal coronal sites from an extended 10-20 system.; Support was not found for modality-specific WM systems. Instead, results suggest that there is a WM circuit that connects dorsal frontal with dorsal parietal areas which is activated during a WM task, regardless of the type of stimuli involved. Data also suggest that left dorsal frontal and parietal areas are sensitive to the level of difficulty of the WM task (0, 1, 2 and 3-back), as well as familiarity of task demands. Overall, results support a model of WM that conceptualizes a unitary WM system rather than a system composed of separate WM circuits for different modalities. More specifically, these data appear to favor the WM model of Petrides (1995) which suggests a ventral/dorsal distinction in frontal activation patterns is related to the executive processes being performed on the information. |
