Review

Motor imagery: if you cant do it, you wont think it

C.-J. Olsson1,2 and L. Nyberg1,2,3

1Integrative Medical Biology, Physiology, Umea University, Umea, Sweden, 2UFBI, Umea Centre for Functional Brain Imaging,Umea University, Umea, Sweden, 3Radiation Sciences, Radiology, Umea University, Umea, SwedenCorresponding author: C.-J. Olsson, Department of Integrative Medical Biology, Umea University, S-901 87 Umea, Sweden.Tel: 146 90786 5186 13, Fax: 146 90 786 6696, E-mail: cj.olsson@physiol.umu.se

Accepted for publication 23 December 2009

Since long, motor imagery has been recognized as a methodfor studying motor representations. In the last few years,important advances regarding the use of motor imageryhave been made. In particular, issues concerning the func-tional equivalence between imagery and action have beenaddressed, and how equivalence aects the use of imagery tostudy motor representations. In this paper, we review recentndings in order to highlight the current state of knowledgeabout motor imagery and its relation to motor action. Three

topics are discussed: (i) the imagery perspective, (ii) taskcomplexity, and (iii) the importance of physical experience.It is shown how theses factors are closely related and howprevious studies may have underestimated to what extentthese factors aect the interpretation of results. Practicalimplications for imagery interventions are considered. It isconcluded that if you cannot perform an action physically,you cannot imagine it in a way that is necessary for a highdegree of functional equivalence.

Slightly over 10 years ago, it was suggested that motorimagery could provide new insights into motor repre-sentations because of its overlap with physical execu-tion, i.e., functional equivalence (Crammond, 1997).There were several papers supporting that motorimagery activates the same brain regions as thoseused during motor performance (see e.g., Jeannerod,1994; Stephan et al., 1995; Porro et al., 1996) andresearchers attempted to nd limitations of imagery inrelation to motor performance improvements as wellas the neural overlap with motor execution. In a seriesof studies, Sirigu et al. (1995, 1996) used physical andmental chronometry tasks and showed how patientswith parietal lesions had impaired imagery. This waselaborated upon by Crammond (1997), and based onthe ndings that the parietal cortex may be importantfor our ability to generate mental movement represen-tations, it was suggested that we may not mentallymove ngers faster than we can move them physically.However, it was also concluded that the basis of thislimitation remains largely unknown.As a result of the suggested functional equivalence

between imagery and action, motor imagery has been awidely used strategy to improve motor performanceboth in rehabilitation (Jackson et al., 2001) and in avariety of sports, e.g., high jump (Olsson et al., 2008),gymnastics (Smith et al., 2007), golf (Brouziyne &Molinaro, 2005), basketball (Pie & Tenenbaum,1996), and swimming (White et al., 1979). The theore-tical basis for motor imagery suggests that beforeexecution of a voluntary movement, the brain has

formed a motor representation that is believed tocomprise the entire movement, including the plan aswell as the intended result (Kandel et al., 2000). More-over, since the motor representation supposedly pre-cedes the execution then it could be detached from theexecution and exist on its own (Jeannerod, 2006).Therefore, it can consciously be accessed during motorimagery and possibly the motor performance will beimproved (Jeannerod, 1994, 1995). Thus, for bothpractical and theoretical reasons, it is important tounderstand how the brain handles motor representa-tions in the imagined state when no execution isperformed. The importance of this has been shown inthe last couple of years, with several papers providingmore details in terms of the specicity of motorrepresentations and how that aects motor imagery.For example, in a recent study, we were able to showhow specic motor representations are (Olsson et al.,2008), and how that specicity aects the use of motorimagery. This leads to the proposal about motorrepresentation specicity, i.e. during motor imagery,specic and not general motor representations are usedand these representations are only created throughtask-specic training. Together with several other stu-dies (e.g., Calvo-Merino et al., 2006; Aglioti et al.,2008), these new ndings have extended our knowledgeabout how motor imagery can be used to study motorrepresentations and what the representations studiedduring motor imagery really represent. In this paper,we review recent studies in order to address issuesregarding similarities between imagery and action. We

Scand J Med Sci Sports 2010: 20: 711715 & 2010 John Wiley & Sons A/S

doi: 10.1111/j.1600-0838.2010.01101.x

711

will focus on three topics that are central to motorimagery and aect how motor imagery may be per-formed. These topics are (i) the imagery perspective, (ii)task complexity, and (iii) the importance of physicalexperience. Careful consideration of these topics, eventhough they are well known and discussed in theliterature, is necessary in order to realize the limitationsregarding the functional equivalence between imageryand action, and thus the borderline conditions for whenimagery can be used to study motor representations.

Imagery perspective

When engaging in imagery, this can be done usingeither a rst- or a third-person perspective. The dier-ence is that in a rst-person perspective, it is importantto feel as if the action is executed (but not actuallyperformed), in comparison with a third-person per-spective that would be as if you watch yourself on TV.By denition, when performing motor imagery, therst-person perspective is used (Jeannerod, 1997).However, the potential similarities, and dierences,between the two perspectives have been the subjectfor several investigations, with some mixed results. Forexample, based on mental chronometry, it has beenproposed that there are not only similarities betweenimagery and action but also that the rst- and third-person perspectives may share common representations(Anquetil & Jeannerod, 2007). However, it has alsobeen suggested that the two perspectives may becontrolled by dierent processes (Farrer & Frith, 2002).The dierent interpretations regarding imagery

perspective can possibly be accounted for by theresults of a study by Calmels et al. (2006). In thisstudy, it was investigated how closely related internaland external imagery are by timing how long it tookfor elite gymnasts to imagine a gymnastics routine incomparison with the time it took to execute it. Theoverall pattern showed no dierences between thedierent perspectives, or between imagery andaction, thus supporting similarities. However, Calmelsand colleagues also investigated whether dierent partsof the routine diered between internal imagery,external imagery, and physical execution. This moredetailed investigation revealed some interesting nd-ings. Although the total time between imagery andaction was statistically similar, in the action condition,the run phase was signicantly slower compared withthe imagined run phase, and the ight phases weresignicantly faster compared with the imagined ightphases. Thus, the total time for an action may be thesame between imagery and performance, but dierentsequences of the action may dier between imaginedperformance and actual execution. Hence, achieving afunctional equivalence between the dierent perspec-tives and action is aected by the level of detail you

investigate. Consequently, if the imagery perspectiveand the functional equivalence is vital in order tomaximize the performance eects, as suggested by,e.g., Holmes and Collins (2001), then the dierencesbetween the two perspectives and its relation to motoraction may be more important than previously ac-knowledged. Therefore, it is important to understandhow the dierent perspectives are related to dierentparts of a task and that the relation between imageryand action may be more complicated than previouslyunderstood. As seen in the above example, imagery isnot only aected by perspective but also by thecomplexity of the task. Thus, complex tasks such asgymnastic routines may cause a dierent neural over-lap compared with more simple tasks, which will be thetopic for the next section.

Task complexity

Task complexity is an important, but perhaps under-estimated, factor when understanding motor imagery.Failures to consider task complexity may have causedsome misinterpretations about the similarity betweenimagery and action. When comparing functionalneural overlap between imagery and action, moststudies have used easy tasks that may be too simpleto distinguish between motor performance and motorimagery (see e.g., Laeur et al., 2002; Orr et al., 2008).It is possible that such tasks may be too simple toreveal the limitations of imagery, which may haveoversimplied the relation between imagery and action.The importance of task complexity was taken into

consideration by Szameitat et al. (2007) when theyused a variety of complex everyday tasks, such asswimming or eating with a fork and knife, to inves-tigate the functional neuroanatomical correlates ofmotor imagery. They concluded that in analogy withsimple tasks, there was an overlap between imageryand execution, supporting that imagery may be usedto improve motor performance for complex tasksbased on the same mechanisms as physical training.However, a potential explanation why this study wasnot able to nd dierences based on task complexitymay be because of the use of everyday tasks. Suchtasks are highly trained, raising the possibility that theresults would have been dierent if the tasks not onlyhad been complex but also novel.This role of task familiarity in relation to task

complexity was highlighted in a study by Olsson et al.(2008). In this study, a group of active high jumpersperformed motor imagery of a high jump duringfMRI scanning. The brain activity for this group wascompared with a group of high jumping novicesmentally performing the same task. Both groupswere given detailed instructions to use an internalperspective. Still, between-group dierences in brain

Olsson & Nyberg

712

activity were profound (see Fig. 1), and it was clearthat only the group with an extensive high jumpingbackground was able to activate motor regions. Bycontrast, the novices activated visual and parietalregions. This led to the conclusion that in order toachieve neural overlap between imagery and action, awell-established motor representation from physical

training is required. Thus, the functional equivalencebetween imagery and action is not only aected bythe imagery perspective but also, as it seems, by taskcomplexity and familiarity.

Importance of physical experience

As pointed out in the previous section, it appears tobe physical experience that is the factor determiningthe limitations of imagery, possibly aecting bothperspective and task complexity. If true, the next stepin understanding the relationship between imageryand action is to understand to what extent physicaltraining inuences the ability to perform motorimagery with a high degree of functional equivalence.That is, can we imagine a motor action that we donot know how to physically perform and thus makethe brain believe that we are actually performing it.When performing imagery with the intention to

improve a behavior, the person envisions and createsan image of the future. Szpunar et al. (2007) testedthe neural correlates underlying our ability to ima-gine future events. The results showed that we useour past events, and similar neural structures, to putourselves in the intended future position. Thus,future events are based on past events, suggestingthat you need to previously have done the task thatyou are about to imagine in order to achieve neuraloverlap between action and imagery. However, itremains unclear how specic one has to be in termsof physical training. For example, would it be en-ough to have the ability to run and jump in order tocreate an imaginary high jump? The Olsson et al.(2008) study, described previously, suggested other-wise. Therefore, the central question that needs to beaddressed is how specic are motor representationsand how that aects the functional equivalence. Inorder to answer this question, we now turn to arelated eld: observation studies.If imagery by denition makes it possible to

consciously access the motor representation, it couldbe viewed as a rather active process. In motorobservation, on the other hand, the process is moreof a passive process in which it is assumed that themotor system is accessed by simply watching motoractions. Support for this comes mainly from thediscovery of the mirror neuron system (see e.g.,Rizzolatti & Craighero, 2004). Nevertheless, one ofthe benets with observation is that you knowexactly what the subjects are watching comparedwith imagery, when you have to rely on the subjectsown imaginary performance with limited control.Further, in the observation literature, several at-tempts have been made to address the issue ofphysical experience. Thus, we acknowledge that thereare dierences between the two conditions, imageryand observation, but we also believe that the simila-

Fig. 1. Comparing brain activity between high jumpers (leftpanel) and high jumping novices (right panel), it was madeclear that even though instructed to use the internal per-spective, only the active high jumpers were able to activatemotor regions. For the high jumpers, activation was seen inregions such as the SMA, cerebellum, and premotor cortex(bilaterally). For the controls, activation was seen in theinferior parietal cortex, superior occipital cortex, and super-ior temporal cortex. It was speculated that in order for afunctional equivalence between imagery and action, the taskmust be well learned, and requiring a well-established motorrepresentation based on physical training. Reproduced withpermission from authors (Olsson et al., 2008).

Motor imagery

713

rities are so strong that we can make parallels betweenthem. For example, during observation, there seems tobe specicity in terms of previous physical experiencethat inuences pattern of brain activity. In a study byAglioti et al. (2008), the task was to predict the fateof an action. More specically, video-clips were shownof a professional basketball player performing freebasket shots. Each clip was interrupted at dierentpositions of the shooter, providing dierent amountsof information to be used for the participants to decidewhether the basketball shot would go in the basket ornot. Some clips were stopped before the basketball leftthe players hand, and in some clips, the full motion ofthe shot was shown. Three groups of participants wereused, one group of elite basketball players, one groupof expert watchers (coaches and journalists), who wereno longer playing basketball but had been playingpreviously, and one group of basketball novices. Theresults showed that only highly trained athletes wereable to use body cues of the shooter to predict thesuccess of a basketball throw; professional watchershad to use the trajectory of the ball to make the samejudgments. This suggests a high degree of specicity ofthe motor representations that is shaped by specicphysical experience. Thus, not only are their resultssupporting that one must train the specic task inorder to come close to accessing the representationwhen watching, but the specic physical experiencemust be relatively recent.In another study using observation to support the

specicity of motor representations, Calvo-Merino etal. (2005) showed how expert dancers were only able toaccess the specic motor system when observing dancemoves within their own repertoire. When instructed toobserve other dance-like moves, capoeira, which weresimilar yet something the participants had not physi-cally performed, the brain activation pattern diered.To further underpin the specicity of motor representa-tions, Del Percio et al. (2008) showed how the corticalactivity of experts was more focused and ecientduring observation compared with novices, and thatthis was dependent on task experience (see also Babiloniet al., 2009). Thus, several studies within the last fewyears have suggested that physical experience aectsobservation and the brain. As such, physical experiencespecic to the task may also be a factor aecting thefunctional equivalence between imagery and action.A related issue, which is not a main issue for the

present paper and yet is necessary to acknowledge, iswhether physical expertise is only due to physicalexperience or whether there are innate genetic con-tributions to athletic performance. If so, this couldalso potentially aect motor imagery. The inuenceof genetic traits in sports performance is acknowl-edged with some empirical support. It has, forexample, been suggested that certain personalityfactors may be vital in order to devote all that time

and eort needed to become a world-class athlete.Moreover, personality may be, at least partly, ge-netically determined (see e.g., Klissouras et al., 2007).However, it is also suggested that exceptional per-formance is inuenced by experience and practiceand only minimally by genetically dened traits(Ericsson et al., 1993; Ericsson, 2007). Despite thedierences between dierent research groups regard-ing the relative contribution of nature vs nurture toathletic performance, the collected eld seems toagree that there is a need for extensive involvementin training in order to achieve exceptional athleticperformance (Baker & Davids, 2007).

Practical implications

Frequently, studies are published that support thenotion of similarities between action and imagery aswell as between action and observation (see e.g.,Gazzola & Keysers, 2009). However, in the future, aswe have argued in this article, we must consider thespecicity of the motor representation in relation to theuse of imagery. Moreover, this issue may be of criticalimportance for studies dealing with mental training forsports or rehabilitation. For example, Beilock andGonso (2008) failed to explain why somehow onlyexperienced golfers were able to ne-tune the motorprocess during mental training, but not novices. Alikely explanation is that only the experienced golferswere able to achieve a functional overlap due to havingvast physical experience, and therefore were able to usethe rst-person perspective on a complex task, resultingin performance improvements. Thus, an apparentlimitation with imagery is that it seems to be of greatimportance to have previously physically executed theaction at a reasonable level. Only then can functionalequivalence be expected and the imagery interventionexpected to be successful. Thus, recent advances inneuroscience research are now oering empirical evi-dence that it is only possible to imagine, with a neuraloverlap to the actual action, if you have a well-established motor representation. Hence, if you cannotperform an action physically, you will not be able tothink it mentally. In practical terms, this means thatwhen deciding upon imagery interventions for rehabi-litation, sports, or even research, we must take theproposed specicity of motor representations morecritically if we want to use motor imagery as a wayof understanding motor representations and to under-stand how it will aect motor performance.

Key words: motor imagery, functional equivalence,mental training, motor representations.

Acknowledgements

This work has been supported by CIF (Centrum for Idrotts-forskning).

Olsson & Nyberg

714

References

Aglioti SM, Cesari P, Romani M, UrgesiC. Action anticipation and motorresonance in elite basketball players.Nat Neurosci 2008: 11(9): 11091116.

Anquetil T, Jeannerod M. Simulatedactions in the rst and in the thirdperson perspectives share commonrepresentations. Brain Res 2007: 1130:125129.

Babiloni C, Del Percio C, Rossini PM,Marzano N, Iacoboni M, Infarinato F,Lizio R, Piazza M, Pirritano M,Berlutti G, Cibelli G, Eusebi F.Judgment of actions in experts: ahigh-resolution EEG study in eliteathletes. NeuroImage 2009: 45(2):512521.

Baker J, Davids K. Sound and fury,signifying nothing? Future directions inthe nature-nurture debate. Int J SportPsychol 2007: 38(1): 434.

Beilock SL, Gonso S. Putting in the mindversus putting on the green: expertise,performance time, and the linking ofimagery and action. Q J Exp Psychol2008: 61(6): 920932.

Brouziyne M, Molinaro C. Mentalimagery combined with physicalpractice of approach shots for golfbeginners. Perceptual Motor Skills2005: 101(1): 203211.

Calmels C, Holmes P, Lopez E, NamanV. Chronometric comparison of actualand imaged complex movementpatterns. J Mot Behav 2006: 38(5):339348.

Calvo-Merino B, Glaser DE, Grezes J,Passingham RE, Haggard P. Actionobservation and acquired motor skills:an fMRI study with expert dancers.Cereb Cortex 2005: 15(8): 12431249.

Calvo-Merino B, Grezes J, Glaser DE,Passingham RE, Haggard P. Seeing ordoing? Inuence of visual and motorfamiliarity in action observation. CurrBiol 2006: 16(19): 19051910.

Crammond DJ. Motor imagery: never inyour wildest dream. Trends Neurosci1997: 20(2): 5457.

Del Percio C, Rossini PM, Marzano N,Iacoboni M, Infarinato F, Aschieri P,Lino B, Fiore A, Toran G, Babiloni C,Eusebi F. Is there a neural eciencyin athletes? A high-resolution EEGstudy. NeuroImage 2008: 42(4): 15441553.

Ericsson KA. Deliberate practice and themodiability of body and mind: towarda science of the structure andacquisition of expert and elite

performance. Int J Sport Psychol 2007:38(1): 434.

Ericsson KA, Krampe RT, Tesch-RomerC. The role of deliberate practice in theacquisition of expert performance.Psychol Rev 1993: 100: 363406.

Farrer C, Frith CD. Experiencing oneselfvs another person as being the cause ofan action: the neural correlates of theexperience of agency. NeuroImage2002: 15(3): 596603.

Gazzola V, Keysers C. The observationand execution of actions share motorand somatosensory voxels in all testedsubjects: single-subject analyses ofunsmoothed fMRI data. Cereb Cortex2009: 19(6): 12391255.

Holmes PS, Collins DJ. The PETTLEPapproach to motor imagery: afunctional equivalence model for sportpsychologists. J Appl Sport Psychol2001: 13(1): 6083.

Jackson PL, Laeur MF, Malouin F,Richards C, Doyon J. Potential role ofmental practice using motor imagery inneurologic rehabilitation. ArchPhysical Med Rehabilitation 2001:82(8): 11331141.

Jeannerod M. The representing brain Neural correlates of motor intentionand imagery. Behav Brain Sci 1994:17(2): 187202.

Jeannerod M. Mental imagery in themotor context. Neuropsychologia1995: 33(11): 14191432.

Jeannerod M. The cogntive neuroscienceof action. Oxford, UK: BlackwellPublishers, 1997.

Jeannerod M. Motor cognition: whatactions tell the self. Oxford, UK:Oxford University Press, 2006.

Kandel ER, Schwartz JH, Jessel TM(eds.). Principles of neural science, 4thedn. New York: McGraw-Hill, 2000.

Klissouras V, Geladas N, Koskolou M.Nature prevails over nurture. Int JSport Psychol 2007: 38(1): 434.

Laeur MF, Jackson PL, Malouin F,Richards CL, Evans AC, Doyon J.Motor learning produces paralleldynamic functional changes during theexecution and imagination ofsequential foot movements.NeuroImage 2002: 16(1): 142157.

Olsson C-J, Jonsson B, Larsson A, NybergL. Motor representations and practiceaect brain systems underlying imagery:an fMRI study of internal imagery innovices and active high jumpers. OpenNeuroimaging J 2008: 2: 513.

Olsson C-J, Jonsson B, Nyberg L.Internal imagery training in active highjumpers. Scand J Psychol 2008: 49(2):133140.

Orr ELR, Lacourse MG, Cohen MJ,Cramer SC. Cortical activation duringexecuted, imagined, and observed footmovements. Neuroreport 2008: 19(6):625630.

Pie JS, Tenenbaum G. Imageryorientation and vividness: their eecton a motor skill performance. J SportBehav 1996: 19(1): 3241.

Porro CA, Francescato MP, Cettolo V,Diamond ME, Baraldi P, Zuiani C,Bazzocchi M, diPrampero PE. Primarymotor and sensory cortex activationduring motor performance and motorimagery: a functional magneticresonance imaging study. J Neurosci1996: 16(23): 76887698.

Rizzolatti G, Craighero L. The mirror-neuron system. Annu Rev Neurosci2004: 27: 169192.

Sirigu A, Cohen L, Duhamel JR, Pillon B,Dubois B, Agid Y, Pierrotdeseilligny C.Congruent unilateral impairments forreal and imagined hand movements.Neuroreport 1995: 6(7): 9971001.

Sirigu A, Duhamel J-R, Cohen L, PillonB, Dubois B, Agid Y. The mentalrepresentation of hand movementsafter parietal cortex damage. Science1996: 273(5281): 15641568.

Smith D, Wright C, Allsopp A, WestheadH. Its all in the mind: PETTLEP-based imagery and sports performance.J Appl Sport Psychol 2007: 19(1):8092.

Stephan KM, Fink GR, Passingham RE,Silbersweig D, Ceballos-Baumann AO,Frith CD, Frackowiak RSJ.Functional anatomy of the mentalrepresentation of upper extremitymovements in healthy subjects.J Neurophysiol 1995: 73(1):373386.

Szameitat AJ, Shen S, Sterr A. Motorimagery of complex everydaymovements. An fMRI study.NeuroImage 2007: 34(2): 702713.

Szpunar KK, Watson JM, McDermottKB. Neural substrates of envisioningthe future. Proc Natl Acad Sci U S A2007: 104(2): 642647.

White KD, Ashton R, Lewis S. Learninga complex skill: eects of mentalpractice, physical practice, and imageryability. Int J Sport Psychol 1979: 10(2):7178.

Motor imagery

715

Copyright of Scandinavian Journal of Medicine & Science in Sports is the property of Wiley-Blackwell and itscontent may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder'sexpress written permission. However, users may print, download, or email articles for individual use.