In the target paper, Dunbar presents data supporting his Social Brain Hypothesis, where an increase in the size of primate groups is associated with structural, behavioral, and cognitive adaptations to social stress. He identifies endorphin-releasing social behaviors (primarily grooming) as the main stress-mitigating mechanism. The pressure of social stress is presented as the driving force behind the evolution of social and metacognitive capacities of the primate brain toward the ability for social mentalizing. This ability is considered crucial for living in large groups.

In this commentary, I want to draw attention to a complementary perspective on the evolution of metacognition. Mitigating stress by prosocial endorphin-releasing behaviors has its counterpart, which is stress tolerance via self-mentalizing. Self-mentalizing may play a role in mitigating social stress alongside with social mentalizing. Hence, metacognition can be seen as co-evolving with the social brain and not only as its emergent property. In this view, the primary function of metacognition is to (indirectly) regulate allostasis, with social mentalizing as an add-on in highly socialized species.

To make this case, I first frame metacognition as self-regulation or self-modeling, since to regulate an extrinsic process means to be its model according to the good regulator theorem (Conant & Ross Ashby, Reference Conant and Ross Ashby1970). Thus, we can see the origin of metacognition in self-modelling, whose primitive forms may be considered as pro-metacognition. Pro-metacognition is already found in bacteria, the simplest biosystem. E. Coli can monitor its level of energy and membrane integrity and transduce this information to its effector protein which controls the bacterium’s behavior (Alexandre, Greer-Phillips, & Zhulin, Reference Alexandre, Greer-Phillips and Zhulin2004). At a higher level of complexity, eucaryotic cells developed the system of Bcl-2 protein family that integrates information about the cumulative level of stress and growth factors to compute the cell’s state of ‘well-being’ (Amaral et al., Reference Amaral, Xavier, Steer and Rodrigues2010). As domain-general models, pro-metacognitive systems are at the top of the information-processing hierarchy and contextualize the domain-specific information in order to model and track the ‘state of the organism.’

The function of pro-metacognition is to keep the organism close to its homeostatic setpoints and to effect the allostatic transition to new setpoints to meet environmental demands. For example, when environmental conditions worsen, cells may go into the states of sporulation or dormancy (Lewis, Reference Lewis2000). This way, the organism can anticipate its future needs and take adaptive actions. An important notion here is the window of stress tolerance, where self-regulatory mechanisms can compensate for deviations from the setpoints without a state transition, such as the aforementioned sporulation or dormancy, so that the organism returns to the initial state of homeostasis. In this view, (brain) metacognition can be seen as serving the same function, that is, integrating/mentalizing domain-specific intero- and exteroceptive information to monitor and model the dynamics of the organism’s well-being and to predict the necessary autonomic and external actions to either maintain its homeostasis or effect an allostatic transition. Social signals are one of those specific domains.

In the human brain, the default mode network (DMN) is often associated with both social and metacognition (Andrews-Hanna, Smallwood, & Spreng, Reference Andrews-Hanna, Smallwood and Spreng2014). Therefore, it seems plausible that DMN initially evolved as a higher-order domain-general network to model the organism’s window of stress tolerance and thus regulate its stress response (analogous to the Bcl-2 protein network), and that in social species, it integrated social cognition. If so, we would expect a different connectivity in DMN-homologous networks of solitary vs social animals. A related hypothesis is that deficiencies in DMN (or DMN-homologs) connectivity with networks specializing in social cognition will leave DMN’s socially-independent functions largely intact.

The hypothesis that the primary DMN’s function is to integrate multimodal sensory information for anticipating and responding to environmental change was suggested a decade ago (Lu et al., Reference Lu, Zou, Gu, Raichle, Stein and Yang2012). It was spurred by the discovery of a DMN homolog in the rat brain. Rats, however, are social animals and may not be a good comparison model to tease DMN’s social from intrinsic metacognitive functions. The primate DMN is functionally connected with such social cognition networks as the mirror neurons system (Molnar-Szakacs & Uddin, Reference Molnar-Szakacs and Uddin2013) and the face processing circuitry (Schwiedrzik et al., Reference Schwiedrzik, Zarco, Everling and Freiwald2015), and has an extensive intrinsic connectivity involved in theory of mind (Spreng & Andrews-Hanna, Reference Spreng and Andrews-Hanna2015). Comparative functional neuroimaging in other mammals may identify similar connectivity patterns in DMN homologs and provide an opportunity for comparison between social and solitary species from close taxa. Examples of such a comparison include rats vs hamsters, wolves vs foxes, chimpanzees vs orangutans, among others. Such studies can identify the ‘minimal’ DMN homolog concerned with only (or mostly) self-modeling metacognition.

Certain psychiatric conditions selectively affect social functioning. Perhaps, the most notable is autism spectrum disorder. There is ample evidence for functional, structural, and developmental abnormalities in the DMN of people with autism (Padmanabhan et al., Reference Padmanabhan, Lynch, Schaer and Menon2017). Also, an aberrant functional connectivity between DMN and the salience network is found in social anxiety disorder (Lucherini Angeletti et al., Reference Lucherini Angeletti, Scalabrini, Ricca and Northoff2021). Nevertheless, individuals with autism or social anxiety can be high-functioning and successful outside social settings.

To summarize, it appears that social and self-metacognition are to some extent dissociable, with self-metacognition likely being the primary higher-order mental capacity in the evolution of the mammalian brain. Further comparative studies can provide confirmation or disconfirmation of this notion.