318gather visual information from the retinaand propagate it through a series of visualand ultimately motor areas. However, thedelays involved in this pathway could lead toserious instability in the control of move-ment (7). An alternative possibility, suggest-ed by computational models, is that thebrain uses a “forward model” that generatesa much faster prediction of the conse-quences of a given motor command (7). Theforward model contains an estimate of thearm’s musculoskeletal properties and of ex-ternal forces such as friction. The modelcombines these estimates with a copy of themotor command being sent to the arm fromM1 (see the figure) and predicts, with mini-mal delay, the actual arm trajectory expect-ed to result from that motor command. Inthe task set by Schwartz et al., the output ofthe forward model could be represented bythe PMv activity that lags M1 activity andencodes the visible (predicted) path of thecursor. By comparing the predicted move-ment with the required movement (the targetcontour) the brain can then make rapid cor-rections to the ongoing arm trajectory. The likely contribution of PMv neuronsto motor control does not, of course, pre-clude the possibility that this area also influ-ences perception. Several important ques-tions remain to be answered in this regard.First, it is necessary to determine whetherthe visual signals encoded in the PMv cor-relate with visual perception within an indi-vidual subject (a macaque monkey). Second,it must be established whether and how in-activation, lesioning, or other manipulationsof premotor areas affect perception. Finally,it must be determined whether differentpopulations of neurons within a single pre-motor structure contribute differentially toaction and perception.References1. A. B. Schwartz, D. W. Moran, G. A. Reina,Science303,380 (2004).2. R. Held, A. Hein,Percept. Mot. Skills8, 87 (1958).3. L. Shen, G. E. Alexander,J.Neurophysiol. 77, 1195(1997).4. A. P. Georgopoulos, J. F. Kalaska, R. Caminiti, J. T.Massey,J. Neurosci.2, 1527 (1982).5. E. Gerardin et al.,Cereb. Cortex10, 1093 (2000).6. M. M. Mesulam,Philos. Trans. R. Soc. London Ser. B354, 1325 (1999).7. M. I. Jordan, Computational Motor Control.TheCognitive Neurosciences,M.S.Gazzaniga, Ed. (MITPress, Cambridge, MA, 1995), pp. 597–609.Humans have acquired six symbolsystems: two that evolved—the ge-netic code and spoken language—and four that we invented: written language,arabic numerals, music notation, and laban-otation (a system for coding choreography).Dobzhansky’s quip “All species are unique,but humans are uniquest” raises the ques-tion: Is it language, the symbol system thatevolved only in humans, that makes hu-mans the “uniquest”? Dobzhansky’s quipraises a more fundamental question: Whatexactly is the nature of human uniqueness?The grammar or syntax of human lan-guage is certainly unique. Like an onion orRussian doll, it is recursive: One instance ofan item is embedded in another instance ofthe same item. Recursion makes it possiblefor the words in a sentence to be widely sep-arated and yet dependent on one another. “If-then” is a classic example. In the sentence“If Jack does not turn up the thermostat inhis house this winter, then Madge and I arenot coming over,” “if ” and “then” are de-pendent on each other even though they areseparated by a variable number of words(1–3). Are animals capable of such recur-sion? In a paper on page 377 of this issue,Fitch and Hauser (4) report that tamarinmonkeys are not capable of recursion.Although the monkeys learned a nonrecur-sive grammar, they failed to learn a grammarthat is recursive. Humans readily learn both.The lack of recursion in tamarins may helpto explain why animals did not evolve recur-sive language, but it leaves open the questionof why they did not evolve nonrecursive lan-guage. Recursion is not, of course, the onlypreexisting faculty on which the evolution oflanguage depends, and when we examinesome of the other factors (listed in the table),we can see why animals did not evolve lan-guage of any kind.Voluntary Control of Sensory-MotorSystemsA laboratory chimpanzee does not call toattract the attention of its trainer; instead, itpounds on a resonant surface. Similarly,when chimpanzees become separated inthe compound, they do not call to one an-other, as humans would, but search silentlyuntil they see one another and then rush to-gether. If, as the evidence suggests, vocal-ization in the chimpanzee is largely non-voluntary (reflexive), speech could nothave evolved. But then why don’t chim-panzees sign to each other? The chim-panzee has voluntary control of its hands.However, sign language depends on theface as well as the hands, and facial ex-pression in the chimpanzee is evidently asreflexive as vocalization. Facial expres-sions play linguistic roles in signing, suchas denoting the boundaries of clauses. Asigner processes emotional facial expres-sion in the right hemisphere, but linguisticfacial expression in the left hemisphere (5).This does not mean, of course, that chim-panzees could not have evolved a languagebased on pounding on resonant surfaces,arranging stones on the ground, and so on.But it does suggest that they could not haveevolved one that is like either speech orsign. (Of course, speech and sign “travel”with the speaker in a way that stones andresonant surfaces do not.)ImitationMany species can copy the object (or loca-tion) chosen by a role model. This is thefirst level of imitation. There is, however, asecond level of imitation when the individ-ual copies not the model’s choice of object,but rather the model’s motor action. Quitea different kettle of fish. Now the individ-ual must form a mental representation ofthe visually perceived action and producean action conforming to the representation(6). Although humans, even as infants, cando this (7), most species cannot, the excep-tion being chimpanzees but they requirehuman training (8–10). Could languageevolve in a species in which the young can-not imitate the action of the speaker?TeachingTeaching, which is strictly human, reversesthe flow of information found in imitation.Unlike imitation, in which the novice ob-serves the expert, the teacher observes thenovice—and not only observes, but alsojudges and modifies (6). Imitation andteaching pair efficiently in humans.Imitation produces a rough copy; teachingsmoothens it. A chimpanzee mother couldnot teach her infant anything