Abstract

Converging anatomical and functional evidence suggests that the cerebellum processes both motor and nonmotor information originating from the primary motor cortex and prefrontal cortex, respectively. However, it has not been established whether the cerebellum only processes prefrontal information where rules specify actions or whether the cerebellum processes any form of prefrontal information no matter how abstract. Using functional magnetic resonance imaging, we distinguish between two competing hypotheses: (1) activity within prefrontal-projecting cerebellar lobules (Crus I and II) will only be evoked by rules that specify action (i.e. first-order rules; arbitrary S–R mappings) and (2) activity will be evoked in these lobules by both first-order rules and second-order rules that govern the application of lower order rules. The results showed that prefrontal-projecting cerebellar lobules Crus I and II were commonly activated by processing both first- and second-order rules. We demonstrate for the first time that cerebellar circuits engage both first- and second-order rules and in doing so show that the cerebellum can contribute to cognitive control independent of motor control.

Introduction

Early theories of cerebellar function proposed that the cerebellum plays a key role in motor learning and in the sensory guidance of action (Holmes 1939; Marr 1969; Albus 1971; Stein and Glickstein 1992; Glickstein 1998; Wolpert et al. 1998; Ito 2000; Doyon et al. 2003). There is also a growing body of functional and anatomical evidence in both humans and nonhuman primates, which suggests that in addition to processing information from the primary motor cortex, the cerebellum also processes “nonmotor” information originating from the prefrontal and posterior parietal cortices (reviewed in Strick et al. 2009; Ramnani 2011). However, it is not yet clear to what end the cerebellum processes information originating from the prefrontal and posterior parietal cortices. Does the cerebellum only process information originating from the prefrontal/parietal cortex when it is paired with an action, or is it the case that the cerebellum will process any form of prefrontal or parietal information, no matter how abstract, in an effort to automate processes within these cortical regions (Ramnani 2006; Balsters and Ramnani 2011)? Anatomical evidence suggests that the latter may be the case, but to our knowledge, no neuroimaging study has specifically investigated whether cerebellar activity is present during the processing of increasingly abstract stimuli.

In this paper, we define increasing levels of abstraction in terms of increasing relational integration, that is, increasing the number of steps or rules necessary to execute a specific response increases the level of abstraction (see Badre and D'Esposito 2009). It has been suggested that increasingly abstract stimuli are processed in increasingly anterior portions within the frontal lobe (Miller and Cohen 2001; Ramnani and Owen 2004; Petrides 2005; Koechlin and Summerfield 2007; Badre and D'Esposito 2009). This rostrocaudal gradient extends from the central sulcus (i.e. primary motor cortex; area 4) through to the anterior prefrontal cortex (area 10), which is active during the most abstract and cognitively demanding tasks such as generating and maintaining subgoals (Ramnani and Owen 2004; Badre and D'Esposito 2007, 2009; Koechlin and Summerfield 2007). In between these two extremes are the premotor cortex (area 6; anterior to the primary motor cortex), which encodes information for preparatory set and is essential to the acquisition of first-order rules (Petrides 1982; Halsband and Passingham 1985; Wise 1985; Toni et al. 1999; Balsters and Ramnani 2008), pre-PMd [area 8; anterior to the dorsal premotor cortex (PMd)], which is selectively activated by more abstract hierarchical mappings (Picard and Strick 2001; Badre and D'Esposito 2007; Badre et al. 2010), and areas 9, 46, and 9/46, which encode increasingly abstract information including rules (Freedman et al. 2001; Wallis et al. 2001) and monitor information in working memory (Petrides 1994; Fuster 1997; Funahashi 2001). These areas are interconnected to form a hierarchically organized network in which executive control is achieved through a cascade of information from areas of the prefrontal cortex through to the primary motor cortex via the premotor system (Koechlin and Summerfield 2007; Badre and D'Esposito 2009). Anatomical evidence from humans and nonhuman primates suggests that each section of this frontal lobe hierarchy is independently connected with the cerebellum.

The architecture of the cortico-cerebellar system has been well characterized in humans and nonhuman primates. It has been suggested that the cerebellum exchanges information with the cerebral cortex within independent sets of closed cortico-cerebellar loops (Middleton and Strick 2000; Kelly and Strick 2003). Kelly and Strick (2003) have characterized 2 distinct cortico-cerebellar loops: the “motor loop” and the “prefrontal loop.” In the motor loop, the primary motor cortex projects to cerebellar cortical lobules V, VI, and HVIIB and HVIII and projects back to the same regions of cortex via dorsal parts of the cerebellar dentate nucleus and the motor thalamus. In the prefrontal loop, area 46 (Walker 1940) of the prefrontal cortex projects to lobule HVIIA (mainly to Crus II and, to a lesser degree, to Crus Ip) via the pontine nuclei, and this area returns projections to the same areas of the prefrontal cortex via ventral parts of the cerebellar dentate nucleus and prefrontal thalamus (Goldman-Rakic and Porrino 1985; Barbas et al. 1991; Middleton and Strick 2001; Kelly and Strick 2003). Connectivity studies in both humans and nonhuman primates have also demonstrated that prefrontal projections to the cerebellum originate from a range of prefrontal territories including regions as anterior as area 10 (Schmahmann and Pandya 1997; Ramnani et al. 2006; Habas et al. 2009; Krienen and Buckner 2009; O'Reilly et al. 2010; Buckner et al. 2011). The existence of connections between rostral portions of the prefrontal cortex and the cerebellum suggests that the cerebellum can automate processing within any cortical region and it is not restricted to rules that specify a motor response.

Unfortunately, there are few clear examples from functional neuroimaging that have found cerebellar activations to increasingly abstract stimuli. While previous studies have shown cerebellar activations related to complex cognitive paradigms such as language (Chen and Desmond 2005a, 2005b; Desmond et al. 2005; Kirschen et al. 2005), playing chess (Atherton et al. 2003; Chen et al. 2003), pegboard puzzles (Kim et al. 1994), and mathematical reasoning [Paced Auditory Serial Addition Task (PASAT); Hayter et al. 2007], these studies were not able to isolate confounding motor responses from cognitive elements of the task. In previous studies (Ramnani and Miall 2003, 2004; Balsters and Ramnani 2008, 2011), we have used conditional motor learning as a method for investigating the acquisition and retrieval of first-order rules without the contaminating effects of the subsequent motor response or feedback. Using this paradigm, we previously demonstrated that regions of the cerebellum interconnected with the prefrontal cortex (Crus I) were active during the acquisition and retrieval of first-order rules (Balsters and Ramnani 2008, 2011). We have previously interpreted these results as evidence that the cerebellum contributes to cognitive control, given that these activations were temporally independent of the subsequent motor processes. However, one caveat to this interpretation is that while the instruction cue is temporally independent of subsequent motor responses, the first-order rule is still linking an arbitrary stimulus with a specific motor effector and as such it is a rule that guides movement. It may be the case that the cerebellum will only process rule-related information where rules specify actions.

This study attempts to address the extent to which the cerebellum processes abstract information originating from the prefrontal cortex using functional magnetic resonance imaging (fMRI). We aim to distinguish between 2 competing hypotheses: the first is that activity within prefrontal-projecting cerebellar lobules will only be evoked by rules that specify action (i.e. first-order rules). The second hypothesis is that activity within prefrontal-projecting cerebellar lobules will be evoked by both rules that specify action and rules that specify another rule (i.e. second-order rules). In our experiment, these second-order rules were devoid of motor information and could only guide the choice of a first-order rule.

Materials and Methods

Participants

Fifteen young (18–30; 9 male) neurologically normal, right-handed subjects participated in this study. Participants gave written informed consent prior to the study, which was approved by the Trinity College Dublin School of Psychology Ethics Committee.

Trial Structure

Subjects were required to execute a delayed-response task (Fig. 1). The same basic trial structure was applied under all conditions, with condition-specific variations explained subsequently. The trial began with the presentation of an instruction cue (500 ms) that signaled which response subjects were required to make. After a variable delay period, subjects saw a “Go!” signal (250 ms) immediately followed by 4 adjacent “hourglass” stimuli of different colors (1000 ms), prompting subjects to execute a response (pressing 1 of the 4 buttons on a response pad held in the right hand). The response was immediately followed by error feedback (a green dot after a correct response and a red dot after an incorrect response; 250 ms). If participants failed to execute a response within a 1000 ms time window, the word “Missed” was displayed instead of these feedback cues.

Figure 1.

Trial structure: each trial was divided into two 4 s periods: the instructed delay period (0–4 s onset latency) contained the instruction cue (green shape with 4 underscores underneath, one of these underscores was replaced with an asterisk for first-order control instruction cues) and the response period (4–8 s onset latency) contained the Go! signal, immediately followed by the trigger cue where the subject made a response (pressing 1 of the 4 buttons) and feedback (green dot for correct response, red dot for incorrect, and missed for no response within the 1000 ms time window). Between each instruction cue and the trigger-related cues is the appropriate response in quotation marks. This information was not presented to subjects.

Figure 1.

Trial structure: each trial was divided into two 4 s periods: the instructed delay period (0–4 s onset latency) contained the instruction cue (green shape with 4 underscores underneath, one of these underscores was replaced with an asterisk for first-order control instruction cues) and the response period (4–8 s onset latency) contained the Go! signal, immediately followed by the trigger cue where the subject made a response (pressing 1 of the 4 buttons) and feedback (green dot for correct response, red dot for incorrect, and missed for no response within the 1000 ms time window). Between each instruction cue and the trigger-related cues is the appropriate response in quotation marks. This information was not presented to subjects.

Conditions

The 4 trial types were embedded in a 2 × 2 factorial design (2 factors, each with 2 levels: Table 1). This experimental design allowed us to compare first- and second-order rules each with their own control conditions.

  • Factor 1: rule (rule and control): instructions either required the participant to recall an arbitrary association between the cue and the response (rule) or the instruction informed responses directly (control).

  • Factor 2: rule order (first-order and second-order): as well as either being a rule or control, the instruction could be either first-order or second-order. First-order instruction cues held all the information necessary for a response at the time of the instruction cue. Second-order instruction cues did not contain enough information for the participant to respond correctly, and the participant would only be able to perform the correct response after viewing the trigger cue.

Table 1

Four conditions embedded in the 2 × 2 factorial design

 Factor 1: rule
 
 Rule Control 
Factor 2: rule order 
 First First-order rule (48 trials) First-order control (48 trials) 
 Second Second-order rule (48 trials) Second-order control (48 trials) 
 Factor 1: rule
 
 Rule Control 
Factor 2: rule order 
 First First-order rule (48 trials) First-order control (48 trials) 
 Second Second-order rule (48 trials) Second-order control (48 trials) 

This 2 × 2 factorial design resulted in 4 condition types (Fig. 1).

  • Condition 1: first-order rule (1stR): instruction cues consisted of a green triangle, hexagon, “bridge,” and 3-quarter circle against a gray background. Underneath each shape, there were 4 adjacent underscores. Each shape had a conditional association with a specific response (pressing button 1, 2, 3, and 4, respectively). Associations between instruction cues and motor responses were acquired during pretraining immediately prior to scanning (see “Pretraining”). This stimulus type has been previously shown to evoke activity within prefrontal-projecting regions of the cerebellum (Balsters and Ramnani 2008, 2011).

  • Condition 2: first-order control (1stC): instruction cues consisted of a simple outlined symmetrical cross. One of the 4 adjacent underscores underneath the cross was replaced with an asterisk signaling the required response.

  • Condition 3: second-order rule (2ndR): instruction cues consisted of a circle, s-like squiggle, square, and “explosion.” Underneath each shape were 4 adjacent underscores. Each instruction cue had a conditional association with a color (pink, blue, yellow, or black) just as each instruction cue within condition 1 (first-order rule) had a conditional association with a motor response. The instruction cue told the subject which color to match at the point of the trigger cue. For example, if the instruction cue was a circle, the participant would have to press a button which corresponded spatially to the pink trigger element. The position of each color stimulus at the point of the trigger cue would change from trial to trial, making it impossible for subjects to prepare the appropriate response prior to the presentation of the trigger cue. As with condition 1, associations between instruction cues and colors were acquired during pretraining immediately before scanning.

  • Condition 4: second-order control (2ndC): this instruction cue used the same image as condition 2. However, all 4 adjacent underscores were presented under the image, and thus subjects were not able to prepare a response at the time of the instruction cue. The required response was specified by the trigger cue (only 1 shape would be present at the trigger cue).

Pretraining

Before participants entered the scanner, they were informed of the stimulus associations and practiced the task for approximately 11 min (approximately 14 trials of each condition and about 3.5 presentations of each instruction cue), in order to learn all the stimulus associations. All participants were able to explicitly describe these associations before entering the scanner. Participants briefly rehearsed the task once more during the structural scan in order to confirm that they understood the task.

Experimental Timing

An important feature of this study was the ability to time-lock activity specifically to instruction cues. A variable delay was introduced between the instruction cue and the Go! signal. As in previous studies (Balsters and Ramnani 2008, 2011), this allowed us to isolate blood oxygen level-dependent (BOLD) activity time-locked to the instruction cue without the contaminating effects of subsequent trial events (Go! signal, trigger cue, motor response, and error feedback). Events in each trial took place across 4 repetition times (TRs) (0–8 s; TR = 2 s, Fig. 1). In order to optimally sample evoked hemodynamic responses (EHRs), we randomly varied the interval between scan onset and instruction cue onset over the range of the first 2 TRs from trial to trial. This achieved an effective temporal sampling resolution much finer than one TR. These intervals were uniformly distributed, ensuring that EHRs time-locked to the instruction cue were sampled evenly across the time period following each type of instruction cue. The Go! signal (along with motor responses and feedback) occurred in the period occupied by the third and fourth TR, and the timing between the third TR and its onset was varied in the same manner (in the range 4–8 s after the onset of the first TR). The range of the variable delay between the onset of instruction cues and the onset of the Go! signal varied from 832 to 6564 ms. Jittering the onset of stimuli relative to the start of each Echo Planar Imaging (EPI) volume additionally guarantees that stimuli are presented during the acquisition of every slice, and as such there is no spatial bias/neglect in our imaging protocol.

Since the instruction cues were temporally uncorrelated with the preceding and subsequent Go! signals, they could be modeled as independent event types. This allowed us to determine activity time-locked to instruction cues without the contaminating effects of the Go! signal and subsequent triggers and responses.

Functional Imaging and Analysis

Apparatus

Subjects lay supine in an MRI scanner with the fingers of their right hand positioned on a 4-button MRI-compatible response box. Stimuli were projected onto a screen behind the subject and viewed in a mirror positioned above the subject's face. Presentation software (Neurobehavioral Systems, Inc., USA) was used for stimulus presentation both inside and outside the scanner. Transistor-Transistor Logic pulses were used to drive the visual stimuli in Presentation.

Data Acquisition

A high-resolution T1-weighted anatomic magnetization-prepared rapid gradient echo image [field of view (FOV) = 230 mm, thickness = 0.9 mm, voxel size = 0.9 mm × 0.9 mm × 0.9 mm] and phase and magnitude maps were acquired first (TE1 = 1.46 ms and TE2 = 7 ms). Each participant then performed a single EPI session containing 782 volumes lasting 26 min. The FOV covered the whole brain, 224 mm × 224 mm (64 × 64 voxels), and 39 axial slices were acquired with a voxel size of 3.5 mm × 3.5 mm × 3.5 mm (0.3 mm slice gap), TR = 2 s, echo time = 30 ms, flip angle = 90°. All MRI data were collected on a Philips 3 T Achieva MRI Scanner (Trinity College Dublin).

Image Preprocessing

Scans were preprocessed using SPM8 (www.fil.ion.ucl.ac.uk/spm). Images were realigned and unwarped using field maps to correct for motion artifacts, susceptibility artifacts, and motion-by-susceptibility interactions (Andersson et al. 2001; Hutton et al. 2002). Images were subsequently normalized to the ICBM EPI template using the unified segmentation approach (Ashburner and Friston 2005). Lastly, a Gaussian kernel of 8 mm full width at half maximum (FWHM) was applied to spatially smooth the image in order to conform to the Gaussian assumptions of a generalized linear model (GLM) as implemented in SPM8 (Friston, Frith, Frackowiak et al. 1995; Friston, Frith, Turner et al. 1995).

Statistical Analysis

First-level single-subject analyses

Seven event types were modeled at the first level. All events were convolved with the canonical hemodynamic response function. Instruction cues for each of the 4 conditions were modeled as 4 separate event types. Trigger cues associated with first-order or second-order instructions were modeled as 2 further event types describing variance associated with the visual trigger, the motor response, and the visually presented outcome. Trials in which responses were incorrect, too early (before the trigger cue), or too late (RT > 1000 ms) were modeled separately as a seventh event type and differentiated from experimental conditions. This seventh event type included the onsets from both the instruction cue and the Go! signal in error trials. Thus, activity time-locked to incorrect trials was excluded from regressors explaining instruction-related activity. The residual effects of head motion were modeled as covariates of no interest in the analysis by including the 6 head motion parameters estimated during the realignment stage of the preprocessing. Prior to the study, a set of planned experimental timings was carefully checked so that they resulted in an estimable GLM, in which the statistical independence of the 7 event types was preserved.

Cerebellar-specific analysis (SUIT)

Given that our hypotheses focussed on the cerebellum, we performed an additional analysis of the cerebellar activity using the SUIT toolbox (http://www.icn.ucl.ac.uk/motorcontrol/imaging/suit.htm). It has been shown that normalization using SUIT produces a more accurate cerebellar normalization across participants (Diedrichsen 2006; Diedrichsen et al. 2009). Similar to the unified segmentation approach, the cerebellum and the brainstem of high-resolution T1 images were isolated (automatically generated masks were manually inspected and any nonbrain tissue removed) and warped into SUIT space. These masks were also used to isolate BOLD signal in these regions and to stop activations from overlying regions such as the visual cortex contaminating cerebellar activations. Parameters describing the nonlinear warping from individual subject space to SUIT space were later applied to first-level contrast images. These contrast images were generated using the same GLMs described earlier; however, these GLMs were run on data that were realigned and unwarped only. Contrast images were warped into SUIT space and smoothed (8 mm FWHM). In order to prevent smoothing from falsely extending cerebellar activation clusters into adjacent structures outside the cerebellum, we applied implicit masking during the smoothing preprocessing step. This effectively preserves the boundaries of the image by masking voxel values of 0 or Not a Number.

Second-level random-effects group analysis

To determine voxels significant at the group level, t-contrasts were incorporated into a random-effects analysis (1- or 2-sample t-tests). One-sample t-tests were used to investigate single contrast images that define main effects or interactions, for example, modeling the main effect of rule at the single-subject level (1stR + 2ndR <> 1stC + 2ndC). This contrast compares the average activity of rule cues with the average activity of control cues. This approach is commonly used in factorial designs, but may not be sufficiently stringent because a significant effect could emerge from a significant difference between only one cue compared with control cues. Two-sample t-tests were used to perform a conjunction analysis (Price and Friston 1997; Friston et al. 2005). A conjunction analysis is a more stringent comparison as it requires that both rule conditions must be significantly different from their respective control conditions (1stR<>1stC && 2ndR<>2ndC).

Given our anatomically specific hypothesis, a small volume correction was used (bilateral Crus I and Crus II gray matter mask) to correct for multiple comparisons. A whole cerebellar gray matter mask was also used for small volume correction (results reported in Supplementary Material). These masks were generated using the atlas of Diedrichsen et al. (2009). SPMs were thresholded at P < 0.001 uncorrected for display purposes, and all results reported survived a correction for multiple comparisons, either False Discovery Rate (FDR) P < 0.05 or small volume correction over Crus I and II.

Localization

Anatomical details of significant signal changes were obtained by superimposing the SPMs on the T1 canonical single-subject image from the Montreal Neurological Institute (MNI) series. Results were checked against normalized T1 images of each subject. The atlas of Duvernoy and Bourgouin (1999) was used as a general neuroanatomical reference. The atlases of Schmahmann et al. (2000) and Diedrichsen et al. (2009) were employed as a specific neuroanatomical reference for cerebellar activations. We used the nomenclature of Schmahmann et al. (2000) to label cerebellar lobules. The SPM anatomy toolbox (Eickhoff et al. 2005) was used to establish cytoarchitectonic probabilities where applicable.

Results

Behavior

Error Rates

Participants' error rates were very low due to pretraining, typically 5.5% across all conditions (mean 6.53, SD 4.78 trials) with less than 2 error trials per condition (1stR: mean 1.27, SD 1.39 trials; 1stC: mean 2.2, SD 2.24 trials; 2ndR: mean 1.87, SD 1.55 trials; 2ndC: mean 1.2, SD 1.32 trials). There was no significant main effect of Factor 1: rule (F1,14 = 0.17, P = 0.69) or Factor 2: rule order (F1,14 = 0.293, P = 0.6). There was a significant interaction (F1,14 = 10.03, P < 0.01) driven by significant differences in the error rate between 1stR and 1stC (P = 0.025).

Reaction Times

Participants were significantly faster at responding to first-order (mean 334.72 ms, SD 49.61 ms) compared with second-order (mean 516.44 ms, SD 42.31 ms) instruction cues (main effect of Factor 2: rule order, F1,14 = 379.01, P < 0.001). Given that first-order instruction cues contained all the information necessary to respond, it is likely that participants were preparing responses when possible during the instructed delay. There was no main effect of rules on reaction time [F1,14 = 1.07, P = 0.32; rules (mean 427.68 ms, SD 43.65 ms) and control (mean 423.48 ms, SD 48.27 ms)] and no significant interaction between rules and rule order (F1,14 = 1.34, P = 0.27; 1stR: mean 334.04 ms, SD 46.05 ms; 1stC: mean 335.4 ms, SD 53.17 ms; 2ndR: mean 521.33 ms, SD 41.25 ms; 2ndC: mean 511.55 ms, SD 43.36 ms). A figure plotting condition-specific reaction times is available in Supplementary Material.

Functional Imaging

Trigger-Related Activity

Sensory and motor areas were active at the time of the trigger cue. This included right inferior frontal gyrus, middle cingulate cortex, bilateral postcentral gyrus (including primary motor and somatosensory cortices), left inferior parietal lobule, and temporal and occipital regions. The largest activations (both in cluster size and in t-value) were in ipsilateral motor lobules of the cerebellum (lobules HVI and HVIIIb).

First-order versus second-order trigger activity

First-order triggers were modeled separately from second-order triggers, given the differences in reaction times between these 2 trial types. Significantly greater activity was present for second-order triggers in the primary visual regions, left supramarginal gyrus, right postcentral gyrus, right middle cingulate cortex, and left hippocampus. Motor lobules of the left and right cerebellum (lobule HVIIIb) were also significantly active. There were no significant activations where first-order triggers were greater than second-order triggers.

Instruction-Related Activity

Main effect of rule

A conjunction analysis was used to localise common areas of activity across the two contrasts: (first-order rules <> first-order control) && (second-order rules <> second-order control). The conjunction analysis showed regions of significant activation commonly seen in studies of cognitive control [left inferior frontal gyrus (pars triangularis) and left inferior parietal lobule; Cabeza and Nyberg 2000] to be more active during rule-based processing. When the main effect of rule was modeled at the first level, the same regions were also active, as well as visual regions (left fusiform gyrus, left middle occipital gyrus, right middle temporal gyrus, and right cuneus), the left superior medial gyrus (area 32), and left hippocampus. The right hemisphere prefrontal-projecting lobule Crus I was also found to be significantly more active during rule processing compared with control conditions (Table 2).

Table 2

Main effect of rule cues (1stR + 2ndR <> 1stC + 2ndC): activity time-locked to instruction cues, FDR-corrected for multiple comparisons (P < 0.05) in a random-effects analysis

Rules <> controls (1stR + 2ndR <> 1stC + 2ndC) Cluster T-value Z-value Coordinates Cytoarchitectonic BA (probability, if available) Most active condition 
Left inferior frontal gyrus (pars triangularis) 1543 7.54 4.69 −46, 22, 20 Area 45 (40%), area 44 (30%) Rules 
Left superior medial gyrus 470 6.06 4.18 −8, 20, 42 Area 32 Rules 
Left inferior parietal lobule 2037 7.39 4.65 −36, −50, 42 hIP1 (40%), hIP3 (40%) Rules 
Left fusiform gyrus 477 6.56 4.36 −46, −60, −18 Area 37 Rules 
Left middle occipital gyrus 669 9.38 5.2 −16, −92, −6 hOC3v (V3v) (60%), area 18 (30%), area 17 (20%) Rules 
Right cuneus 657 8.95 5.09 20, −98, 10 Area 18 (40%), area 17 (40%) Rules 
Left hippocampus 180 6.16 4.22 −26, −26, −12 Hipp (FD) (80%) and Hipp (SUB) (50%) Rules 
Right cerebellar hemisphere 991 8.14 4.87 30, −60, −34 Crus I (34%) Rules 
Right middle temporal gyrus 224 6.53 4.35 52, −70, 22 IPC (PGp) (60%) Control 
1stR <> 1stC 
 Left inferior frontal gyrus (pars triangularis) 643 7.65 4.72 −46, 28, 14 Area 45 (40%) 1stR 
 Left middle cingulate cortex 304 6.38 4.3 −6, 4, 42 Area 6 (10%) 1stR 
 Left posterior cingulate cortex 631 7.35 4.63 −4, −32, 26 Area 23 1stR 
 Left inferior parietal lobule 420 6.32 4.28 −38, −56, 40 hIP1 (50%) 1stR 
 Right precuneus 173 4.81 3.63 8, −68, 42 SPL (7A) (20%) 1stR 
 Left caudate nucleus 104 5.3 3.85 −12, 2, 12 n/a 1stR 
 Right caudate nucleus 236 6.33 4.28 10, 2, 0 n/a 1stR 
 Right cerebellum 44 4.87 3.67 26, −78, −50 Crus II (70%) 1stR 
 Right middle temporal gyrus 245 6.46 4.33 54, −66, 16 IPC (PGp) (50%) 1stC 
 Left lingual gyrus 302 5.45 3.93 −18, −82, −16 hOC3v (30%), hOC4v (40%) 1stC 
2ndR <> 2ndC 
 Left inferior frontal gyrus (pars triangularis) 156 5.33 3.88 −46, 26, 14 Area 45 (50%) 2ndR 
 Left posterior cingulate cortex 239 5.02 3.74 −2, −32, 32 Area 23 2ndR 
 Left precentral gyrus 724 5.97 4.14 −52, 0, 36 Area 6 (10%) 2ndR 
 Left lingual gyrus 10 876 9.3 5.18 −14, −88, −12 hOC3v (60%), area 18 (40%) 2ndR 
 Left hippocampus 394 5.27 3.85 −24, −26, −10 Hipp (SUB) (30%), (FD) (20%) 2ndR 
 Right cerebellum 1205 5.54 3.97 30, −60, 32 HVI (81%), Crus I (19%) 2ndR 
 Right cerebellum 45 4.22 3.68 10, −84, −38 Crus II (84%) 2ndR 
Rules <> controls (1stR + 2ndR <> 1stC + 2ndC) Cluster T-value Z-value Coordinates Cytoarchitectonic BA (probability, if available) Most active condition 
Left inferior frontal gyrus (pars triangularis) 1543 7.54 4.69 −46, 22, 20 Area 45 (40%), area 44 (30%) Rules 
Left superior medial gyrus 470 6.06 4.18 −8, 20, 42 Area 32 Rules 
Left inferior parietal lobule 2037 7.39 4.65 −36, −50, 42 hIP1 (40%), hIP3 (40%) Rules 
Left fusiform gyrus 477 6.56 4.36 −46, −60, −18 Area 37 Rules 
Left middle occipital gyrus 669 9.38 5.2 −16, −92, −6 hOC3v (V3v) (60%), area 18 (30%), area 17 (20%) Rules 
Right cuneus 657 8.95 5.09 20, −98, 10 Area 18 (40%), area 17 (40%) Rules 
Left hippocampus 180 6.16 4.22 −26, −26, −12 Hipp (FD) (80%) and Hipp (SUB) (50%) Rules 
Right cerebellar hemisphere 991 8.14 4.87 30, −60, −34 Crus I (34%) Rules 
Right middle temporal gyrus 224 6.53 4.35 52, −70, 22 IPC (PGp) (60%) Control 
1stR <> 1stC 
 Left inferior frontal gyrus (pars triangularis) 643 7.65 4.72 −46, 28, 14 Area 45 (40%) 1stR 
 Left middle cingulate cortex 304 6.38 4.3 −6, 4, 42 Area 6 (10%) 1stR 
 Left posterior cingulate cortex 631 7.35 4.63 −4, −32, 26 Area 23 1stR 
 Left inferior parietal lobule 420 6.32 4.28 −38, −56, 40 hIP1 (50%) 1stR 
 Right precuneus 173 4.81 3.63 8, −68, 42 SPL (7A) (20%) 1stR 
 Left caudate nucleus 104 5.3 3.85 −12, 2, 12 n/a 1stR 
 Right caudate nucleus 236 6.33 4.28 10, 2, 0 n/a 1stR 
 Right cerebellum 44 4.87 3.67 26, −78, −50 Crus II (70%) 1stR 
 Right middle temporal gyrus 245 6.46 4.33 54, −66, 16 IPC (PGp) (50%) 1stC 
 Left lingual gyrus 302 5.45 3.93 −18, −82, −16 hOC3v (30%), hOC4v (40%) 1stC 
2ndR <> 2ndC 
 Left inferior frontal gyrus (pars triangularis) 156 5.33 3.88 −46, 26, 14 Area 45 (50%) 2ndR 
 Left posterior cingulate cortex 239 5.02 3.74 −2, −32, 32 Area 23 2ndR 
 Left precentral gyrus 724 5.97 4.14 −52, 0, 36 Area 6 (10%) 2ndR 
 Left lingual gyrus 10 876 9.3 5.18 −14, −88, −12 hOC3v (60%), area 18 (40%) 2ndR 
 Left hippocampus 394 5.27 3.85 −24, −26, −10 Hipp (SUB) (30%), (FD) (20%) 2ndR 
 Right cerebellum 1205 5.54 3.97 30, −60, 32 HVI (81%), Crus I (19%) 2ndR 
 Right cerebellum 45 4.22 3.68 10, −84, −38 Crus II (84%) 2ndR 

Note: Cluster size indicates the number of voxels active in each cluster. X-coordinates with a negative value represent activity in the left hemisphere. Activations highlighted in bold were also present in more stringent conjunction analysis (1stR<>1stC && 2ndR<>2ndC; P < 0.05, FDR-corrected). The final column indicates the most active condition, that is, “Rules” indicates that rule cues (1stR + 2ndR) were more active than control cues (1stC + 2ndC). 1stR, first-order rule; 1stC, first-order control; 2ndR, second-order rule; 2ndC, second-order control.

Cerebellar activations found using SUIT normalization overlapped with the results using whole-brain analysis (Supplementary Material). However, these results were now more clearly within Crus I (56% and 100% probability of being in Crus I compared with 34% using whole-brain approach), and an additional activation was present in right hemisphere Crus II. In addition, the dominant activation with Crus I now passed the significance threshold in the more stringent conjunction analysis, further suggesting that this activation was evoked by the rule-based content present in both rule instruction cues (Fig. 2 and Table 3). A comparison between activation maps using the unified segmentation approach and SUIT is available in Supplementary Material. This result exclusively supports the second of our two hypotheses that the cerebellum processes prefrontal activity regardless of whether it contains any motor information.

Table 3

Main effect of rule cues (1stR + 2ndR <> 1stC + 2ndC) using SUIT normalization: activity time-locked to instruction cues, small volume corrected using a Crus I and Crus II mask

Rules <> controls (1stR + 2ndR <> 1stC + 2ndC) Cluster T-value Z-value Coordinates Cerebellar lobule (probability, if available) Most active condition 
Right cerebellum, Crus I 390 6.03 4.17 16, −76, −27 Crus I (56%) Rules 
 Right cerebellum, Crus I Same cluster 6.03 4.17 34, 74, 25 Crus I (100%) Rules 
 Right cerebellum, Crus II 43 4.75 3.61 28, 78, 53 Crus II (55%) Rules 
1stR <> 1stC 
 Right cerebellum, Crus I 10 5.36 3.89 44, 66, 33 Crus I (99%) 1stR 
 Left cerebellum, Crus I 12 4.56 3.51 28, 84, 55 Crus II (46%) 1stR 
2ndR <> 2ndC 
 Right cerebellum, Crus I 15 5.39 3.9 44, 42, 33 Crus I (48%) 2ndR 
 Right cerebellum, Crus I 125 4.69 3.58 34, 66, 29 Crus I (99%) 2ndR 
 Right cerebellum, Crus I 24 4.36 3.41 12, 78, 23 Crus I (71%) 2ndR 
1stR <> 2ndR 
 Left cerebellum, Crus I 223 8.01 4.83 46, 68, 31 Crus I (100%) 2ndR 
 Right cerebellum, Crus I 38 4.48 3.47 32, 64, 31 Crus I (79%) 2ndR 
 Left cerebellum, Crus I 15 4.25 3.35 32, 82, 25 Crus I (62%) 2ndR 
 Right cerebellum, Crus I 22 4.14 3.29 8, 80, 25 Crus I (78%) 2ndR 
Rules <> controls (1stR + 2ndR <> 1stC + 2ndC) Cluster T-value Z-value Coordinates Cerebellar lobule (probability, if available) Most active condition 
Right cerebellum, Crus I 390 6.03 4.17 16, −76, −27 Crus I (56%) Rules 
 Right cerebellum, Crus I Same cluster 6.03 4.17 34, 74, 25 Crus I (100%) Rules 
 Right cerebellum, Crus II 43 4.75 3.61 28, 78, 53 Crus II (55%) Rules 
1stR <> 1stC 
 Right cerebellum, Crus I 10 5.36 3.89 44, 66, 33 Crus I (99%) 1stR 
 Left cerebellum, Crus I 12 4.56 3.51 28, 84, 55 Crus II (46%) 1stR 
2ndR <> 2ndC 
 Right cerebellum, Crus I 15 5.39 3.9 44, 42, 33 Crus I (48%) 2ndR 
 Right cerebellum, Crus I 125 4.69 3.58 34, 66, 29 Crus I (99%) 2ndR 
 Right cerebellum, Crus I 24 4.36 3.41 12, 78, 23 Crus I (71%) 2ndR 
1stR <> 2ndR 
 Left cerebellum, Crus I 223 8.01 4.83 46, 68, 31 Crus I (100%) 2ndR 
 Right cerebellum, Crus I 38 4.48 3.47 32, 64, 31 Crus I (79%) 2ndR 
 Left cerebellum, Crus I 15 4.25 3.35 32, 82, 25 Crus I (62%) 2ndR 
 Right cerebellum, Crus I 22 4.14 3.29 8, 80, 25 Crus I (78%) 2ndR 

Note: Cluster size indicates the number of voxels active in each cluster. Coordinates are in SUIT space, which is similar but not identical to the MNI space. X-coordinates with a negative value represent activity in the left hemisphere. Activations highlighted in bold were also present in more stringent conjunction analysis (1stR<>1stC && 2ndR<>2ndC; P < 0.05, FDR-corrected). The final column indicates the most active condition, that is, Rules indicates that rule cues (1stR + 2ndR) were more active than control cues (1stC + 2ndC). 1stR, first-order rule; 1stC, first-order control; 2ndR, second-order rule; 2ndC, second-order control.

Figure 2.

Cerebellar activations for main effect of rule: activation for rule instruction cues greater than control instruction cues within the right hemisphere Crus I and Crus II overlayed on SUIT template. Left and right of the activations are plots of parameter estimates for medial Crus I, lateral Crus I, and lateral Crus II (arrows indicate which plots correspond to which activation). Activation in medial Crus I (circled) additionally survived a more stringent conjunction analysis of main effect of rules.

Figure 2.

Cerebellar activations for main effect of rule: activation for rule instruction cues greater than control instruction cues within the right hemisphere Crus I and Crus II overlayed on SUIT template. Left and right of the activations are plots of parameter estimates for medial Crus I, lateral Crus I, and lateral Crus II (arrows indicate which plots correspond to which activation). Activation in medial Crus I (circled) additionally survived a more stringent conjunction analysis of main effect of rules.

First-order rule versus control

A large number of regions were more active for first-order rule compared with first-order control. A number of these regions were also seen in the main effects of rules contrast described earlier. These included the left inferior frontal gyrus (pars triangularis), the left inferior parietal lobule, bilateral caudate nucleus, and medial cortical regions (middle and posterior cingulate cortices and precuneus). A small volume correction for prefrontal-projecting cerebellar lobules found a significant cluster in right Crus II specific to first-order rule compared with first-order control. This cerebellar activation cluster is spatially separate from the previously mentioned cerebellar cluster found in the whole-brain analysis (Table 2). Our SUIT analysis showed activations in medial Crus I and lateral Crus II, which overlapped with the activation described earlier (SUIT analysis of main effect of rules). We also note that the Crus II activation mentioned earlier in the whole brain and SUIT analyses overlaps with the Crus II activation found in the SUIT main effect of rules analysis (Table 3).

Second-order rule versus control

Significantly greater activity for second-order rules compared with controls was found in the visual cortex, left inferior frontal gyrus (pars triangularis), left precentral gyrus, left posterior cingulate cortex, and left hippocampus. A small volume correction for prefrontal-projecting cerebellar clusters showed 2 significant cerebellar activations: the first was in the right cerebellar lobule Crus I and overlapped with the cerebellar activation described in the main effects of rule contrast. The second activation was in the right cerebellar lobule Crus II. This cluster is spatially separate from the 2 whole-brain cerebellar activations mentioned earlier (Table 2). As in the whole-brain analysis, our SUIT analysis showed activation in the right cerebellar lobule Crus I overlapping with the activation described in the main effects of rule contrast. This activation is now more certainly in Crus I (99%), compared with the activation found in the whole-brain analysis (19%) (Table 3).

Main effect of rule order

A conjunction analysis was also used to compare (first-order rules <> second-order rules) && (first-order control <> second-order control). Only one significant difference was found within the left supramarginal gyrus, extending to the postcentral gyrus. When rule order was modeled at the first level, additional significant clusters were found, including the left and right precentral gyri (area 4a and area 4p, respectively), the left middle and right superior orbital gyri (area 11), and the left insula. An activation cluster was also present in the right hemisphere motor-projecting cerebellar lobule HVI (90%) (Table 4). The SUIT analysis showed no significant differences in the cerebellum.

Table 4

Main effect of rule order (1stR + 1stC <> 2ndR + 2ndC): activity time-locked to instruction cues, FDR-corrected for multiple comparisons (P < 0.05) in a random-effects analysis

First-order <> second-order (1stR + 1stC <> 2ndR + 2ndC) Cluster T-value Z-value Coordinates Cytoarchitectonic BA (probability, if available) Most active condition 
Left insula lobe 150 5.87 4.1 −38, −4, 12 OP3 (20%) First order 
Left postcentral Gyrus/supramarginal gyrus 1111 6.03 4.17 −58, −24, 40 IPC (PFt) (50%), area 2 (50%) First order 
Right superior orbital gyrus 147 5.06 3.75 24, 38, −14 Area 11 Second order 
Left middle orbital gyrus 156 6.39 4.3 −24, 40, −10 Area 11 Second order 
Left precentral gyrus 148 3.94 3.18 −42, −6, 34 Area 4p (20%) Second order 
Right precentral gyrus 231 6.35 4.29 42, −10, 40 Area 4a (40%) Second order 
1stR <> 2ndR 
 Right inferior frontal gyrus (pars opercularis) 140 6.32 4.28 56, 8, 12 Area 44 (60%) 1stR 
 Left insula 180 7.94 4.81 −38, −4, 12 Op 3 (20%) 1stR 
 Right supramarginal gyrus 415 6.44 4.32 58, −26, 42 IPC (PFt) (50%) 1stR 
 Left supramarginal gyrus 815 5.9 4.11 −56, −30, 38 IPC (PFt) (50%) 1stR 
 Left inferior frontal gyrus (pars triangularis) 503 6.93 4.49 −44, −6, 34 Area 44 (30%) 2ndR 
 Right inferior frontal gyrus (pars orbitalis) 238 5.94 4.13 36, 28, −6 Area 47 2ndR 
 Left lingual gyrus 2599 6.74 4.43 −20, −70, −2 hOC3v (40%) 2ndR 
First-order <> second-order (1stR + 1stC <> 2ndR + 2ndC) Cluster T-value Z-value Coordinates Cytoarchitectonic BA (probability, if available) Most active condition 
Left insula lobe 150 5.87 4.1 −38, −4, 12 OP3 (20%) First order 
Left postcentral Gyrus/supramarginal gyrus 1111 6.03 4.17 −58, −24, 40 IPC (PFt) (50%), area 2 (50%) First order 
Right superior orbital gyrus 147 5.06 3.75 24, 38, −14 Area 11 Second order 
Left middle orbital gyrus 156 6.39 4.3 −24, 40, −10 Area 11 Second order 
Left precentral gyrus 148 3.94 3.18 −42, −6, 34 Area 4p (20%) Second order 
Right precentral gyrus 231 6.35 4.29 42, −10, 40 Area 4a (40%) Second order 
1stR <> 2ndR 
 Right inferior frontal gyrus (pars opercularis) 140 6.32 4.28 56, 8, 12 Area 44 (60%) 1stR 
 Left insula 180 7.94 4.81 −38, −4, 12 Op 3 (20%) 1stR 
 Right supramarginal gyrus 415 6.44 4.32 58, −26, 42 IPC (PFt) (50%) 1stR 
 Left supramarginal gyrus 815 5.9 4.11 −56, −30, 38 IPC (PFt) (50%) 1stR 
 Left inferior frontal gyrus (pars triangularis) 503 6.93 4.49 −44, −6, 34 Area 44 (30%) 2ndR 
 Right inferior frontal gyrus (pars orbitalis) 238 5.94 4.13 36, 28, −6 Area 47 2ndR 
 Left lingual gyrus 2599 6.74 4.43 −20, −70, −2 hOC3v (40%) 2ndR 

Note: Cluster size indicates the number of voxels active in each cluster. X-coordinates with a negative value represent activity in the left hemisphere. Activations highlighted in bold were also present in more stringent conjunction analysis (1stR<>2ndR && 1stC<>2ndC; P < 0.05, FDR-corrected). The final column indicates the most active condition, that is, “First order” indicates that first-order cues (1stR + 1stC) were more active than second-order cues (2ndR + 2ndC). 1stR, first-order rule; 1stC, first-order control; 2ndR, second-order rule; 2ndC, second-order control.

First-order rule versus second-order rule

First-order rules evoked significantly greater activations in the right inferior frontal gyrus (pars opercularis), left insula, and bilateral supramarginal gyrus. Second-order rules evoked significantly greater activations in a different portion of the right inferior frontal gyrus (pars orbitalis), as well as left inferior frontal gyrus (par triangularis), and visual cortex. When comparing these activations, it is clear to see that second-order rules were processed in more anterior regions of the prefrontal cortex compared with first-order rules (Fig. 3). This is in keeping with theories of a rostrocaudal functional gradient within the prefrontal cortex and mirrors the results of other studies (Koechlin and Summerfield 2007; Badre and D'Esposito 2009). There were no significant differences between first- and second-order rules in the cerebellum (Table 4).

Figure 3.

(A) Image from Badre and D'Esposito (2007) showing frontal lobe hierarchy. Superimposed on this figure: A (blue) refers to first-order rules, B (purple) refers to second-order rules, C (green) refers to third-order rules, and D (red) refers to fourth-order rules. (B) Results from this study showing first-order rules > second-order rules (blue) and second-order rules > first-order rules (green).

Figure 3.

(A) Image from Badre and D'Esposito (2007) showing frontal lobe hierarchy. Superimposed on this figure: A (blue) refers to first-order rules, B (purple) refers to second-order rules, C (green) refers to third-order rules, and D (red) refers to fourth-order rules. (B) Results from this study showing first-order rules > second-order rules (blue) and second-order rules > first-order rules (green).

While the whole-brain analysis did not find any significant differences in the cerebellum for first- versus second-order rules, our SUIT analysis showed a number of significant cerebellar activations. This included activations in left and right cerebellar lobules Crus I and Crus II. Activation was also present in vermal lobule VI for first-order rules > second-order rules, most likely reflecting the difference in motor preparation at the time of the instruction cue (Table 3). A figure illustrating the activation and parameter estimates for each condition is presented in Supplementary Material.

Interaction

The only region showing a significant interaction was the primary visual cortex. Activity in this region was greatest for second-order rules. Neither whole-brain nor SUIT analyses showed any significant interactions within the cerebellum. Given that there were no significant interactions within prefrontal-projecting cerebellar lobules, our first hypothesis (activity within prefrontal-projecting cerebellar lobules will only be evoked by rules that specify action) was not supported.

Discussion

In this study, we investigated whether the cerebellum responds only when rules specify the properties of action or whether the cerebellum additionally processes rules relating to cognitive control independent of such action properties. We focussed our analysis on activity time-locked to the instruction cue, thus isolating cognitive processes specifically related to rule-based processing (translation of symbolic information into additional rules or actions) and removing additional confounding processes (i.e. motor responses and processing of feedback). Our results support the hypothesis that prefrontal-projecting cerebellar lobules (Crus I and Crus II) process rule-based information, regardless of whether or not rules specify actions.

Differences in Processing First- and Second-Order Rules in the Frontal Lobes

While a number of studies have proposed a rostrocaudal functional hierarchy in the frontal lobes (Koechlin et al. 2003; Badre and D'Esposito 2007, 2009; Koechlin and Summerfield 2007; Badre et al. 2009, 2010; Race et al. 2010), there are other studies that suggest that cytoarchitectonic subdivisions of the prefrontal cortex do not necessarily lead to functional subdivisions (Duncan and Owen 2000; Duncan 2001; Rowe et al. 2008). The adaptive coding model of the prefrontal cortex (Duncan 2001) suggests that any region of the prefrontal cortex is capable of integrating nearly any kind of information due to the extremely plastic nature of prefrontal neurons. Examples of this can be seen from electrophysiological recordings of the primate prefrontal cortex (Rao et al. 1997; Freedman et al. 2001) and functional neuroimaging (Duncan and Owen 2000; Rowe et al. 2008). Both Badre and D'Esposito (2007) and Rowe et al. (2008) compared dimensions of cognitive control with task competition using fMRI and arrived at different results. The work of Badre and D'Esposito (2007) showed a rostrocaudal hierarchy of cognitive control with increasingly abstract stimuli activating increasingly anterior regions of the prefrontal cortex. However, Rowe et al. (2008) specifically compared action selection (the choice between action alternatives in the absence of a specified rule) with rule selection (deciding to respond using either height or brightness as a rule compared with being specified to use highest, lowest, lightest, or darkest) and found a spatial overlap in the prefrontal cortex rather than a hierarchy of prefrontal regions. The results of our study are comparable with the results of Badre and D'Esposito (2007) along with other studies (Koechlin et al. 2003; Badre et al. 2010; Race et al. 2010), showing more anterior prefrontal activations for second-order compared with first-order rules (Fig. 3). The spatial overlap between the results of Badre and D'Esposito (2007) and this study is presented in Figure 3. First-order rules appear to activate more dorsal and caudal regions of the frontal lobe, whereas second-order rules appear to activate more rostral and ventral portions of the prefrontal cortex, most likely pre-PMd as seen in other studies (Badre and D'Esposito 2007; Badre et al. 2010). Rowe et al. (2008) suggest that differences between their results and the results of Koechlin et al. (2003) and Badre and D'Esposito (2007) may be due to differences in the experimental design. Koechlin et al. (2003) used a block design, whereas Badre and D'Esposito (2007) had an event-related design, and the levels of task abstraction (cognitive complexity) were kept constant during fMRI sessions. Rowe et al. (2008) presented all trial types pseudo-randomly within a single fMRI session. However, we also included all trial types pseudo-randomly intermixed within a single fMRI session and found a result similar to that of Badre and D'Esposito (2007). It may be the case that the additional temporal jittering of instruction cues removed some of the noise of subsequent cognitive and motor confounds and increased the signal of cognitive processes of interest, thus giving us results different from Rowe et al. (2008). It is also possible that the temporal jittering improved the temporal resolution of our fMRI response (Josephs and Henson 1999), perhaps enough to detect temporal as well as spatial dynamics of the prefrontal hierarchy (Koechlin and Summerfield 2007; Race et al. 2010).

Differences in Processing First- and Second-Order Rules in the Cerebellar Cortex

In a previous study (Balsters and Ramnani 2008), we used an identical conditional motor task comparing symbolic and direct instruction cues that could be used to prepare an action and found an activation cluster specific to symbolic instruction cues (identical to first-order rule stimuli in this study) within right hemisphere Crus I. The activation cluster found in Balsters and Ramnani (2008) is spatially consistent with the cluster presented in Figure 2, replicating our previous results that prefrontal-projecting cerebellar lobule Crus I is involved in processing first-order rules. Given that this cluster was also active for the processing of second-order rules, we can further suggest that this activation is not due to the integration of motor effectors and sensory cues, but rather relates to the abstract translation of a sensory stimulus into a future response, even when that future response is another rule and not a motor response.

Activity within left inferior frontal gyrus (pars triangularis) and right cerebellar lobule Crus I has been found in studies of language and verbal working memory (Kirschen et al. 2005; Stoodley and Schmahmann 2009). However, we would argue that processes that engage these brain regions are not restricted to verbal working memory, but a wide range of processes that includes the acquisition of rules. Lesion studies have repeatedly shown that rule learning, and the implementation of previously learned rules, is severely impaired after lesions to the ventrolateral prefrontal cortex (Murray et al. 2000; Passingham et al. 2000; Bussey et al. 2002). Similarly, a number of neuroimaging studies have shown activations in the inferior frontal gyrus during conditional motor learning (Toni and Passingham 1999; Toni, Ramnani et al. 2001; Toni, Rushworth et al. 2001; Bunge et al. 2003; Brass and von Cramon 2004). Given that our experimental design matched requirements for verbal working memory under experimental and control conditions, our results are consistent with the interpretation that the effects are related to rule-based information processing rather than verbal working memory.

As hypothesized, rule-based processes occurring at the presentation of a symbolic cue elicited activity within prefrontal-projecting cerebellar lobules (primarily Crus I but also Crus II). It is important to reiterate that while Kelly and Strick (2003) showed that connections with Crus II were more abundant than those with Crus I, both clearly have access to information from the prefrontal cortex. Other studies have investigated monosynaptic connections from prefrontal cortical regions to the pontine nuclei (Schmahmann and Pandya 1997), but to our knowledge, no other tracer study has investigated the topography of anatomical connections between the prefrontal cortex and the cerebellar cortex in greater detail than that of Kelly and Strick (2003). However, there are increasingly detailed studies investigating cortico-cerebellar connectivity in humans using resting-state fMRI (Habas et al. 2009; Krienen and Buckner 2009; O'Reilly et al. 2010; Buckner et al. 2011). These studies mostly provide support for the view that the system is similarly organized in the human brain compared with the nonhuman primates. Krienen and Buckner (2009) and O'Reilly et al. (2010) recently reported that resting-state activity in Crus II could be explained by fluctuations in resting-state activity in the dorsolateral prefrontal cortex, consistent with the findings of Kelly and Strick (2003) in capuchin monkeys. Interestingly, Krienen and Buckner (2009) showed that activity within parts of Crus I and Crus II covaried with the dorsolateral prefrontal cortex, but, in addition, there are adjacent areas of Crus I in which activity corresponds to medial portions of the prefrontal cortex. Our main effect of rules activated regions in the inferior frontal gyrus (area 45), which Schmahmann and Pandya (1997) have shown send projections to the pontine nuclei. Similarly, Ramnani et al. (2004) have demonstrated that fiber pathways in humans originating in the inferior frontal gyrus pass through the anterior segments of the cerebral peduncle before penetrating the pons. In addition to structural evidence, Buckner et al. (2011) provide functional evidence that resting fluctuations within ventral portions of the prefrontal cortex covary with resting fluctuations in the cerebellar lobule Crus I. The frontal and cerebellar activations reported in Buckner et al. (2011) appear to spatially overlap with the results of this study, and it is likely that rule-based processing activates this cortico-cerebellar circuit.

It has been suggested that eye movements explain much of the cerebellar activity associated with cognitive tasks in neuroimaging experiments and that these areas are connected with the frontal eye fields (Glickstein and Doron 2008). Although the cerebellar cortex does indeed have connections with the frontal eye fields, these are in addition to a number of other prefrontal regions that connect with the cerebellum, but make no known contributions to the kinematics of eye movements. For example, Glickstein et al. (1985) have shown dense projections to the pontine nuclei from areas 24 and 25, which are well known for their involvement in decision making and the regulation of mood, respectively (Devinsky et al. 1995; Mayberg et al. 2005; Lozano et al. 2008; Hamani et al. 2009). In our study, participants had to initiate a visual search to respond to second-order instruction cues. However, there are a number of reasons why this would not impact on the results presented here. First, visual searches were made at the time of the trigger, not at the time of the instruction cue (rule-related activity was time-locked to the instruction cues, not to the trigger cues). Given that second-order instruction cues specified a response at the time of the trigger, it is possible that there may have been eye movement preparation during second-order instruction cues that was not present for first-order instruction cues. However, our comparison of interest was between rule and control instruction cues, not first- and second-order instruction cues. The eye movement demands were equal for both first-order rules compared with first-order controls and second-order rules compared to second-order controls. This is supported by the fact that we find no evidence of activity in the eye movement circuits in this study, further suggesting that activity is unlikely to be related to eye movement demands. We do not agree that cerebellar activity in neuroimaging studies can be explained purely by eye movements, partly because most studies provide adequate experimental control for this confound. There are also, for example, studies that make cognitive demands in the total absence of any visual demands (see Hayter et al. 2007, in which there were no visual demands at all; in this experiment, stimuli were auditory and responses were verbal).

One caveat to this study is that while participants could not prepare a specific effector for second-order rules, they may have been preparing a response at the level of the whole hand or multiple digits. We would argue against this interpretation, given that the contrast for main effect of rules failed to show a significant difference in reaction times or highlight any preparatory activity within the premotor cortex (a small volume correction using a cytoarchitectonic mask of area 6; Geyer 2004). However, without a direct measure of muscle movements in the hand, we cannot completely discount this possibility. Bischoff-Grethe et al. (2002) neatly disambiguate cerebellar contributions to motor control from shifts in attention using a combination of response and no-response conditions. However, conditional motor learning paradigms require the subject to make a specific response to an instruction cue. A condition without a motor response would be qualitatively different from the other conditions in this study. The differences between the experimental conditions presented in this study and this hypothetical control condition would extend well beyond motor preparation and as such would not offer improved experimental control.

This study provides strong support for the cerebellar contributions to cognitive control; however, future research should investigate cortico-cerebellar interactions during decision making. Both Ramnani (2006) and Ito (2008) have endeavored to expand cortico-cerebellar models of motor control such as those of Kawato, Miall, and Wolpert (Kawato and Wolpert 1998; Wolpert and Kawato 1998; Wolpert et al. 1998). Ramnani (2006) predicted that activity would be greatest within the frontal lobe during the formation of novel rules or motor sequences and that activity within these regions would decrease during learning (Jueptner et al. 1997; Boettiger and D'Esposito 2005). This decrease in activity within the frontal lobe would coincide with an increase in interconnected cerebellar territories, representing the acquisition and automatization of prefrontal and motor processes. This would also lead to a shift in the locus of control, such that the cerebellum would output well-rehearsed motor and prefrontal skilled processes rather than its cortical counterpart. However, this has not been supported by previous neuroimaging studies, which mostly show a decrease in cerebellar activity during learning and automaticity (Imamizu et al. 2000; Doyon et al. 2002; Penhune and Doyon 2005; Balsters and Ramnani 2011). This could relate to the decreases in complex and simple spikes seen in electrophysiology studies of the cerebellum during skill acquisition (Gilbert and Thach 1977; De Zeeuw and Yeo 2005; Medina and Lisberger 2008; Lepora et al. 2009), or it could suggest that the cerebellum is involved in adapting and tuning cortical processes but does not act as a storage for these processes (Doyon et al. 2003; Debas et al. 2010). This is an area that requires further investigation, and we would suggest that future studies investigate cortico-cerebellar connectivity (possibly using dynamic causal modeling) to try and establish how neocortical and connected cerebellar areas interact during learning (Apps et al. 2009; Saalmann et al. 2009).

Supplementary Material

Supplementary material can be found at: http://www.cercor.oxfourdjournals.org/.

Funding

J.H.B. is supported by a fellowship from the Irish Research Council for Science Engineering and Technology (IRCSET). The authors wish to acknowledge research infrastructure funding support via the HEA PRTLI Cycle 3 program of the Irish Government's National Development Plan 2002–2006 and EU Framework funding via TCIN. The project forms part of a doctoral thesis by J.H.B., supervised by N.R. at Royal Holloway, University of London, UK. N.R. was supported by the BBSRC UK (Grant Ref. BB/E003451/1). During doctoral work, J.H.B. was supported by an RHUL-SGUL Strategic Alliance Studentship award to N.R.

Notes

The authors would like to thank Sojo Joseph for help acquiring the MRI data. The MRI protocol was developed by C.M. Kerskens with support of Sojo Joseph at the TCIN. Conflict of Interest: None declared.

References

Albus
JS
A theory of cerebellar function
Math Biosci
 , 
1971
, vol. 
10
 (pg. 
25
-
61
)
Andersson
JL
Hutton
C
Ashburner
J
Turner
R
Friston
K
Modeling geometric deformations in EPI time series
NeuroImage
 , 
2001
, vol. 
13
 (pg. 
903
-
919
)
Apps
MAJ
Lesage
E
Turner
S
Ramnani
N
Learning-related effective connectivity in the cortico-cerebellar system
Soc Neurosci Abstr
 , 
2009
Ashburner
J
Friston
KJ
Unified segmentation
NeuroImage
 , 
2005
, vol. 
26
 (pg. 
839
-
851
)
Atherton
M
Zhuang
J
Bart
WM
Hu
X
He
S
A functional MRI study of high-level cognition. I. The game of chess
Brain Res Cogn Brain Res
 , 
2003
, vol. 
16
 (pg. 
26
-
31
)
Badre
D
D'Esposito
M
Functional magnetic resonance imaging evidence for a hierarchical organization of the prefrontal cortex
J Cogn Neurosci
 , 
2007
, vol. 
19
 (pg. 
2082
-
2099
)
Badre
D
D'Esposito
M
Is the rostro-caudal axis of the frontal lobe hierarchical?
Nat Rev Neurosci
 , 
2009
, vol. 
10
 (pg. 
659
-
669
)
Badre
D
Hoffman
J
Cooney
JW
D'Esposito
M
Hierarchical cognitive control deficits following damage to the human frontal lobe
Nat Neurosci
 , 
2009
, vol. 
12
 (pg. 
515
-
522
)
Badre
D
Kayser
AS
D'Esposito
M
Frontal cortex and the discovery of abstract action rules
Neuron
 , 
2010
, vol. 
66
 (pg. 
315
-
326
)
Balsters
JH
Ramnani
N
Cerebellar plasticity and the automation of first-order rules
J Neurosci
 , 
2011
, vol. 
31
 (pg. 
2305
-
2312
)
Balsters
JH
Ramnani
N
Symbolic representations of action in the human cerebellum
NeuroImage
 , 
2008
, vol. 
43
 (pg. 
388
-
398
)
Barbas
H
Henion
TH
Dermon
CR
Diverse thalamic projections to the prefrontal cortex in the rhesus monkey
J Comp Neurol
 , 
1991
, vol. 
313
 (pg. 
65
-
94
)
Bischoff-Grethe
A
Ivry
RB
Grafton
ST
Cerebellar involvement in response reassignment rather than attention
J Neurosci
 , 
2002
, vol. 
22
 (pg. 
546
-
553
)
Boettiger
CA
D'Esposito
M
Frontal networks for learning and executing arbitrary stimulus–response associations
J Neurosci
 , 
2005
, vol. 
25
 (pg. 
2723
-
2732
)
Brass
M
von Cramon
DY
Decomposing components of task preparation with functional magnetic resonance imaging
J Cogn Neurosci
 , 
2004
, vol. 
16
 (pg. 
609
-
620
)
Buckner
RL
Krienen
FM
Castellanos
A
Diaz
JC
Yeo
BT
The organization of the human cerebellum estimated by intrinsic functional connectivity
J Neurophysiol
 , 
2011
, vol. 
106
 (pg. 
2322
-
2345
)
Bunge
SA
Kahn
I
Wallis
JD
Miller
EK
Wagner
AD
Neural circuits subserving the retrieval and maintenance of abstract rules
J Neurophysiol
 , 
2003
, vol. 
90
 (pg. 
3419
-
3428
)
Bussey
TJ
Wise
SP
Murray
EA
Interaction of ventral and orbital prefrontal cortex with inferotemporal cortex in conditional visuomotor learning
Behav Neurosci
 , 
2002
, vol. 
116
 (pg. 
703
-
715
)
Cabeza
R
Nyberg
L
Imaging cognition II: an empirical review of 275 PET and fMRI studies
J Cogn Neurosci
 , 
2000
, vol. 
12
 (pg. 
1
-
47
)
Chen
SH
Desmond
JE
Cerebrocerebellar networks during articulatory rehearsal and verbal working memory tasks
NeuroImage
 , 
2005a
, vol. 
24
 (pg. 
332
-
338
)
Chen
SH
Desmond
JE
Temporal dynamics of cerebro-cerebellar network recruitment during a cognitive task
Neuropsychologia
 , 
2005b
, vol. 
43
 (pg. 
1227
-
1237
)
Chen
X
Zhang
D
Zhang
X
Li
Z
Meng
X
He
S
Hu
X
A functional MRI study of high-level cognition. II. The game of GO
Brain Res Cogn Brain Res
 , 
2003
, vol. 
16
 (pg. 
32
-
37
)
Debas
K
Carrier
J
Orban
P
Barakat
M
Lungu
O
Vandewalle
G
Hadj Tahar
A
Bellec
P
Karni
A
Ungerleider
LG
, et al.  . 
Brain plasticity related to the consolidation of motor sequence learning and motor adaptation
Proc Natl Acad Sci USA
 , 
2010
, vol. 
107
 (pg. 
17839
-
17844
)
Desmond
JE
Chen
SH
Shieh
PB
Cerebellar transcranial magnetic stimulation impairs verbal working memory
Ann Neurol
 , 
2005
, vol. 
58
 (pg. 
553
-
560
)
Devinsky
O
Morrell
MJ
Vogt
BA
Contributions of anterior cingulate cortex to behaviour
Brain
 , 
1995
, vol. 
118
 
Pt 1
(pg. 
279
-
306
)
De Zeeuw
CI
Yeo
CH
Time and tide in cerebellar memory formation
Curr Opin Neurobiol
 , 
2005
, vol. 
15
 (pg. 
667
-
674
)
Diedrichsen
J
A spatially unbiased atlas template of the human cerebellum
NeuroImage
 , 
2006
, vol. 
33
 (pg. 
127
-
138
)
Diedrichsen
J
Balsters
JH
Flavell
J
Cussans
E
Ramnani
N
A probabilistic MR atlas of the human cerebellum
NeuroImage
 , 
2009
, vol. 
46
 (pg. 
39
-
46
)
Doyon
J
Penhune
V
Ungerleider
LG
Distinct contribution of the cortico-striatal and cortico-cerebellar systems to motor skill learning
Neuropsychologia
 , 
2003
, vol. 
41
 (pg. 
252
-
262
)
Doyon
J
Song
AW
Karni
A
Lalonde
F
Adams
MM
Ungerleider
LG
Experience-dependent changes in cerebellar contributions to motor sequence learning
Proc Natl Acad Sci USA
 , 
2002
, vol. 
99
 (pg. 
1017
-
1022
)
Duncan
J
An adaptive coding model of neural function in prefrontal cortex
Nat Rev Neurosci
 , 
2001
, vol. 
2
 (pg. 
820
-
829
)
Duncan
J
Owen
AM
Common regions of the human frontal lobe recruited by diverse cognitive demands
Trends Neurosci
 , 
2000
, vol. 
23
 (pg. 
475
-
483
)
Duvernoy
HM
Bourgouin
P
The human brain: surface, three-dimensional sectional anatomy and MRI
 , 
1999
Wein
Springer-Verlag
Eickhoff
SB
Stephan
KE
Mohlberg
H
Grefkes
C
Fink
GR
Amunts
K
Zilles
K
A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data
NeuroImage
 , 
2005
, vol. 
25
 (pg. 
1325
-
1335
)
Freedman
DJ
Riesenhuber
M
Poggio
T
Miller
EK
Categorical representation of visual stimuli in the primate prefrontal cortex
Science
 , 
2001
, vol. 
291
 (pg. 
312
-
316
)
Friston
KJ
Frith
CD
Frackowiak
RS
Turner
R
Characterizing dynamic brain responses with fMRI: a multivariate approach
NeuroImage
 , 
1995
, vol. 
2
 (pg. 
166
-
172
)
Friston
KJ
Frith
CD
Turner
R
Frackowiak
RS
Characterizing evoked hemodynamics with fMRI
NeuroImage
 , 
1995
, vol. 
2
 (pg. 
157
-
165
)
Friston
KJ
Penny
WD
Glaser
DE
Conjunction revisited
NeuroImage
 , 
2005
, vol. 
25
 (pg. 
661
-
667
)
Funahashi
S
Neuronal mechanisms of executive control by the prefrontal cortex
Neurosci Res
 , 
2001
, vol. 
39
 (pg. 
147
-
165
)
Fuster
JM
The prefrontal cortex-anatomy physiology, and neuropsychology of the frontal lobe
 , 
1997
Philadelphia (PA)
Lippincott-Raven
Geyer
S
The microstructural border between the motor and the cognitive domain in the human cerebral cortex
Adv Anat Embryol Cell Biol
 , 
2004
, vol. 
174
 
I-VIII
(pg. 
1
-
89
)
Gilbert
PF
Thach
WT
Purkinje cell activity during motor learning
Brain Res
 , 
1977
, vol. 
128
 (pg. 
309
-
328
)
Glickstein
M
Cerebellum and the sensory guidance of movement
Novartis Found Symp
 , 
1998
, vol. 
218
 (pg. 
252
-
266
discussion 266-271, 332-253
Glickstein
M
Doron
K
Cerebellum: connections and functions
Cerebellum
 , 
2008
, vol. 
7
 (pg. 
589
-
594
)
Glickstein
M
May
JG
Mercier
BE
Corticopontine projection in the macaque: the distribution of labelled cortical cells after large injections of horseradish peroxidase in the pontine nuclei
J Comp Neurol
 , 
1985
, vol. 
235
 (pg. 
343
-
359
)
Goldman-Rakic
PS
Porrino
LJ
The primate mediodorsal (MD) nucleus and its projection to the frontal lobe
J Comp Neurol
 , 
1985
, vol. 
242
 (pg. 
535
-
560
)
Habas
C
Kamdar
N
Nguyen
D
Prater
K
Beckmann
CF
Menon
V
Greicius
MD
Distinct cerebellar contributions to intrinsic connectivity networks
J Neurosci
 , 
2009
, vol. 
29
 (pg. 
8586
-
8594
)
Halsband
U
Passingham
RE
Premotor cortex and the conditions for movement in monkeys (Macaca fascicularis)
Behav Brain Res
 , 
1985
, vol. 
18
 (pg. 
269
-
277
)
Hamani
C
Mayberg
H
Snyder
B
Giacobbe
P
Kennedy
S
Lozano
AM
Deep brain stimulation of the subcallosal cingulate gyrus for depression: anatomical location of active contacts in clinical responders and a suggested guideline for targeting
J Neurosurg
 , 
2009
, vol. 
111
 (pg. 
1209
-
1215
)
Hayter
AL
Langdon
DW
Ramnani
N
Cerebellar contributions to working memory
NeuroImage
 , 
2007
, vol. 
36
 (pg. 
943
-
954
)
Holmes
G
The cerebellum of man
Brain
 , 
1939
, vol. 
62
 (pg. 
1
-
30
)
Hutton
C
Bork
A
Josephs
O
Deichmann
R
Ashburner
J
Turner
R
Image distortion correction in fMRI: a quantitative evaluation
NeuroImage
 , 
2002
, vol. 
16
 (pg. 
217
-
240
)
Imamizu
H
Miyauchi
S
Tamada
T
Sasaki
Y
Takino
R
Putz
B
Yoshioka
T
Kawato
M
Human cerebellar activity reflecting an acquired internal model of a new tool
Nature
 , 
2000
, vol. 
403
 (pg. 
192
-
195
)
Ito
M
Mechanisms of motor learning in the cerebellum
Brain Res
 , 
2000
, vol. 
886
 (pg. 
237
-
245
)
Ito
M
Control of mental activities by internal models in the cerebellum
Nat Rev Neurosci
 , 
2008
, vol. 
9
 (pg. 
304
-
313
)
Josephs
O
Henson
RN
Event-related functional magnetic resonance imaging: modelling, inference and optimization
Phil Trans R Soc Lond B Biol Sci
 , 
1999
, vol. 
354
 (pg. 
1215
-
1228
)
Jueptner
M
Stephan
KM
Frith
CD
Brooks
DJ
Frackowiak
RS
Passingham
RE
Anatomy of motor learning. I. Frontal cortex and attention to action
J Neurophysiol
 , 
1997
, vol. 
77
 (pg. 
1313
-
1324
)
Kawato
M
Wolpert
D
Internal models for motor control
Novartis Found Symp
 , 
1998
, vol. 
218
 (pg. 
291
-
304
discussion 304-297
Kelly
RM
Strick
PL
Cerebellar loops with motor cortex and prefrontal cortex of a nonhuman primate
J Neurosci
 , 
2003
, vol. 
23
 (pg. 
8432
-
8444
)
Kim
SG
Ugurbil
K
Strick
PL
Activation of a cerebellar output nucleus during cognitive processing
Science
 , 
1994
, vol. 
265
 (pg. 
949
-
951
)
Kirschen
MP
Chen
SH
Schraedley-Desmond
P
Desmond
JE
Load- and practice-dependent increases in cerebro-cerebellar activation in verbal working memory: an fMRI study
NeuroImage
 , 
2005
, vol. 
24
 (pg. 
462
-
472
)
Koechlin
E
Ody
C
Kouneiher
F
The architecture of cognitive control in the human prefrontal cortex
Science
 , 
2003
, vol. 
302
 (pg. 
1181
-
1185
)
Koechlin
E
Summerfield
C
An information theoretical approach to prefrontal executive function
Trends Cogn Sci
 , 
2007
, vol. 
11
 (pg. 
229
-
235
)
Krienen
FM
Buckner
RL
Segregated fronto-cerebellar circuits revealed by intrinsic functional connectivity
Cereb Cortex
 , 
2009
, vol. 
19
 (pg. 
2485
-
2497
)
Lepora
NF
Porrill
J
Yeo
CH
Evinger
C
Dean
P
Recruitment in retractor bulbi muscle during eyeblink conditioning: EMG analysis and common-drive model
J Neurophysiol
 , 
2009
, vol. 
102
 (pg. 
2498
-
2513
)
Lozano
AM
Mayberg
HS
Giacobbe
P
Hamani
C
Craddock
RC
Kennedy
SH
Subcallosal cingulate gyrus deep brain stimulation for treatment-resistant depression
Biol Psychiatr
 , 
2008
, vol. 
64
 (pg. 
461
-
467
)
Marr
D
A theory of cerebellar cortex
J Physiol
 , 
1969
, vol. 
202
 (pg. 
437
-
470
)
Mayberg
HS
Lozano
AM
Voon
V
McNeely
HE
Seminowicz
D
Hamani
C
Schwalb
JM
Kennedy
SH
Deep brain stimulation for treatment-resistant depression
Neuron
 , 
2005
, vol. 
45
 (pg. 
651
-
660
)
Medina
JF
Lisberger
SG
Links from complex spikes to local plasticity and motor learning in the cerebellum of awake-behaving monkeys
Nat Neurosci
 , 
2008
, vol. 
11
 (pg. 
1185
-
1192
)
Middleton
FA
Strick
PL
Basal ganglia and cerebellar loops: motor and cognitive circuits
Brain Res Brain Res Rev
 , 
2000
, vol. 
31
 (pg. 
236
-
250
)
Middleton
FA
Strick
PL
Cerebellar projections to the prefrontal cortex of the primate
J Neurosci
 , 
2001
, vol. 
21
 (pg. 
700
-
712
)
Miller
EK
Cohen
JD
An integrative theory of prefrontal cortex function
Annu Rev Neurosci
 , 
2001
, vol. 
24
 (pg. 
167
-
202
)
Murray
EA
Bussey
TJ
Wise
SP
Role of prefrontal cortex in a network for arbitrary visuomotor mapping
Exp Brain Res
 , 
2000
, vol. 
133
 (pg. 
114
-
129
)
O'Reilly
JX
Beckmann
CF
Tomassini
V
Ramnani
N
Johansen-Berg
H
Distinct and overlapping functional zones in the cerebellum defined by resting state functional connectivity
Cereb Cortex
 , 
2010
, vol. 
20
 (pg. 
953
-
965
)
Passingham
RE
Toni
I
Rushworth
MF
Specialisation within the prefrontal cortex: the ventral prefrontal cortex and associative learning
Exp Brain Res
 , 
2000
, vol. 
133
 (pg. 
103
-
113
)
Penhune
VB
Doyon
J
Cerebellum and M1 interaction during early learning of timed motor sequences
NeuroImage
 , 
2005
, vol. 
26
 (pg. 
801
-
812
)
Petrides
M
Frontal lobes and behaviour
Curr Opin Neurobiol
 , 
1994
, vol. 
4
 (pg. 
207
-
211
)
Petrides
M
Motor conditional associative-learning after selective prefrontal lesions in the monkey
Behav Brain Res
 , 
1982
, vol. 
5
 (pg. 
407
-
413
)
Petrides
M
Dehaene
S
Duhamel
GR
Hauser
M
Rizzolatti
G
The rostral-caudal axis of cognitive control within the lateral frontal cortex
From monkey brain to human brain
 , 
2005
Cambridge, MA
MIT Press
(pg. 
293
-
314
)
Picard
N
Strick
PL
Imaging the premotor areas
Curr Opin Neurobiol
 , 
2001
, vol. 
11
 (pg. 
663
-
672
)
Price
CJ
Friston
KJ
Cognitive conjunction: a new approach to brain activation experiments
NeuroImage
 , 
1997
, vol. 
5
 (pg. 
261
-
270
)
Race
EA
Badre
D
Wagner
AD
Multiple forms of learning yield temporally distinct electrophysiological repetition effects
Cereb Cortex
 , 
2010
, vol. 
20
 (pg. 
1726
-
1738
)
Ramnani
N
Frontal lobe and posterior parietal contributions to the cortico-cerebellar system
 , 
2011
Cerebellum
Ramnani
N
The primate cortico-cerebellar system: anatomy and function
Nat Rev Neurosci
 , 
2006
, vol. 
7
 (pg. 
511
-
522
)
Ramnani
N
Behrens
TE
Johansen-Berg
H
Richter
MC
Pinsk
MA
Andersson
JL
Rudebeck
P
Ciccarelli
O
Richter
W
Thompson
AJ
, et al.  . 
The evolution of prefrontal inputs to the cortico-pontine system: diffusion imaging evidence from macaque monkeys and humans
Cereb Cortex
 , 
2006
, vol. 
16
 (pg. 
811
-
818
)
Ramnani
N
Miall
RC
A system in the human brain for predicting the actions of others
Nat Neurosci
 , 
2004
, vol. 
7
 (pg. 
85
-
90
)
Ramnani
N
Miall
RC
Instructed delay activity in the human prefrontal cortex is modulated by monetary reward expectation
Cereb Cortex
 , 
2003
, vol. 
13
 (pg. 
318
-
327
)
Ramnani
N
Owen
AM
Anterior prefrontal cortex: insights into function from anatomy and neuroimaging
Nat Rev Neurosci
 , 
2004
, vol. 
5
 (pg. 
184
-
194
)
Ramnani
N
Rudebeck
P
Behrens
T
Johansen-Berg
H
Matthews
PM
The organisation of prefrontal projections to the cerebellum in the human brain: an in-vivo diffusion imaging study
 , 
2004
Organ. Hum. Brain Mapp. Abstr
Rao
SC
Rainer
G
Miller
EK
Integration of what and where in the primate prefrontal cortex
Science
 , 
1997
, vol. 
276
 (pg. 
821
-
824
)
Rowe
J
Hughes
L
Eckstein
D
Owen
AM
Rule-selection and action-selection have a shared neuroanatomical basis in the human prefrontal and parietal cortex
 , 
2008
, vol. 
18
 
Cereb Cortex
(pg. 
2275
-
2285
)
Saalmann
Y
Balsters
JH
Wright
MJ
Ramnani
N
Learning rule changes between the prefrontal cortex and cerebellum
 , 
2008
, vol. 
47
 
Organization for Human Brain Mapping Abstracts
pg. 
S98
 
Schmahmann
JD
Doyon
J
Toga
A
Evans
A
Petrides
M
MRI atlas of the human cerebellum
 , 
2000
San Diego (CA)
Academic Press
Schmahmann
JD
Pandya
DN
Anatomic organization of the basilar pontine projections from prefrontal cortices in rhesus monkey
J Neurosci
 , 
1997
, vol. 
17
 (pg. 
438
-
458
)
Stein
JF
Glickstein
M
Role of the cerebellum in visual guidance of movement
Physiol Rev
 , 
1992
, vol. 
72
 (pg. 
967
-
1017
)
Stoodley
CJ
Schmahmann
JD
Functional topography in the human cerebellum: a meta-analysis of neuroimaging studies
NeuroImage
 , 
2009
, vol. 
44
 (pg. 
489
-
501
)
Strick
PL
Dum
RP
Fiez
JA
Cerebellum and nonmotor function
Annu Rev Neurosci
 , 
2009
, vol. 
32
 (pg. 
413
-
434
)
Toni
I
Passingham
RE
Prefrontal-basal ganglia pathways are involved in the learning of arbitrary visuomotor associations: a PET study
Exp Brain Res
 , 
1999
, vol. 
127
 (pg. 
19
-
32
)
Toni
I
Ramnani
N
Josephs
O
Ashburner
J
Passingham
RE
Learning arbitrary visuomotor associations: temporal dynamic of brain activity
NeuroImage
 , 
2001
, vol. 
14
 (pg. 
1048
-
1057
)
Toni
I
Rushworth
MF
Passingham
RE
Neural correlates of visuomotor associations. Spatial rules compared with arbitrary rules
Exp Brain Res
 , 
2001
, vol. 
141
 (pg. 
359
-
369
)
Toni
I
Schluter
ND
Josephs
O
Friston
K
Passingham
RE
Signal-, set- and movement-related activity in the human brain: an event-related fMRI study
Cereb Cortex
 , 
1999
, vol. 
9
 (pg. 
35
-
49
)
Walker
AE
A cytoarchitectural study of the prefrontal area of the macaque monkey
J Comp Neurol
 , 
1940
, vol. 
73
 (pg. 
59
-
86
)
Wallis
JD
Anderson
KC
Miller
EK
Single neurons in prefrontal cortex encode abstract rules
Nature
 , 
2001
, vol. 
411
 (pg. 
953
-
956
)
Wise
SP
The primate premotor cortex: past, present, and preparatory
Annu Rev Neurosci
 , 
1985
, vol. 
8
 (pg. 
1
-
19
)
Wolpert
DM
Kawato
M
Multiple paired forward and inverse models for motor control
Neural Netw
 , 
1998
, vol. 
11
 (pg. 
1317
-
1329
)
Wolpert
DM
Miall
C
Kawato
M
Internal models in the cerebellum
Trends Cogn Sci
 , 
1998
, vol. 
2
 (pg. 
338
-
347
)