The “second-person neuroscience” proposed by Schilbach et al. in the target article proffers the intriguing idea that social cognition during real-time interactions with another individual may be fundamentally different from passive observations of another's actions. Understanding the contribution of neural processes to ongoing interactions with complex beings is a fascinating research direction, with potential implications for the treatment of disorders attended by social deficits, as well as for ethics and public policy.

Several decades of neuroscientific research have sketched out the neural circuits that may translate perceptual information about other individuals into purposeful action. Specifically, regions of the human and nonhuman primate brain including the superior temporal sulcus and fusiform face area contribute to social identification (Tsao et al. Reference Tsao, Moeller and Freiwald2008). The ventromedial prefrontal cortex, orbitofrontal cortex, and striatum appear to play a role in translating knowledge of others into motivational signals (Azzi et al. Reference Azzi, Sirigu and Duhamel2012; Burke et al. Reference Burke, Tobler, Schultz and Baddeley2010; Cooper et al. Reference Cooper, Kreps, Wiebe, Pirkl and Knutson2010). The anterior cingulate cortex and fronto-insular cortex contribute to empathy and other-regarding cognition (Chang et al. Reference Chang, Gariépy and Platt2013; Decety Reference Decety2010; Gu et al. Reference Gu, Liu, Guise, Naidich, Hof and Fan2010). The so-called mentalizing and mirroring networks appear to participate in action and intention understanding (Becchio et al. Reference Becchio, Cavallo, Begliomini, Sartori, Feltrin and Castiello2012; Rizzolatti & Sinigaglia Reference Rizzolatti and Sinigaglia2010). Circuits connecting these areas could translate social perceptual information into appropriate actions via decision-making mechanisms (Baumgartner et al. Reference Baumgartner, Fischbacher, Feierabend, Lutz and Fehr2009, Reference Baumgartner, Knoch, Hotz, Eisenegger and Fehr2011; Knoch et al. Reference Knoch, Schneider, Schunk, Hohmann and Fehr2009).

To better understand the neural mechanisms underlying social cognition, we propose that social neuroscience needs to ground its predictions and hypotheses in a formal framework such as that provided by behavioral game theory (Dorris & Glimcher Reference Dorris and Glimcher2004; Gintis Reference Gintis2009; Kosfeld et al. Reference Kosfeld, Heinrichs, Zak, Fischbacher and Fehr2005; Lee Reference Lee2008; Platt & Glimcher Reference Platt and Glimcher1999; Tomlin et al. Reference Tomlin, Kayali, King-Casas, Anen, Camerer, Quartz and Montague2006). Schilbach et al. criticize game theoretical approaches for not recreating the dynamics of everyday real-life social encounters, but this common opposition has been rebutted before (Gintis Reference Gintis2009). Game theoretical frameworks are general and open, allowing formal delineation of specific hypotheses while not imposing restrictions on the behaviors that are being described. Formal approaches borrowed from economics, game theory, and behavioral ecology have been extremely useful in describing decisions in dynamic foraging or social environments (Chang et al. Reference Chang, Winecoff and Platt2011; Hayden et al. Reference Hayden, Pearson and Platt2011; Lee Reference Lee2008; Sugrue et al. Reference Sugrue, Corrado and Newsome2004).

These approaches can be extended to describe the dynamics of interacting individuals, with several advantages. First, they allow us to generate empirically testable and mathematically formalizable predictions about the neural mechanisms that could underlie decisions in complex social environments. Second, they allow for comparative analyses of decision processes in humans and other animals with respect to the demands placed on them in specific physical and social environments (Heilbronner et al. Reference Heilbronner, Rosati, Stevens, Hare and Hauser2008; Kacelnik & Bateson Reference Kacelnik and Bateson1996; Stephens et al. Reference Stephens, McLinn and Stevens2002).

Schilbach et al. also raise the concern that classical game theory paradigms involve mainly one-shot interactions or turn-taking. Although this structure is often used for simplicity, we contend that continuous interactions in interactive games can also be effectively described using a similar theoretical framework (Braun et al. Reference Braun, Ortega and Wolpert2009; Debreu Reference Debreu1952). Such mathematical tools would help translate some of the intuitive aspects of Schilbach et al.'s approach into concrete experimental predictions.

Second-person neuroscience would also benefit from broadening its inquiry to the interactions of nonhuman animals (Chang et al. Reference Chang, Winecoff and Platt2011; Fujii et al. Reference Fujii, Hihara and Iriki2007; Washburn et al. Reference Washburn, Hopkins and Rumbaugh1990). Social complexity appears to have favored the evolution of higher social cognition in animals that have brains similar to ours, like macaques (Azzi et al. Reference Azzi, Sirigu and Duhamel2012; Barsalou, Reference Barsalou2005; Chang et al. Reference Chang, Barter, Ebitz, Watson and Platt2012; Rudebeck et al. Reference Rudebeck, Buckley, Walton and Rushworth2006; Tsao et al. Reference Tsao, Moeller and Freiwald2008) and in animals that have very different brains as well, like scrub jays and rooks (Bird & Emery Reference Bird and Emery2010; Emery & Clayton Reference Emery and Clayton2001). We know from research in macaques, sheep, and mice that social cognition in mammals appears to rely on neural circuits that are similar, and perhaps homologous, to those in humans (Azzi et al. Reference Azzi, Sirigu and Duhamel2012; Barsalou, Reference Barsalou2005; Jeon et al. Reference Jeon, Kim, Chetana, Jo, Ruley, Lin, Rabah, Kinet and Shin2010; Rudebeck et al. Reference Rudebeck, Buckley, Walton and Rushworth2006; Sanchez-Andrade & Kendrick Reference Sanchez-Andrade and Kendrick2009; Tsao et al. Reference Tsao, Moeller and Freiwald2008). One possible explanation is that we inherited those circuits from a common ancestor that possessed some level of social complexity. Alternatively, similar constraints applying to neural circuits could also have caused them to evolve in similar ways to support similar functions. How such functions are accomplished by neural circuits in animals with brains that are very different from our own – such as birds – remains an open question.

We agree with Schilbach et al. that studying the neural processes mediating live interaction between real agents is crucial for the maturation of social neuroscience as a discipline. What we propose is to supplement this approach with formal game theory and value-based analysis of preferences in humans and nonhuman animals. In our lab, for example, we study pairs of monkeys interacting both in economical and interactive games (Chang et al. Reference Chang, Winecoff and Platt2011; Chang et al. Reference Chang, Gariépy and Platt2013). Estimating preferences allows us to quantify how much monkeys value certain options (e.g., giving juice to another monkey). Game theory will allow us to generate predictions of the equilibriums that could develop over time between two interacting individuals (see Braun et al. Reference Braun, Ortega and Wolpert2009). Understanding the neural processes that underlie social cognition in such animals could powerfully inform our understanding of the evolutionary origins of our own social abilities.