J. M. Kilner and C. D. Frith* The Wellcome Trust Centre for Neuroimaging, 12 Queen Square, London WC1N 3BG, United Kingdom

S

ocial interaction depends on the ability to infer beliefs and intentions in the minds of others. Little is known about the neural basis of our ability to ‘‘read’’ the intentions of others, but a likely candidate is the mirror-neuron system (MNS). Mirror neurons discharge not only during action execution but also during action observation. It is this property that makes these neurons a possible neural substrate for action understanding. The notion that actions are intrinsically linked to perception was proposed by William James, who suggested that ‘‘every mental representation of a movement awakens to some degree the actual movement which is its object’’ (1). The implication is that observing, imagining, preparing, or in any way representing an action excites the motor program used to execute that same action (2, 3). Distinguishing Self from Other Mirror neurons were originally discovered in macaque monkeys (4). However, a variety of subsequent studies have found homologous areas in the human brain that are similarly activated when observing and executing movements (5–7). Of particular relevance here is that studies employing electroencephalography or magnetoencephalography have shown a modulation of cortical oscillatory activity at both ⬇10 and ⬇20 Hz during periods of movement observation that is similar to that observed during movement execution (8, 9). The work of Caetano et al. (10) in this issue of PNAS focuses on modulations of such oscillatory activity, recorded from human subjects by using magnetoencephalography, when executing an action (banging a drum), as well as when observing the action (seeing someone else bang a drum) or hearing the action (hearing the sound of a drum being hit) performed by another person. First, Caetano et al. showed that the oscillatory activity at ⬇10 Hz, which is thought to originate in the primary somatosensory cortex (S1), showed a similar pattern of modulation in all conditions but that this modulation lasted ⬇600 ms longer when the subjects performed the actions compared with when they either observed or listened to them. The authors suggest that this difference could reflect that fact that when executwww.pnas.org兾cgi兾doi兾10.1073兾pnas.0702937104

ing an action there is a proprioceptive signal caused by the action that does not occur when either observing or listening to an action. The authors make the interesting suggestion that this signal would enable the MNS to solve the problem of correct agency attribution. It is argued that the MNS can decode the intention of an observed action because when observing an action the MNS is active in an identical manner to when executing the action. Although such a system might enable the intention of the action to be decoded, it creates the problem of agency attribution (see ref. 11). If the MNS is identically active during action execution and action observation, then the MNS cannot attribute

Mirror neurons respond both to observation of actions and to actionrelated sounds. agency to the action because it cannot distinguish between self and other. Caetano et al. suggest that this problem can be resolved by the presence or absence of a proprioceptive signal that prolongs the modulations of cortical activity of ⬇10 Hz. Second, Caetano et al. (10) demonstrate that oscillatory activity at ⬇20 Hz showed an increase in amplitude at the same latency after movement execution in all conditions irrespective of whether the movement was performed, observed, or heard. This is in agreement with the fact that mirror neurons respond both to observation of actions and to actionrelated sounds (12). However, one notable difference between the results of Caetano et al. and those of Kohler et al. (12) is that the mirror neurons described by Kohler et al. are in ventral premotor area F5, whereas it is now well established that the synchronous oscillatory activity at ⬇20 Hz, described by Caetano et al., originates in the primary motor cortex, M1 (13–16). This is interesting because primary motor cortex has never been considered part of the MNS.

ventral premotor area F5, inferior parietal lobule, area PF, and the superior temporal sulcus (STS) (17, 18). Of these three areas, mirror neurons have only been reported in two areas, F5 and area PF (4). Although neurons in the STS are not mirror neurons because they do not discharge during action execution, the area is often considered part of the MNS because the neurons in the STS respond selectively to biological movements, both in monkeys (19) and in humans (20). In addition, the STS is reciprocally connected to the inferior parietal lobule, area PF (21), and therefore provides a sensory input to areas of the MNS with mirror neurons (see ref. 17). Importantly, however, in relation to the work of Caetano et al. (10), to date, mirror neurons have not been reported in M1. This creates a potential interpretational problem in relating modulations in ⬇20 Hz that are generated in M1 to a MNS that does not include M1. In humans, at least, it does appear beyond doubt that activity in M1 is modulated during action observation. In addition to the report of Caetano et al. (10), there is work dating back over 50 years (4, 8, 9) showing that cortical oscillations that originate in M1 are modulated when subjects are observing actions. Furthermore, when the motor cortex is stimulated by using transcranial magnetic stimulation, the motor evoked potentials generated in contralateral hand muscles are augmented when subjects observe actions involving the hands compared with control conditions (22). Finally, observed actions have been shown to have a measurable interference effect on simultaneously executed actions (23). So why is M1 active during action observation? One possibility is that activity in the hand area of M1 is not functional but is simply modulated as a consequence of the strong reciprocal cortico-cortical connections with ventral premotor areas, F5 (24, 25). Indeed, in cebus monkeys area F5 provides the largest single cortico-cortical Author contributions: J.M.K. and C.D.F. wrote the paper. The authors declare no conflict of interest. See companion article on page 9058.

Primary Motor Cortex and the MNS The MNS is often considered to consist of three reciprocally connected areas,

*To whom correspondence should be addressed. E-mail: [email protected]. © 2007 by The National Academy of Sciences of the USA

PNAS 兩 May 22, 2007 兩 vol. 104 兩 no. 21 兩 8683– 8684

COMMENTARY

A possible role for primary motor cortex during action observation

input to M1 (25). The strength of these connections is such that activity in F5 can influence the corticospinal drive to hand muscles by modulating the sensitivity to stimulation of neurons in M1 (26). According to this explanation, activity in M1 is modulated during action observation simply because activity in F5 is itself modulated at this time. An alternative explanation is that M1 is functionally active during action observation. Recently, several studies have demonstrated that, during action execution, F5 and M1 code the executed action in different coordinate systems. The majority of F5 neurons code an exe-

cuted action in an extrinsic reference framework, which defines the relative position of the object and the hand in space, whereas neurons in M1 code the action in an intrinsic framework, based on muscle and joint space that is related to shaping of the hand and digits (27– 29). This transformation is believed to be mediated by the reciprocal connections between the two areas (27–30). It is possible that such a transformation would also be necessary to decode the intentions of an observed action because the action can be described both at a kinematic level that describes the shape of the hand and the movement of the

arm in space and time (i.e., in an extrinsic framework), and at a muscle level that describes the pattern of muscle activity required to execute the action (i.e., an intrinsic framework) (see ref. 18). This explanation would suggest that M1 should be considered part of the MNS and that neurons in M1 might well be active during both action execution and action observation. Although to date there is no report of mirror neurons in M1, this account suggests that there may be a significant population of mirror neurons in this area.

1. James W (1890) Principles of Psychology (Holt, New York). 2. Jeannerod M (1994) Behav Brain Sci 17:187–202. 3. Prinz W (1997) Eur J Cognit Psychol 9:129–154. 4. Rizzolatti G, Craighero L (2004) Annu Rev Neurosci 27:169–192. 5. Buccino G, Binkofski F, Fink GR, Fadiga L, Fogassi L, Gallese V, Seitz RJ, Zilles K, Rizzolatti G, Freund HJ (2001) Eur J Neurosci 13:400–404. 6. Decety J, Gre`zes J, Costes N, Perani D, Jeannerod M, Procyk E, Grassi F, Fazio F (1997) Brain 120:1763–1777. 7. Gre`zes J, Decety J (2001) Hum Brain Mapp 12:1–19. 8. Cochin S, Barthelemy C, Roux S, Martineau J (1999) Eur J Neurosci 11:1839–1842. 9. Hari R, Fross N, Avikainen E, Kirveskari E, Salenius S, Rizzolatti G (1998) Proc Natl Acad Sci USA 95:15061–15065. 10. Caetano G, Jousmaki V, Hari R (2007) Proc Natl Acad Sci USA 104:9058–9062.

11. Gallagher S (2007) Soc Neurosci, in press. 12. Kohler E, Keysers C, Umilta MA, Fogassi L, Gallese V, Rizzolatti G (2002) Science 297:846–848. 13. Gastaut H (1952) Rev Neurol (Paris) 87:176–182. 14. Jasper H, Penfield W (1949) Arch Psychiatr Z 183:163–174. 15. Murthy VN, Fetz EE (1992) Proc Natl Acad Sci USA 89:5670–5674. 16. Salmelin R, Hari R (1994) Neuroscience 60:537– 550. 17. Keysers C, Perrett DI (2004) Trends Cognit Sci 8:501–507. 18. Kilner JM, Friston KJ, Frith CD (2007) NeuroReport 18:619–623. 19. Frith CD, Frith U (1999) Science 286:1692–1695. 20. Oram MW, Perrett DI (1994) J Cognit Neurosci 6:99–116. 21. Harries MH, Perrett DI (1991) J Cognit Neurosci 3:9–24.

22. Fadiga L, Fogassi L, Pavesi G, Rizzolatti G (1995) J Neurophysiol 73:2608–2611. 23. Kilner JM, Paulignan Y, Blakemore S-J (2003) Curr Biol 13:522–525. 24. Matelli M, Camarda R, Glickstein M, Rizzolatti G (1986) J Comp Neurol 251:281–298. 25. Dum RP, Strick PL (2005) J Neurosci 25:1375– 1386. 26. Shimazu H, Maier MA, Cerri G, Kirkwood PA, Lemon RN (2004) J Neurosci 24:1200 – 1211. 27. Kakei S, Hoffman DS, Strick PL (2001) Nat Neurosci 4:1020–1025. 28. Kakei S, Hoffman DS, Strick PL (2003) Neurosci Res 46:1–10. 29. Kurata K, Hoshi E (2002) J Neurophysiol 88:3118– 3132. 30. Umilta MA, Brochier TG, Spinks RL, Lemon RN (2007) J Neurophysiol, in press.

8684 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0702937104

J.M.K. and C.D.F. are supported by the Wellcome Trust.

Kilner and Frith