Evidence of a modality-dependent role of the cerebellum in working memory? An fMRI study comparing verbal and abstract n-back tasks

Abstract

In working memory (WM), functional imaging studies demonstrate cerebellar involvement indicating a cognitive role of the cerebellum. These cognitive contributions were predominantly interpreted as part of the phonological loop within the Baddeley model of WM. However, those underlying investigations were performed in the context of visual verbal WM which could pose a bias when interpreting the results. The aim of this fMRI study was to address the question of whether the cerebellum supports additional aspects of WM in the context of higher cognitive functions. Furthermore, laterality effects were investigated to further disentangle the cerebellar role in the context of the phonological loop and the visuospatial sketchpad. A direct comparison of verbal and abstract visual WM was performed in 17 young volunteers by applying a 2-back paradigm and extracting the % change in BOLD signal from the fMRI data. To minimize potential verbal strategies, Attneave and Arnoult shapes of non-nameable objects were chosen for the abstract condition. The analyses revealed no significant differences in verbal vs. abstract WM. Moreover, no laterality effects were demonstrated in both verbal and abstract WM. These results provide further evidence of a broader cognitive involvement of the cerebellum in WM that is not only confined to the phonological loop but also supports central executive subfunctions. The fact that no lateralization effects are found might be attributed to the characteristics of the n-back paradigm which emphasizes central executive subfunctions over the subsidiary slave systems.

Introduction

It has been repeatedly demonstrated that the cerebellum is essentially involved in the control and integration of motor activity. However, over the last two decades, evidence has also been generated with regard to a cerebellar role in cognition (Desmond and Fiez, 1998, Fiez, 2001b, Marien et al., 2001, Schmahmann, 1991, Schmahmann, 2004, Schmahmann and Sherman, 1998). Studies on patients with cerebellar lesions and neuroimaging studies on healthy volunteers revealed a cerebellar contribution to executive functions such as planning, temporal sequencing, attention, learning and memory as well as an involvement in language processes (for review see e.g. Ackermann et al., 2007, Bellebaum and Daum, 2007, Ben-Yehudah et al., 2007, Haarmeier and Thier, 2007, Hokkanen et al., 2006, Leggio et al., 2008, Marien et al., 2001, Timmann and Daum, 2007). Accordingly, a number of studies focusing on working memory (WM) reported neuronal activity in the cerebellum in addition to the well-known fronto-parietal neocortical network (Cabeza and Nyberg, 2000, Desmond et al., 1997, Fiez, 2001a, Gruber, 2001, Gruber and von Cramon, 2003, Hautzel et al., 2002Hautzel et al., 2003, Krause et al., 2006, Owen et al., 2005, Paulesu et al., 1993). This neuronal network activated by WM tasks is discussed predominantly in the highly influential work of Baddeley (1986), Baddeley and Hitch (1974), and Repovs and Baddeley (2006). From this theoretical perspective, the cerebellum and Broca's area were interpreted to form a subvocal rehearsal process (Chen and Desmond, 2005a, Chen and Desmond, 2005b; Desmond et al., 1997, Paulesu et al., 1993). In combination with the phonological store, this subvocal rehearsal builds the phonological loop — one of the content-specific components used to maintain different types of information (Baddeley, 2000). The main function of the subvocal rehearsal process is to actively refresh the information items in the phonological store. However, only a few studies tackled the issue of the cerebellar function in WM directly (Chen and Desmond, 2005b, Chen and Desmond, 2005a, Desmond et al., 1997, Desmond et al., 2005, Hokkanen et al., 2006, Mathiak et al., 2004). Desmond and Fiez postulate a more prominent role of the cerebellum in WM when general task demands and/or memory load increase, respectively (Desmond and Fiez, 1998). However, studies specifically focusing on WM and the cerebellum have tested cerebellar function primarily within verbal WM (Chen and Desmond, 2005aChen and Desmond, 2005b, Chiricozzi et al., 2008, Desmond et al., 1997Desmond et al., 2005, Fiez, 2001a, Paulesu et al., 1993). Consequently, the results have been discussed in the context of the phonological loop and its subsystems, leading to the conclusion that parts of the cerebellum subserve the rehearsal system, while others belong to the phonological store (Chiricozzi et al., 2008, Desmond et al., 1997). In addition, by applying exclusively visual verbal WM, the Desmond group revealed a right-sided laterality effect with the involvement of lobules VI and crus1 in subvocal rehearsal processes, while right lobules VIIb and VIII were found to contribute to the phonological store (Chen and Desmond, 2005aChen and Desmond, 2005b, Desmond et al., 1997, 2003Desmond et al., 2005, Kirschen et al., 2005). However, the work of Kirschen et al. also presented evidence of an activation of the contralateral lobules as well, despite the fact that these left-sided contributions were less extended and less significant. Studies in lesioned patients failed to replicate a clear-cut right-sided lateralization of WM functions in the cerebellum. Concerning the visuospatial sketchpad, a study has yet to address the question of a cerebellar contribution to this WM subsystem directly but data derived from patients with unilateral left-sided cerebellar damage point towards a higher impairment in a visuospatial task compared to a right-sided lesion within the cerebellum (Hokkanen et al., 2006). Overall, the issue of cerebellar laterality in WM requires further evaluation.

In addition to verbal WM, other modalities employing object, spatial or abstract shape stimuli also induce cerebellar activations (Courtney et al., 1996, Hautzel et al., 2002, Nystrom et al., 2000, Pessoa and Ungerleider, 2004). This could be indicative of a more general role of the cerebellum within WM beyond its function in subordinate routines closely related to motor processes like inner speech. Evidence of extended cerebellar contributions to WM can be found in studies by Mathiak et al. testing WM of time (Mathiak et al., 2004) and by Gottwald et al. investigating attention and other central executive functions in patients with cerebellar lesions (Gottwald et al., 2003, 2004). Moreover, Appollonio et al. interpreted WM impairments after cerebellar damage as secondary to an impairment of executive functions because after controlling for executive measures, memory scores were no longer different to those of healthy controls (Appollonio et al., 1993). Finally, the cerebellar cognitive affective syndrome (CCAS) summarizes an association of lesions to the posterior cerebellar lobe with deficits of executive functions (planning, set shifting, verbal fluency, abstract reasoning, and working memory) (Schmahmann and Caplan, 2006, Schmahmann and Sherman, 1998).

Given such divergent evidence concerning the role of the cerebellum, the aim of the fMRI study presented here was twofold: (1) to directly address the question of whether the cerebellum processes the predominantly language-related aspects of WM or whether it is also involved in higher-order cognitive WM processes, and (2) to evaluate laterality effects of the cerebellum at the hemisphere and the lobular levels for both the verbal and abstract WM separately.

In order to maximize the differences of speech input, we performed a direct intra-individual comparison of verbal WM and abstract shape WM using non-nameable polygons as stimuli for the latter constructed according to method 1 described by Attneave and Arnoult (1956). Furthermore, to ensure adequate comparability between modalities and underlying processes and particularly to stress the central executive subfunctions of time coding und updating, an n-back WM task was chosen instead of a delayed match-to-sample paradigm (Peters et al., 2005, Smith and Jonides, 1997, Wager and Smith, 2003). If the cerebellum is predominantly involved in subvocal rehearsal processes, we hypothesized an intense activation of neocerebellar structures during the verbal WM task but not while performing the abstract WM paradigm. In addition, predominantly right-sided cerebellar activations should be evident as a result of the crossed cerebello-cortical diaschisis (Botez-Marquard et al., 1994, Broich et al., 1987, Pantano et al., 1986). If the abstract WM recruits cerebellar regions in the context of the visuospatial sketchpad in analogy to verbal WM and the phonological loop, then a left-sided lateralization in the abstract WM activation patterns should result. On the other hand, if higher cognitive functions summarized in the central executive are the major cause of cerebellar activation, both tasks should reveal comparable signal increases in the cerebellum. From the neocortical perspective, both WM paradigms have been proven to activate the central executive core areas in prefrontal and parietal cortices bilaterally and to the same extent (Hautzel et al., 2002). If this alternative hypothesis of cerebellar involvement beyond the subordinate WM slave systems and particularly beyond the phonological loop holds true, then the resulting neuronal activations should be symmetrically distributed over both cerebellar hemispheres.