Significance

The present research investigates whether right and left-handers show differences in terms of brain activity (i.e., area involved) when they perform or imagine a movement with the right, left, or both hands. The study was performed with functional magnetic resonance imaging and for the first time the same paradigm was applied to both right and left-handers. Our results show significant differences in brain activity between the groups during both motor imagery and movement execution. Such results could be used to exploit new clinical assessments as well to ameliorate rehabilitation planning (e.g., personalized treatment).

Highlights We are the first to investigate brain activity during different motor imagery (MI) and motor execution (ME) tasks in both left- and right-handed participants. We found significant differences (p < 0.05) in brain activity between left- and right-handed participants during both MI and ME. During MI in both groups, the highest peak was found over the ipsilateral hemisphere for MI task. During MI, only right-handers showed the involvement of the posterior cerebellum. MI is influenced by the way people habitually perform motor actions. 1 INTRODUCTION

Nine out of 10 individuals in the world (Baraldi et al., 1999; Kim et al., 1993; Singh et al., 1998) show right-hand dominance. Although handedness is the primary manifestation of hemispheric lateralization, it is still debatable how this preference is reflected in the motor organization of the brain. Studies over the past decades have found agreement on the activation of the primary motor and sensory areas, as well as the motor association areas and cerebellum for complex movement production. Additionally, several functional magnetic resonance imaging (fMRI) studies revealed contralateral and ipsilateral activation within motor-related areas (Baraldi et al., 1999; Kim et al., 1993; Singh et al., 1998). Although the main observed activity for unimanual movement is on the contralateral primary motor cortex (M1) and ipsilateral cerebellum, activations have also been detected on the same side of the performing hand. In the study of Kim et al. (1993), results showed that the right motor cortex was activated during contralateral movements in both right and left-handers. Left motor cortex activity was instead found during ipsilateral movements in both groups and seemed more pronounced in right-handed participants. Nevertheless, there is also evidence of no difference in brain activation in left and right-handers (Dassonville et al., 1997) and results supporting that only left-handers have bilateral activation in the lateral premotor cortex (Kawashima et al., 1997). A further question of interest is whether these differences are present during motor imagery (MI) tasks.

MI is defined as an internal representation of simple or complex movements in the absence of physical action, meaning that it is not accompanied by any kind of peripheral muscular activity (Annett, 1995; Jeannerod, 1995; Porro et al., 1996). Jeannerod (1995) first differentiated it from the broader class of mental imagery, a quasi-perceptual experience occurring in the absence of perceptual input (Cattaneo & Silvanto, 2015). According to the author, MI is part of motor representation, a broader phenomenon related to planning and preparing movements. More precisely, if motor representation usually is a nonconscious process, MI represents conscious access to the content of a movement. Thus, MI and motor representation would share functional properties (i.e., motor images), suggesting a functional equivalence in the causal role of movement generation. Several neuroimaging studies attempted to disentangle this question finding mixed results (Decety, 1996; Fitts & Peterson, 1964; Jeannerod, 1994, 1995; Viviani & Schneider, 1991). Overlapping activations were found in primary motor cortex (M1), supplementary motor area (SMA), premotor cortex (PMC), inferior parietal lobule (IPL), superior parietal lobule (SPL), primary somatosensory (S1) cortex, and cerebellum (Gerardin, 2000; Guillot et al., 2008; Hanakawa et al., 2003; Lotze et al., 1999; Michelon et al., 2006; Roth et al., 1996; Solodkin et al., 2004; Szameitat et al., 2007a). Concerning subcortical areas, ipsilateral cerebellar activity over lobules V and VIII (van der Zwaag et al., 2013) has been reported for movement execution (ME) of unimanual tasks with lobules IV, V, and VIII considered representing the cerebellar homunculus (Mottolese et al., 2013). While anterior cerebellar lobules project to contralateral cerebral cortical motor-related areas, the cerebellar hemispheres are mainly connected to contralateral cerebral association networks (Bostan et al., 2013; Buckner, 2013; Buckner et al., 2011). Thus, if anterior lobules (lobules I–V) are associated with the execution of ipsilateral simple, repetitive movements, functional heterogeneity was reported for the superior-posterior cerebellum. In detail, higher-order cognition is associated with Crus I and Crus II and a combination of complex motor and cognitive functions is linked to lobules VI and VIIb. Furthermore, there is evidence that the posterior cerebellum (lobules VI through IX) is involved in the inhibition of ME (Lotze et al., 1999) and lobule VI in cognitively demanding tasks (Stoodley et al., 2012) and MI (Sakai et al., 1999; for a Review see Hétu et al., 2013). Although further evidence suggests that MI and ME networks are distinct (Gao et al., 2011; Gerardin, 2000; Jeannerod, 2001; Stephan & Frackowiak, 1996; Vigneswaran et al., 2013), it is generally accepted that the distribution of activation tends to be similar.

What it is still lacking is understanding how MI differs across individuals on the base of their experience both at the cortical and subcortical levels.

With specific regard to hand movements and handedness, the question is whether MI neural correlates vary between right and left-handers and to which extent this difference is resembling the one found for ME. In this view, MI should be considered “body-specific” (Casasanto, 2009), meaning that it is influenced by how people habitually perform motor actions. Casasanto (2009) found that during MI for complex hand actions, the activation of cortical areas involved in motor planning and execution was left-lateralized in right-handers but right-lateralized in left-handers. Thus, MI reflects the difference found for ME between right and left-hander, supporting that long-term motor history (i.e., a preference to execute an action with one hand) also influences MI. This is in line with the “simulation hypothesis” (Decety, 1996; Jeannerod, 1994, 1995), where MI of action involves the recruitment of the same neural networks in the motor system that are engaged when the movement is actually being executed. An alternative position defines MI as a more abstract implementation of general kinematic rules of biological motion, such as the selection of an action's goal (Fitts & Peterson, 1964; Viviani & Schneider, 1991). Hence, the motor plan generated during MI is abstracted away from the individual motor experience or specific effectors and occurs at the level of goal of the imagined action (Rijntjes et al., 1999). Rijntjes et al. (1999) showed that movement parameters of highly trained movement are stored in secondary sensorimotor cortices of the extremity with which it is usually performed (i.e., dorsal and ventral lateral premotor cortices). These areas are, therefore, functionally independent from the primary representation of the effector. This issue is still in debate and several approaches have been used to address this question (e.g., EEG, fMRI, brain-damaged patients, behavioral techniques), finding mixed results in support of one of the two hypotheses.

Overall, most research supports the “simulation hypothesis” (Decety, 1996; Jeannerod, 1994, 1995) reporting EEG and fMRI studies which were able to discriminate MI tasks according to the specific effector imagined. For instance, the results of the study of Perruchoud et al. (2016) illustrate how our brain is specialized in representing visual and sensorimotor aspects of our body. More in detail, they highlighted the difference between hand and body MI in recruiting two distinct brain networks (i.e., local and global bodily representations).

Furthermore, different fMRI studies suggest that hand posture influences MI (Nico, 2004; Shenton et al., 2004). Ehrsson et al. (2003) showed that MI of hand movements activates the hand sections of the contralateral motor-related areas (i.e., M1, dPM, PCC, and SMA). Consistent with this hypothesis, Szameitat et al. (2007) demonstrated that during MI of complex everyday movements different lateralization in right-handers was found when imagining actions with the right hand as compared to actions with the left hand. Willems et al. (2009) demonstrated differential and opposite lateralization for the two groups during MI of action verbs, suggesting that right- and left-handers represent action verb meanings from an egocentric perspective, which reflects the way they perform these actions with their dominant hand. Additional findings from Gentilucci et al. (1998) postulated differences in the MI mechanism between left and right-handers, suggesting that the former rely more on a pictorial hand representation, whereas the latter on a pragmatic one.

Thus, fMRI studies corroborate the hypothesis that right and left-handed individuals show different sensorimotor processing reflected in distinct brain activity patterns during MI tasks. With specific regard to handedness, it implies that activation of motor areas during MI is different in right- and left-handed individuals.

Although these differences have been highly investigated, the majority of fMRI studies concern only right-handed participants. Furthermore, to the best of our knowledge, there is no study directly investigating the difference between right and left-handers' MI neural correlates per se and especially in the cerebellum. To achieve a deeper understanding of handedness effects on MI in both right and left-handed individuals, the present study used fMRI to compare neural correlates associated with the execution and imagination of a simple task (i.e., squeezing a ball) with the dominant, non-dominant, and both hands.

2 MATERIALS AND METHODS 2.1 Participants

Fifty-one healthy participants, 26 right-handed (13 males and 13 females) and 25 left-handed (12 males and 13 females), between 19 and 32 years of age (mean = 24.61 years; SD = 3.09) volunteered to participate in the experiment for compensation of 20 euros (Table 1). Exclusion criteria included any major medical illness that could impact brain function, such as neurological or psychiatric conditions. Besides, no participants had any contraindications to MRI such as having chronic diseases, being under medication, being pregnant, having metallic or electrically conductive implants or prostheses, having tattoos on the head or neck area, nicotine patches, or cosmetic eye manipulations. All participants were informed about the purpose of the study before giving their written consent. The study was approved by the local ethics committee (Medical University of Graz) and is in accordance with the ethical standards of the Declaration of Helsinki. Handedness assessment was calculated for each participant through the Edinburgh Handedness Inventory (EHI; Oldfield, 1971) and Hand Dominance Test (HDT; Steingrüber & Lienert, 1971). At the end of the experiment, they were also asked to fill the second version of the Vividness of Movement Imagery Questionnaire (VMIQ-2; Kinesthetic part only; see Extended Data Figure 6) to assess participants’ kinesthetic MI ability.

TABLE 1. Demographic characteristics of the sample (N = 51) Number 51 Age M = 24.61 (SD = 3.09) Handedness Right = 26 Male = 13 Female = 13 Left = 25 Male = 12 Female = 13 Note Data are presented as sample size (N) for number and handedness. Right = right-handed; Left = left-handed. Age is reported in terms of mean (M) and standard deviation (SD). 2.2 Anti-COVID-19 measures

This study took place in accordance with the recommended measures of the Austrian Ministry of Social Affairs. All employees in the laboratory wore mouth and nose protection (MNS) and maintained a distance of 1 m from each other. The maximum number of study participants present in the laboratory was one person at any time and all employees checked their state of health daily using a COVID-19 questionnaire (see Extended Data Figure 7). To clarify whether there was a suspected corona case, we request the COVID-19 questionnaire to each participant who was allowed to enter the laboratory only if such a questionnaire was available and checked. Participants were picked up at a meeting point outside the university building. This was followed by the review of the questionnaire as well as the participant side setting up of MNS. Participants were always accompanied when entering and leaving the laboratory, and participants were not allowed to meet at any time. Participants wore MNS when they entered the laboratory and were asked to wash their hands. The person authorized to carry out the MR measurement wore MNS (FFP2) and a face shield. Participants kept their MNS on until immediately before attaching the head coil and then removed it themselves (at the end of the measurement, it was put back on before leaving the scanner room). After the measurement, all objects that were in direct contact or in close proximity to the test participants were disinfected (Incidin TM Liquid; required minimum exposure time 10 min sufficient for the surface disinfection). The scanner room has an exhaust system. All other rooms were ventilated for 5 min. The next participant was allowed to enter the laboratory only when the cleaning and disinfection were completed.

2.3 Procedure

Each session lasted around one and a half hours and took place in two separate rooms at the MRI Lab Graz. The whole experiment included two phases: the preparation phase and the fMRI scanning. The first part (i.e., preparation phase) took place in a separate room equipped with a computer, a table, and chairs. Here, the participants were asked to read an information letter in which they were given further information about the study, fill and sign the informed consent and the participant protocol.

In order to assess hand dominance, each participant completed the Hand Dominance Test (HDT; Steingrüber & Lienert, 1971). Then, to familiarize themselves with the experimental conditions, all participants were instructed to squeeze a ball several times at their own pace. Next, the concept of MI was explained verbally and through a written definition which provides the precise difference between kinesthetic, internal, and external visual imagery. Then, they were instructed to “imagine repeatedly squeezing the ball with kinesthetic perspective at your own pace during the task periods of ‘Imagery’ and not to change the pace during the experiment.” Once they felt ready, a short version of the paradigm was shown, including both the execution and imagery tasks presented on the computer.

The second phase started with the preparation according to the safety measurements for undergoing an MRI session. Then, the participant entered the scanner room where he performed in a single session both the imagery and execution tasks. Participants were lying on the scanner table and cushions were used to reduce head motion. Earplugs were also provided to reduce MRI scanner noise. Verbal instructions were repeated through a microphone from the outside of the scanner room, and once the participant felt ready, scanning started.

The first block included 3.47 min of T1 structural MRI (Figure 1). The second block was the MI task. Both conditions (i.e., MI and ME) had the same visual input. To create the stimulus material, we took photographs of a model naturally squeezing a ball using the right hand, left hand, and both hands (Figure 1). Visual stimuli were presented on an LCD screen (NNL MR compatible 32-inch monitor) with Presentation Version 20.0 Build 07.26.17 (Neurobehavioral System, Inc., www.neurobs.com). Participants looked at the screen located at the backside of the scanner through a mirror system. Each participant performed first the imagination task and then the execution task. We decided not to randomize the order of the tasks (i.e., MI, ME) since previous studies showed that just a few minutes of motor exercise on the eve of the imagery task leads to stronger cerebral activation in motor-related areas (Wriessnegger et al., 2014). Thus, to prevent ME influence on the cerebral activation, each participant performed the two tasks in the following order: (1) “Imagery,” where participants imagined squeezing a foam ball (7 cm diameter) with the right, left, or both hands without holding the ball; (2) “Execution,” where participants squeezed the foam ball (7 cm diameter) with the right, left, or both hands while holding the ball. Only at the beginning of the MI task, the word “Imagery” appeared on the screen. Each trial started with a fixation cross, followed by a picture showing a right hand, left hand, or both hands indicating which condition had to be executed. The order of trials for each condition (left/right/both hands) was randomized across participants. Thirty trials of 7 s for each condition (left/right/both hands) were randomized across participants. A jitter interval between 5 and 9 s of resting time (fixation cross) took place between trials. Thus, the duration of the task was not the same for all the participants. The MI task lasted for approximately 20 min. Then, participants could rest for 5 min while DTI data were acquired. ME task followed. Before the ME run started, the experimenter entered the scanner room to provide the balls to the participant. Only at the beginning of the ME task, the word “Execution” appeared on the screen. ME task included 30 trials of 7 s for each condition (left/right/both hands) randomized across participants. A jitter interval between 5 and 9 s of resting time took place between trials. Thus, the duration of the task was not the same for all the participants. The ME task lasted for approximately 10 min. After the experiment, the participants filled out the modified version of the VMIQ-2 concerning the kinesthetic modality only.

Timeline of the experiment. The trial started with 3.47 min of T1 structural MRI followed by the motor imagery (MI) task (approximately 20 min). DTI data were then acquired (5 min), followed by the motor execution (ME) task (approximately 10 min). Detailed timing of the motor imagery task is here displayed. Only at the beginning of the MI task, the word “Imagery” appeared on the screen. Each trial started with a fixation cross (3 s) followed by a picture showing a right hand, left hand, or both hands indicating which condition had to be executed (7 s). A jitter interval between 5 and 9 s of resting time (fixation cross) took place between trials. Trials for each condition (left/right/both hands) were randomized across participants. The same procedure was applied for the ME task with the only difference that at the beginning the word “Execution” was displayed

2.4 Paradigm validation

To control for active movements during the MI task, participants were observed through a monitor while no electromyogram (EMG) was recorded over participants’ hands. To validate the MI task, we performed a multivariate pattern analysis with ProNTo v.2.1 (Schrouff et al., 2013). A support vector machine (SVM) classification was performed to find differences between whole-brain functional maps of left and right MI. We used the “Leave One Subject per Group Out” Cross-Validation scheme. Results indicate a very good classification for the two different imagery tasks (See Figure 8 in the Extended Data). With a total accuracy of 93.62% and a balanced accuracy for left MI of 91.49 and 95.74% for right MI, we can be quite certain that all participants were engaged in the task.

2.5 Image acquisition

The experiment was conducted using a 3T MRI scanner (SIEMENS Magnetom VIDA, Syngo MR XA20) at the MRI Lab Graz. A 64-channel head coil was used for acquisition. High-resolution T1-weighted anatomical 3D scans were recorded and reconstructed in sagittal plane (MPRAGE sequence; voxel size 1.0 × 1.0 × 1.0 mm3, number of slices = 192, repetition time (TR) = 1,600 ms, echo time (TE) = 2.36 ms, flip angle = 9°). Functional images were acquired using SIEMENS SMS EPI (Simultaneous Multislice Echoplanar Imaging) sequence with the following parameters: voxel size = 1.8 × 1.8 × 1.8 mm3, number of slices = 78, TR = 2,300 ms, TE = 30.00 ms, flip angle = 72° SMS factor = 3, parallel imaging with acceleration factor = 2 (total PAT = 6) using an interleaved multi-slice mode for each experimental session (i.e., Imagery and Execution). Functional images were acquired using AP phase encoding direction. Field maps were acquired with inverse coding direction. Multishell DTI images were acquired using SIEMENS SMS EPI (Simultaneous Multislice Ecoplanar Imaging) sequence with voxel size = 2.5 × 2.5 × 2.5 mm3, number of slices = 57, TR = 2,800 ms, TE = 95.00 ms, SMS factor = 3, parallel imaging with acceleration factor = 2 (total PAT = 6) using fat-saturation. A custom diffusion vector set with 100 directions and the following b-values (b = 0; 1,000; 3,000 s/mm2) was used.

2.6 Data analysis

The software packages used for preprocessing was fMRIPrep 20.1.1 (Esteban, Blair, et al., 2018; Esteban, Markiewicz, et al., 2018; RRID:SCR_016216), which is based on Nipype 1.5.0 (Esteban, Markiewicz, et al., 2018; Gorgolewski et al., 2011; RRID:SCR_002502).

fMRI first-level data analysis was performed using Fitlins (version: 0.6.2, 2017–2019, Center for Reproducible Neuroscience; https://github.com/poldracklab /fitlins). For each participant, a random-effects analysis (RFX) of the activation during each of the three experimental conditions (i.e., right-hand, left-hand, and both hand execution) was performed for the two runs (i.e., MI and ME). To test for condition effects at the first level, a general linear model was constructed with the following regressors of interest: left, right, and both hands. Additionally, we also modeled regressors of no interest, such as framewise displacement, three axial translation, three axial rotation, five a_comp_cor, and eight drifts. The contrasts of interest were the experimental condition effects (left > right; both > right; both > left; both > fixation; left > fixation; and right > fixation) for a total number of six contrasts for both the execution and imagery runs. The activation–fixation contrast images based on the SPM model were then taken to a second-level random-effects analysis to test for group effects.

GLM Flex Fast2 (Schultz, 2017) based on MATLAB2019a (Mathworks, Inc., Natick, MA) was used for the second-level analysis. This tool has the advantage of using a common notation and can handle between- and within-participant designs. The contrasts from Fitlins were entered into a random-effects group analysis. A t test was performed for each contrast with p 

Due to problems with the Fitlins model estimation, we had to exclude two participants for the ME and four participants for the MI task. The final sample included 45 participants, 25 right-handed and 20 left-handed (Table 2). Post hoc contrasts were specified in GLM FlexFast2 for both the between and within-group analysis. A 2 × 3 within-participant design (t test, n = 45; p < 0.05) was specified to investigate if there was any difference for the two tasks (i.e., MI and ME) and two conditions (i.e., left and right) for a total number of four contrasts per group (i.e., MEleft>MEright; MIleft>MIright; MEleft>MIleft; and MEright>MIright).

TABLE 2. Demographics characteristics of the final sample (N = 45) Number 45 Handedness Right = 25 Male = 12 Female = 13 Left = 20 Male = 11 Female = 9 Note Data are presented as sample size (N) for number and handedness. Right = right-handed; Left = left-handed.

A 2 × 3 between-participant design (t test, n = 45; p < 0.05) was specified to investigate any difference between the left and right groups for the two tasks (i.e., MI and ME) and three conditions (i.e., left, right, and both hands). All the contrasts were then thresholded with a FWE voxel correction (p < 0.05) with bspmview (Spunt, 2016) based on SPM8/SPM12 operating in MATLAB 2019b.

3 RESULTS 3.1 Behavioral results 3.1.1 Edinburgh handedness inventory (EHI)

On the Edinburgh Handedness Inventory (EHI), participants' results were the following: mean = 11.08, SD = 75.69, range −100 to −100. Twenty-six participants were right-handed (N = 26, 12 female, mean age 24.23, range 21–32 years, EHI score: mean = 76.92, SD = 14.49, range 50–100), three were mixed-handed (N = 3, 1 female, mean age 22.67, range 21–24 years, EHI score: mean = 11.67, SD = 32.53, range −20 to 45), and 22 were left-handed (N = 22, 12 female, mean age 25.32, range 19–30 years, EHI score: mean = −66.82, SD = 39.56, range −100 to 90).

3.1.2 Hand dominance test (HDT)

On the Hand Dominance Test (HDT), participants' results were the following: mean = 101, SD = 75.69, range 93 to 108. Twenty-five participants were right-handed (N = 25, 11 female, mean age 24.36, range 21–32 years, HDT score: mean = 101, SD = 7.3, range 93–108), seven were mixed-handed (N = 7, 3 female, mean age 25, range 21–30 years, HDT score: mean = 82.29, SD = 4.57, range 78–92), and 19 were left-handed (N = 19, 11 female, mean age 24.79, range 19–30 years, HDT score: mean = 73.32, SD = 4.59, range 62–81).

3.1.3 Correlation analysis

A Pearson correlation analysis was run between the EHI and the HDT for assessing if self-report measurements were correlated with the results of participants’ performances. A significant correlation (r = 0.866) was found between EHI and the HDT scores (Figure 2). Furthermore, given the nature of the data for the EHI and HDT, a Cohen's kappa analysis was performed. According to the result (wk = 0.3271640), we can state that our tests, EHI and EDT, have a fair agreement.

Correlation plot between the Edinburgh Handedness Inventory (EHI) and the Hand Dominance Test (HDT). The score for each participant for the EHI are displayed on the y axis and those for the HDT on the x axis. Higher scores indicate right-handed participants for both the EHI and the HDT. Green lines define the cutoff scores for groups; the red line displays the regression line between the two test scores. A significant correlation (r = 0.866) was found between EHI and the HDT scores

Our final sample was divided into two groups, namely right and left-handed. Since the correlation between the two tests was high, we decided to divide our groups according to the result of the self-report EHI. Hence, three participants were scored as mixed handers and when asked “Do you consider yourself right or left-handed?,” they all replied “left-handed.” For this reason, we decided to include them in the left-handed group.

3.1.4 VMIQ-2

On the VMIQ-2, participants’ results were the following: mean = 24.25, SD = 6.87, range 13 to 40, mode = 29. The lower the score, the more vivid is the imagination. Being the test formed by 12 items with a rating score between 1 and 5, the minimum score achievable is 12 and the maximum is 60. Despite the subjectivity of this measurement, we think that the VMIQ-2 can be considered as a good self-report index about participants’ imagery ability.

3.2 fMRI

All participants performed the task sequences correctly at their own pace. No overt movement was observed during the imagery tasks. A complete overview of the results from the GLM-Flex analysis for both main effects and interactions is displayed in Table 3 in the Extended Data.

Voxel-wise FWE-corrected threshold images were computed with “bspmview” and for each contrast, we obtained a table showing all local maxima separated by more than 20 mm. Regions were automatically labeled using the Anatomy Toolbox atlas (Eickhoff et al., 2005) (see Tables 4–6 in the Extended Data).

Contrast images were then displayed on mrview based on MRtrix3 (Tournier et al., 2019). We first describe the between-group results comparing MI and ME for right, left, and both hands (i.e., MIleft>0; MIright>0; MIboth>0; MEleft>0; MEright>0; and MEboth>0). In the described contrasts, zero is the fixation cross. Within-group results are then displayed, comparing right and left-hand tasks for right and left-handers separately (i.e., MEleft>MEright and MIleft>MIright). Then, to make a direct comparison of the two tasks (i.e., MI and ME) in both the groups, we discuss the contrasts between ME and MI of the same hand (i.e., MEleft>MIleft and MEright>MIright) with a specific focus of the cerebellum.

3.2.1 Between group

To verify the hypothesis of whether there was any significant difference between the left and right-handed group, a two-sample t test (p < 0.05, FWE corrected for voxel) for all the contrast of interest was run.

Among MI conditions, significant group effects (p < 0.05) were found for the following contrasts, namely MIleft>0 (Figure 3a), MIright>0 (Figure 3b), and MIboth>0 (Figure 3c). Here again, zero is the fixation cross.

Between-group contrasts for motor imagery and motor execution (ME). (a–c) Between-group contrasts (t test, n = 45; p < 0.05) for motor imagery (MI). (d–f) Between-group contrasts (t test, n = 45; p < 0.05) for ME. Contrasts are shown for left hand (a,d), right hand (b,e), and both hand (c,f) conditions in comparison with “fixation” (0). Suprathreshold (FWE voxel p < 0.05) voxels for right-handers are shown in blue and in red for left-handers. Position MNI coordinates [12 −86 12]; right hemisphere (R) is displayed on the left side, left hemisphere (L) on the right side

For the left MI task, the right-handed group showed a predominantly left-lateralized activation pattern (Figure 3a, blue) compared to the left-handed group. The activation included the left lingual gyrus and left cerebellum (Crus 1) and a small cluster on the right inferior temporal gyrus. In the left-handed group, left MI yielded a more bilateral activation pattern (Figure 3a, red), including the right calcarine gyrus and the left posterior-medial frontal and left superior occipital gyrus.

In the right-handed group, “right MI” comparison with “fixation” induced a bilateral activation pattern (Figure 3b, blue), including the right calcarine gyrus, left lingual gyrus, and fusiform gyrus, while the left-handed group (Figure 3b, red) showed a right-lateralized activation pattern over the calcarine gyrus. The direct comparison of these two contrasts revealed a similar pattern for the two groups, showing a bilateral activation for MI of the dominant hand and a lateralized activation over the ipsilateral hemisphere for MI of the non-dominant hand.

The comparison of “both MI” and “fixation” condition revealed for the right-handed group (Figure 3c, blue) maxima suprathreshold activity over the right lingual gyrus and smaller clusters over the left lingual gyrus, middle occipital and inferior occipital gyrus. The left-handed group (Figure 3c, red) showed the maxima locus over the right calcarine gyrus and smaller clusters over the bilateral cerebellum (VI).

Among the ME conditions, significant group effects (p < 0.005) were found for the following contrasts, namely MEleft>0 (Figure 3d), MEright>0 (Figure 3e), and MEboth>0 (Figure 3f). In the right-handed group, “left ME” comparison with “fixation” induced a predominantly right-lateralized activation pattern (Figure 3d, blue), with a global peak on the Rolandic operculum, followed by the right calcarine and fusiform gyrus and right cuneus. In the left-handed group, “left ME” comparison with “fixation” (Figure 3d, red) yielded a more spread and bilateral activation pattern, including the cerebellar vermis, on the right hemisphere the highest peak over the superior temporal calcarine, precentral, middle temporal and fusiform gyrus, cerebellum (IX), and IFG (pars triangularis) while for the left hemisphere over the IFG (pars orbitalis), middle and inferior temporal gyrus and MCC.

In the right-handed group, “right ME” comparison with “fixation” (Figure 3e, blue) induced a predominantly left-lateralized activation pattern over the cerebellum (VI), calcarine and posterior-medial frontal gyrus. Also, for this contrast, the left-handed group yielded a more spread and bilateral activation pattern, including for the right hemisphere the superior occipital and temporal gyrus, cerebellum (VI) fusiform, and lingual gyrus, while for the left hemisphere over the calcarine and lingual and middle orbital gyrus and cerebellum (III). The direct comparison of these two contrasts revealed a different pattern for the two groups, showing a bilateral activation for ME of both the dominant and non-dominant-hands in left-handers and a more lateralized activation over the contralateral hemisphere for ME of both the dominant and non-dominant hands for right-handers.

The comparison of “both ME” and “fixation” condition revealed for the right-handed group (Figure 5f, blue) maxima suprathreshold activity over the right lingual and calcarine gyrus. The left-handed group (Figure 3f, red) showed the maxima locus over the bilateral cerebellum (VI-crus 1) and left pre- and postcentral gyrus.

3.2.2 Within-group

Among ME conditions, significant within-group effects (p < 0.005) were found for the contrast MEleft>MEright for both the groups, showing contralateral involvement of the cortical motor areas and ipsilateral cerebellum (for a complete description, see Extended Data, Tables 4.3 and 4.4).

Among MI conditions, significant within-group effects (p < 0.005) were found for the contrast MIleft>MIright for both the groups. In the left-handed group (Figure 4a), right-hand MI (blue) induced a bilateral activation over the calcarine gyrus while for left-hand MI (red), over the bilateral lingual gyrus. A result of interest is that for both the conditions, the highest peak was found over the ipsilateral hemisphere for MI task.

Within-group contrasts for motor imagery. Within-group contrasts (t test, n = 45; p < 0.05) for motor imagery (MI) for the left-handed group (a) and right-handed group (b). Suprathreshold (FWE voxel p < 0.05) voxels for right-hand MI in comparison with “fixation” are shown in blue and for left-hand MI in comparison with “fixation” in red. Position MNI coordinates [12 −86 12]; right hemisphere (R) is displayed on the left side and left hemisphere (L) on the right side

In the right-handed group (Figure 4b), both right- and left-hand MI induced a bilateral activation over the lingual gyrus with the highest peak over the ipsilateral hemisphere for MI task.

3.2.3 The cerebellum

The within-group comparison of MI and ME for the conditions of right and left hands with a specific focus on the cerebellum is described below.

Left-handers' (Figure 5; panel a,b) within-group analysis (i.e., MEleft>MIleft and MEright>MIright) showed for “left ME” (green) significant pattern (FWE voxel p < 0.05) over the cerebellar vermis (4/5), and left cerebellum (IV-V-VIII) while “right ME” (red) over the right cerebellum (VIII). No significant activity was found over the cerebellum for the reverse contrasts (i.e., MEleft<MIleft and MEright<MIright).

Within-group contrasts with focus on the cerebellum. (a,b) Within-group contrasts for the left-handed group (t test, n = 45; p < 0.05) for MEleft>MIleft and MEright>MIright. (c,d) Within-group contrasts for the right-handed group (t test, n = 45; p < 0.05) for MEleft>MIleft and MEright>MIright. Suprathreshold (FWE voxel p < 0.05) voxels in comparison with fixation for ME right are shown in red, MI right in blue, MEleft in green, and MIleft in purple. Right hemisphere (R) is displayed on the left side and left hemisphere (L) on the right side

Right-handers' (Figure 5; panel C, D) within-group analysis (i.e., MEleft>MIleft and MEright>MIright) showed for “left ME” (green) significant pattern (FWE voxel p < 0.05) over the cerebellar vermis (3) and left cerebellum (IV-V-VIII) while “right ME” (red) over the cerebellar vermis (6), right (IX) and left (Crus1) cerebellum. The reverse contrasts (i.e., MEleft<MIleft and MEright<MIright) revealed for “left MI” (purple) significant pattern (FWE voxel p < 0.05) over the cerebellar vermis (3) and right cerebellum (IX) while for “right MI” (blue) over the right (VIII) and left (VIII; Crus 2) cerebellum.

4 DISCUSSION

It is widely recognized that MI and ME rely on partly overlapping mechanisms. However, the exact nature of their relationship still lacks a proper definition. More in detail, it is not well understood how MI differs across individuals based on their experience. With specific regard to hand movements, the question is whether MI neural correlates vary between right and left-handers and to which extent this difference is resembling the one found for ME. Several neuroimaging studies have addressed this issue, finding mixed results (Decety, 1996; Fitts & Peterson, 1964; Jeannerod, 1994, 1995; Viviani & Schneider, 1991). The majority corroborate the hypothesis that right and left-handed individuals show different patterns of brain activity during MI tasks. Although these differences are well recognized, most of the fMRI studies concern only right-handed participants (Ehrsson et al., 2003; Szameitat et al., 200