We agree with Schilbach et al.'s suggestion that investigation of social cognition is too often restricted to passive observation, and social cognition is entirely different when one is interacting and emotionally engaged with others. We applaud their proposal for further investigation into second-person information but suggest their outlook for future research is limited. We encourage investigation beyond the traditional parameters of social knowing. We suggest less emphasis on social “cognition” and more on a direct interaction between systems responsible for immediate social perception, affective processing and action. We propose a shift in focus, away from corticocentricity – a hierarchical interpretation of the brain with the cortex as the centre for “higher cognition,” responsible for top-down monitoring of subcortical structures (Parvizi Reference Parvizi2009) – and toward social understanding as a dynamic system supported by bottom-up processes. Recent theories suggest that behavior is modulated by affective and conative processing of physical and social environmental stimuli (Berridge Reference Berridge2003; Panksepp Reference Panksepp1998, Siviy & Panksepp Reference Siviy and Panksepp2011).

Due to the homology of affective systems among mammals, foundations of the human social mind can be investigated by studying subcortical systems, as done with other animals. In doing so, a biological component is added to social behavior without a restrictive emphasis on higher order processes. If adopting a functional and evolutionary framework, then canine social play is an ideal behavioral phenotype to consider for investigations into social knowing. Canine social play is a dynamic social interaction involving emotionally engaged conspecifics, modulated by subcortical processes.

Schilbach et al. identify three aspects of social interaction that are important for research in social neuroscience. Each aspect is evident in canine social play. First, a social encounter involves two players – an initiator and a responder. During the social play routine one dog elicits a response from another. The responder perceives and responds to socially relevant cues. The play bow is one cue that signals the invitation to play plus the intention to maintain play. The encounter requires frequent reappraisals; if one partner becomes too aggressive, play stops and that partner must modify their behavior in order for play to resume. The bow can signal reappraisal of a preceding action, and reinforce the intention to play (Bekoff Reference Bekoff2008; Burghardt Reference Burghardt2005). Consistent modification of approach and response relies on rapid recognition of the state of the other and flexibility in action output.

The playful interaction is dynamic, rich in anticipation, preparation, and revision, consisting of reciprocities based on perception of socially relevant information prompting a response. In keeping with Schilbach et al.'s description of a social interaction, reciprocities that occur during a play encounter may be sequential, but not predictable. Social play can be considered a system that evolves in response to the environment. It is a stable system as there may be a repertoire of behaviors, and unstable because the expression of the repertoire is nonlinear – it is responsive, and dynamic (Fentress & Gadbois Reference Fentress, Gadbois and Blass2001; Fentress & McLeod Reference Fentress, McLeod and Blass1986; Smith & Thelen, Reference Smith and Thelen2003; Thelen & Smith Reference Thelen and Smith1994).

Second, the play interaction demands cooperation. Play is a voluntary encounter built on shared intentions and negotiations to initiate and maintain play at nonaggressive levels (Bekoff Reference Bekoff2004; Reference Bekoff2008). In a cooperative manner, threats of aggression are absent when large partners self-handicap and dominants exhibit role reversals (Bekoff Reference Bekoff2004; Burghardt Reference Burghardt2005). Critical to the interaction is the perception of sensory cues that indicate more than agency. It involves recognition of the state of the other, intentions, reconciliation, and consolation. We suggest social play is indicative of empathic responding between conspecifics. It may involve a same or similar mechanism that supports post-conflict affiliative (conciliatory) behavior (de Waal Reference de Waal2008; Palagi & Cordoni Reference Palagi and Cordoni2009).

Third, play emerges in the young without training and continues throughout the life span, even after maturation of the motivational system for aggression and sexual maturation. Because play is observed in the wild among wolves and feral dogs, it is not considered an artifact of domestication (Bauer & Smuts Reference Bauer and Smuts2007). Play may be an intrinsic function supported by a play-specific motivational system (Panksepp Reference Panksepp1998). Social isolates show high levels of play-seeking behavior (Burghardt Reference Burghardt2005), higher than controls (Panksepp Reference Panksepp1998). Decorticates show normal levels of play solicitation (Pellis et al. Reference Pellis, Pellis and Whitshaw1992), however, if damage extends to the limbic structures, then play declines (Panksepp et al. Reference Panksepp, Normansell, Cox and Siviy1994). Lack of motivation to play has been associated with maladaptive forms of behavior such as depression (Burgdorf et al. Reference Burgdorf, Panksepp and Moskal2011). The suggested benefits of play include recognition of social hierarchy, development of motor coordination, and the acquisition and refinement of social skills (Bekoff Reference Bekoff2008; Miklosi Reference Miklosi2008; Panksepp Reference Panksepp2011). Moreover, play feels good and is associated with a release of endorphins. We concur with the authors' suggestion to consider the influence of reward-related processes on action control due to the hedonic experience of social stimuli.

Play is a source of fundamental pleasure and play as reward is a strong incentive. Research supports the idea of a subcortical locus of control for play, with stimuli predicting play activating the mesolimbic dopaminergic pathway (Panksepp Reference Panksepp1998; Siviy Reference Siviy, Bekoff and Beyers1998). Neurotransmitter release in a proposed PLAY circuit may modulate the playful interaction with noradrenergic and opiod pathways influencing attention and pleasure respectively (Panksepp Reference Panksepp1998; Reference Panksepp2011; Siviy Reference Siviy, Bekoff and Beyers1998; Siviy & Panksepp Reference Siviy and Panksepp2011). Rats emit ultrasonic vocalizations (50–55 kHz range) during play and when anticipating play, and the mesolimbic dopaminergic system controls production of these vocalizations (Burgdorf et al. Reference Burgdorf, Wood, Kroes, Moskal and Panksepp2007).

Fundamental pleasure may be supported by a three-part subcortical reward system based on sensory pleasure (liking), incentive motivation in response to cues (wanting) and associations between cues and pleasure to follow (learning) (Kringelbach & Berridge Reference Kringelbach and Berridge2009). Berridge and colleagues identified distinct neural representations of each component of this system for a common reward. Opioid stimulation in the nucleus accumbens increased both liking and wanting signals represented by firing in the ventral pallidum. Dopamine stimulation increased wanting signals and behaviors, but not liking and wanting. Neither affected neurochemical modulation of Pavlovian signals (Smith et al. Reference Smith, Berridge and Aldridge2011). Separate neural firing may indicate that motivation and liking components of the system can be modulated independent of learning supported by cognitive representation. Fundamental pleasure experienced during social interaction with conspecifics and supported by subcortical structures may not require cortical interfacing for higher order processing. As Kringelbach and Berridge (Reference Kringelbach and Berridge2009) suggest, wanting matched with liking may “facilitate engagement with the world.”

Schilbach et al. suggest the need for a naturalistic social interaction for investigation of the neural mechanisms supporting social understanding, yet their examples involve laboratory-controlled interaction with virtual characters and an emphasis on higher order cognitions. A study of second-person neuroscience could benefit from an examination of bottom-up mechanisms (i.e., affective and conative processes) that have been well demonstrated in other species and contexts (e.g., canine play behavior).