Trial-to-trial Adaptation: Parsing out the Roles of
Cerebellum and BG in Predictive Motor Timing
Ovidiu V. Lungu1,2,3,4,5, Martin Bares1,2,6,7, Tao Liu1,2, Christopher M. Gomez8,
Ivica Cechova6, and James Ashe1,2
Abstract
■ We previously demonstrated that predictive motor timing
(i.e., timing requiring visuomotor coordination in anticipation
of a future event, such as catching or batting a ball) is impaired
in patients with spinocerebellar ataxia (SCA) types 6 and 8 rel-
ative to healthy controls. Specifically, SCA patients had difficul-
ties postponing their motor response while estimating the target
kinematics. This behavioral difference relied on the activation of
both cerebellum and striatum in healthy controls, but not in cer-
ebellar patients, despite both groups activating certain parts of
cerebellum during the task. However, the role of these two key
structures in the dynamic adaptation of the motor timing to target
kinematic properties remained unexplored. In the current paper,
we analyzed these data with the aim of characterizing the trial-by-
trial changes in brain activation. We found that in healthy controls
alone, and in comparison with SCA patients, the activation in
bilateral striatum was exclusively associated with past successes
INTRODUCTION
Precise timing is essential in everyday life, as it is an inte-
gral part of many motor skills and behaviors (Hore &
Watts, 2005; Iacoboni, 2001) and its disruption in a num-
ber of neurological disorders can have a devastating ef-
fect on those afflicted with the condition (Harrington &
Haaland, 1999). Studies of motor timing have shown that
this important function is controlled primarily by the cere-
bellum and BG (Ivry & Spencer, 2004; Matell & Meck, 2004;
Mauk & Buonomano, 2004; Gibbon, Malapani, Dale, &
Gallistel, 1997; Ivry & Keele, 1989); however, most of these
experiments employed rhythmic activity, time estimation,
or time production paradigms. Although this particular
type of motor timing may be well suited to investigate
the neural structures implicated in temporal processing
1 Luo, Chang, & Li, 2009; Aron & Poldrack, 2005; Mostofsky
Veterans Affairs Medical Center, Minneapolis, MN, 2University
of Minnesota, 3Université de Montréal, 4Research Center of the
Montreal Geriatric Institute affiliated with the Université de
Montréal, Montréal, 5Donald Berman Maimonides Geriatric
Centre, Montréal, Québec, Canada, 6Masaryk University, Brno,
Czech Republic, 7St. Anne’s Hospital, Brno, Czech Republic,
8
University of Chicago
© 2016 Massachusetts Institute of Technology
and that in the left putamen, with maintaining a successful perfor-
mance across successive trials. In healthy controls, relative to SCA
patients, a larger network was involved in maintaining a successful
trial-by-trial strategy; this included cerebellum and fronto-parieto-
temporo-occipital regions that are typically part of attentional
network and action monitoring. Cerebellum was also part of a net-
work of regions activated when healthy participants postponed
their motor response from one trial to the next; SCA patients
showed reduced activation relative to healthy controls in both
cerebellum and striatum in the same contrast. These findings
support the idea that cerebellum and striatum play complemen-
tary roles in the trial-by-trial adaptation in predictive motor
timing. In addition to expanding our knowledge of brain struc-
tures involved in time processing, our results have implications
for the understanding of BG disorders, such as Parkinson disease
where feedback processing or reward learning is affected. ■
alone, most everyday motor activities actually require pre-
dictive timing: visuomotor coordination in response to
anticipated changes of stimuli in the environment (i.e.,
catching or hitting a ball, grabbing moving objects, turning
the steering wheel, etc.). In this context, predictive motor
timing can be envisioned as having two components, the
temporal processing per se and holding or postponing
the motor response until the appropriate moment to act.
To date, postponement of motor responses has been a
measure of interest mainly in response inhibition para-
digms (e.g., suppression of a prepotent response tenden-
cy), using go/no-go or Stroop-like tasks. Some of these
studies, which included healthy controls or patients with
movement disorders, have found striatum as well as cere-
bellum to be involved in response inhibition in different
capacities ( Vink, Kaldewaij, Zandbelt, Pas, & du Plessis,
2015; Ye et al., 2015; van der Salm et al., 2013; Chao,
et al., 2003). Typically, striatal activation was related to the
actual response inhibition and expectations about the
probability of a stop signal to occur ( Vink et al., 2015;
van der Salm et al., 2013; Aron & Poldrack, 2005), whereas
cerebellar activation was mainly associated with “go” trials
(Mostofsky et al., 2003). However, in most studies on
Journal of Cognitive Neuroscience 28:7, pp. 920–934
doi:10.1162/jocn_a_00943
response inhibition, participants react to external cues
(“go” or “no-go” signals), whereas in predictive motor
timing, these “go/no-go” cues are self-generated. Thus,
there is a scarcity of studies investigating the neural sub-
strate of predictive motor timing with a focus on the spe-
cific roles played by cerebellum and BG, in particular.
There is a debate in the literature as to whether cere-
bellum and striatum are involved in timing of short or
long intervals, as well as in continuous or discontinuous
tasks. For instance, both cerebellum and BG were shown
to be associated with the perception of short temporal
intervals ( Jahanshahi, Jones, Dirnberger, & Frith, 2006;
Ferrandez et al., 2003; Nenadic et al., 2003; Harrington,
Haaland, & Hermanowicz, 1998; O’Boyle, Freeman, &
Cody, 1996). However, the striatum is also known to be
activated during processing of long temporal intervals
(Meck, 2005; Meck & Benson, 2002). Thus, although
the role of cerebellum seems confined almost exclusively
to the estimation of millisecond to second-range inter-
vals, the BG are engaged in the neural control of a wider
range of time (e.g., seconds), perhaps due to its strong
connections with the frontal regions involved in higher
cognitive processes such as attention, memory, control
(Coull, Cheng, & Meck, 2011).
Indeed, in a previous study (Bares et al., 2011), we
found that both striatum and cerebellum were necessary
in a predictive motor timing task that required temporal
processing over a period of several seconds. However,
given that predictive motor timing requires both tempo-
ral processing and postponement of motor response
until the right time to act, one cannot dissociate the ac-
tivation due to temporal processing from that due to
postponement of the motor response per se, if only
the current trials are taken into account in the analysis.
In contrast, if a trial-by-trial analysis is employed, account-
ing for changes in brain activity in the current trial as a
function of the outcome in the immediately preceding
trial, the roles of the cerebellum and BG in predictive
motor timing could be better teased apart. To do so, in
the current paper, we used the neuroimaging data col-
lected in a previous study (Bares et al., 2011), and we con-
ducted a new analysis, which focused on the trial-by-trial
changes in the brain activity as healthy controls and SCA
patients performed a predictive motor timing task. We
hypothesized that changes in striatum would reflect both
the cognitive requirements of the task as well as postpone-
ment of motor action, whereas cerebellum would be
involved primarily in correct temporal estimation.
METHODS
Participants
For this study, patients diagnosed as having a genetically
defined spinocerebellar ataxia (SCA) type 6 (n = 5) or
type 8 (n = 4) were recruited from the ataxia outpatient
clinic at the University of Minnesota (five women, mean
age = 49.1 years, SD = 10.7 years), together with 12
healthy controls (HC; age-matched) with no neurolog-
ical or other health problems (five women, mean age =
51.5 years, SD = 12.5 years). The patients (Table 1) with
SCA types 6 and 8 have clinical syndromes involving
predominantly the cerebellum and are well characterized
both clinically and genetically (Maschke et al., 2005). Before
testing, the clinical status of the patients was scored on the
International Cooperative Ataxia Rating Scale of the World
Federation of Neurology (Trouillas et al., 1997). All partici-
pants were right-handed (Oldfield, 1971) and did not have
clinically apparent depression on the Montgomery and
Asberg Depression Rating Scale (Montgomery & Asberg,

Table 1. Detailed Description of the SCA Patients

WFN a
Disease
Age SCA Duration Walking Total Posture, Limb Oculomotor
No. (years) Sex Type (years) Aid Score Gait Ataxia Speech Dysfunction

1 63 F 6 11 Walker 50 17 23 5 5
2 43 F 6 15 Cane 41 16 16 5 4
3 30 F 8 11 Cane 25 10 7 4 4
4 43 M 8 10 None 28 11 11 3 3
5 47 M 8 20 Chair 54 27 18 4 5
6 45 M 8 15 Walker 42 20 13 4 5
7 52 F 6 11 None 13 4 5 2 2
8 62 M 6 8 None 23 7 8 6 2
9 55 F 6 11 Cane 24 14 3 2 5

M = male; F = female; SCA = spinocerebellar ataxia; WFN = World Federation of Neurology score.
a
Range 0–100. The higher the total score, the more severe ataxia (Trouillas et al., 1997).

Lungu et al. 921

1979). None of the participants had a history of color blind-
ness or evidence of cognitive decline. All participants were
remunerated for their participation after giving informed
consent. The study was approved by the institutional
review board of the University of Minnesota.

Behavioral Task
We employed the same experimental paradigm as that re-
ported elsewhere (Bares, Lungu, Husarova, & Gescheidt,
2010; Bares et al., 2007). Specifically, participants per-
formed a predictive motor timing task in which they
had to respond via button-press to intercept a circular
target that moved from left to right on a computer
screen. Their action launched a small fireball from a virtual
cannon located on the lower right side of the screen. The
fireball moved upward, at constant speed (20.0 cm/sec),
and served to intercept the moving target (Figure 1A).
The objective was to launch the fireball at the optimal
time to hit the target when it passed through the inter-
ception zone (not visible to the participants), which was
always in the same location on the upper right side of the
screen. If the participant were successful, both the fire-
ball and target appeared to explode (the red dots in
Figure 1B); if the participant failed to intercept the target,
no explosion animation occurred. The diameter of the
target was 1 cm, and the diameter of the fireball was
0.3 cm. The target was launched from the left side of
the screen at three different angles of movement relative
to the horizontal plane of the screen (0°, 15°, and 30°)
and followed a linear trajectory (the dotted yellow lines
in Figure 1A, not visible to the participants) toward the
interception zone. The target could have one of three dif-
ferent movement types, constant velocity, deceleration,
and acceleration, and three different travel speeds, slow
(complete travel time across the screen: 3.5 sec), me-
dium (complete travel time across the screen: 3.0 sec),
and fast (complete travel time across the screen: 2.5 sec).
Figure 1. The prediction motor timing task. (A) A target ball (green) was
The target movement was thus characterized by any com-
launched from the left side of the screen on one of the three linear
bination of the three variables (type, speed, angle), giving a trajectories (yellow dotted lines, not visible to the participants) at
total of 27 different kinematic combinations. various speeds, angles, and acceleration rates toward a “target” zone
For the actual experiment, blocks of trials were orga- (the intersection point of the three trajectories). Participants launched a
nized on the basis of movement type (either constant, cannonball (lower right side of the screen) by pressing a button. The
cannonball traveled upward at a constant speed. The goal was to intercept
acceleration, or deceleration) within which the combina-
the target ball in the target zone. (B) If the action was correctly timed, the
tion of speed and angle was randomized. There were six cannonball intercepted the target ball and both exploded (red dots).
blocks of trials, two for each movement type, containing (C) The distribution of trial types by blocks during the motor timing task.
54 trials bounded by 20-sec break periods. The total du- There were six blocks, two for each type of movement (accelerated,
ration of the interception task was 18 min 27 sec. Each constant, decelerated) presented in pseudorandom order across
participants. Within each block, there were 54 trials resulting in six
combination of speed and angle was presented six times
presentations of each of the nine trial types (originating from all
during each block of a given movement type, for a total combination of the three types of speeds and three angles).
of 12 presentations of a particular trial type (movement
type, speed, and angle) during the entire experiment.
Before the imaging session, participants were given front of a computer monitor for the practice block; dur-
one block to practice the task. This block had 54 trials ing scanning, they performed the task lying supine in the
during which each combination of a particular trial type scanner, seeing the screen through a mirror mounted on
was presented twice and the total duration of the prac- the coil and pressing the button on an MR-compatible
tice was 3 min 26 sec. Participants were seated 60 cm in button box.

922 Journal of Cognitive Neuroscience Volume 28, Number 7

Analysis of Behavioral Data
The dependent variable in behavioral analysis was the out-
come of each individual trial, which was binary (Hit vs.
Error); therefore, we could not use parametric methods
to analyze the behavioral results. As such, we employed
nonparametric test designed to handle nominal data,
such as test of proportion or chi-square to compare the
distribution of trial types among each group and the dif-
ference between groups. To account for inflated chi-
square values due to the high number of trials, we used
Phi and Cramer’s V corrections. For each trial type, partic-
ipants were successful in their attempt provided that they
had pressed the button during an optimal temporal win-
dow while the target was moving on the screen. The lower
bound of this temporal window varied between 605 and
1924 msec and its upper bound varied between 789 and
2100 msec from the trial onset, depending on the trial
type. The median optimal hitting point (middle point of
the temporal window), for all trials, was at 1467 msec from
the trial onset. Whenever participants pressed the button
too early or too late relative to the boundaries of the tem-
poral window, their action resulted in an error. Thus, trials
were classified as hits, early errors, and late errors based
on whether participants pressed the button within, be-
fore, or after this temporal window, respectively. Trials
in which participants did not press the button by the time
the target cleared the screen were classified as omissions.
To demonstrate that participants actively adapted their
strategy on a trial-by-trial basis, we performed the following
steps. First, we calculated the middle point in the optimal
time window for each trial type. Then, for each trial, we
computed the difference between the actual RT and this
middle point, and we expressed this difference in percent-
age relative to the window’s temporal width. As such, suc-
cessful trials (hits) had values between −50% and +50%
(depending on whether the button press occurred before
or after the middle point in the optimal time window); con-
versely, early errors had values below −50% and late errors,
above +50%. This variable allowed for the quantification of
the magnitude of deviation from the “optimal hit-point” in-
dependent of the target parameters, primarily the speed
(i.e., the width of the optimal window was smaller for the
trials in which target moved fast). Next, a discrete variable
with five categories was used to classify the speed change
across successive trials: extreme decrease (from 3.5 to
2.5 sec), average decrease (from 3.5 to 3 sec or from 3
to 2.5 sec), no change (same speed), average increase
(from 2.5 to 3 sec or from 3 to 3.5 sec), and extreme
increase (from 2.5 to 3.5 sec) in speed.
Similarly, we computed a discrete variable with five cat-
egories representing the magnitude of change in target’s
trajectory angle across successive trials. Finally, evidence
for the trial-by-trial strategy was obtained using random-
effects GLM models with the speed or angle change
category and group as independent factors and the devi-
ation from the optimal point as the dependent variable.
In addition, the same dependent variable was employed
in a random-effects GLM with the group, current and pre-
vious trial type (based on outcome: hit, early error, and
late error) as independent factors.
Imaging Parameters of the fMRI Experiment
All participants were scanned during the task using a 3-T
Whole Body MR System (MAGNETOM Trio, Siemens
Medical Systems, Erlangen, Germany). Before the fMRI
run, 144 or 160 (depending on the participant’s head
width) FLASH structural images were acquired in slices
of 1-mm thickness in sagittal plane (256 × 256 mm),
yielding a spatial resolution of 1 × 1 × 1 mm for the an-
atomical volume. The imaging parameters of the anatom-
ical sequence had a repetition time and echo time of 13
and 4.92 msec, respectively, with a flip angle of 25°. Then,
whole-brain fMRI was performed using an EPI sequence
measuring BOLD signal. A total of 30 functional slices per
volume were acquired for all participants in all runs. These
slices, which were acquired in transversal plane with a 30°
positive tilt around y axis (to cover the whole brain and
the cerebellum), were in ascending order and inter-
leaved. They had a thickness of 3 mm, an in-plane reso-
lution of 3 × 3 mm (matrix size = 64 × 64), and a field of
view of 192 × 192 mm, with a 1-mm gap between them
to avoid cross-talking. A complete scan of the whole brain
was acquired in 2000 msec (repetition time); the flip an-
gle was 75°, and echo time was 30 msec. A total number
of 556 volumes were acquired during the functional run.
Preprocessing of fMRI Data
Brain Voyager QX (Brain Innovation B.V., Maastricht, the
Netherlands) software was used for fMRI data preprocess-
ing and analysis. The functional bidimensional images of
every participant were preprocessed to correct for the dif-
ference in time slice acquisition (slice scan time correc-
tion). In addition to linear detrending, a high-pass filter
of three cycles per time course (frequency domain) was
applied to the corrected 2-D slices. Then, the functional
series was preprocessed to correct for possible motion
artifacts in any plane of the tridimensional space and to en-
sure that the movements in any plane did not exceed
3 mm. Later, these functional images were then used to re-
construct the 3-D functional volume for every participant
and every run, which was aligned with the corresponding
3-D anatomical volume, and both were normalized to stan-
dard Talairach space (Talairach & Tournoux, 1988). Finally,
spatial smoothing using a Gaussian kernel at 7 mm FWHM
was applied to the 3-D functional data.
Analysis of fMRI Data
We used a rapid event-related block design for our exper-
iment. Each behavioral event, which lasted between 2.5
and 3.5 sec (corresponding to the maximum travel time
of the moving target across the screen), was classified
Lungu et al. 923
according to the behavioral outcome (hit, early error, and
late error) in both the current and previous trials. As
such, we had nine events (early error after early error,
hit after early error, late error after early error, early error
after hit, hit after hit, late error after hit, early error after
late error, hit after late error, and late error after late er-
ror). These events describe all possible combinations of
behavioral outcomes from one trial to the next. These
predictors were entered as fixed factors in single partici-
pant GLM; then, the parameter estimates of this GLM
model were subsequently entered into a second level
of analysis corresponding to a random-effect GLM model
that was used for group analysis (Penny & Holmes, 2003).
The statistical parameters of this latter model were esti-
mated voxelwise for the entire brain, and activation maps
were computed for various contrasts between the predic-
tors. In analyzing the neuroimaging data, we first charac-
terized brain activation patterns for various statistical
contrasts only for the healthy individuals, and then we
performed group comparisons using the same contrasts.
We were interested in the following contrasts: (1) brain
activation during current trial as a function of outcome in
the previous trial (hit vs. error); (2) brain activation dur-
Figure 2. (A) The proportion of hits, early errors, and late errors for the two groups in the current trials. (B) The proportion of hits, early errors, and
late errors in the current trials as a function of previous trial type for the two groups.
924 Journal of Cognitive Neuroscience
ing successive successful trials versus successive errors
(hits followed by hits vs. errors followed by errors, of
the same kind), and (3) brain activation during hits and
late errors versus the early errors in the current trials
when they followed early errors. The first contrast will
reveal the impact of immediate past outcome on the cur-
rent brain activity; the second contrast will highlight brain
regions involved in maintaining a successful strategy; the
last contrast will highlight the brain activation related to
the postponement of the motor response from one trial
to the next, regardless of trial duration. When displaying
the activation maps corresponding to these contrasts for
the HC group only, we employed the false discovery rate
(q < 0.05) correction, for the whole brain, at the millime-
ter isotropic resolution and a minimum cluster size of
108 adjacent significant voxels (corresponding to a
volume of 108 mm3). For group differences, we displayed
the activation map using the same cluster size and a
statistical threshold for each voxel in the cluster of at least
p < .001 (uncorrected), adjusted to correspond to p <
.05 family-wise error. It is worth noting that, in most
cases, this threshold turned out to be more conservative
than the false discovery rate of q < 0.05. In addition, we
Volume 28, Number 7
also conducted additional detailed ROI analyses in vari-
ous clusters of interest that we obtained as a result
of the above-mentioned contrasts. For this, we used a
GLM random-effects analysis with group and type of trial
as independent factors and the BOLD signal expressing
the magnitude of the contrast of interest averaged over
all voxels in the ROI as dependent measure. To this
end, we employed the SPSS (Statistical Package for Social
Sciences, SPSS, Inc., Chicago, IL) software.
RESULTS
Behavioral Results
As reported previously, SCA patients performed worse
than healthy controls (Figure 2A) as revealed by both
the proportion and chi-square tests (Bares et al., 2011).
Not only was the proportion of hits significantly lower
in patients than in healthy controls (χ2 = 321.63, df =
1, N = 6804; Phi = 0.21, and Cramer’s V = 0.21; p <
.001), but the patients had a smaller proportion of hits
than early (χ2(1) = 104.40, p < .001) and late errors
(χ2(1) = 16.83, p < .001). In contrast, healthy controls
had a higher proportion of hits than any of the other
two error types (χ2(1) = 191.51, p < .001 for early errors
and χ2(1) = 293.45, p < .001 for late errors).
The trial-by-trial performance of the two groups is
shown in Figure 2B. The proportion and chi-square tests
performed both within each group as well as comparing
the two groups revealed that healthy controls had a sig-
nificantly higher proportion of hits in the current trials
than SCA patients, regardless of the outcome in the
previous trial (χ2(1) = 78.98, p < .001 after early errors;
χ2(1) = 137.15, p < .001 after hits and χ2(1) = 84.72, p <
.001 after late errors). This suggests that the performance
difference between healthy controls and SCA patients re-
ported earlier (Bares et al., 2011) is a very stable one, and
it does not depend, at least for healthy controls, on the
outcome in the previous trial. In addition, the healthy
controls were more likely to succeed than to make either
an early error or a late error regardless of the previous
trial type trial (all ps < .01). In contrast, the SCA patients

Figure 3. (A) The change in deviation from the optimal hitting point across successive trials (red and blue bars, left vertical axis) and average hit ratio
(dotted lines, right vertical axis) as a function of change in target speed. (B) The change in deviation from the optimal hitting point across successive
trials as a function of outcomes in the current and preceding trials, presented separately for each of the two groups.

Lungu et al. 925

Figure 4. Brain activation comparing current trials preceded by hits versus errors in HC only and in the comparison HC > SCA. The bar graphs
represent the change in BOLD signal (mean; SEM ) for each trial type and group in clusters from the left and right putamen.

tended to produce more early errors than hits regardless
of previous trial type; early errors were more frequent
than hits after hits and late errors (all ps < .01), but these
proportions were the same when the previous trial was
an early error (χ2(1) = 2.75, p = .1).
The random-effects GLM assessing the change in devi-
ation from the optimal hitting point revealed a significant
main effect for the change in speed, F(4, 76) = 109.57,
p < .001, as well as an interaction effect Group × Change
in speed, F(4, 76) = 24.88, p < .001 (red and blue bars

926 Journal of Cognitive Neuroscience Volume 28, Number 7

in Figure 3A). This indicated that the more extreme the
change in speed across successive trials, the larger the
change in deviation from the optimal hitting point in
the current trial. In addition, this effect had a significantly
larger magnitude among SCA patients than healthy con-
trols. Although this result reflects participants’ trial-to-trial
adaptation to the change in speed, it does not inform us
about the success of the adaptation strategy. In this con-
text, the same GLM model applied to hit ratios revealed
a significant interaction between group and hit ratio, F(4,
76) = 2.59, p < .05 (the points and dotted lines in
Figure 3A). This suggests that healthy controls main-
tained their successful strategy despite the change in
speed from one trial to the next (e.g., successful adapta-
tion), whereas SCA patients performed significantly
worse as the speed of the target ball decreased across
successive trials. This apparently counterintuitive finding
reflects a deficit in SCA capacity to postpone their action
long enough to obtain a hit.
Regarding the change in angle across successive trials,
the GLM analysis revealed a significant main effect for an-
gle change, F(4, 76) = 4.41, p < .005, and a significant
Group × Angle change interaction effect, F(4, 76) =
2.72, p < .05. This indicates that, for healthy controls,
changes in angle did not affect the change in deviation
from the optimal hitting point in the current trial, whereas
SCA patients tended to anticipate or delay their reaction
depending on whether the angle decreased or increased
from one trial to the next, respectively. However, the hit
ratio within each group and the group differences in hit
ratio were not significantly affected by the angle change
across successive trials [F(4, 76) = 2.02, p = .1 for the main
effect of angle and F(4, 76) = 0.84, p = .51 for the Group ×
Angle change interaction effect].
The GLM analysis with group, current and previous
trial type (hit, early error, and late error) as independent
factors revealed that, for both groups, the change in de-
viation from the optimal hitting point in the current trial
depended on the outcome in both the previous and cur-
rent trials, F(4, 66) = 4.45, p < .005 (Figure 3B). Specif-
ically, the magnitude of the deviation from the optimal
hitting point increased as the discrepancy in outcome
across successive trials increased (i.e., from early to late
errors or vice versa; see Figure 3B), in line with the find-
ings related to the changes in speed alone.
In summary, these behavioral results indicate not only
that the two groups are different in terms of their general
performance but also that their trial-by-trial strategy is dif-
ferent: Specifically, the SCA patients tend to make more
early errors than the other type, suggesting that they have
trouble suppressing or holding their motor response long
enough to improve their task performance.
Imaging Results
Given that in the neuroimaging data analysis we relied
on the natural jitter between different event types, cre-
ated by the variability in participants’ performance (e.g.,
occurrence of hits, early and late errors), it was important
to assess the magnitude of the ISI between events. On

Table 2. Brain Activation during Current Trials When It Was Preceded by a Hit versus an Error

Activated Regions Talairach Coordinates (Peak of Activation) Volume (mm3) Maximum t Value
Healthy Controls Only (df = 11)
Right putamen 21 8 4 2967 10.52
Right parahippocampal gyrus (BA 35) 18 −16 −8 332 9.79
Right anterior cerebellum (culmen) 3 −37 4 177 8.15
Left precuneus (BA 7) 0 −76 49 231 8.37
Left precentral gyrus (BA 6) −18 −19 67 173 9.33
Left putamen −18 8 1 3093 9.06

Healthy Controls vs. SCA Patients (df = 19)

Right premotor cortex (BA 6) 63 5 16 189 5.71
Right inferior parietal lobule (BA 40) 42 −46 61 227 4.41
Right putamen 24 5 19 2257 5.54
Right superior parietal lobule (BA 7) 18 −52 64 244 4.98
Left precuneus (BA 7) 0 −70 55 2086 6.40
Left putamen −27 2 −2 2578 6.14
Left superior parietal lobule (BA 7) −21 −58 64 202 4.61

Lungu et al. 927

Figure 5. Brain activation
comparing current trials in
successive hits versus
successive errors of the same
kind (early after early and late
after late errors, respectively) in
HC group only and in the
comparison HC > SCA. The bar
graphs represent the change in
BOLD signal (mean; SEM ) for
each trial type and group in the
cluster from the left putamen.

average, the ISI for all participants and all trials (grouped
by trial type, 27 categories) was 28.67 sec, with a median
of 16.5 sec.
Bilateral anterior putamen was strongly activated in
healthy controls whenever the previous trial was a hit,
regardless of the outcome in the current trial (Figure 4;
Table 2). The same contrast, but comparing the two
groups, revealed overlapping activation in the same loca-
tion in bilateral putamen (Figure 4; Table 2). In both of
these clusters, the brain activation was significantly higher
in healthy controls than in SCA patients in the current trial,
when the previous trial was a hit, regardless of the out-
come in the current trial [F(1, 19) = 28.58, p < .001;
η2 = 0.60, observed power = 0.92 for right putamen
and F(1, 19) = 33.34, p < .001; η2 = 0.64, for a post hoc
power = 0.99 for left putamen] (bar graphs in Figure 4).
This finding suggests that, in healthy individuals, the puta-
men acts as a relay of past successes.

928 Journal of Cognitive Neuroscience Volume 28, Number 7

 Mastering the predictive motor timing task requires
one to maintain a successful performance in the current
trial after a hit in the previous trial. Comparing the brain
activity during a successful (hits after hits) versus a failed
strategy (errors after errors, of the same kind) yielded
activation in left putamen (lentiform nucleus) both when
considering only then healthy controls as well as when
comparing them with the SCA patients (Figure 5;
Table 3). A detailed ROI analysis in the left putamen clus-
ter (bar graph in Figure 5) indicates that the difference
between the two groups is specific to trials where a suc-
cessful strategy was maintained (i.e., hits after previous
hits) and not to trials where the motor strategy was main-
tained but resulted in a failure (successive errors of the
same kind) [F(2, 38) = 9.54, p < .001; η2 = .33; observed
power = 0.97; for the Group × Trial type interaction].
However, the same contrast performed in the healthy
controls group only revealed activation in a vast network
including bilateral striatum, cerebellum, and many froto-
parieto-temporo-occipital regions that are typically
considered to be part of the attentional and action mon-
itoring networks (Table 3). A detailed ROI analysis in-
dicated that in the right putamen and left dentate
nucleus there were significant Group × Trial type inter-
action effects [F(2, 38) = 5.79, p < .01; η2 = .23; ob-
served power = 0.4; for right putamen and F(2, 38) =
3.80, p < .05; η2 = .17; observed power = 0.66; for left
dentate]. These effects did not have a large enough mag-
nitude to surpass the statistical threshold imposed brain-
wise when comparing the two groups.
Finally, we found significant group differences regard-
ing the postponement of motor response from one trial
to the next in the right posterior putamen and right
anterior cerebellar lobe (Figure 6; Table 4), but only

Table 3. Brain Activation Comparing the Activity in the Current Trials in Successive Hits with that in Successive Errors of the Same
Kind (Early after Early and Late after Late Errors, Respectively)
Activated Regions Talairach Coordinates (Peak of Activation) Volume (mm3) Maximum t Value

Healthy Controls Only (df = 11)

Right posterior cerebellum (declive) 24 −61 −20 3434 7.97
Right inferior occipital gyrus (BA 18) 36 −82 −5 2794 9.34
Right middle temporal gyrus (BA 39) 30 −61 22 372 7.34
Right middle occipital gyrus (BA 19) 30 −79 10 143 5.44
Right putamen 18 8 4 2590 9.52
Right thalamus 24 −16 16 201 7.71
Right inferior occipital gyrus (BA 18) 27 −88 −14 112 5.81
Right anterior cingulate cortex (BA 32) 21 29 22 350 10.55
Right mid-cingulate gyrus (BA 24) 9 −4 49 121 5.84
Left posterior cingulate gyrus (BA 31) 0 −34 34 3476 7.13
Left posterior cingulate gyrus (BA 30) −9 −52 7 901 7.93
Left paracentral lobule (BA 31) 0 −19 43 794 6.60
Left anterior cerebellum (dentate nucleus) −15 −55 −26 167 6.00
Left putamen −24 2 −2 3412 9.90
Left medial frontal gyrus (BA 8) −18 26 40 141 8.33
Left middle occipital gyrus (BA 18) −27 −82 −2 504 6.45
Left precuneus (BA 7) −24 −73 40 241 6.29
Left fusiform gyrus (BA 19) −21 −61 −5 1317 7.32
Left anterior cerebellum (culmen) −30 −34 −20 144 7.57
Left middle temporal gyrus (BA 39) −36 −73 10 646 7.94

Healthy Controls vs. SCA Patients (df = 19)

Left putamen −27 2 −2 396 7.75

Lungu et al. 929

Figure 6. Brain activation comparing current hits and late errors versus early errors following early errors in HC group only and in the comparison
HC > SCA. The bar graphs represent the change in BOLD signal (mean; SEM ) for each trial type and group in clusters from the right cerebellum
and putamen.
the cerebellar cluster surpassed the statistical threshold
when considering only the healthy controls. In all regions
found when comparing the groups, healthy controls had
a higher activation level than SCA patients ( p < .001). A
detailed analysis of the BOLD signal in cerebellar and
striatal clusters showed that, indeed, these regions are
specifically activated in healthy individuals whenever
the motor response was postponed for a longer duration
(i.e., transition from early errors to hits or late errors) as
compared with a shorter one (i.e., from a previous hit or
late error to a current early error) [F(1, 19) = 5.44, p <
.001 for the Group × Current trial type interaction for
right putamen cluster and F(1, 19) = 15.51, p < .001
for right cerebellum]. This finding indicates that the suc-
cessful postponement of a motor response during the
estimation of target kinematics depends on a network
that includes both the cerebellum and striatum. Thus, we
expanded our previous results (Bares et al., 2011), which
did not account for the previous trial outcome, and we
showed that these two key structures also act in tandem
930 Journal of Cognitive Neuroscience
when dynamic adaptation of motor response is required
from one trial to the next.
DISCUSSION
In the current study, we investigated the brain activation
specific to trial-by-trial adaptation of motor performance
in an interception task requiring precise predictive motor
timing in healthy participants and SCA patients with
abnormal cerebellar function. Our behavioral results
showed that SCA patients were not only worse than
healthy controls in terms of their general performance
in the task, but also that their trial-by-trial strategy was
different due to their tendency to make more early errors
than the other type. In addition, patients were affected
more than healthy controls by the changes in speed from
one trial to the next, reacting with a greater magnitude of
change in their motor strategy and with a significant wors-
ening of performance as the speed decreased in succes-
sive trials. This result suggests that cerebellar dysfunction
Volume 28, Number 7
Table 4. Brain Activation Comparing Current Hits and Late Errors versus Early Errors following Early Errors
Activated Regions Talairach Coordinates (Peak of Activation) Volume (mm3) Maximum t Value

Healthy Controls Only (df = 11)

Right middle temporal gyrus (BA 21) 54 −13 −8 142 5.56
Right inferior occipital gyrus (BA 18) 30 −85 −5 799 5.89
Right anterior cerebellum (culmen) 33 −46 −17 118 5.29
Right middle occipital gyrus (BA 18) 27 −88 13 228 6.09
Right cerebellum (anterior lobe) 21 −46 −23 110 5.33
Right posterior cingulate (BA 23) 3 −49 25 264 5.42
Left anterior cingulate (BA 24) −6 32 −2 234 7.57
Left middle occipital gyrus (BA 18) −27 −85 −2 718 9.33
Left middle occipital gyrus (BA 19) −42 −73 7 1115 10.07

Healthy Controls vs. SCA Patients (df = 19)

Right middle temporal gyrus (BA 21) 54 −13 −14 317 5.38
Right inferior temporal gyrus (BA 20) 57 −25 −17 182 5.55
Right fusiform gyrus (BA 19) 45 −70 −11 113 4.82
Right middle occipital gyrus (BA 18) 24 −82 −2 420 4.81
Right cerebellum (culmen) 21 −46 −23 252 4.99
Right anterior cingulate gyrus (BA 24) 18 −7 40 270 5.12
Right putamen 25 −13 13 116 5.99
Right precuneus (BA 7) 3 −61 61 210 4.57
Left precuneus (BA 7) −18 −70 49 304 5.05
Left inferior occipital gyrus (BA 17) −18 −91 −8 155 4.53
Left middle occipital gyrus (BA 18) −27 −85 −2 372 5.27

is associated with impairment in suppressing or holding a
motor response long enough to improve task perfor-
mance. In probing the neural substrate of this behavior,
we focused on the dynamic interaction between the cere-
bellum and BG because the joint activation of both struc-
tures is necessary for successful performance in healthy
individuals (Bares et al., 2011). Our principal findings related
to BG–cerebellum interaction during predictive motor
behaviors are the following: (1) reward-based activation
of the BG was reduced in SCA patients, which suggests that
it is dependent on cerebellar input; (2) trial-to-trial adjust-
ments required to maintain a successful strategy were re-
lated to activation of a large neuronal network in healthy
controls, including clusters in cerebellum, bilateral puta-
men, but also regions from the visual attention and action
monitoring networks (Table 3); and (3) the postponement
of the motor response during evaluation of target’s kine-
matic properties, the key to success in the motor timing
task, was associated with joint activation of the right cere-
bellum and putamen in addition to other temporo-occipital
regions.
The increased striatal activation in the current trials,
following hits compared with errors in the preceding
trials (Figure 4), indicates that the striatum is involved
in conveying the reward value or positive feedback from
one trial to the next. This is consistent with the role of
the striatum in providing a feed-forward model for the
future actions (Husarova et al., 2013; Bastian, 2006), a
conveyer of positive feedback in motor learning (Wachter,
Lungu, Liu, Willingham, & Ashe, 2009) or reward value, in
general (Berns, McClure, Pagnoni, & Montague, 2001).
Although reciprocal connections between the cerebellum
and BG have been recently documented (Bostan, Dum, &
Strick, 2013), we know little of the functional significance
of this connection. The two structures may have a funda-
mentally different relation to reward during learning with
the striatum being activated primarily for immediate
rewards whereas the cerebellum is involved with future
rewards (Tanaka et al., 2004). Our data suggest that imme-
diate reward-related activation in the BG during motor
behavior is dependent in part on normal input from the
cerebellum. This interpretation was supported by the

Lungu et al. 931

results of a post hoc effective connectivity analysis be-
tween bilateral clusters in cerebellum and striatum, which
indicated that following hits, but not errors, the informa-
tion “flowed” from cerebellar to striatal clusters. In addi-
tion, the fact that in all contrasts we found significant
between-group differences in different parts of putamen
confirms our hypothesis that striatum sits at the cross-
roads between cognitive and motor processes in this mo-
tor timing task.
We chose a behavioral task that not only tested the
ability of participants to integrate a prediction about
the perception of target movement with precise timing
of the motor response but was also very similar to natu-
ralistic everyday behaviors, which require the integration
of sensory and motor prediction, such as in catching,
shooting, playing tennis, and many other activities. How-
ever, the prediction of perceptual events is rarely an end
in itself and is often part of a greater goal that involves
concomitant temporal processing and the control of mo-
tor behavior. Motor control literature posits the existence
of adaptive internal models that are used by our brain to
make predictions about the state of the body and the en-
vironment and calibrate movements to accurately reach a
goal (Shadmehr, Smith, & Krakauer, 2010). Although our
task and experimental design are different from those
typically employed in the motor adaptation and motor
control domains, our results are consistent with reports
from this literature. Specifically, our cerebellar findings
are consistent with previous results, which indicate that
the cerebellum is involved in integrating sensory informa-
tion and constructing predictions, as well as predictive
control of motor commands that can be further proc-
essed by the cerebral cortex (Shadmehr et al., 2010;
Tseng, Diedrichsen, Krakauer, Shadmehr, & Bastian,
2007). Consistent with this integrative approach is also
the activity in the parahippocampal gyrus, which is a
known site for implicit spatial scene recognition (Bastin
et al., 2013; Howard, Kumaran, Olafsdottir, & Spiers,
2011; Rajimehr, Devaney, Bilenko, Young, & Tootell,
2011). It is conceivable that, in our predictive timing task,
participants had to quickly and implicitly recognize vari-
ous target movement patterns to facilitate the fast imple-
mentation of the strategy from one trial to the next. It is
interesting to note that the strategy that enabled the op-
timal trial-by-trial adaptation depended on the relative
change in target kinematic parameters across successive
trials (Figure 3). This is consistent with previous findings
indicating that humans employ adaptive strategies based
on the statistical properties of the environment (Semrau,
Daitch, & Thoroughman, 2012; Fine & Thoroughman,
2007).
Action-monitoring processes, such as response inhibi-
tion and error detection, are also important parts in our
interception task, especially for the trial-by-trial adjust-
ments used to improve performance. A number of corti-
cal and some subcortical structures of brain are involved
in these functions; several frontal regions, especially the
932 Journal of Cognitive Neuroscience
(right) inferior frontal gyrus, SMA, pre-SMA with included
DLPFC, and the anterior cingulate gyrus, play an impor-
tant role in response inhibition (Mayer et al., 2012; Sharp
et al., 2010; Nakata et al., 2008; Aron et al., 2007; Aron &
Poldrack, 2005; Aron, Robbins, & Poldrack, 2004), and
deficits in response inhibition have also been demon-
strated after lesions of the BG (Thoma, Koch, Heyder,
Schwarz, & Daum, 2008). Impairments in response inhi-
bition are a major component of many of the cognitive
disturbances seen in Parkinson disease in particular and
the tendency toward compulsive behaviors (Gauggel,
Rieger, & Feghoff, 2004; Lees & Smith, 1983). In this
context, it is interesting to note that obsessive com-
pulsive behaviors are part of the cognitive syndrome
seen in some patients with disease of the cerebellum
(Schmahmann, 2004). Our finding that joint activation
of the striatum and cerebellum was associated with supe-
rior performance in response inhibition suggests that re-
sponse inhibition in cerebellar disorders may be related
primarily to disruption of cerebellar-striatal networks.
The fact that the documented paired activation was ipsi-
lateral may appear to be puzzling when viewed from the
perspective of the known direct anatomical connections
between these structures. However, recent data have
shown there are ipsilateral and contralateral disynaptic
connections from the BG to cerebellum via the thalamus
(Hoshi, Tremblay, Feger, Carras, & Strick, 2005) and re-
ciprocal connections from the cerebellum to the BG via
pontine nuclei (Bostan, Dum, & Strick, 2010). Therefore,
we see no inconsistency with known anatomy in docu-
menting ipsilateral paired activation, especially given
the fact that the effective connectivity analysis revealed
strong ipsilateral functional connections between these
structures, with information “flowing” from cerebellum
to striatum. Furthermore, we know so little about dynam-
ic brain activity during complex behaviors that it may not
be prudent to dismiss activation that appears not to con-
form to our concept of how the brain works.
It is worth mentioning that in all contrasts we found no
region in which activity was significantly higher in SCA
patients than in healthy controls, indicating a general hy-
poactivation during motor timing for this clinical popula-
tion, which was also seen when analyzing the trial-by-trial
strategy. This may be the result of inherent limitations of
our study design. One might argue that brain structures
and systems outside the cerebellum may also be affected
in SCA (Maschke et al., 2005) and thus confound the in-
terpretation of our results. Although it is impossible to
completely discount this objection, the phenotypes in
the participants we studied (SCA types 6 and 8) are gen-
erally regarded as having a rather pure cerebellar syn-
drome (Maschke et al., 2005). We rather believe that
our results reflect perturbations in networks that involve
cerebellum, especially its connection with striatum. In
support of this conjecture is the fact that, although SCA
type 8 is associated with pathological abnormalities in the
cerebellum alone, some patients with this disorder also
Volume 28, Number 7
have Parkinsonian features, presumably as a result of the
disrupted cerebellum–striatum connections. In SCA type
6 individuals, the nigrostriatal dysfunction is described,
as well.
Feed-forward control and correctly estimating future
actions or states of the motor system are critical for fast
and coordinated movements. Here we showed that cer-
ebellar dysfunction in spinocerbellar ataxia patients ex-
tends to BG dopaminergic circuits, as well, leading to a
failure of predictive feed-forward control mechanism
and inaccurate estimation of motor commands con-
sequences (Ebner & Pasalar, 2008). The striatum is the
critical input structure of the BG, and it is crucial to both
motor control and learning (Humphries & Prescott,
2010). Our results may have also implications for under-
standing not only for the diseases of the cerebellum but
also for the diseases of BG with dopamine dysfunction
like Parkinson disease, where feedback processing or
reward learning is observed and the fine movement coor-
dination is affected.
Acknowledgments
This work was supported by NIH grant NS40106, MH065598,
the Department of Veterans Affairs, the Brain Sciences Chair,
a Proshek-Fulbright grant, the Academia Medica Pragensis
Foundation, and a “CEITEC-Central European Institute of Tech-
nology” project (CZ.1.05/1.1.00/02.0068) from the European
Regional Development Fund.
Reprint requests should be sent to James Ashe, Department of
Neuroscience, University of Minnesota, 4-134 Molecular and
Cellular Biology Building, 420 Washington Ave. S.E., Minneapolis,
MN 55455, or via e-mail: ashe@tc.umn.edu.
REFERENCES
Aron, A. R., Durston, S., Eagle, D. M., Logan, G. D., Stinear,
C. M., & Stuphorn, V. (2007). Converging evidence for a
fronto-basal-ganglia network for inhibitory control of action
and cognition. Journal of Neuroscience, 27, 11860–11864.
Aron, A. R., & Poldrack, R. A. (2005). The cognitive
neuroscience of response inhibition: Relevance for genetic
research in attention-deficit/hyperactivity disorder.
Biological Psychiatry, 57, 1285–1292.
Aron, A. R., Robbins, T. W., & Poldrack, R. A. (2004). Inhibition
and the right inferior frontal cortex. Trends in Cognitive
Sciences, 8, 170–177.
Bares, M., Lungu, O. V., Husarova, I., & Gescheidt, T. (2010).
Predictive motor timing performance dissociates between
early diseases of the cerebellum and Parkinson’s disease.
Cerebellum, 9, 124–135.
Bares, M., Lungu, O., Liu, T., Waechter, T., Gomez, C. M., &
Ashe, J. (2007). Impaired predictive motor timing in patients
with cerebellar disorders. Experimental Brain Research,
180, 355–365.
Bares, M., Lungu, O. V., Liu, T., Waechter, T., Gomez, C. M., &
Ashe, J. (2011). The neural substrate of predictive motor timing
in spinocerebellar ataxia. Cerebellum, 10, 233–244.
Bastian, A. J. (2006). Learning to predict the future: The
cerebellum adapts feedforward movement control. Current
Opinion in Neurobiology, 16, 645–649.
Bastin, J., Committeri, G., Kahane, P., Galati, G., Minotti, L.,
Lachaux, J. P., et al. (2013). Timing of posterior parahippocampal
gyrus activity reveals multiple scene processing stages.
Human Brain Mapping, 34, 1357–1370.
Berns, G. S., McClure, S. M., Pagnoni, G., & Montague, P. R.
(2001). Predictability modulates human brain response to
reward. Journal of Neuroscience, 21, 2793–2798.
Bostan, A., Dum, R., & Strick, P. (2010). The basal ganglia
communicate with the cerebellum. Proceedings of the
National Academy of Sciences, U.S.A., 107, 8452–8456.
Bostan, A. C., Dum, R. P., & Strick, P. L. (2013). Cerebellar
networks with the cerebral cortex and basal ganglia.
Trends in Cognitive Sciences, 17, 241–254.
Chao, H. H., Luo, X., Chang, J. L., & Li, C. S. (2009). Activation
of the pre-supplementary motor area but not inferior prefrontal
cortex in association with short stop signal reaction time—
An intra-subject analysis. BMC Neuroscience, 10, 75.
Coull, J. T., Cheng, R. K., & Meck, W. H. (2011).
Neuroanatomical and neurochemical substrates of timing.
Neuropsychopharmacology, 36, 3–25.
Ebner, T. J., & Pasalar, S. (2008). Cerebellum predicts the future
motor state. Cerebellum, 7, 583–588.
Ferrandez, A. M., Hugueville, L., Lehericy, S., Poline, J. B.,
Marsault, C., & Pouthas, V. (2003). Basal ganglia and
supplementary motor area subtend duration perception:
An fMRI study. Neuroimage, 19, 1532–1544.
Fine, M. S., & Thoroughman, K. A. (2007). Trial-by-trial
transformation of error into sensorimotor adaptation changes
with environmental dynamics. Journal of Neurophysiology,
98, 1392–1404.
Gauggel, S., Rieger, M., & Feghoff, T. A. (2004). Inhibition
of ongoing responses in patients with Parkinson’s disease.
Journal of Neurology, Neurosurgery, and Psychiatry,
75, 539–544.
Gibbon, J., Malapani, C., Dale, C. L., & Gallistel, C. (1997).
Toward a neurobiology of temporal cognition: Advances and
challenges. Current Opinion in Neurobiology, 7, 170–184.
Harrington, D. L., & Haaland, K. Y. (1999). Neural
underpinnings of temporal processing: A review of focal
lesion, pharmacological, and functional imaging research.
Reviews in the Neurosciences, 10, 91–116.
Harrington, D. L., Haaland, K. Y., & Hermanowicz, N. (1998).
Temporal processing in the basal ganglia. Neuropsychology,
12, 3–12.
Hore, J., & Watts, S. (2005). Timing finger opening in overarm
throwing based on a spatial representation of hand path.
Journal of Neurophysiology, 93, 3189–3199.
Hoshi, E., Tremblay, L., Feger, J., Carras, P., & Strick, P. (2005).
The cerebellum communicates with the basal ganglia. Nature
Neuroscience, 8, 1491–1493.
Howard, L. R., Kumaran, D., Olafsdottir, H. F., & Spiers, H. J.
(2011). Double dissociation between hippocampal and
parahippocampal responses to object-background context
and scene novelty. Journal of Neuroscience, 31, 5253–5261.
Humphries, M. D., & Prescott, T. J. (2010). The ventral
basal ganglia, a selection mechanism at the crossroads
of space, strategy, and reward. Progress in Neurobiology,
90, 385–417.
Husarova, I., Mikl, M., Lungu, O. V., Marecek, R., Vanicek, J.,
& Bares, M. (2013). Similar circuits but different connectivity
patterns between the cerebellum, basal ganglia, and
supplementary motor area in early Parkinson’s disease
patients and controls during predictive motor timing.
Journal of Neuroimaging, 23, 452–462.
Iacoboni, M. (2001). Playing tennis with the cerebellum.
Nature Neuroscience, 4, 555–556.
Ivry, R. B., & Keele, S. W. (1989). Timing functions of the
cerebellum. Journal of Cognitive Neuroscience, 1, 136–152.
Lungu et al. 933
Ivry, R. B., & Spencer, R. M. (2004). The neural representation
of time. Current Opinion in Neurobiology, 14, 225–232.
Jahanshahi, M., Jones, C. R., Dirnberger, G., & Frith, C. D.
(2006). The substantia nigra pars compacta and temporal
processing. Journal of Neuroscience, 26, 12266–12273.
Lees, A. J., & Smith, E. (1983). Cognitive deficits in the early
stages of Parkinson’s disease. Brain, 106, 257–270.
Maschke, M., Oehlert, G., Xie, T. D., Perlman, S., Subramony,
S. H., Kumar, N., et al. (2005). Clinical feature profile of
spinocerebellar ataxia type 1–8 predicts genetically defined
subtypes. Movement Disorders, 20, 1405–1412.
Matell, M. S., & Meck, W. H. (2004). Cortico-striatal circuits
and interval timing: Coincidence detection of oscillatory
processes. Brain Research, Cognitive Brain Research,
21, 139–170.
Mauk, M. D., & Buonomano, D. V. (2004). The neural basis
of temporal processing. Annual Review of Neuroscience,
27, 307–340.
Mayer, A. R., Teshiba, T. M., Franco, A. R., Ling, J., Shane, M. S.,
Stephen, J. M., et al. (2012). Modeling conflict and error
in the medial frontal cortex. Human Brain Mapping,
33, 2843–2855.
Meck, W. H. (2005). Neuropsychology of timing and time
perception. Brain and Cognition, 58, 1–8.
Meck, W. H., & Benson, A. M. (2002). Dissecting the brain’s
internal clock: How frontal-striatal circuitry keeps time and
shifts attention. Brain and Cognition, 48, 195–211.
Montgomery, S. A., & Asberg, M. (1979). A new depression scale
designed to be sensitive to change. British Journal of
Psychiatry, 134, 382–389.
Mostofsky, S. H., Schafer, J. G., Abrams, M. T., Goldberg, M. C.,
Flower, A. A., Boyce, A., et al. (2003). fMRI evidence that
the neural basis of response inhibition is task-dependent.
Brain Research, Cognitive Brain Research, 17, 419–430.
Nakata, H., Sakamoto, K., Ferretti, A., Gianni Perrucci, M.,
Del Gratta, C., Kakigi, R., et al. (2008). Somato-motor
inhibitory processing in humans: An event-related
functional MRI study. Neuroimage, 39, 1858–1866.
Nenadic, I., Gaser, C., Volz, H. P., Rammsayer, T., Hager, F.,
& Sauer, H. (2003). Processing of temporal information and
the basal ganglia: New evidence from fMRI. Experimental
Brain Research, 148, 238–246.
O’Boyle, D. J., Freeman, J. S., & Cody, F. W. (1996). The
accuracy and precision of timing of self-paced, repetitive
movements in subjects with Parkinson’s disease. Brain,
119, 51–70.
Oldfield, R. C. (1971). The assessment and analysis of
handedness: The Edinburgh inventory. Neuropsychologia,
9, 97–113.
Penny, W. D., & Holmes, A. J. (2003). Random-effects analysis.
In R. S. J. Frackowiak, K. J. Friston, C. Frith, R. Dolan,
K. J. Friston, C. J. Price, et al. (Eds.), Human brain
function (2nd ed., pp. 843–850). London: Academic Press.
Rajimehr, R., Devaney, K. J., Bilenko, N. Y., Young, J. C., &
Tootell, R. B. (2011). The “parahippocampal place area”
934 Journal of Cognitive Neuroscience
responds preferentially to high spatial frequencies in humans
and monkeys. PLoS Biology, 9, e1000608.
Schmahmann, J. D. (2004). Disorders of the cerebellum: Ataxia,
dysmetria of thought, and the cerebellar cognitive affective
syndrome. Journal of Neuropsychiatry and Clinical
Neurosciences, 16, 367–378.
Semrau, J. A., Daitch, A. L., & Thoroughman, K. A. (2012).
Environmental experience within and across testing days
determines the strength of human visuomotor adaptation.
Experimental Brain Research, 216, 409–418.
Shadmehr, R., Smith, M. A., & Krakauer, J. W. (2010). Error
correction, sensory prediction, and adaptation in motor
control. Annual Review of Neuroscience, 33, 89–108.
Sharp, D. J., Bonnelle, V., De Boissezon, X., Beckmann, C. F.,
James, S. G., Patel, M. C., et al. (2010). Distinct frontal
systems for response inhibition, attentional capture, and
error processing. Proceedings of the National Academy
of Sciences, U.S.A., 107, 6106–6111.
Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic
atlas of the human brain. New York: Thieme.
Tanaka, S. C., Doya, K., Okada, G., Ueda, K., Okamoto, Y.,
& Yamawaki, S. (2004). Prediction of immediate and future
rewards differentially recruits cortico-basal ganglia loops.
Nature Neuroscience, 7, 887–893.
Thoma, P., Koch, B., Heyder, K., Schwarz, M., & Daum, I.
(2008). Subcortical contributions to multitasking and
response inhibition. Behavioural Brain Research,
194, 214–222.
Trouillas, P., Takayanagi, T., Hallett, M., Currier, R. D.,
Subramony, S. H., Wessel, K., et al. (1997). International
Cooperative Ataxia Rating Scale for pharmacological
assessment of the cerebellar syndrome. The Ataxia
Neuropharmacology Committee of the World Federation
of Neurology. Journal of the Neurological Sciences,
145, 205–211.
Tseng, Y. W., Diedrichsen, J., Krakauer, J. W., Shadmehr, R.,
& Bastian, A. J. (2007). Sensory prediction errors drive
cerebellum-dependent adaptation of reaching. Journal
of Neurophysiology, 98, 54–62.
van der Salm, S. M., van der Meer, J. N., Nederveen, A. J.,
Veltman, D. J., van Rootselaar, A. F., & Tijssen, M. A. (2013).
Functional MRI study of response inhibition in myoclonus
dystonia. Experimental Neurology, 247, 623–629.
Vink, M., Kaldewaij, R., Zandbelt, B. B., Pas, P., & du Plessis, S.
(2015). The role of stop-signal probability and expectation
in proactive inhibition. European Journal of Neuroscience,
41, 1086–1094.
Wachter, T., Lungu, O. V., Liu, T., Willingham, D. T., &
Ashe, J. (2009). Differential effect of reward and punishment
on procedural learning. Journal of Neuroscience,
29, 436–443.
Ye, Z., Altena, E., Nombela, C., Housden, C. R., Maxwell, H.,
Rittman, T., et al. (2015). Improving response inhibition in
Parkinson’s disease with atomoxetine. Biological Psychiatry,
77, 740–748.
Volume 28, Number 7