Cerebrocerebellar networks during articulatory rehearsal and verbal working memory tasks

Converging evidence has implicated the cerebellum in verbal working memory. The current fMRI study sought to further characterize cerebrocerebellar participation in this cognitive process by revealing regions of activation common to a verbal working task and an articulatory control task, as well as regions that are uniquely activated by working memory. Consistent with our model's predictions, load-dependent activations were observed in Broca's area (BA 44/6) and the superior cerebellar hemisphere (VI/CrusI) for both working memory and motoric rehearsal. In contrast, activations unique to verbal working memory were found in the inferior parietal lobule (BA 40) and the right inferior cerebellum hemisphere (VIIB). These findings provide evidence for two cerebrocerebellar networks for verbal working memory: a frontal/superior cerebellar articulatory control system and a parietal/inferior cerebellar phonological storage system.

Introduction

Current models of verbal working memory (Baddeley, 1986, Baddeley, 1992) postulate a phonological loop for the short-term maintenance of verbal information. This loop is assumed to be composed of two components, a phonological store, which can hold speech-related information for 1–2 s, and an articulatory control process, which serves to subvocally refresh the contents of the phonological store. Evidences from both neuropsychological and neuroimaging studies have suggested that left prefrontal regions corresponding to Broca's area are likely involved in the articulatory control process, whereas left inferior parietal cortex is the likely locus of the phonological store (Awh et al., 1996, Fiez et al., 1996, Jonides et al., 1998, Paulesu et al., 1993, Smith and Jonides, 1998).

In a previous investigation (Desmond et al., 1997), we identified two cerebellar regions activated in verbal working memory: a bilateral superior region located in the hemisphere of lobule VI/Crus I, and an inferior region located in right hemispheral VIIB. For both a verbal working memory and a motoric rehearsal task (one that had similar letter encoding and subvocal articulatory requirements as the verbal working memory task but no storage requirements for the verbal information), load-dependent activation was observed in the superior cerebellum. In contrast, load-dependent inferior cerebellar activation was observed only for the working memory task. We then formulated a cerebrocerebellar model of verbal working memory based on this pattern of activation, as well as neuroanatomical evidence of corticopontine and pontocerebellar patterns of projections. Specifically, several studies in primates have indicated that corticopontine projections from frontal regions tend to be distributed along medial portions of the pontine nuclei (Brodal, 1978, Leichnetz et al., 1984, Schmahmann, 1996, Schmahmann and Pandya, 1997a, Schmahmann and Pandya, 1997b, Wiesendanger et al., 1979), whereas parietal and temporal regions project to lateral pontine regions (Brodal, 1978, Brodal and Bjaalie, 1997, Leichnetz et al., 1984, May and Andersen, 1986, Schmahmann, 1996, Schmahmann and Pandya, 1989, Schmahmann and Pandya, 1991, Weber and Yin, 1984, Wiesendanger et al., 1979). Medial pontine nuclei in turn have been shown to project to superior portions of cerebellar cortex, including the simplex lobule (hemispheric lobule VI) and Crus I (superior part of hemispheric lobule VIIA), whereas lateral regions of pontine nuclei project to the paramedian lobule (hemispheric lobule VIIB) (Brodal, 1979, Brodal, 1982, Glickstein et al., 1994). We hypothesized that the superior cerebellar activation reflected the articulatory control process from frontal regions and that the inferior cerebellar activation reflected phonological storage from parietal regions. Because the study of Desmond et al. (1997) was confined to the cerebellum, a comparison of activations in the working memory and motoric rehearsal tasks in both neocortex and cerebellum was not possible.

Thus, the purpose of the present study was to extend these previous findings and characterize regions of activation common to verbal working memory and motoric rehearsal, as well as regions that are unique to verbal working memory, in both neocortex and cerebellum. Based on the above model of cerebrocerebellar interactions for verbal working memory (Desmond, 2001, Desmond et al., 1997), we predicted that the high or low load comparison for both verbal working memory and motoric rehearsal would activate Broca's area and superior cerebellum, reflecting common use of the articulatory control system in both tasks. In contrast, because phonological storage is required only for the verbal working memory task, we expected left parietal and right inferior cerebellar activations to be unique to the verbal working memory task.