Attentional Activation of the Cerebellum Independent of Motor Involvement

See allHide authors and affiliations

Science  28 Mar 1997:
Vol. 275, Issue 5308, pp. 1940-1943
DOI: 10.1126/science.275.5308.1940


The cerebellum traditionally has been viewed as a neural device dedicated to motor control. Although recent evidence shows that it is involved in nonmotor operations as well, an important question is whether this involvement is independent of motor control and motor guidance. Functional magnetic resonance imaging was used to demonstrate that attention and motor performance independently activate distinct cerebellar regions. These findings support a broader concept of cerebellar function, in which the cerebellum is involved in diverse cognitive and noncognitive neurobehavioral systems, including the attention and motor systems, in order to anticipate imminent information acquisition, analysis, or action.

The human cerebellum has more neurons than the remainder of the brain combined (1). It is physiologically connected, by monosynaptic or multisynaptic pathways, with all major subdivisions of the central nervous system (CNS), including the cerebrum, basal ganglia, diencephalon, limbic system, brainstem, and spinal cord (26). It is, therefore, one of the busiest intersections in the human brain. Nonetheless, for more than a century, neurologists and neuroscientists alike have held the view that the singular function of the human cerebellum is to help coordinate movement (6).

Controversy over this long-established position has emerged because of evidence from recent functional neuroimaging and neurobehavioral studies (2, 714). These studies show that the cerebellum may be involved in a variety of nonmotor functions, including sensory discrimination (7), attention (2, 810), working memory (11), semantic association (12), verbal learning and memory (13), and complex problem solving (14). However, in almost all of these studies, movement or motor planning were necessary components of the sensory or cognitive experimental task (15). In the face of such critical confounding factors, the traditional concept remains largely unmoved. One recent advance over the more traditional view suggests that the cerebellum modulates the motor control system in the service of acquiring high-quality sensory information (7, 16). Although this model incorporates a sensory role for the cerebellum, it still construes the cerebellum as a device whose function is motor control.

Missing from experiments to date is a single design that addresses two crucial questions. First, is the cerebellum involved in cognitive operations that do not involve the motor system for learning, planning, or guiding movements? Second, if there is such cognitive cerebellar involvement, is it colocalized to the same region (or regions) involved in movement when movement is required, or is it localized to a separate region within the cerebellum? We used functional magnetic resonance imaging (fMRI) to examine the differential involvement of the human cerebellum in three tasks: (i) a visual attention task that neither required motor learning nor made use of or guided motor operations, (ii) a motor task, and (iii) a task combining these two. We found evidence of a classic double dissociation in structure and function between areas of the cerebellum: Visual attention activates one anatomic location within the cerebellar cortex, whereas motor performance activates a distinctly different location. Moreover, attention activation can occur independently of motor involvement.

Six right-handed, healthy, normal volunteers (three male, three female) ranging in age from 23 to 29 years (mean ± SD, 25.8 ± 2.1 years) participated after informed consent. During the Attention task, circles, squares, or triangles in red, green, or blue were presented one at a time at a single spatial location in the center of foveal vision (17). This task tested the ability to attend selectively to targets (squares or red shapes) within a visual dimension (form or color). Subjects were instructed to silently count each target stimulus, which required attention to visual stimuli in the absence of a motor response. In the Motor task, subjects were instructed to execute repeatedly a self-paced movement of the right hand in the absence of visual stimuli. This movement was then used in the Attention-with-Motor task, which was identical to the Attention task, with one exception: Rather than silently counting target stimuli, subjects were instructed to respond to each target using movement of the right hand. Within each of the three tasks, a task activation condition was alternated with a baseline control condition. As a control for visual sensory stimulation, both the Attention task and the Attention-with-Motor task were alternated with passive visual stimulation, during which the subjects were instructed to observe the same set of visual stimuli but not selectively attend or respond to targets. The Motor task was alternated with rest (18).

During all three tasks, a time series of 128 gradient-echo echo-planar (EPI) images per slice was acquired (19) at five coronal slice locations through the cerebellum. Three slices at comparable locations within the cerebellum for all six subjects were analyzed (Fig. 1A). For each slice in each subject, the number of significantly activated voxels (20) during the three tasks was calculated within two regions of interest (ROIs) (Fig. 1B) defined a priori (21). The location of both ROIs was determined and drawn using standard cerebellar landmarks (22) on a single EPI image of each slice for each subject before the calculation of activations. The two ROIs were collapsed across the three slices to create two volumes of interest (VOIs), an Attention VOI and a Motor VOI. The percent volume active during each task was then calculated for the two VOIs.

Fig. 1.

(A) Approximate positions of slices 1 to 3 used for data analysis, shown on a midsagittal anatomical magnetic resonance (MR) image from a single subject. Slice 1 is the most anterior slice. Intracerebellar landmarks were used to guide the choice of slice positions in order to obtain images from comparable anatomical locations within the cerebellum across subjects. (B) Locations of ROIs (21, 22) shown on an anatomical MR image of slice 1 from a single subject (pf, primary fissure; hf, horizontal fissure). The Attention ROI included the left posterior quadrangular lobule (QuP) and the left superior semilunar lobule (SeS). The Motor ROI included the right anterior vermis (AVe), the right central lobule (C), and the right anterior quadrangular lobule (QuA).

In all subjects, the cerebellum was active during the Attention task, which was performed in the absence of movement or motor planning (23). It was also active during the Motor task, which involved movement but demanded no selective attention. Moreover, the two tasks differentially engaged the Attention and Motor VOIs. The Attention hotspot—the maximally activated voxel in the Attention VOI during the Attention task in each subject (mean r = 0.56; mean percent signal change = 1.58)—was not active during the Motor task [mean r = 0.15; mean percent signal change = 0.38; matched-pairs t5 (comparing r values) = 4.9, P < 0.01]. In contrast, the Motor hotspot—the maximally activated voxel in the Motor VOI during the Motor task in each subject (mean r = 0.58; mean percent signal change = 1.37)—was not active during the Attention task [mean r = 0.16; mean percent signal change = 0.28; matched-pairs t5 (comparing r values) = 7.29, P < 0.01].

A closer look at the time course of activation underlying the above differences reveals the sharp distinction between attention and motor activation in the cerebellum (Fig. 2). At the onset of the Motor task, which was performed in the absence of the visual sensory stimulation used in the Attention task, there was a transient increase in activation in the Attention hotspot (Fig. 2A). This suggests that the initiation of the required simple motor action involved some degree of attention, whereas sustaining the simple actions did not. In contrast, during the Attention task, which was performed without any motor planning or execution, there was no increase in activation in the Motor hotspot (Fig. 2B), suggesting that neither the initiation nor the sustained execution of the Attention task required the use of those cerebellar regions most involved in the Motor task. These results highlight the functional independence of cerebellar activation by attention: Motor activation required attention, but attention activated the cerebellum regardless of whether there was visual sensory input or motor output.

Fig. 2.

Intertask comparisons within the Attention (A) and Motor (B) hotspots. For each hotspot, the time-course signal data for each subject were averaged, collapsed across the four cycles between task activation and baseline control conditions, and plotted in terms of percent change in MR signal (thick line, Attention task; thin line, Motor task).

Repeated-measures analysis of variance (ANOVA) of the percent volume active in the two VOIs during the three tasks (Fig. 3A) resulted in a statistically significant task × VOI interaction (F2,10 = 5.81, P < 0.05). Follow-up comparisons demonstrated that there was significantly greater activation in the Attention VOI during the Attention task (F1,5 = 10.35, P < 0.05), and, conversely, significantly greater activation in the Motor VOI during the Motor task (F1,5 = 6.95, P < 0.05). Both VOIs were activated during the Attention-with-Motor task, and the difference between the two was not significant.

Fig. 3.

(A) Percent (median of six subjects) of the two VOIs activated during each of the three tasks (Motor VOI, striped bars; Attention VOI, solid bars). (B to D) Percent of the two ROIs activated at each slice location during the three tasks. During the Attention task (B), the extent of activation in the Attention ROI (solid bars) was greatest in the posterior slices, falling off in the most anterior slice, whereas the Motor ROI (striped bars) was only 5% active in both slices 1 and 2, with no activation in slice 3. Conversely, during the Motor task (C), the extent of activation in the Attention ROI was minimal, while in the Motor ROI, the extent of activation was greatest in the most anterior slice, falling off in a gradient toward posterior slices. During the Attention-with-Motor task (D), the extent of activation in the two ROIs approximated a summation of results from the other two tasks (for example, the Motor ROI in slice 2 was 5.3% active during the Attention task, 20.2% active during the Motor task, and 25.8% active during the Attention-with-Motor task), with the exception of the Motor ROI in slice 1.

Together, these results reflect a double dissociation between these two areas of the cerebellum with respect to their involvement in visual selective attention and movement. This dissociation is emphasized by the differential extent of activation within the ROIs across slices, with motor activity greatest in the most anterior slice and attention activity greatest in the more posterior slices (Fig. 3, B to D). The dissociation is most clearly demonstrated by the functional maps (Fig. 4) showing the differential neuroanatomical localization of these two distinct types of activity. This double dissociation is of theoretical importance because it shows that the cerebellum is not designed to perform a single neurobehavioral function, such as motor control or attention, but instead is a system composed of different regions that influence distinctly different neurobehavioral functions.

Fig. 4.

Functional maps (32) demonstrating the most common sites of activation across subjects overlaid on an averaged coronal anatomical image of the cerebellum. Yellow, overlap of three or more subjects; blue, any two subjects. (A and B) During the Attention task, the most common site of activation was in the left superior posterior cerebellum [the posterior portion of the quadrangular lobule (QuP) and the superior portion of the semilunar lobule (SeS); approximate Talairach coordinates of center of mass, x = −37, y = −63, z = −22]. (C and D) During the Motor task, the most common site was in the right anterior cerebellum [the anterior portion of the quadrangular lobule (QuA), the central lobule (C), and the anterior vermis (AVe); approximate Talairach coordinates of center of mass, x = 7, y = −51, z = −12]. (E and F) During the Attention-with-Motor task, common sites of activation were in a combination of the areas from the other two tasks, with the addition of the posterior vermis (Pve). SeI, inferior portion of the semilunar lobule; Gr, gracile lobule; other abbreviations are as in Fig. 1.

Such cerebellar influences, though differentially localized, might serve comparable, if not also complementary, goals. For instance, the cerebellum may modulate attention and sensory responsiveness (2, 9, 24) as well as movements that reposition sensory receptors (7, 16) or track the trajectories of sensory information (25). These are complementary preparatory actions that optimize the acquisition and analysis of relevant sensory information during a search for a known and expected target stimulus or during exploration of a novel environment.

Our results demonstrate that such cerebellar preparatory influences can occur independently of motor involvement. In the Attention task, attention to sensory information alone was sufficient to activate the cerebellum, and engagement of the motor system was not necessary to produce cerebellar activation. Cerebellar attention activation occurred even though no motor learning was required; no motor response selection, error detection, or error correction was required; no imagined motor action was required; and no guidance of motor systems was required. In sum, these findings are contrary to the expectation of traditional theories of the cerebellum as a motor control system (6). These findings, in concert with other evidence about the cerebellum [for example, its homogeneous anatomical structure (26), its widespread connections with virtually all levels of the CNS (26), and its apparent involvement in a wide range of neurobehavioral functions (2, 714, 16, 21, 2325, 27, 28)], highlight the need for a new conception of cerebellar function.

Such a conception, consistent with our findings, is provided by a recent hypothesis (2, 9, 29) that the cerebellum influences a variety of neurobehavioral systems—including sensory (24), motor (7, 16, 25), attention, and other cognitive and noncognitive systems—in order to accomplish its prime function, which is to learn to predict and prepare for imminent information acquisition, analysis, or action (30). As would be predicted by this hypothesis, cerebellar activation has been reported to be highest in the early stages of learning novel information, responses, or skills, or when nonmotor and motor sequences of information that are difficult to predict (such as the random sequences used in our experiment) must be processed (27, 28). Successful anticipation of imminent real-time events improves the potential for the effective and timely directing of cognitive and noncognitive resources (31) to facilitate the learning of new information; it also improves the rapid, accurate, and effortless coordination of previously learned cognitive and noncognitive operations. We suggest that the human cerebellum may play a key role in the learning and smooth coordination of such anticipatory operations.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
View Abstract

Navigate This Article