Keven & Akins (K&A) offer a novel and convincing hypothesis explaining why neonate primates protrude their tongues in response to various types of stimulation (including adults protruding their tongues at them). Because oral movements required for suckling mature early, they also come under voluntary control early, making tongue protrusion and retraction (TP/R) one of the few motor acts available to newborns. Their hypothesis makes sense in terms of mammalian phylogeny and evolution, and in terms of nervous system development. Essentially, in the same way that “to a man with a hammer everything looks like a nail,” early in development the infant's repertoire is so limited that a wide variety of stimuli become affordances for TP/R. These stimuli include seeing others protrude their tongues (putative “imitation”), but also include seeing flashing lights or toys or hearing arousing music. Crucially, K&A's hypothesis explains not just why TP/R is observed early in development, but also explains why it mysteriously disappears shortly thereafter: As the infant's motor repertoire diversifies, a wider response repertoire is available, and the infant moves on to more mature responses. I find K&A's hypothesis and arguments both reasonable and compelling.
Given that TP/R is the only well-replicated “imitative” neonate action from Meltzoff and Moore's (Reference Meltzoff and Moore1977) study, and the only action documented more recently in neonates of several nonhuman primate species, K&A's hypothesis should prompt careful reexamination of the literature on neonatal imitation. We know that human babies will eventually become imitators: this is a robust and distinctive feature of Homo sapiens. But with other primates the opposite is true, and in adult macaques there is little evidence for imitation.
Meltzoff and Moore's original (1977) study was astounding not because it demonstrated imitation in humans, but because it seemed to show that the connections between human visual perception and motor control were present at birth. But in nonhuman primates, neonatal TP/R remains the only strong evidence we have of any form of direct imitation in macaques, and the best evidence for chimpanzees (Ferrari et al. Reference Ferrari, Visalberghi, Paukner, Fogassi, Ruggiero and Suomi2006b; Myowa-Yamakoshi et al. Reference Myowa-Yamakoshi, Tomonaga, Tanaka and Matsuzawa2004; Paukner et al. Reference Paukner, Ferrari and Suomi2011). To the extent that this apparent evidence does not in fact demonstrate imitation, the only accepted example of macaque imitation has just disappeared.
The significance attached to “imitation” has waxed and waned over time, and a daunting empirical and theoretical literature exists debating and refining terminology (reviewed by Whiten & Ham Reference Whiten and Ham1992). In the early days of animal behavior, imitation – “learning to do an act from seeing it done” – was considered a boring low-level form of behavior. This prejudice was perhaps spurred by such English sayings as “monkey see, monkey do” or the German “nachaffen” (“after ape”) meaning “to imitate.” But accumulating evidence made clear that much apparent animal imitation is purely in the eye of the human beholder. In many circumstances where we would expect monkeys to imitate each other (e.g., learning to crack nuts with stones by watching a skilled monkey), they fail to show true imitation but rather show simpler behaviors like “stimulus enhancement” (simply observing that rocks and nuts together can lead to food). Each monkey still has to figure out, for itself, precisely how to hold and swing the stones and position the nut (Visalberghi Reference Visalberghi1987). Such studies led imitation in primates to be seen today as a sophisticated cognitive achievement (cf. Fitch et al. Reference Fitch, Huber and Bugnyar2010; Visalberghi & Fragaszy Reference Visalberghi, Fragaszy, Parker and Gibson1990; Voelkel & Huber Reference Voelkel and Huber2000).
Perhaps the biggest reason that evidence for or against monkey imitation is important is that it concerns its implications for the literature on mirror neurons, which were discovered in macaques. Mirror neurons (sometimes called “monkey see, monkey do” neurons [Carey Reference Carey1996]) are frontal neurons in macaques that fire both when the monkey performs some action and when it sees that same action performed. Such neurons appear to provide a computational substrate for motor imitation. But the catch is that – at the time of the discovery of mirror neurons – the behavioral evidence indicated that macaques do not, in fact, imitate. Although this ugly fact did not stop people from inferring that human mirror neurons play a key role in imitation, it was awkward from an evolutionary viewpoint: Just what are these mirror neurons doing in macaques, if not supporting imitation?
For mirror neuron enthusiasts, the 2006 discovery of apparent neonatal imitation in macaques was thus a great relief. Finally, it seemed, a behavioral function for macaque mirror neurons had been found, filling an otherwise uncomfortable lacuna in the theoretical edifice built upon mirror neurons. This is important, given the huge scope of explanations based on mirror neurons today, extending to speech perception, language evolution, autism research, empathy, and other major issues in cognitive neuroscience (skeptically reviewed by Hickok Reference Hickok2014). K&A's hypothesis calls such extensions sharply into question, by offering a simpler explanation of TP/R. Indeed K&A's hypothesis seems preferable to imitative hypotheses because it explains the disappearance of “imitation” during maturation that remains unexplained by the mirror neuron/imitation hypothesis.
The next and crucial step will be to design empirical tests pitting the two hypotheses against one another. I hope that researchers studying primate neonatal “imitation” and mirror neurons will rise to this challenge: The most obvious evidence in favor of K&A would come from single-unit recordings in neonatal macaques, in area F5 where mirror neurons are classically found. If such recordings find no evidence of mirror neuron involvement in the tongue protrusion response, it would be strong evidence in favor of K&A's new hypothesis.
In summary, I applaud K&A for providing a plausible alternative hypothesis for the widely accepted “neonatal imitation” interpretation of the TP/R response, and I am impressed by the breadth and depth of data that they have brought to bear in evaluating and supporting their hypothesis. Although the jury is still out, K&A provide one more reason for skepticism about neonatal imitation in general and monkey imitation in particular, as well as for circumspection about cognitive explanations that rely heavily on mirror neurons.
Keven & Akins (K&A) offer a novel and convincing hypothesis explaining why neonate primates protrude their tongues in response to various types of stimulation (including adults protruding their tongues at them). Because oral movements required for suckling mature early, they also come under voluntary control early, making tongue protrusion and retraction (TP/R) one of the few motor acts available to newborns. Their hypothesis makes sense in terms of mammalian phylogeny and evolution, and in terms of nervous system development. Essentially, in the same way that “to a man with a hammer everything looks like a nail,” early in development the infant's repertoire is so limited that a wide variety of stimuli become affordances for TP/R. These stimuli include seeing others protrude their tongues (putative “imitation”), but also include seeing flashing lights or toys or hearing arousing music. Crucially, K&A's hypothesis explains not just why TP/R is observed early in development, but also explains why it mysteriously disappears shortly thereafter: As the infant's motor repertoire diversifies, a wider response repertoire is available, and the infant moves on to more mature responses. I find K&A's hypothesis and arguments both reasonable and compelling.
Given that TP/R is the only well-replicated “imitative” neonate action from Meltzoff and Moore's (Reference Meltzoff and Moore1977) study, and the only action documented more recently in neonates of several nonhuman primate species, K&A's hypothesis should prompt careful reexamination of the literature on neonatal imitation. We know that human babies will eventually become imitators: this is a robust and distinctive feature of Homo sapiens. But with other primates the opposite is true, and in adult macaques there is little evidence for imitation.
Meltzoff and Moore's original (1977) study was astounding not because it demonstrated imitation in humans, but because it seemed to show that the connections between human visual perception and motor control were present at birth. But in nonhuman primates, neonatal TP/R remains the only strong evidence we have of any form of direct imitation in macaques, and the best evidence for chimpanzees (Ferrari et al. Reference Ferrari, Visalberghi, Paukner, Fogassi, Ruggiero and Suomi2006b; Myowa-Yamakoshi et al. Reference Myowa-Yamakoshi, Tomonaga, Tanaka and Matsuzawa2004; Paukner et al. Reference Paukner, Ferrari and Suomi2011). To the extent that this apparent evidence does not in fact demonstrate imitation, the only accepted example of macaque imitation has just disappeared.
The significance attached to “imitation” has waxed and waned over time, and a daunting empirical and theoretical literature exists debating and refining terminology (reviewed by Whiten & Ham Reference Whiten and Ham1992). In the early days of animal behavior, imitation – “learning to do an act from seeing it done” – was considered a boring low-level form of behavior. This prejudice was perhaps spurred by such English sayings as “monkey see, monkey do” or the German “nachaffen” (“after ape”) meaning “to imitate.” But accumulating evidence made clear that much apparent animal imitation is purely in the eye of the human beholder. In many circumstances where we would expect monkeys to imitate each other (e.g., learning to crack nuts with stones by watching a skilled monkey), they fail to show true imitation but rather show simpler behaviors like “stimulus enhancement” (simply observing that rocks and nuts together can lead to food). Each monkey still has to figure out, for itself, precisely how to hold and swing the stones and position the nut (Visalberghi Reference Visalberghi1987). Such studies led imitation in primates to be seen today as a sophisticated cognitive achievement (cf. Fitch et al. Reference Fitch, Huber and Bugnyar2010; Visalberghi & Fragaszy Reference Visalberghi, Fragaszy, Parker and Gibson1990; Voelkel & Huber Reference Voelkel and Huber2000).
Perhaps the biggest reason that evidence for or against monkey imitation is important is that it concerns its implications for the literature on mirror neurons, which were discovered in macaques. Mirror neurons (sometimes called “monkey see, monkey do” neurons [Carey Reference Carey1996]) are frontal neurons in macaques that fire both when the monkey performs some action and when it sees that same action performed. Such neurons appear to provide a computational substrate for motor imitation. But the catch is that – at the time of the discovery of mirror neurons – the behavioral evidence indicated that macaques do not, in fact, imitate. Although this ugly fact did not stop people from inferring that human mirror neurons play a key role in imitation, it was awkward from an evolutionary viewpoint: Just what are these mirror neurons doing in macaques, if not supporting imitation?
For mirror neuron enthusiasts, the 2006 discovery of apparent neonatal imitation in macaques was thus a great relief. Finally, it seemed, a behavioral function for macaque mirror neurons had been found, filling an otherwise uncomfortable lacuna in the theoretical edifice built upon mirror neurons. This is important, given the huge scope of explanations based on mirror neurons today, extending to speech perception, language evolution, autism research, empathy, and other major issues in cognitive neuroscience (skeptically reviewed by Hickok Reference Hickok2014). K&A's hypothesis calls such extensions sharply into question, by offering a simpler explanation of TP/R. Indeed K&A's hypothesis seems preferable to imitative hypotheses because it explains the disappearance of “imitation” during maturation that remains unexplained by the mirror neuron/imitation hypothesis.
The next and crucial step will be to design empirical tests pitting the two hypotheses against one another. I hope that researchers studying primate neonatal “imitation” and mirror neurons will rise to this challenge: The most obvious evidence in favor of K&A would come from single-unit recordings in neonatal macaques, in area F5 where mirror neurons are classically found. If such recordings find no evidence of mirror neuron involvement in the tongue protrusion response, it would be strong evidence in favor of K&A's new hypothesis.
In summary, I applaud K&A for providing a plausible alternative hypothesis for the widely accepted “neonatal imitation” interpretation of the TP/R response, and I am impressed by the breadth and depth of data that they have brought to bear in evaluating and supporting their hypothesis. Although the jury is still out, K&A provide one more reason for skepticism about neonatal imitation in general and monkey imitation in particular, as well as for circumspection about cognitive explanations that rely heavily on mirror neurons.