Human Bain Mapping 30:2157 2172 (2009) Bain Activity Duing Visual Vesus Kinesthetic Imagey: An fmri Study Aymeic Guillot, 1,2 * Chistian Collet, 1 Vo An Nguyen, 3 Fancine Malouin, 4,2 Caol Richads, 4,2 and Julien Doyon 3,2 1 Cente of Reseach and Innovation in Spot, Mental Pocesses and Moto Pefomance, Univesity Claude Benad Lyon I, Univesity of Lyon, Villeubanne, Fance 2 Regeneative Medicine and Nanomedicine Initiative Pogam, Canadian Institutes of Health Reseach, Canada 3 Depatment of Psychology, Functional Neuoimaging Unit, Univesity of Monteal Geiatic Institute, Univesity of Monteal, Monteal, Quebec, Canada 4 Depatment of Rehabilitation, Laval Univesity and Cente fo Intedisciplinay Reseach in Rehabilitation and Social Integation, Quebec City, Canada Abstact: Although thee is ample evidence that moto imagey activates simila ceebal egions to those solicited duing actual movements, it is still unknown whethe visual (VI) and kinesthetic imagey (KI) ecuit compaable o distinct neual netwoks. The pesent study was thus designed to identify, though functional magnetic esonance imaging at 3.0 Tesla in 13 skilled images, the ceebal stuctues implicated in VI and KI. Paticipants wee scanned in a peceptual contol condition and while physically executing o focusing duing moto imagey on eithe the visual o kinesthetic components of an explicitly known sequence of finge movements. Subjects imagey abilities wee assessed using well-established psychological, chonometic, and new physiological measues fom the autonomic nevous system. Compaed with the peceptual condition, physical executing, VI, and KI esulted in ovelapping (albeit non-identical) bain activations, including moto-elated egions and the infeio and supeio paietal lobules. By contast, a divegent patten of inceased activity was obseved when VI and KI wee compaed diectly: VI activated pedominantly the occipital egions and the supeio paietal lobules, wheeas KI yielded moe activity in moto-associated stuctues and the infeio paietal lobule. These esults suggest that VI and KI ae mediated though sepaate neual systems, which contibute diffeently duing pocesses of moto leaning and neuological ehabilitation. Hum Bain Mapp 30:2157 2172, 2009. VC 2008 Wiley-Liss, Inc. Key wods: fmri; kinesthetic imagey; moto pefomance; visual imagey; moto cognition Contact gant sponsos: The Canadian Institutes of Health Reseach, The Foundation Simone & Cino Del Duca. *Coespondence to: Aymeic Guillot, Cente de Recheche et d Innovation su le Spot, Univesité Claude Benad Lyon I, 27-29 Boulevad du 11 Novembe 1918, 69622 Villeubanne Cedex, Fance. E-mail: aymeic.guillot@univ-lyon1.f Received fo publication 4 Octobe 2007; Revised 26 May 2008; Accepted 28 July 2008 DOI: 10.1002/hbm.20658 Published online 25 Septembe 2008 in Wiley InteScience (www. intescience.wiley.com). INTRODUCTION Moto imagey is defined as the mental simulation of a specific action without any coesponding moto output, hence equiing a epesentation of the body as the geneato of acting foces [Jeanneod, 1994]. To date, a lage body of eseach has been caied out to investigate the neuophysiological coelates of MI and to claify the elationships between MI and both moto pepaation and execution using vaious complementay methods [fo eview, VC 2008 Wiley-Liss, Inc.
Guillot et al. see Guillot and Collet, 2005a]. Findings fom these expeiments have povided evidence that the execution of a movement and MI show seveal paallel chaacteistics. Fist, the time couse of mentally simulated actions has been found to be highly coelated with the time taken to execute the same movement [e.g., Decety et al., 1989]. This tempoal equivalence, howeve, is not systematic as seveal factos (e.g., task duation and difficulty, instuctions given to subjects, o the way in which attention is focused on a specific aspect of the mental image) may lead to an ove- o undeestimation of the imagined movement duation [Guillot and Collet, 2005b]. Second, peipheal activity of the autonomic nevous system (ANS) has also been found to show simila esponses pio and duing movement duing both MI and the moto execution of actions. Fo example, heat ate and pulmonay fequencies ae known to covay with the degee of imagined effot [Decety et al., 1991, 1993; Fusi et al., 2005]. Futhemoe, inceases in ventilation and systolic blood pessue have also been epoted duing MI of dumbbells lifting [Wang and Mogan, 1992; Wuyam et al., 1995], wheeas simila electodemal and themovascula esponses have been elicited duing MI and moto pefomance [Guillot et al., 2004, 2005]. Finally, the advent of bain mapping techniques like positon emission tomogaphy and functional magnetic esonance imaging (fmri) have shown that goaldiected actions, whethe executed o imagined ecuit simila (albeit non identical) neual substates [Decety et al., 1994; Guillot et al., 2008; Lotze et al., 1999; Mellet et al., 1998]. Although the physical execution of hand movements inevitably activates the Rolandic egion, ealy positon emission tomogaphy studies looking at the functional anatomy of MI have epoted an incease of activity in moto-elated aeas like pemoto and supplementay moto aea (SMA), but not in the pimay moto cotex pe se [Decety et al., 1990, 1994; Ingva and Philipson, 1977; Roland et al., 1980]. Fo example, some authos have epoted activations in diffeent sites within the posteio pat of the SMA [Stephan et al., 1995], wheeas othes have povided evidence of an activation in the pe-sma egion [Deibe et al., 1998; Geadin et al., 2000; Lafleu et al., 2002], hence, demonstating that medial cotical stuctues also play a cucial ole in MI [Lotze and Halsband, 2006]. Some fmri studies have shown that not only seconday moto aeas ae ecuited duing MI, but that the pimay moto cotex can also be pat of the netwok involved in imagined actions [Ehsson et al., 2003; Kuhtz- Bushbeck et al., 2003; Poo et al., 1996; Solodkin et al., 2004]. Indeed, MI has been shown to activate the anteio pat of the pimay moto aea [Lotze and Halsband, 2006] duing MI of moving diffeent body pats (foot, hand, and tongue), thus suggesting that the pimay moto cotex may also be involved duing MI. Yet the ole of the pimay moto cotex duing MI emains contovesial as some authos failed to find any peak of activation [Binkofski et al., 2000; Hanakawa et al., 2003], wheeas othes epoted fleeting involvement [Dechent et al., 2004]. Shama et al. [2006] have suggested that such discepancies in esults may be due, at least patially, to methodological diffeences and difficulties in monitoing compliance. In a second aticle, Shama et al. [2008] have shown that the cluste distibution in the anteio pat of the pimay moto cotex was significantly educed duing MI as compaed with the physical execution, wheeas that of the posteio pat was simila. Accodingly, they poposed that the ole of this aea and its subdivisions ae nonexecutive and may be elated to spatial encoding. Apat fom moto-elated aeas, othe egions like supeio paietal lobule have been found to be activated duing mental simulation of actions [Decety et al., 1994; Stephan et al., 1995; Wolbes et al., 2003]. Similaly, Nai et al. [2003] have shown that the pecuneus is engaged duing bimanual action sequences that ely on emembeing and executing the coect odeing of tasks components as well as on pocessing the sensoy consequences of action, wheeas othes [Binkofski et al., 2000; Schubet et al., 1998] have epoted that cotical stuctues aound the intapaietal sulcus ae also ecuited duing MI of self-paced finge movements. Such findings ae consistent with studies in both healthy subjects and patients with paietal lesions, which have confimed the cucial ole of the supeio and infeio paietal cotices in the geneation of mental images [Siigu et al., 1996; Suchan et al., 2002]. Finally, ceebella and basal ganglia activations have consistently been epoted duing MI [Decety et al., 1994; Guillot et al., 2008; Lafleu et al., 2002; Naito et al., 2002; Pasons et al., 1998], although activations in the ceebellum wee moe posteio and infeio duing this condition than duing the actual movements [Lotze and Halsband, 2006; Lotze et al., 1999]. Studies on basal ganglia dysfunctions have also emphasized the ole of these subcotical stuctues [especially the putamen; Li, 2000] in the neual netwok mediating MI [Dominey et al., 1995]. Taken togethe, the esults epoted ealie povide evidence that MI and moto pefomance shae simila behavioal, physiological, and anatomical chaacteistics. It is now well-established, howeve, that MI may be subdivided into two diffeent modalities: visual imagey (VI) and kinesthetic imagey (KI). On one hand, VI equies self-visualization of a movement fom a fist- (intenal VI) o thid-peson (extenal VI) pespective. The fist-peson pespective coesponds to the epesentation of a movement as if the individual takes pat to the action himself, hence suggesting that he/she would visualize the movement like having a camea on his/he head. By contast, the thid-peson pespective coesponds to the epesentation of the movement as if the subject was a spectato and that somebody (himself o anothe peson) pefomed the action. On the othe hand, KI equies one to feel the movement and to peceive muscle contactions and stetching mentally. Despite accumulated evidence epoted ealie that MI and moto pefomance shae common neual substate, little is known, howeve, with espect to the possible specialization in neual systems 2158
Neual Substates of Moto Imagey duing these two diffeent foms of MI. In a fist study, looking at this issue duing actual and imagined finge movements, Binkofski et al. [2000] have shown that the anteio pat of the intapaietal sulcus was moe active duing KI, wheeas the posteio pat was moe involved duing VI. Futhemoe, when bilateal activations wee epoted duing KI in the opecula potion of the vental pemoto cotex (BA 44), a lack of activation in the paietal aeas was obseved duing KI of finge movements. Solodkin et al. [2004] have also investigated the effective connectivity in netwoks associated with physical execution, VI, and KI of hand movements. Even though VI and KI wee found to shae simila neual substates including the connection fom the supeio paietal lobule to the SMA, the neual netwoks undelying these two MI conditions wee not identical. Especially, the main diffeence was found in the inputs fom the supeio paietal lobule and the SMA to the pimay moto cotex, which wee opposite to those obseved duing moto pefomance. The esults of the latte two studies [Binkofski et al., 2000; Solodkin et al., 2004] suggest that VI and KI may ecuit quite diffeent neual substates. Howeve, some issues egading the methodology used in these studies hinde the conclusions that one can each. Fist, the authos did not indicate whethe they contolled the quality of MI and the compliance of the subjects with MI instuctions. Second, in the study by Solodkin et al. [2004], VI and KI wee tested in two independent goups, thus allowing a possible bias with egad to the intepetation of the diffeent neual netwoks mediating MI. Final, and most impotantly, individual MI abilities wee not systematically taken into consideation, and wee patially measued qualitatively, i.e., subjectively using mental otation tests o post-test questionnaies, but not quantitatively. The pesent fmri study was thus devised to examine whethe the same goup of healthy subjects with vey good to excellent MI abilities (assessed using new quantitative physiological measues) does ecuit compaable o distinct bain activations duing VI and KI of hand movements. Although simila ceebal stuctues wee expected to be activated duing the physical pefomance and MI, VI, and KI wee each hypothesized to involve diffeent cotical aeas. Because, compaed with VI, KI shaes physiological chaacteistics that ae close to movement execution pe se [Solodkin et al., 2004], including the electomyogaphic activity [Guillot et al., 2007], the excitability of the cotico-spinal system [Hashimoto and Rothwell, 1999], and the autonomic changes associated [Decety et al., 1991], it was expected that the moto systems and the elated paietal egions would be moe involved duing KI of complex hand movements. By contast, a stonge incease in signal intensity and/o additional stuctues eceiving visual affeences would be seen duing VI. MATERIALS AND METHODS Distinguishing the neual netwoks mediating VI and KI equied selecting subjects who showed vey good to excellent MI abilities. A seies of tests wee thus administeed befoe the fmri study to select subjects who wee able to each this high level of pefomance in MI. Psychological, behavioal, and neuophysiological tests wee combined to evaluate MI capacity within a lage sample of subjects, as suggested by Guillot and Collet [2005a] and Lotze and Halsband [2006]. The fmri expeiment was then pefomed a few days late using a subset of these subjects. Paticipants Subjects Peselection Fifty healthy, ight-handed, voluntees (24 men: mean age 26.4 yeas, SD 5 3.4, age ange 21 34 and 26 women: mean age 25.6 yeas, SD 5 3.9, age ange 20 35) without neuological complications paticipated in a peselection session. This study was appoved by the Local Ethics Committee fom the Univesity of Monteal Geiatic Institute. All paticipants gave thei infomed consent and wee paid fo thei paticipation. Behavioal tasks Fou tests wee combined to evaluate MI ability. Fist, each paticipant completed the evised vesion of the Movement Imagey Questionnaie [MIQ-R, Hall and Matin, 1997]. The MIQ-R is made up of eight items to evaluate both intesubject diffeences in VI (fou items) and KI (fou items), as well as within-subject diffeences (VI vs. KI). Paticipants wee fist equested to ead desciptions of movements to be pefomed physically and then to imagine themselves pefoming the same movement. A scoe based on the difficulty associated with epesenting each movement mentally was assigned using a seven point ating scale. Second, movement duations wee ecoded wheeas the paticipants physically pefomed and imagined thee moto actions that equied the ability to use both VI and KI: (i) a sequence of 16 hythmic steps pefomed within a squae dawn on the floo, (ii) a seies of five consecutive complete flexion-extension (squats) of the lowe limbs, and (iii) the maintenance of a position with the knee at 908 of flexion against a wall duing 12 s. Subjects had to pefom the fome action using VI and the latte movement using KI. By contast, the seies of flexion-extension actions wee completed using a combination of both MI types. The ability to peseve tempoal chaacteistics between the actual movement and duing imagey was measued [Guillot and Collet, 2005b; Malouin et al., 2007]. Fo each moto sequence, subjects wee equied to stat and stop the time on mental initiation of the fist body movement and at the end of the sequence, espectively. Thid, ANS activity (see late) was simultaneously and continuously ecoded duing each tial, hence allowing a moe quantitative measue of the subjects mental capacities and epesentations [Guillot et al., 2004; 2159
Guillot et al. Roue et al., 1999]. Fou tials wee pefomed at andom unde each condition (physical and MI pefomance), each being sepaated fom the next tial by a est peiod (lasting at least 20 s) fo the physiological measue to ecove its baseline level. In both conditions, subjects wee acoustically isolated. They wee asked to close thei eyes duing MI and to open them to acknowledge the next assignment. Finally, to veify that they pefomed MI as they wee instucted to, paticipants wee equied to descibe the natue of the images they attempted to fom afte the MI session and to scoe thei effot using a fou point ating scale (1 5 vey difficult to imagine/feel and 4 5 vey easy to imagine/feel). Autonomic nevous system ecodings Highe bain functions may be investigated though autonomic nevous system (ANS) effectos activity at the peipheal level [Hugdahl, 1996], as cental opeations (e.g., planning and pogamming) ae paalleled by ANS esponses that epesent nonconscious physiological mechanisms of mental pocesses [Collet et al., 1999]. Among ANS effectos, sweat glands ae innevated by sympathetic endings only. An incease of the subjects level of aousal (such as duing MI) elicits sweat elease and, consequently, a decease in skin esistance (SR). Skin esistance was ecoded using two 30 mm 2 unpolaizable Ag/AgCl electodes (Clak Electomedical Instuments, Ref. E243) placed on the second phalanx of the second and thid digits of the nondominant hand and held by adhesive tape [Fowles et al., 1981]. A conductive paste (TECA ef: 822 201210) was used to impove skin/electode contact. Resistance was ecoded with the constant cuent method [Boucsein, 1993] using a 15 la cuent (density 5 0.5 la/mm 2 ). As esponse amplitude depends on the pestimulation value [Fuedy and Sche, 1989], a moe eliable index was taken without efeing to that initial value (tonic level). The ohmic petubation duation was measued at the beginning of the sudden dop elicited by MI and was ended when the slope, wheeas ecoveing basal level showed no fluctuation and esembled that obseved befoe stimulation [Venet-Mauy et al., 1995]. Response latency efeed to the time lapse fom the onset of the stimulus to the initiation of the esponse. Based on fequency of esponse latencies to simple stimuli, it is common to use a 1 2 s o 1 3 s latency windows [Levinson and Edelbeg, 1985]. Any esponse onset within 1 and 3 s following stimulus onset was thus consideed as being elicited by that stimulus. Moto imagey ability The fou well-established imagey ability measues mentioned ealie wee used to evaluate individual MI abilities. Autonomic nevous system scoe. The numbe of SR esponses was fist calculated and epesented on a 0 12 scale: 0 indicating no SR esponse on each MI tial and 12 indicating that each MI tial elicited a SR esponse. In addition, the subject s level of aousal was assessed though SR basal tonic evolution acoss the MI session. Deschaumes-Molinao et al. [1991] found that high images have simila physiological activation duing both MI and moto pefomance, SR tonic level being one of the most eliable indicatos of aousal vaiations [Collet et al., 1996]. To give equal impotance to these two factos, the evolution of the aousal level was gaded between 25 (subjects elaxed thoughout MI), 22 (inceased elaxation by steps), 0 (no adjustment, the activation level emaining stable duing MI), 12 (inceased activation by steps duing MI), and 15 (inceasing activation egulaly). Theeby, the autonomic nevous system (ANS) scoe consisted of the sum of the two peceding measues (numbe of SR esponses 1 aousal level scoe): the minimal scoe subjects could obtain being 25 and the maximal scoe being 12. MIQ-R scoe. This measue was calculated by adding the scoes assigned by the subjects to each MI test-item. The minimal scoe subjects could obtain was 8, and the maximal scoe was 56. Autoestimation scoe. This scoe was the mean of all atings given by subjects when evaluating the vividness of each MI tial on a 4-point scale. Mental chonomety scoe. This scoe was the mean of the absolute time diffeences between actual and imagined tials. This diffeence scoe was subtacted fom the global imagey scoe as it was invesely popotional to the subjects ability to peseve the tempoal chaacteistics of each movement duing MI, suggesting theefoe a difficulty to imagine the action. Following Roue et al. s [1999] ecommendations, a global imagey scoe was finally calculated fo each paticipant, using this simple fomula: (ANS scoe 1 MIQ-R scoe 1 Autoestimation scoe) 2 (Mental chonomety scoe). Paticipants in the frmi Study Among the 50 voluntees who took pat in the peselection study, only 13 (six men: mean age 25.5 yeas, SD 5 2.4 and seven women: mean age 23.9 yeas, SD 5 2.8) wee ated as good to excellent images (as defined by at least one SD above the mean global imagey scoe) and wee theefoe selected fo the fmri expeiment. All wee also found to outpefom the othe subjects fo each ating of the global imagey scoe. None of the subjects wee eithe a musician o a pofessional typist in ode to eliminate subjects with pe-existing skills equiing highly coodinated finge dexteities. Finge sequence task Paticipants wee fist asked to lean a sequence of eight moves using finges 2 5 of the left hand, until they wee able to pefom them explicitly fom memoy within a 6 s- peiod. Within the finge sequence, the ode of finge movements was pseudoandomly geneated such that each 2160
Neual Substates of Moto Imagey Figue 1. Schematic epesentation of the expeimental block-design. Afte eading the instuctions, the paticipants wee subjected to a 30 s-peiod duing which they eithe physically pefomed the movement, mentally eheased the sequence using visual o kinesthetic imagey, o emained motionless while closing thei eyes. Simila stat and stop sound signals wee used in each condition. finge was used twice. They physically pefomed the sequence on a few occasions to get both sensoimoto feedback and visual guidance to enhance the memoization pocess. The subjects pefomance was assessed inside and outside the scanne by using a fou keys keyboad (Electical Geodesics, Eugene, OR) that was MR-compatible. The keyboad allowed ecoding of the subjects esponse accuacy and timing. Paticipants wee equied to keep thei finges on the keys to minimize amplitude vaiation and the amount of foce equied to pess the keys. They wee instucted to tap the sequence at a comfotable and self-paced speed, while making as few eos as possible. Speed tests, howeve, wee also scheduled to check that the paticipants wee able to coectly pefom and imagine the movement within a peiod not exceeding 6s. Following intoduction to the sequence, paticipants wee scanned duing a block-design paadigm including six uns. The physical execution, VI, KI, and peceptual contol conditions wee systematically followed by a estpeiod, in ode fo the subjects to ead the instuctions. Fo each un, subjects thus altenated eight countebalanced epochs of 30 s of expeimental conditions and 10 s of est (see Fig. 1). Physical execution. The paticipants executed the finge sequence opposition task that was explicitly leaned befoe the scanning session by using the fou keys keyboad. Visual imagey. The subjects wee equested to visualize the finge sequence without any movement using the fistpeson pespective. They wee equested to close thei eyes, and thus did not have any visual guidance o sensoimoto feedback. Subjects wee then equied to keep thei finges on the buttons of the sensitive fou key keyboad to check that thee was not any actual movement and the am stayed immobile along thei body on the laboatoy bench. Kinesthetic imagey. The paticipants wee asked to peceive body sensations of the finge sequence without any movement. The hand position on the keyboad was simila to the VI condition. Futhemoe, they wee also equested to close thei eyes to avoid visual guidance. Peceptual contol condition. The subjects wee specifically instucted to emain motionless while listening to distinct high- and low-tones sounds. This contol condition was chosen to contol fo the same stat and stop sound signals that wee used in the othe expeimental conditions. The ode of administation of each expeimental and contol condition was countebalanced acoss the six uns. Instuctions wee given on a compute sceen that could be seen though a mio embedded within the head-coil. Afte eading the instuctions, the subjects wee equied to close thei eyes. Each 30 s-peiod of the expeimental conditions (physical execution, VI and KI) wee composed of five tials (eithe physically o mentally pefomed evey 6 s). Two sepaate sounds wee used to indicate the beginning of each tial (high tones) and end of a un (low tones) duing the expeimental and peceptual contol conditions, using MR-compatible headphones (MR confon HP-SI01, Gemany). The low-pitch tones also indicated to subjects to open thei eyes and to emain in a esting awake state until the next assignment which appeaed on the compute sceen (est-peiod). Befoe the scanning session began, all paticipants wee given a few tials inside the scanne until they physically pefomed five successive coect finge sequences. They wee also asked to pefom five MI tials to become familia with the pesentation of the auditoy stimuli and the appaatus itself. To do so, they wee asked to close thei eyes, so that they did not have any futhe visual guidance duing the expeiment o any sensoimoto feedback to ensue that the taining conditions would be vey simila to those of the following scanning session. Functional imaging Blood oxygen level-dependent signal was ecoded using a 3-T whole-body TRIO system (Siemens, Elangen, Gemany) located at the Functional Neuoimaging Unit, Univesity of Monteal Geiatic Institute. The subject s head was immobilized using foam cushions. The potocol lasted 90 min and included (i) one localize (scout) to 2161
Guillot et al. localize functional axial slices, six functional uns, and one high-esolution anatomical scan [sagittal T1-weighted; epetition time: 13 ms; echo time: 4.92 ms; one slab divided into 160 slices; matix size: 256 3 256; voxel size: 1 3 1 3 1mm 3 ; patial Fouie imaging 7/8; bandwidth 140 Hz pe voxel]; (ii) Foty-thee oblique axial gadient echo-plana imaging T2*-weighted images [epetition time: 4.5 s; echo time: 30 ms; a: 908; bandwidth: 1 562 Hz pe voxel; field of view: 192 3 192 mm 2 ; voxel size: 1.5 3 1.5 3 2.5 mm 3 ; patial Fouie imaging 6/8; matix size: 128 3 128]. Fo each seies, 75 echo-plana imaging volumes wee acquied ove 5 min and 37.5 s. Behavioal ecodings Behavioal dependent vaiables (key pesses, movement fequency, total sequence speed, and eaction times) wee automatically ecoded based on the subjects esponses using a home-made MATLAB-witten outine. Fo each paticipant, this softwae compaed the sequence of key pesses poduced by the subject to the coect sequence template to be pefomed, hence allowing detecting any discodance between the eal and expected taps within the given sequence. Functional magnetic esonance imaging data analysis Functional data analyses wee pefomed with the toolbox Statistical Paametic Mapping (SPM2, Wellcome Depatment of Cognitive Neuoscience, London). Motion coection in the functional images was done fist using the SPM ealignment. This estimates a set of six igid-body tansfomation paametes fo each image by finding the paametes that minimize the mean squaed diffeence between it and a efeence image. The middle image of the last un of the session was used as the efeence image fo each subject. The anatomical image was ealigned to the mean functional image with the SPM coegiste method [Collignon et al., 1995]. The functional and anatomical images wee fist nomalized to the MNI coodinates (avg. 152 T1.mnc template with a final voxel size of 1.5 3 1.5 3 2.5 mm 3 ) using the 4th degee B-spline intepolation method, and finally into the standad popotional steeotaxic space of Talaiach and Tounoux [1988]. The scans wee smoothed using a Gaussian kenel set at 8-mm full width at half-maximum. Statistical analysis was done using the geneal linea model to descibe the data in tems of expeimental and confounding effects, as well as esidual vaiability. Six egessos wee used (x, y, z, pitch, oll, and yaw). Single subject analyses wee fist pefomed with a fist-level analysis (fixed-effect), which use withinsubject vaiance and povide fo infeences that genealize to the subjects studied. Then goup analyses wee done with andom-effects analyses, which involve taking the contasts of paametes estimated fom a fist-level (fixedeffect) analysis and enteing them into a second-level (andom-effect) analysis. To identify the location of bain aeas involved in each task, one sample t-tests wee used to contast (i) the actual, VI and KI conditions with the peceptual contol condition (i.e., physical execution vs. peceptual contol, VI vs. peceptual contol, and KI vs. peceptual contol), (ii) the physical condition with each MI condition (physical execution vs. VI and physical execution vs. KI), and (iii) the VI condition vs. KI condition. Compaisons of the functional data between uns and between sessions wee assessed at the P < 0.005 uncoected statistical theshold fo multiple compaisons. In these maps, activated clustes wee consideed significant if thei spatial extent was >10 voxels. The esults ae pesented as mean (standad deviation) values. RESULTS Pe-Expeiment The mean MIQ-R scoe in the good image goup diffeed significantly fom that of the othe subjects (t 5 2.77, P < 0.01), the mean scoes being 45.54 (2.7) and 38.25 (9.2), espectively. Good images also assigned a significantly highe scoe (3 [0.2]) in evaluating the vividness of thei MI than did poo images (2.56 [0.6], t 5 2.2, P < 0.05). The mean scoe based on the ANS activity (numbe of esponses 1 aousal level scoe) stongly diffeed between the good image goup (7.08 [3.9]) and the othe subjects (20.2 [2.5], t 5 5.99, P < 0.0001). An example of SR esponse compaison in good and poo images is pesented in Figue 2. By contast, the absolute mean time diffeence between the actual and imagined movement duations did not diffe (t 5 21.2, P > 0.05, NS) in the two goups, even though the good images wee found to be faste. The mean diffeence time was 2.22 s (1.5) in the good images and 3.26 s (2.8) in the othe subjects. Finally, the global imagey scoes in the good image goup diffeed significantly fom those of the othe subjects (t 5 4.8, P < 0.0001) as thei means wee 53.4 (4.7) and 39.9 (9.5), espectively. Inteestingly, thee was no significant gende diffeence on the global imagey scoe o on any of the subscoes composing it, and thus thee was no need to subdivide the subject s goup by gende. Functional Magnetic Resonance Imaging Study Behavioal data Duing the fist un, subjects made 9% of eos on aveage and the mean time to physically complete the eight item finge sequence opposition task was 2.98 s (0.5). Pefomance emained stable fom un to un. Indeed, subjects made 11.4% eos on aveage duing the five uns and completed the coect sequence in 2.94 s (0.12). Thee was no leaning effect fom the fist to the last un of the expeiment. 2162
Neual Substates of Moto Imagey supeio egions, the netwok including the pimay sensoy aeas (BA 1-2-3), the infeio paietal lobule (BA 40), and the ight supeio paietal egion (BA 5). Finally, bilateal activations wee found in the anteio and posteio pats of the putamen coesponding most pobably to the associative and sensoimoto egions, in the globus pallidus as well as the ight caudate nucleus. Activations wee also located in both the anteio and posteio ceebella egions [Lobules IV, V, VI, and Cus I; see Schmahmann et al., 1999]. Visual imagey vs. peceptual contol condition and physical execution Figue 2. Skin esistance esponses duing moto imagey. In the good image goup, a esponse (indicated by the dotted line) was ecoded duing each moto imagey tial, hence attesting to mental wok. Convesely, in the poo images, the lack of skin esistance esponses attests the difficulty to fom an accuate epesentation of action. The stong incease in skin esistance thoughout the session also indicated that the subject becomes too elaxed and was not able to keep an adequate aousal level. MI, moto imagey. Functional magnetic esonance imaging data Physical execution vs. peceptual contol condition. When the peceptual contol condition was subtacted fom the physical execution condition, the esults evealed significant activations in the expected moto netwok including cotical moto-elated aeas as well as subcotical stuctues compising the basal ganglia and ceebellum (Table I, Fig. 3). Moe specifically, peaks of activations wee located in the ight pimay moto aea (BA 4) and bilateally in the lateal dosal pemoto cotex (BA 6) and in medial sufaces of the fontal cotex including both SMA and pe-sma egions. Significant signal changes wee also obseved in the infeio fontal aeas (BA 44) bilateally and in the left cingulate cotex (BA 24). In the paietal lobe, data showed activations in both infeio and When the VI condition was fist compaed with the peceptual contol condition, an impotant set of activated aeas was found in the occipital egions bilateally. This netwok included both the pimay (BA 17) and associative visual aeas coesponding to BA 18 and 19 (Table I, Fig. 3). Additional activations wee located bilateally in the infeio paietal lobule (BA 40), as well as in the ight supeio (BA 7) and infeio (BA 31) paietal aeas. In the fontal lobe, the ight pimay moto aea (BA 4), and both the lateal (BA 6) and medial pemoto aeas (anteio and posteio pats of the SMA) wee activated bilateally. Activation was also found in the left dosolateal (BA 46) and ight ostal (BA 10) pefontal aeas. Finally, bilateal activations wee found in the putamen (anteio and posteio pats), globus pallidus and ceebella hemisphees (Lobules IV, V, VI, and VIII). When VI was then contasted to the physical execution condition, activations wee obseved in the left infeio pefontal aea (BA 45) and the pefontal cotex (BA 24 32). A cluste of activations was also found in the supeio pat of the lateal pemoto aea (BA 6; z coodinate of t-value maxima 5 66 and 60 in the left and ight hemisphees, espectively). The supeio paietal aeas (BA 5 7) wee also activated bilateally. Finally, a lage activation of the occipital egion was obseved, including the bilateal pimay and associative aeas (BA 17-18-19). Kinesthetic imagey vs. peceptual contol condition and physical execution When KI was diectly compaed with the peceptual contol condition (Table I, Fig. 3), activations wee obseved bilateally in the medial and lateal pemoto aeas, pefontal egions, infeio and supeio paietal aeas, basal ganglia and ceebella cotex. The ight pimay moto cotex (BA 4), and both the lateal (BA 6) and medial pemoto aeas (anteio and posteio pats of the SMA) wee also activated bilateally. In addition, bilateal activation was located in the infeio (BA 44/45) and dosolateal pefontal aeas (BA 9/46). An impotant bilateal netwok of activated egions including the pimay sensoy aeas (BA 1 3) and both supeio (BA 5 7) and infeio (BA 40) paietal lobules wee ecuited as well. 2163
TABLE I. Coodinates of peak activations fo the physical execution, visual imagey, and kinesthetic imagey conditions vesus the peceptual contol condition Anatomical aeas Hemisphee Physical execution vs. contol Visual imagey vs. contol Kinesthetic imagey vs. contol x y z t value x y z t value x y z t value Occipital cotex Pimay visual aea (BA 17) L 211 286 2 3.26 R 16 265 42 4.42 Pe-stiate cotex (BA 18) L 214 268 24 3.93 R 20 272 29 3.6 Pe-stiate cotex (BA 19) L 212 262 26 4.55 R 36 269 214 3.36 Paietal cotex Infeio paietal lobule (BA 31) R 8 272 23 3.23 Infeio paietal lobule (BA 40) L 236 240 47 3.86 251 230 32 6.25 239 241 41 9.85 R 62 222 20 7.43 34 236 49 8.1 46 229 46 10.02 L 250 230 51 3.8 260 222 26 3.66 R 46 229 46 6.15 40 240 57 6.74 Paietal aeas (BA 1,2,3) L 245 227 39 4.5 261 0 6 5.52 R 39 221 51 9.75 60 227 40 3.31 R 47 224 43 6.83 Supeio paietal lobule (BA 5) R 38 244 63 4.14 39 240 60 6.25 Supeio paietal lobule (BA 7) L 230 248 49 4.85 R 21 259 58 4.21 18 244 52 3.85 Moto and pemoto cotex Pimay moto cotex (BA 4) R 41 217 56 11.4 36 215 52 4.36 36 215 52 3.76 Latela pemoto aea (BA 6) L 239 29 57 6.68 236 211 43 5.6 228 27 56 6.48 R 27 212 59 7.09 24 211 54 7.61 27 28 61 7.29 L 227 28 50 3.41 R 42 26 57 4.29 Medial pemoto aea (BA 6) Pe-SMA L 28 26 61 4.72 29 29 61 8.94 29 29 61 5.4 R 8 26 61 9.64 9 26 59 6.08 9 22 63 3.33 SMA pope L 23 23 53 8.83 23 23 66 5.91 24 25 64 7.45 R 5 1 55 3.68 5 23 66 3.06 L 25 1 55 3.73 Vental pemoto cotex (BA 44) L 246 21 8 7.03 245 2 8 7.96 R 57 4 19 8.28 46 2 9 6.43 Vental pemoto cotex (BA 45) L 261 11 19 7.69 Pefontal cotex Dosolateal pefontal aea (BA 9) L 257 4 22 9.07 R 60 7 25 7.06 Dosolateal pefontal aea (BA 46) L 240 33 12 3.47 242 46 23 3.61 Rostal pefontal aea (BA 10) R 39 44 26 5.59 Obito-fontal cotex (BA 13) L 240 9 11 9.02 Limbic egions L 239 12 8 8.25 Cingulate cotex (BA 24) L 22 0 37 5.72 Subcotical egions Caudate nucleus L 213 11 21 5.12 R 18 224 18 3.81 20 226 18 3.18 Globus pallidus L 224 210 22 3.94 22 213 3 5.28 221 22 0 4.58 R 24 215 4 5.13 220 211 3 3.22 24 29 1 4.75 Anteio putamen L 225 5 10 3.78 220 3 5 6.35 227 9 12 6.16 R 23 6 13 4.66 20 5 9 5.77 20 5 9 4.45 Posteio putamen L 226 22 25 4.5 227 29 12 6.69 228 25 11 8.79 R 30 214 8 5.58 27 29 12 4.15 22 23 0 6.89 Ceebellum Anteio (lobule IV) L 212 247 215 11.92 226 230 224 4.33 211 249 218 4.1 L 24 248 0 3.59 24 249 25 3.45 Anteio (lobule V) L 221 244 221 15.33 228 236 226 3.96 210 255 22 5.08 R 8 257 22 4.14 L 25 255 22 13.58 215 251 220 4.67 L 212 249 28 13.21 Anteio (lobule VI) L 235 252 218 7.44 234 246 225 3.32 215 266 28 5.37 R 35 251 220 5.88 40 251 220 6.66 38 250 223 6.18 L 23 268 29 3.41 Anteio (lobule VIII) 24 262 225 3.86 Posteio (lobule V) L 25 259 215 7.58 Posteio (lobule VI) L 220 267 214 10.67 224 257 217 5.25 232 255 217 7.03 R 24 261 215 7.74 7 270 217 4.7 6 269 217 4.71 L 214 267 212 10.02 236 259 217 4.19 218 267 214 5.04 Posteio (lobule VIIb) 242 255 220 3.9 Posteio (Cus 1) R 44 255 22 3.39 38 263 236 3.46
Neual Substates of Moto Imagey SPM activation maps compaing the diffeent expeimental conditions (physical execution, VI, and KI) with the peceptual contol condition. Voxels ae displayed on a pogessive gay scale. The maps ae displayed on Talaiach space as a maximum intensity pojection Figue 3. (all voxels activated ae visible as if viewed in tanspaency though thebain)viewedfomtheightside(sagittal),theback(coonal), and the top (tansvese) of the bain. [Colo figue can be viewed in the online issue, which is available at www.intescience.wiley.com.] Inceased activity was also obseved bilateally in the anteio and posteio pats of the putamen, in the caudate nucleus and in the globus pallidus. Finally, activations wee also found in the ceebella hemisphees (Lobules IV, V, VI, VIIb, and Cus I) and in the ight infeio tempoal lobule (BA 22). When KI was contasted to the physical execution condition, bilateal activations wee found in the lateal pemoto (BA 6), vental pemoto (BA 44), and pefontal cotex (BA 46). A lage activation was also obseved in the infeio (BA 40) and supeio (BA 7) paietal lobules, bilateally. Finally, two peaks of activation wee also obseved in the occipital cotex (BA 19) and in the ight tempoal lobe (BA 21 22), espectively. Visual imagey vs. kinesthetic imagey Most impotantly, when VI and KI conditions wee compaed diectly, both contast analyses evealed activations in the lateal pemoto cotex, but VI ecuited aeas that wee moe posteio and supeio than duing KI, wheeas activity in the anteio and posteio pats of the SMA was stonge duing KI than duing VI (Table II; Fig. 4). Futhemoe, VI showed exclusive occipital activations (BA 17-18-19) and manifested moe pominent blood oxygen level-dependent signal that was clusteed within the supeio paietal egions (BA 5-7; Fig. 4). In the infeio paietal lobule (BA 40), the contasts between VI and KI showed that activation was located in a moe lateal and anteio pat duing KI and in a moe posteio pat duing VI. Finally, KI showed stonge activation in many fontal aeas (BA 9, 24), including the infeio fontal aeas (BA 44). Bilateal basal ganglia (putamen and caudate nucleus) and ceebella activations (Lobule VI, VIIb, and Cus I) wee also exclusively seen when the KI condition was compaed to the VI condition (Fig. 5). 2165
Guillot et al. TABLE II. Coodinates of peak activations duing visual vesus kinesthetic imagey conditions Anatomical aeas Hemisphee Visual imagey vs. kinesthetic imagey Kinesthetic imagey vs. visual imagey x y z t value x y z t value Occipital cotex Pimay visual aea (BA 17) R 15 277 12 3.29 Pestiate cotex (BA 18) R 8 272 21 3.06 Pestiate cotex (BA 19) L 238 277 9 3.06 R 27 280 40 3.97 Paietal cotex Infeio paietal lobule (BA 40) L 263 230 29 5.24 R 59 241 32 6.05 Supeio paietal lobule (BA 5) R 3 240 57 5.61 Supeio paietal lobule (BA 7) L 218 254 55 4.30 R 27 276 45 4.19 Moto and pemoto cotex Lateal pemoto aea (BA 6) L 232 27 50 3.56 253 25 9 6 R 26 29 53 4.33 56 1 17 4.51 Medial pemoto aea (BA 6) Pe-SMA R 4 9 58 3.75 SMA pope R 11 26 67 3.92 Vental pemoto cotex (BA 44) L 258 6 19 6.61 R 56 6 17 4.84 Pefontal cotex Dosolateal pefontal aea (BA 9) L 247 21 36 3.27 Limbic egions Cingulate cotex (BA 24) R 10 5 36 4.3 Subcotical egions Caudate nucleus R 20 224 21 4.55 Putamen POST L 229 212 1 3.2 R 29 28 11 4.16 Ceebellum Anteio (lobule VI) L 232 251 220 3.7 Posteio (lobule VI) L 223 265 217 3.14 R 22 261 220 3.82 Posteio (Cus I) 242 255 220 3.9 Posteio (Lbule VIIb) R 38 263 236 3.46 DISCUSSION The main pupose of this study was to detemine whethe good images ecuit compaable o distinct neual substates duing both VI and KI of complex hand movements. We hypothesized that diffeent types of MI might be mediated by distinct neual netwoks involving the ceebal egions elated to the pedominant sensoy systems suppoting the MI content. Accodingly, the VI task was thought to equie pimaily the visual peception of the subject s finges movement in his mind s eye, wheeas KI was believed to be associated with moto functions pe se, hence equiing subjects to imagine o to feel the movement while consideing the body as a geneato of foces. Visual egions wee thus expected to be moe activated duing VI, wheeas the moto systems and the sensoimoto aeas wee assumed to be pimaily ecuited duing KI. A citical issue elated to the classical studies dealing with MI, howeve, is that the subject s compliance with instuctions has aely been contolled adequately. This paticula methodological concen is impotant to conside as subjects may encounte difficulty to dissociate VI fom KI, even though they have eceived specific instuctions and undestood the diffeence between MI based on visual and kinesthetic cues. To veify in the pesent study that subjects pefomed MI as instucted to, we used both subjective and objective measues to evaluate the individual accuacy of MI, including physiological measues fom the ANS. The combination of these complementay techniques is a valid pocedue to contol the quality of MI, hence guaanteeing the high-mi abilities of the paticipants in the fmri study [Guillot and Collet, 2005a; Lotze and Halsband, 2006]. Ideally, simultaneous electomyogaphic (EMG) data should have been used to ecod the muscula quiescence duing MI and to ensue that the patten of ceebal activation obseved duing MI was not due to any movement. Howeve, fmri activations duing MI of finge movements have peviously been found independent fom the degee of EMG activation [Poo et al., 1996], hence suggesting that fmri changes can be mainly attibutable to the activity of intacotical cicuits and not due to a 2166
Figue 4. Activation maps of the lateal pemoto aea duing the visual and kinesthetic imagey conditions. When the visual imagey condition was contasted with the kinesthetic imagey condition (uppe ow, x 5229; y 529; z 5 53), the activation was located moe posteioly and supeioly than when the kinesthetic imagey condition was compaed to the visual imagey condition (lowe ow, x 5259; y 5 8; z 5 7). Activation maps of the exclusive bain activations duing both visual and kinesthetic imagey conditions. When the visual imagey condition was contasted to the kinesthetic imagey condition (left side), selective activation was found in the supeio paietal lobule (BA 5 7; X 521) and occipital cotex (BA 17 19, Figue 5. Z 5 12 and Z 5 40). When the visual imagey condition was subtacted fom the kinesthetic imagey condition (ight side), exclusive activation was located in the putamen (Z 5 1), the ceebella cotex (Z 5223), and the infeio pefontal aea (BA 44, Z 5 13).
Guillot et al. muscula contaction. Futhemoe, we did not obseve any movement duing MI, as no behavioal data wee ecoded while the hand emained on the fou key keyboad. As expected and consistent with ealie neuoimaging studies, when we compaed the physical execution and peceptual contol conditions [Decety et al., 1994; Geadin et al., 2000; Lotze et al., 1999, 2003; Solodkin et al., 2004], activations wee found in the moto-elated egions including the moto and pemoto cotices, as well as the SMA and the supeio and infeio paietal lobules. The anteio and posteio pats of the putamen and both the anteio and posteio egions of the ipsilateal ceebellum wee also ecuited. The esults evealed that both MI and moto pefomance shae common neual substates. The pesent study thus confims the functional anatomy of MI and suppots that MI and moto pefomance diffe in the final moto output stage, which is not expessed duing MI [Geadin et al., 2000]. Moto Imagey vs. Peceptual Contol Condition Even if we postulated that diffeent types of MI may esult in diffeent pattens of activation, the ceebal netwoks elated to the two MI tasks wee thought to shae similaities, in ode fo the subject to geneate a mental epesentation of the finge sequence. We compaed the two MI conditions with the contol task and obseved activations within the ight pimay moto cotex (M1). Such activation is in accodance with ecent fmri studies that have epoted significant changes in M1 activity duing MI of hand movements [Kutz-Buschbeck et al., 2003; Poo et al., 1996; Shama et al., 2008]. As shown in pevious studies, inceased activity was also found in pe-sma and SMA egions duing both VI and KI [Geadin et al., 2000; Lafleu et al., 2002; Lotze and Halsband, 2006; Stephan et al., 1995]. The latte activations ae consistent with the well-known contibution of the pe-sma to seveal stages duing movement planning [Deibe et al., 1992; Humbestone et al., 1997; Shima and Tanji, 1998], as well as the SMA pope, which is usually active duing the initiation and execution of actions [Deibe et al., 1992; Passingham, 1997]. Ou findings ae also in ageement with those of Amado and Fied [2004] who showed though depth electode ecodings that SMA neuons espond moe stongly duing MI of a finge opposition sequence than duing the physical execution of the same action, hence highlighting the cucial ole of this stuctue in mediating MI. Futhemoe, the esults of the pesent study evealed inceased activity in the dosolateal pefontal cotex (BA 46), which is known to be ecuited when the tempoal initiation of individual movements is unde the subjects volition, i.e., deciding when to initiate finge movements [Jahanshahi et al., 1995] as well as which finge to move [Fith et al., 1991]. Lastly, the activations obseved in the supeio paietal lobule (BA 7) and in subcotical egions including the putamen and ceebellum ae consistent with esults fom ealie expeiments which have assessed the neual netwoks activated by these two diffeent foms of MI [Binkofski et al., 2000; Decety et al., 1994; Geadin et al., 2000; Lotze et al., 1999; Solodkin et al., 2004]. In addition to the similaity in the patten of activity duing VI and KI pesented ealie, the compaison of both VI and KI with the peceptual contol condition also povided evidence of diffeences in bain activations. Indeed, the occipital egions (including the pimay visual aea and the pestiate cotex) wee only ecuited duing VI, wheeas the infeio and supeio paietal lobules(ba1 3and40),aswellastheventalpemoto cotex (BA 44), wee only seen duing KI. Ou esults futhe confim pevious findings suggesting that the lateal pat of the pemoto cotex (BA 6) is active duing both MI conditions [Decety et al., 1994; Geadin et al., 2000; Stephan et al., 1995] and that VI ecuits pedominantly the dosal pat of this aea, wheeas KI activates its vental pat. Inteestingly, bain activity mediating VI and KI involved many egions with somatotopic oganization. Pevious studies have epoted homuncula oganization duing MI [Ehsson et al., 2003]. Fo example, latealization has been obseved in the pimay moto cotex as well as in the SMA pope, wheeas minimal lateal oganization has been epoted in othe egions including the pemoto cotex, the pe-sma and the cingulate cotex [Michelon et al., 2006]. Electoencephalogaphic studies have also found evidence fo such lateal oganization [Pfutschelle and Neupe, 1997; Pfutschelle et al., 1999]. In the pesent findings, howeve, we did not obseve a clea patten fo lateal oganization in the moto systems duing both VI and KI, with the exception of the pimay moto cotex. By contast, we moe often found bilateal pattens of activation in egions with somatotopic oganization. As suggested by Michelon et al. [2006], this may be elated to the epetition of the complex finge sequence MI task, so that paticipants elied less on moto simulation than on memoy etieval, which may have educed the latealization patten of activity. Visual Imagey vs. Kinesthetic Imagey As expected, a divegent patten of inceased activity was obseved when VI and KI wee compaed diectly. Fist, the involvement of the pimay visual aea and the pestiate cotex duing VI confims the esults of pevious neuoimaging studies [Jackson et al., 2006; Mellet et al., 1998; Solodkin et al., 2004; Thompson and Kosslyn, 2000; Zacks et al., 2002], as such aeas have been shown to be ecuited when subjects visualize hand movements, but not when they peceived body sensations associated with KI. This demonstates that VI of body pat movements, compaed with KI, shaes common occipital substates with visual peception [Andesen et al., 1985]. Second, consistent with neuoimaging and neuopsychological studies of healthy subjects as well as patients with paietal lobe lesions [Deibe et al., 1998; Geadin et al., 2000; Siigu 2168
Neual Substates of Moto Imagey et al., 1996], we found activations of the visual pathways including the supeio paietal stuctues (BA 7) and the pecuneus, which ae known to be involved in the geneation of mental images. These stuctues eceive affeents fom multisensoy inputs, including the lowe visual field, to elaboate the egocentic epesentation of space [Iacoboni et al., 1999; Lacquaniti and Caminiti, 1998], and have been found to be ecuited duing voluntay attention shifting and mental imagey tasks [Cavanna and Timble, 2006; Malouin et al., 2004; Suchan et al., 2002]. Thus, the pesent esults show that supeio occipital and paietal egions, which ae known to be involved in spatial mental imagey and mental navigation even in the absence of any visual input [Mellet et al., 1998], may also paticipate to the mental epesentation of specific body pat movement duing VI. When contasted to the VI condition, KI was found to elicit bilateal activations of the infeio paietal lobule (BA 40), a egion known to have a cucial ole duing MI [Decety et al., 1994; Geadin et al., 2000; Nai et al., 2003; Stephan et al., 1995; Suchan et al., 2002]. Futhe, the contast also showed activations in seveal moto-elated egions, including the putamen, the caudate nucleus, and the ceebella hemisphees. The latte esults suppot the findings by Jackson et al. [2006] who epoted a geate activation within the moto system duing the fist-peson pespective than duing the thid-peson pespective (consideing that KI is close to a fist- than to a thid-peson pespective). This finding suggests that the paallel chaacteistics between MI and physical execution of movements is based on the integation of a moto pogam and its coesponding sensoy feedback, which is moe diectly available duing KI than duing VI [Koch et al., 2004; Pinz, 1997; Wulf and Pinz, 2001]. Finally, when each MI conditions was diectly contasted, the vental pemoto cotex (BA 44), which is usually active duing MI [Binkofski et al., 2000; Buccino et al., 2006; Geadin et al., 2000; Gafton et al., 1996; Solodkin et al., 2004] was also activated, but only duing KI. The vental pemoto cotex coesponds to the human analogue of the so-called mio neuons egion [F5 aea in the monkey cotex Rizzolati et al., 1996, 2001], whee it has been shown that obseved actions ae eflected in the moto epesentation of the same action in the obseve. The mio neuons have been shown to espond at highe levels when stong kinesthetic feedback is peceived [Buccino et al., 2001; Koski et al., 2002]. The hemodynamic changes obseved in this egion duing KI confim the latte esults, as KI involves moe kinesthetic components than VI. Howeve, the pesent esults do not eplicate those of Binkofski et al. [2000] who epoted that the vental pat of the lateal pemoto cotex is moe ponounced duing VI. The diffeences between the task design of the two studies may be the main explanation, as ou finge sequence was moe complex than that used by Binkofski et al. [2000]. Especially, significantly stonge pemoto activity has been found duing imagey of complex sequential compaed with simple epetitive finge movements [Kuhtz-Buschbeck et al., 2003]. To conclude, ou esults suppot ou theoy-diven account suggesting that VI of a finge sequence task makes efeence to the visual popeties of visual peception, wheeas KI includes in geate extend moto simulation pocesses closely elated to the fom and timing of actual movements [Michelon et al., 2006]. These findings thus demonstate that diffeent neual pocessing may be elated to the VI and KI conditions. Although Solodkin et al. [2004] investigated the diffeences between VI and KI in two independent goups, i.e., one pefoming VI and the othe using KI, ou esults ae based on ANS ecodings and behavioal tests, and fmri scanning that wee done in the same subjects, with good to excellent individual MI abilities, hence leading us to compae diectly the neual activity elicited by VI and KI of finge movements. Even though these two conditions shae common activations, the neual substates mediating the ability to pefom a specific MI type wee found to be quite diffeent, as the two MI conditions elicited sepaate pattens of neual activity. The diffeences and similaities in the pattens of activation mediating the two types of MI thus povide evidence that subjects ae able to favo one sensoy modality to fom mental images, albeit they emain able to have a geneal mental epesentation of the movement. Such a finding has stong theoetical and pactical implications both in moto leaning and ehabilitation pocesses. Fist, neuoanatomical evidence of neual netwoks mediating MI is povided with egad to the use of MI based on visual o kinesthetic sensoy cues. This finding cooboates the impotant numbe of behavioal and psychological studies elated to the effect of MI in diffeent stages of moto leaning, which have suggested that VI o KI may be pefomed selectively with egads to the chaacteistics of the moto skill [fo eview, see Guillot and Collet, 2008]. Second, when consideing the ehabilitation pocess following stoke, the pesent findings suggest that pactitiones should pay moe attention to the MI execises they ecommend to thei patients. In some cases, the seveity of the lesion is such that pefoming a moto task is vey difficult, sometimes impossible, thus pecluding ealy paticipation in an active ehabilitation pogam. Recent data suggest, howeve, that MI could be used as a theapeutic tool to pevent mis-epai by keeping the emaining wellfunctioning stuctues active, by impoving neuonal plasticity, and thus to peseve moto functions [Jackson et al., 2004; Johnson-Fey, 2004; Liu et al., 2004; Malouin et al., 2004; Page et al., 2001; Pascual-Leone et al., 1995]. When consideing the selective activations we obseved duing VI and KI, ou esults may also help to select goups of patients that could benefit fom this theapeutic appoach. The fist step would then be to investigate thei abilities to use VI and/o KI afte cotical and subcotical lesions, with the aim to detemine the optimal taining conditions fo leaning how to use MI in neuological ehabilitation [Malouin et al., 2007]. 2169