In the target article, Cook et al. claim that the genetic account of mirror neurons (MNs) is problematic and they propose an associativist alternative, arguing that “sensorimotor learning plays a crucial, inductive role in the development of MNs” (sect. 9.3). However, several models of sensorimotor learning for MNs have addressed neurophysiological findings (e.g., Bonaiuto & Arbib Reference Bonaiuto and Arbib2010; Bonaiuto et al. Reference Bonaiuto, Rosta and Arbib2007; Keysers & Perrett Reference Keysers and Perrett2004; Oztop & Arbib Reference Oztop and Arbib2002; Oztop et al. Reference Oztop, Kawatob and Arbib2013), so it seems the article is directed only to those who take MNs metaphorically and/or are unfamiliar with the primary literature.
It is true that any account proposing a rigid genetic fixation of MNs is incompatible with available evidence. Yet, Cook et al. fail to provide evidence that proponents of the genetic hypothesis are committed to such problematic accounts and some of their criticism seems to attack implausible “straw men” instead of existing accounts. For example, they claim that the “fact that these MNs respond maximally to unnatural stimuli – stimuli to which the evolutionary ancestors of contemporary monkeys could not possibly have been exposed – is hard to reconcile with the genetic hypothesis” (sect. 4.1, para. 3). The fact that “audiovisual” MNs are responding to “unnatural sounds” such as metal striking metal, plastic crumpling, or paper tearing associated with actions is allegedly problematic for the genetic hypothesis because no such sound/action connections existed for our distant ancestors and evolution could not have “acted on them.” The real issue is: “What would genetics specify?” Given that natural selection can only act on present conditions and pass currently beneficial traits on to the offspring, there can be a time lag between what an organism encounters in its environment and what it has been “genetically equipped” to deal with. So if our distant ancestors encountered any novel sound-action combination, a rigid mechanism that encodes only correlations as specific as those mentioned by Cook et al. would be of no use. One should reasonably expect then, that it is the ability to acquire MNs adapted to changing circumstances that is genetically specified, not what MNs code.
Cook et al. conclude that we can “get reliable information about the function of MNs only by applying an approach based on developmental history, system-level theory, and rigorous experimentation” (sect. 9.3). Given this reasonable conclusion, it is surprising that Cook et al. failed to pay adequate attention to recent work linking MNs to language acquisition and evolution (e.g., Arbib Reference Arbib2005; Reference Arbib2010; Reference Arbib2011; Arbib et al. Reference Arbib, Liebal and Pika2008; Corballis Reference Corballis2010; Corina & Knapp Reference Corina and Knapp2008; Gentilucci & Corballis Reference Gentilucci and Corballis2006; Ramachandran Reference Ramachandran2000; for a skeptical view, see Bickerton Reference Bickerton2007). Language acquisition seems to offer an excellent opportunity to gather evidence against a rigid genetic hypothesis (e.g., language acquisition accounts defended by: Chomsky Reference Chomsky1981; Reference Chomsky1995; Reference Chomsky2012; Legate & Yang Reference Legate and Yang2002; Lightfoot Reference Lightfoot1999; McGilvray Reference McGilvray and Stainton2006; Pietroski & Crain Reference Pietroski, Crain and McGilvray2005; Pinker Reference Pinker1994). According to this framework, all humans possess “some innate mental state common to the species that provides the basis for acquisition of knowledge of grammar” (Chomsky Reference Chomsky1981, p. 224). The interesting question is how the intricate details of linguistic knowledge might be genetically encoded. A healthy infant can acquire any human language. Therefore, she needs to be able to imitate both sounds that are the same as, and very different from, those her ancestors have acquired. Further, the English acquired by an infant born in 2013 differs greatly from that of an infant born in the year 848. The differences between those sounds are arguably as great as the difference between the sound of branches breaking (natural) and plastic crumbling (artificial) discussed above. It is indeed implausible that any genetically fixated mechanism could underwrite such highly flexible imitation. Such considerations lead to the abandonment of the proposal that the “innate endowment consists of a system of principles, each with certain possibilities of parametric variation” (Chomsky Reference Chomsky1981, p. 224). The minimalist program (Chomsky Reference Chomsky1995), that replaced the Principles and Parameters account, focuses on “powerful third factor effects” (Chomsky Reference Chomsky2012, p. 46), allegedly constraining language acquisition. But neither of these frameworks seems to provide a satisfactory account for language acquisition (Behme Reference Behme2014).
For this reason, alternatives have been suggested (e.g., Elman et al. Reference Elman, Bates, Johnson, Karmiloff-Smith, Parisi and Plunkett1996; MacWhinney Reference MacWhinney2004; Sampson Reference Sampson2002; Tomasello Reference Tomasello2003). Relevant here is one proposal (Arbib Reference Arbib2005; Reference Arbib2010) that involves mirror neurons. It assumes that Broca's area evolved atop an already existing mirror system for grasping with its capacity to generate and recognize a set of actions. Possibly, in language acquisition “mirror neurons for words encode recognition of the articulatory form … but must be linked to other neural networks for the encoding of meaning” (Arbib Reference Arbib2010, p. 18). It is of course implausible that genetically pre-programmed MNs could be implicated in the highly flexible imitation required by language acquisition. Instead, the evidence suggests that there are quasi-mirror neurons ready to become mirror neurons for novel actions demonstrated by others but which, prior to imitation, do not have this capacity.
Regarding the adaptive value of the language related MN-system, it has been suggested that language evolution was a gradual process that provided us step by step with brain mechanisms supporting (i) the ability to recognize performance as a set of familiar movements, (ii) complex action recognition, and (iii) mechanisms for complex imitation (Oztop et al. Reference Oztop, Kawatob and Arbib2013, p. 52). There is no a priori reason to question that such improvements in cognitive abilities could have been selected for. For an adequate evaluation of the function of MNs we need to keep in mind that they did not evolve in isolation but as part of an embodied cognitive system. The “[a]ctivity seen in [human] mirror systems involves not only mirror neurons but other cell types as well… and [s]uch activity may reflect widespread influence of prefrontal cortex and ventral pathways as well as the classic STS→IPL→IFG pathway” (Arbib Reference Arbib2010, p. 14). Under a more holistic analysis Cook et al.'s claim that “there is no positive evidence that MNs are a genetic adaptation or exaptation, or that their development has been canalized, for action understanding” (sect. 9, my emphasis) seems too strong, again missing the point that genetics may specify how neurons may learn, not what they learn.
You have
Access
In the target article, Cook et al. claim that the genetic account of mirror neurons (MNs) is problematic and they propose an associativist alternative, arguing that “sensorimotor learning plays a crucial, inductive role in the development of MNs” (sect. 9.3). However, several models of sensorimotor learning for MNs have addressed neurophysiological findings (e.g., Bonaiuto & Arbib Reference Bonaiuto and Arbib2010; Bonaiuto et al. Reference Bonaiuto, Rosta and Arbib2007; Keysers & Perrett Reference Keysers and Perrett2004; Oztop & Arbib Reference Oztop and Arbib2002; Oztop et al. Reference Oztop, Kawatob and Arbib2013), so it seems the article is directed only to those who take MNs metaphorically and/or are unfamiliar with the primary literature.
It is true that any account proposing a rigid genetic fixation of MNs is incompatible with available evidence. Yet, Cook et al. fail to provide evidence that proponents of the genetic hypothesis are committed to such problematic accounts and some of their criticism seems to attack implausible “straw men” instead of existing accounts. For example, they claim that the “fact that these MNs respond maximally to unnatural stimuli – stimuli to which the evolutionary ancestors of contemporary monkeys could not possibly have been exposed – is hard to reconcile with the genetic hypothesis” (sect. 4.1, para. 3). The fact that “audiovisual” MNs are responding to “unnatural sounds” such as metal striking metal, plastic crumpling, or paper tearing associated with actions is allegedly problematic for the genetic hypothesis because no such sound/action connections existed for our distant ancestors and evolution could not have “acted on them.” The real issue is: “What would genetics specify?” Given that natural selection can only act on present conditions and pass currently beneficial traits on to the offspring, there can be a time lag between what an organism encounters in its environment and what it has been “genetically equipped” to deal with. So if our distant ancestors encountered any novel sound-action combination, a rigid mechanism that encodes only correlations as specific as those mentioned by Cook et al. would be of no use. One should reasonably expect then, that it is the ability to acquire MNs adapted to changing circumstances that is genetically specified, not what MNs code.
Cook et al. conclude that we can “get reliable information about the function of MNs only by applying an approach based on developmental history, system-level theory, and rigorous experimentation” (sect. 9.3). Given this reasonable conclusion, it is surprising that Cook et al. failed to pay adequate attention to recent work linking MNs to language acquisition and evolution (e.g., Arbib Reference Arbib2005; Reference Arbib2010; Reference Arbib2011; Arbib et al. Reference Arbib, Liebal and Pika2008; Corballis Reference Corballis2010; Corina & Knapp Reference Corina and Knapp2008; Gentilucci & Corballis Reference Gentilucci and Corballis2006; Ramachandran Reference Ramachandran2000; for a skeptical view, see Bickerton Reference Bickerton2007). Language acquisition seems to offer an excellent opportunity to gather evidence against a rigid genetic hypothesis (e.g., language acquisition accounts defended by: Chomsky Reference Chomsky1981; Reference Chomsky1995; Reference Chomsky2012; Legate & Yang Reference Legate and Yang2002; Lightfoot Reference Lightfoot1999; McGilvray Reference McGilvray and Stainton2006; Pietroski & Crain Reference Pietroski, Crain and McGilvray2005; Pinker Reference Pinker1994). According to this framework, all humans possess “some innate mental state common to the species that provides the basis for acquisition of knowledge of grammar” (Chomsky Reference Chomsky1981, p. 224). The interesting question is how the intricate details of linguistic knowledge might be genetically encoded. A healthy infant can acquire any human language. Therefore, she needs to be able to imitate both sounds that are the same as, and very different from, those her ancestors have acquired. Further, the English acquired by an infant born in 2013 differs greatly from that of an infant born in the year 848. The differences between those sounds are arguably as great as the difference between the sound of branches breaking (natural) and plastic crumbling (artificial) discussed above. It is indeed implausible that any genetically fixated mechanism could underwrite such highly flexible imitation. Such considerations lead to the abandonment of the proposal that the “innate endowment consists of a system of principles, each with certain possibilities of parametric variation” (Chomsky Reference Chomsky1981, p. 224). The minimalist program (Chomsky Reference Chomsky1995), that replaced the Principles and Parameters account, focuses on “powerful third factor effects” (Chomsky Reference Chomsky2012, p. 46), allegedly constraining language acquisition. But neither of these frameworks seems to provide a satisfactory account for language acquisition (Behme Reference Behme2014).
For this reason, alternatives have been suggested (e.g., Elman et al. Reference Elman, Bates, Johnson, Karmiloff-Smith, Parisi and Plunkett1996; MacWhinney Reference MacWhinney2004; Sampson Reference Sampson2002; Tomasello Reference Tomasello2003). Relevant here is one proposal (Arbib Reference Arbib2005; Reference Arbib2010) that involves mirror neurons. It assumes that Broca's area evolved atop an already existing mirror system for grasping with its capacity to generate and recognize a set of actions. Possibly, in language acquisition “mirror neurons for words encode recognition of the articulatory form … but must be linked to other neural networks for the encoding of meaning” (Arbib Reference Arbib2010, p. 18). It is of course implausible that genetically pre-programmed MNs could be implicated in the highly flexible imitation required by language acquisition. Instead, the evidence suggests that there are quasi-mirror neurons ready to become mirror neurons for novel actions demonstrated by others but which, prior to imitation, do not have this capacity.
Regarding the adaptive value of the language related MN-system, it has been suggested that language evolution was a gradual process that provided us step by step with brain mechanisms supporting (i) the ability to recognize performance as a set of familiar movements, (ii) complex action recognition, and (iii) mechanisms for complex imitation (Oztop et al. Reference Oztop, Kawatob and Arbib2013, p. 52). There is no a priori reason to question that such improvements in cognitive abilities could have been selected for. For an adequate evaluation of the function of MNs we need to keep in mind that they did not evolve in isolation but as part of an embodied cognitive system. The “[a]ctivity seen in [human] mirror systems involves not only mirror neurons but other cell types as well… and [s]uch activity may reflect widespread influence of prefrontal cortex and ventral pathways as well as the classic STS→IPL→IFG pathway” (Arbib Reference Arbib2010, p. 14). Under a more holistic analysis Cook et al.'s claim that “there is no positive evidence that MNs are a genetic adaptation or exaptation, or that their development has been canalized, for action understanding” (sect. 9, my emphasis) seems too strong, again missing the point that genetics may specify how neurons may learn, not what they learn.