In the target article, Cook et al. suggest that the evolutionary origins and maintenance of the mirror neuron system (MNS) lie in “domain-general processes of associative learning in the course of individual development, and, although they may have psychological functions, their psychological functions do not necessarily have a specific evolutionary purpose or adaptive function” (target article, Abstract). We agree that the excitement surrounding the discovery of mirror neurons (MNs) has led to an inordinate focus on their role in social-cognitive functions and how these functions might play a role in the evolution of the MNS. However, we strongly disagree with the authors' claims that the known social-cognitive roles the MNS plays in primate cognitive and behavioral functioning have not and do not affect the MNS in an evolutionary context and that the associative account “separates questions about their origin and function” (sect. 1, para. 4).

The target article describes a lower-level biological mechanism (associative learning) that, as Cook et al. argue, fully accounts for the phylogenetic and ontogenetic development of mirror neurons. We assert that the initial evolution, further evolution, and evolutionary maintenance of the MNS is likely jointly influenced by such lower-level biological mechanisms and by the well-documented role that the MNS plays in social-cognitive functions. One example of this joint influence can be observed in individuals with an intact versus an impaired MNS who are more able to attract reproductive partners, reproduce, and protect and provide for their offspring within the complex social structures of primate societies (e.g., Howlin & Moss Reference Howlin and Moss2012).

We agree that associative learning is likely a critical mechanism for both the development and the evolution of mirror neurons. However, given that associative learning mechanisms exist in species that do not have a MNS, some alternative mechanism must interact with associative learning in order to produce the evolutionary pressure required for the origin and maintenance of the MNS in humans. To avoid directly addressing the evolutionary advantages the social-cognitive functions of the MNS confer, Cook et al. use a “straw man” argument. They attack the most extreme proposal of the role of social-cognitive functions in the evolution of the MNS – evolutionary selection via action understanding. The “associative learning in vivo” and “evolutionary selection based upon action understanding” accounts represent polar extremes, both of which are unlikely to reflect reality. Simultaneously, however, the adaptive advantages of the social-cognitive capacities (e.g., action perception, processing, and prediction) ascribed to the MNS enhance individuals' reproductive fitness, creating precisely the evolutionary pressure that the authors propose has not, and does not, exist.

Cook et al.'s depiction of the role of developmental research in elucidating biological/genetic versus environmental/learning influences on the MNS is concerning. We agree that evidence for neonatal imitation is limited and, even if it is present, is unlikely to be driven by MNS mechanisms since cortical regions that contain MNs are not fully developed at birth. However, the postnatal developmental timeline of the MNS neither rules out genetic/biological and evolutionary processes nor demonstrates the role of associative learning. It is well known that frontal and association cortices that house MNs undergo striking synaptic development and myelination between 8-months and 3-years of age (Huttenlocher Reference Huttenlocher2002; Imada et al. Reference Imada, Zhang, Cheour, Taulu, Ahonen and Kuhl2006; Locke et al. Reference Locke, Bekken, McMinn-Larson and Wein1995). Developmental EEG evidence similarly indicates protracted cortical development in these regions (Hagne Reference Hagne1968; Southgate et al. Reference Southgate, Johnson, Osborne and Csibra2009), with continuing maturation until late childhood or adolescence (Martineau & Cochin Reference Martineau and Cochin2003). Therefore, biological factors may explain protracted MNS development.

Cook et al. also dismiss EEG mu suppression as an index of MNS functioning too quickly. The strong relationship between the mu rhythm and action observation/execution can be traced back to 1954, when Gastaut and Bert reported that the mu rhythm was consistently reduced when stationary subjects “identified themselves with an active person represented on a screen” (see also Pineda Reference Pineda2005). We also recently published a re-analysis of pooled data from four published studies, including a total of 66 individuals with autism spectrum disorders (ASD), demonstrating that, across the age-span from 6–17 years, there was significantly less mu suppression in individuals with ASD compared with matched controls during action observation, but not during self-movement (Oberman et al. Reference Oberman, McCleery, Hubbard, Bernier, Wiersema, Raymaekers and Pineda2013). Although source estimation indicates that the generator of the mu rhythm is in the postcentral gyrus rather than premotor or primary motor cortex (Hari & Salmelin Reference Hari and Salmelin1997), the possible downstream modulation of motor cortex by the MNS is tangential to their mirror properties. Cook et al. also ignore recent studies showing that the same stimuli that elicit mu suppression also activate MN regions (as indicated by BOLD response; Perry & Bentin Reference Perry and Bentin2009) and modulate a TMS-induced motor evoked potential (Lepage et al. Reference Lepage, Saint-Amour and Théoret2008), suggesting that all three indices are likely capturing the same underlying cortical mechanism.

In summary, we argue, contrary to Cook et al., that the origins and evolution of mirror neurons are unlikely to be driven by associative learning alone, but, rather, to be a consequence of a combination of evolutionary, biological, developmental, social-cognitive, and experience-based influences. Indeed, we speculate that the MNS is not functionally fixed, but rather a currently evolving, flexible, semi-modular neural network that interacts with multiple other neural systems, including the motor and social-motivation systems (Oberman et al. Reference Oberman, Ramachandran and Pineda2008). The functioning of such a system at any point in development should be viewed as a snapshot of a dynamic system that is constantly modulated by these influences and interactions with other systems (Johnson Reference Johnson2011; Johnson et al. Reference Johnson, Halit, Grice and Karmiloff-Smith2002). Environmental and biological influences unfold simultaneously and interactively, not separately and sequentially, and their relative roles can only be disentangled with careful measurement and calculation (Dobkins et al. Reference Dobkins, Bosworth and McCleery2009; Smit et al. Reference Smit, Boomsma, Schnack, Hulshoff Pol and de Geus2012). Cook and colleagues attack theories that argue for the evolution of the MNS based upon its proposed role in action understanding (Rizzolatti & Fadiga Reference Rizzolatti and Fadiga1998; Rizzolatti et al. Reference Rizzolatti, Fadiga, Gallese and Fogassi1996), but we believe that the theory proposed by Cook et al. arguing that associative learning mechanisms alone can account for the origins and development of the MNS is equally as unlikely. Both models ignore the reciprocal relationships between evolutionarily adaptive psychological mechanisms and the lower-level biological mechanisms that are required for their existence.