Continuous manipulation of mental representations is compromised in cerebellar degeneration

Abstract

We introduce a novel perspective on how the cerebellum might contribute to cognition, hypothesizing that this structure supports dynamic transformations of mental representations. In support of this hypothesis, we report a series of neuropsychological experiments comparing the performance of individuals with degenerative cerebellar disorders on tasks that either entail continuous, movement-like mental operations or more discrete mental operations. In the domain of visual cognition, the cerebellar disorders group exhibited an impaired rate of mental rotation, an operation hypothesized to require the continuous manipulation of a visual representation. In contrast, the cerebellar disorders group showed a normal processing rate when scanning items in visual working memory, an operation hypothesized to require the maintenance and retrieval of remembered items. In the domain of mathematical cognition, the cerebellar disorders group was impaired at single-digit addition, an operation hypothesized to primarily require iterative manipulations along a mental number-line; this group was not impaired on arithmetic tasks linked to memory retrieval (e.g. single-digit multiplication). These results, obtained in tasks from two disparate domains, point to a potential constraint on the contribution of the cerebellum to cognitive tasks. Paralleling its role in motor control, the cerebellum may be essential for coordinating dynamic, movement-like transformations in a mental workspace.

Keywords: cerebellum; mental imagery; mental operations; spinocerebellar ataxia.

Figure 1

Visual cognition tasks employed in Experiments 1a and b. In the mental rotation experiment, participants judged if a letter stimulus was normal (e.g. ‘R’, right key press) or mirror-reflected (e.g. ‘Я’, left key press). On most trials, the stimulus was rotated relative to the upright orientation (depicted example involves 135° rotation). The same mental rotation task was used in Experiments 1a and b for the continuous condition. Two memory search tasks were used for the non-continuous, control conditions. In the visual memory search task (Experiment 1a), a sequence of stimuli (1–5 images) was presented (1 s per image). After a maintenance period (3 s), a probe stimulus appeared, and the participant judged whether it was a member of the memory set (right key) or not (left key). Sequences varied in length from one to five items. In the visuospatial working memory task (Experiment 1b), a sequence of circles (2–5 items) was presented at random locations on a ring (1 s per target). After a maintenance period (2 s), a probe stimulus was presented, and the participant indicated the ordinal position of the probe. Responses in all tasks were followed by feedback (1 s), and a 2 s inter-trial interval.

Figure 2

Reaction time analysis for Experiment 1a. Cerebellar degeneration is associated with a slower rate of mental rotation but does not impact the rate of search through visual working memory. (A) Median RT as a function of stimulus orientation in the mental rotation task for the CD group and the control group. (B) Estimated rate of rotation from the regression analysis (slope of RT function). (C) Median RT as a function of set size in the visual memory search task. (D) Estimated search rate from the regression analysis. Mean regression lines are displayed in A and C. Shaded error bars denote 1 SEM. n.s. = not significant. *P < 0.05.

Figure 3

Reaction time analysis for Experiment 1b. Replication of selective impairment on the continuous transformation condition. In the mental rotation task (A and B), the CD group showed slower mental rotation speeds relative to controls. Mental rotation rates are plotted with the absolute rotation magnitude of the stimulus on the x-axis, and the median change in RTs for each rotation condition on the y-axis. In the search task (C and D), the two groups showed comparable spatial memory search speeds. The length of the test sequence (set size) is plotted on the x-axis, and the median change in RTs for each set size on the y-axis. Mean regression lines are displayed in A and C. Shaded error bars denote 1 SEM. n.s. = not significant. *P < 0.05.

Figure 4

Mathematical cognition tasks employed in Experiments 2a and b. Participants made a speeded response to verify whether the equation was true or false. Addition problems (top row, tested in Experiments 2a and b) and multiplication problems (bottom row, tested only in Experiment 2b) involved either non-identical operands (left column) or identical operands (right column). Non-identical addition constitutes the continuous condition, whereas the remaining three conditions constitute non-continuous conditions.

Figure 5

Reaction time analysis for Experiment 2a. Cerebellar degeneration associated with slower rate for addition problems putatively linked to continuous transformations along a number line, but not problems assumed to depend on memory retrieval. Median RT is plotted as a function of the maximum operand for equations with non-identical (A) and identical operands (C). Thin lines denote functions for each participant. The rate of adding non-identical (C) or identical operands (D) is estimated by the slope of each individual’s RT function. Dots denote individuals. Mean regression lines are displayed in A and C. Shaded error bars denote 1 SEM. n.s. = not significant. *P < 0.05.

Figure 6

Reaction time analysis for Experient 2b. Replication of selective impairment in CD group on arithmetic problems associated with continuous transformations along a number line. Median RT is plotted as a function of the maximum operand for addition (A) and multiplication (C), using equations with non-identical operands only. Thin lines denote functions for each participant. The rate for addition (C) or multiplication (D), is estimated by the slope of each individual’s RT function. Dots denote individuals. Mean regression lines are displayed in A and C. Shaded error bars denote 1 SEM. n.s. = not significant. *P < 0.05.