The Association Between Eye Movements and Cerebellar Activation in a Verbal Working Memory Task

Abstract

It has been argued that cerebellar activations during cognitive tasks may masquerade as cognition, while actually reflecting processes related to movement planning or motor learning. The present study investigated whether the cerebellar load effect for verbal working memory, that is, increased activations in lobule VI/Crus I and lobule VIIB/VIIIA, is related to eye movements and oculomotor processing. Fifteen participants performed an fMRI-based Sternberg verbal working memory task. Oculomotor and cognitive task demands were manipulated by using closely and widely spaced stimuli, and high and low cognitive load. Trial-based quantitative eye movement parameters were obtained from concurrent eye tracking. Conventional MRI analysis replicated the cerebellar load effect in lobules VI and VIIB/VIIIa. With quantitative eye movement parameters as regressors, analysis yielded very similar activation patterns. While load effect and eye regressor generally recruited spatially distinct neocortical and cerebellar regions, conjunction analysis showed that a small subset of prefrontal areas implicated in the load effect also responded to the eye regressor. The present results indicate that cognitive load-dependent activations in lateral superior and posteroinferior cerebellar regions in the Sternberg task are independent of eye movements occurring during stimulus encoding. This is inconsistent with the notion that cognitive load-dependent cerebellar activations merely reflect oculomotor processing.

Keywords: cerebellum; cognition; eye tracking; fMRI; working memory.

Figure 1.

Time course of stimulus presentation in the Sternberg verbal working memory task, illustrating each of the 8 types of encoding stimuli that each subject received.

Figure 2.

High versus low cognitive load in the conventional analysis (see Table 1 for peak coordinates). Coronal slices from Talairach y = +46 to −90 mm are depicted. Positive activations (high > low) are shown in red; negative activations (low > high) are shown in blue; P < 0.001−0.00001.

Figure 3.

High versus low cognitive load with eye regressor (see Table 1 for peak coordinates). Coronal slices from Talairach y = +46 to −90 mm are depicted. Positive activations (high > low) are shown in red; negative activations (low > high) are shown in blue; P < 0.001−0.00001.

Figure 4.

Eye regressor (see Table 2 for peak coordinates). Coronal slices from Talairach y = +46 to −90 mm are depicted. Positive activations are shown in red; negative activations are shown in blue; P < 0.001−0.00001.

Figure 5.

Signal increase in superior colliculus associated with the eye regressor as revealed by explorative analysis at P < 0.001−0.00001.

Figure 6.

Surface rendering of regions with signal increase the load effect (red) and the eye regressor (green) and shared regions (yellow; circled) as revealed by conjunction analysis. P < 0.001−0.00001.