Reason is Social
The unique human ability to think probably evolved so that we can socialize effectively.
The most recent generation of evolutionary anthropology research has brought us closer than ever before in understanding the origins of human cognition. Unlike any other animal, humans are capable of causal reasoning, language, technical innovation, and analysis of abstract concepts—instances of the unique overarching capability of “reason.” However, there have emerged two general categories of narratives of human cognitive development that, arguably, are incompatible. The first category posits that activities involving the material world enabled and guided the development of reason. For example, when a rock is used to remove a sharp flake from a piece of flint, the advanced cognizing first occurs in the environment, externally to the human body. Engagement with non-human environmental features thus provides the foundation for reason (Osvath & Gardenfors, 2005; Malafouris, 2020; Osiurak, 2020), and may be considered the “technology-first” hypothesis. The alternative standpoint, whose proponents have pioneered research in “shared intentionality” and “theory of mind” psychology, implies that reasoning is a socially-derived capability. Referred to by various researchers as the “interactionist” viewpoint, “thinking through other minds,” or the “Vygotskian intelligence hypothesis,” reason evolved directly from human interactions with each other rather than individual interactions with the environment (Moll & Tomasello, 2005; Tomasello et al., 2007; Veissiere et al., 2020).
Anthropologists commonly refer to human advancement from an evolutionary point of view as “cumulative cultural evolution” (Mesoudi & Thornton, 2018). Continuous generations of humans created a cumulative, or “ratcheting,” effect, in which previous developments can be taken for granted and compounded upon in subsequent generations. This posit, however, reflects a “chicken-and-egg” type of debate as to whether innovation preceded culture, or culture preceded innovation. The technology-first hypothesis suggests that individual hominin interactions with their environments form the origin of this cumulative effect, which later guided and was bolstered by communicative capabilities. If the capacity for engaging rationally with the environment (using reason) on an individual scale originated before social cognition, cumulative cultural evolution may have simply abetted the prevalence and growth of reason in humans (Osiurak, 2020). The alternate perspective is that reason evolved through the process of optimizing interpersonal interactions. A growing consensus is suggesting that the development of social reasoning led to the individual capacity for reason, which reinforced cumulative cultural evolution into today’s advanced technologies and institutions.
While sparse evidence spanning millions of years makes it difficult to draw accurate conclusions, an interdisciplinary approach brings to light critical perspectives that together paint the most resolute picture of the evolution of reason. This paper will comprehensively indicate, by synthesizing evidence from archaeology, anthropology, psychology, education, and philosophy of logic, that the capacity for reason primarily evolved through hominin social interactions.
Evolutionary advantages for hominin collaboration
Among the first notable divergences of hominins from the other Great Apes was the transition to bipedalism. While walking on two legs is not unique to humans (think of birds or kangaroos), there is no other evidence of primate bipedalism. Hominins who walked on two legs—a transition which was likely sudden—likely had significant behavioral changes. Relatively quickly, they began to traverse longer distances, gather and store larger quantities of food (DeSilva, 2021). The transformation to a bipedal species created a completely different set of tasks, capabilities, and necessities—many of which may have led, directly or indirectly, to an evolutionary mechanism selecting for social skills. While determining these individual behaviors is highly speculative, it provides a broad scope of possibilities that may be related to each other or to the actual behaviors of early hominins.
Searching. Hominins began to to traverse longer distances than previously possible while foraging for plant-based food sources. Searching is a social coordination problem that strongly rewards cooperative mechanisms (Kuroda & Kameda, 2019). Say, in the process of foraging, that Bob finds a wonderful plum tree—but his friend Joe is 400 feet away, still searching for lunch. Bob might have shouted to Joe to come over and gather fruits for their family or community. Hominins splitting up for more efficient searching would also become more vulnerable to predators—establishing communication mechanisms for collective risk monitoring may have been crucial for the process of foraging.
Birth Assistance. The transition to bipedalism brought forth various difficulties, one of which was a significant change in the practice of giving birth. Pelvic adaptations to walking upright altered the shape of the birth canal, necessitating the fetus to be rotated in the process of birth. This prompts a need for a midwife or some type of assistant in the process of birth—an interpersonal activity completely unique to hominins. Indeed, as Martin Packer and Michael Cole have indicated, social customs like birth assistance or healing (detailed below) may have formed into larger institutions, which supported the development of collective “deontic” niches of hominin obligation to others (Packer & Cole, 2019).
Healing. As bipedal hominins began to travel increasingly long distances, the health of each leg was crucial for the ability of a hominin to keep up with the group—while extremely long distances can be traversed on two legs, the human stamina for one-legged hopping is minimal; thus, if one ankle is sprained, or one ACL is torn, recovery is crucial. In contrast, species that thrive on four legs can much more easily transport themselves when a single leg is incapacitated. This suggests that interpersonal injury support systems evolved early in the hominin lineage. Indeed, archeological evidence of healed leg and ankle injuries from the earliest species of bipedal hominins supports this claim (DeSilva, 2021).
These transitions in the ancestral hominin environment necessitated the use of unique social skills, providing strong evidence that these advanced social skills preceded the broader development of reason. Each of these social situations, immediately prompted by bipedalism, preceded tool use by millions of years—supporting the interactionist viewpoint. Yet how might they have evolved into those other cognitive abilities, from which diverse cultural practices, impactful technologies, and social institutions surely emerged?
Shared intentionality precedes reason in ontogeny
Psychological research indicates that infants develop a perception of others’ interests before they have the generalized representational capabilities that may be prerequisite to the ability to reason. ****Tomasello’s landmark 2005 investigation into shared intentionality provides the backbone for the claim that social cognition is a foundational origin for advanced human cognition (Tomasello, 2005). Tomasello ties together the relevant empirical evidence indicating that non-human primates fall short in capacities for shared intentionality. Shared intentionality forms the basis for collective intentionality, from which culture emerges. As elaborated upon below, only after cultural norms are established do individuals begin using reason to evaluate their own thought processes.
Moll & Tomasello’s 2007 paper continues this reasoning with the case that “the unique aspects of human cognition were driven by, or even constituted by, social cooperation” (Moll & Tomasello, 2007). It hints at distinctions between the three categories of symbols, artifacts, and other humans as primary entities upon which reasoning is constructed, indicating that symbolic representations necessary for collaboration later facilitated development of culture and tools: “unique aspects of human cognition—the cognitive skills needed to create complex technologies, cultural institutions and systems of symbols, for example—were driven by, or even constituted by, social cooperation” (Moll & Tomasello, 2007, p. 1). The earliest instances of uniquely-advanced intuitions appearing in human ontogeny occur in the interactions that human infants have with their parents, which lead to more robust joint intentionality, in which individual perspectives in a “joint intentional frame” can then be swapped and altered. This capacity specifically develops through triadic engagement, in which both the infant and the infant’s parent are jointly focused on a third entity—such as a physical object, an event, or a goal.
Extensive child development research confirms Tomasello’s concept of shared intentionality in ontogeny. In a modern analysis of Piaget’s developmental stages, the triadic nature of conceptual reasoning is emphasized: infants only begin to use reasoning in social contexts (Muller & Carpendale, 2000). Young students learn to reason most effectively in the classroom in a variety of contexts through “exploratory talk” rather than studying independently, which only occurs after social norms are established (Mercer et al., 1999). This reflects the large body of research supporting social cognition and social learning theory as a whole, which suggests that students learn to effectively use reason through observing others in social contexts (Bandura et al., 1961, Csibra & Gergely, 2006, Frith & Frith, 2011). If significant portions of human reasoning capabilities are learned through interactions with others, rather than intrinsically developed in ontogeny, reason as a whole likely rests upon foundations for collaborative engagement. Collective intentionality, consisting of scaled-up joint intentional frames, may have been sufficient and necessary for the transfer of advanced reasoning abilities that cumulative cultural evolution has enabled.
The reasoning mechanism
Let us now transition from speculations about the development of reason from shared intentionality to a lower-level analysis of the cognitive mechanism of reason.
The transition from inference to reasoning
Humans do not operate based on logical deduction (Magnani, 2005; Holldobler et al., 2011, Mercier & Sperber, 2017). Humans have intuitions, and manage to use logic by filtering out illogical intuitions. Logical thought is essentially a distillation from a large space of intuitions with varying levels of validity. Reasoning only occurs and is evaluated after a conclusion is developed—a process called “abductive” reasoning. To see how human’s advanced intuitive capabilities may have evolved—granting the ability to develop reasons for potentially logical conclusions—we can incrementally investigate the transition from primitive inference to advanced intuition.
What is inference?
Inference—the process of acting based upon an internal understanding of one’s surroundings—is mutual to humans and other animals. In humans, inference encapsulates perception, intuition, and reasoning. We use basic, perceptive inference when our nervous system triggers sweating on a sunny day, just like an elephant uses its trunk to shower water onto itself. “System 1” inference, which renowned psychologist Daniel Kahneman also calls “intuition,” is a quick, automatic, human mechanism that guides behavior—in contrast to a slower, more conscious “System 2” which guides rational thought (Kahneman, 2011). The properties and semantics associated with each system are illustrated below in Figure 1.
Figure 1: A breakdown of the functions comprising human inference.
Kahneman, **D. (2011) Thinking, Fast and Slow. Farrar, Straus and Giroux.
Active inference, a modern approach to inferential cognitive processing, posits a closed, continuous feedback loop between an organism and its surroundings. Thus, sensory input informs a statistical model of the environment, and action serves to strengthen this internal predictive engine. Through this process, each action of an organism minimizes a quantity called “free energy,” which is an abstract representation of the difference between observed and expected sensory input. Essentially, every action of an organism is a realization of what the organism expects to happen next, and a strengthening of the organism’s internal predictions (Friston et al., 2016).
Are we each other’s primary affordances?
Active inference lends itself to individual entities with which an agent interacts—such as a tree, a rock, or a fruit—which then generate patterns of sensory input. These entities are called “affordances.” An organism is simultaneously biased toward interacting with the features of its environment, or “epistemic resources” that give it the most information that can be used in the organism’s internal predictive engine. Thus, the ultimate affordance to a hominin, which both leads to the greatest range of possible outcomes and provides the greatest amount of information, is another hominin.
Veissiere and his colleagues characterize this active inference interpretation of social interactions as “thinking through other minds” (TTOM), an extension of the conventional theory of mind. TOOM is a probabilistic version of shared intentionality (Vessiere et al, 2020). The first representational affordance for hominins was most likely the representation of another hominin—along with its potential interests and behaviors. This strongly suggests that social intuitions may be a logical next step in cognitive ability from the basic inferential capabilities shared by animals. From the necessity of the interpersonal affordance being encoded as a representation of the other hominin’s internal representations, the ability for generalized metarepresentation may have developed.
The cultural niche to which hominins adapt, or the “cultural affordance,” is thus a scaled up manifestation of interpersonal TTOM. Both implicitly learned and explicitly explained aspects of enculturation are facilitated through TTOM (although a majority of transmission of cultural information is implicit, at the individual and group levels). Reason itself may thus be an entirely conventional, culturally-learned capability underlying explicit cultural transmission of information from epistemic affordances.
Intuitions: bridging the gap between inference and reason
Hugo Mercier and Dan Sperber, in The Enigma of Reason (2017), put forth an approach that is slightly different yet largely complementary to Veissiere’s account. They draw a more complex line between inference and reason. By their account, intuitions are a specific category of inferences, “the output of which happens to be experienced as intuitions” (Mercier & Sperber, 2017, p. 66).
Mercier and Sperber believe that the intuitions are simply metacognized inferences—essentially, an inference that can be used in conscious reasoning. Once intuitions emerge, they can then be evaluated for likelihood of accuracy, logical consistency, and benefits to the individual or group. Reasons (plural), which are the fundamental unit comprising the capacity for reason, are metarepresentations derived subsequently from these intuitive inferences (Mercier & Sperber, 2017). This reflects the aforementioned concept that an argument can only be evaluated for logical validity after it is made; humans do not actively derive logical conclusions from all evidence they are presented.
Thus, reasoning is more likely to have developed in evolution for a communicative purpose, rather than for contriving ways to manipulate the environment—so that humans can agree on plans and roles in collaborative activities, and determine whether to trust the testimony of one another. Reasoning does not need to be perfectly logical if it is irrespective of natural rules of causality; it only needs to serve as a loose enough approximation for logic such that arguments pass the bar of being seemingly intuitive. Hominins used reasoning to justify and explain their actions to others, to develop the level of mutual trust required for the collaborative activities to which they were adapting.
Recursive thought: the glue binding reasoning together
Veissiere and his colleagues, as well as Mercier and Sperber, would agree that reasoning is a highly advanced version of inference. Both the active inference perspective and the intuitions perspective would also concur that reasoning requires an advanced capacity for representation stretching beyond the predictive capacity of inference. For Veissiere, this is the ability for a hominin to integrate interactions with conspecifics within inference; hominins are a novel kind of epistemic affordance presenting an exponentially greater range of information and environment-impacting capabilities. For Mercier and Sperber, this is the mere capacity for metarepresentation, which enables some facts to rationalize or undermine conclusions.
Tomasello’s A Natural History of Human Thinking (2014) suggests that reasoning about objective facts stems from “socially recursive inferences,” which integrate one person’s worldview into that of another (Tomasello, 2014). While Great Apes in general have basic intuitions for causal inference, the ability to generally reason may entail a recursive function that enables any intentional state to be embedded within another. This reflects the same concept as metarepresentation and TTOM. The recursive function effectively unites the intentional states of individuals, lending itself to the establishment of a collective intentionality—from which a general capacity for reasoning may emerge, as indicated by various phenomena in language and culture (Corballis, 2011). Thus the human capacity to evaluate intentional states for objective validity—which approaches, but is not derived from, raw logical consistency—evolved from the socially-derived capacity for recursive thought.
Conclusion, discussion, and implications
The underpinnings of reasoning most likely developed in hominin evolution in order to enable effective collaboration, which was initially incentivized by the need for healing, assisted birth, or collaborative foraging in the ancestral environment. Substantial research from child development psychology and education is complementary to this claim—indicating that uniquely advanced social skills develop in humans before reasoning capabilities, and that the process of children learning to reason effectively is guided by social interactions. An intimate look at cognitive inference demonstrates that reasoning is built upon a specific type of inference called intuition. Intuitions most likely developed to facilitate interpersonal interactions, because humans present uniquely high epistemic value to each others’ internal active inference mechanisms. A metarepresentational capacity for processing interpersonal interactions generalizes to the recursive thought mechanism, which may have been necessary and sufficient for the development of the general reasoning capability.
Multiple takeaways can be drawn from this analysis that are practical in other fields. The mechanism of reason is extremely applicable to the development of advanced artificial intelligence, and the specific conclusion that reason is fundamentally social in nature supports the practice of reinforcement learning with human feedback (RLHF), a technique used to train large generative machine learning models. Additionally, the social foundations of reason may be quite valuable for educators seeking to teach students to reason effectively. Rather than teaching mere facts—from which no human can automatically deduct conclusions—educators might prioritize attuning students' intuitions about methods logic so that they can think critically for themselves and with others. This subject is also relevant to building group cohesion in any large social context: communities and shared intuitions go hand-in-hand. Ideological pluralism is thus perhaps unnatural, violating the cultural norm upon which reasoning is based—extensive amounts of rationalization are needed to fit a variety of opinions, backgrounds, and ideologies into a collectively agreed-upon narrative; humans must work to substantially extend their intrinsic capabilities in order to espouse the tenets of a pluralist community.
It is critical to acknowledge the speculative nature of generalizing conclusions from sparse anthropological evidence and theories of evolutionary psychology. While this argument is a synthesis of available research, substantial supporting evidence from evolutionary anthropology, neurology, and other fields would be required to validate it. Further research into the origins of reason may involve additional archeological discoveries, robust genetic analysis of early hominins, comparative behavioral psychology of primates, child development and educational research, functional neuroscience, and studies of the reasoning capabilities of artificially intelligent agents. The nature of consciousness and its exact relation to reason have yet to be revealed. In addition, reason likely developed over the span of millions of years, so an extremely wide variety of evolutionary factors could have triggered specific aspects within its development. Nevertheless, a comprehensive, multidisciplinary analysis yields the most accurate possible general argument for how advanced human cognition originated.
References
Bandura. A. et al. (1961). Transmission of aggression through imitation of aggressive models. Journal of Abnormal and Social Psychology, 63, p. 575-582.
Corballis, M. (2011) The Recursive Mind. Princeton University Press.
Csibra, G., & Gergely, G. (2006). Social learning and social cognition: The case for pedagogy. Processes of Change in Brain and Cognitive Development. Attention and Performance, XXI. p. 249-274.
DeSilva, J. (2021). First Steps: How Upright Walking Made Us Human. William Collins.
Friston, K. et al. (2016) Active inference and learning. Neuroscience & Biobehavioral Reviews 68, p. 862-879. https://doi.org/10.1016/j.neubiorev.2016.06.022
Frith, C. & Frith, U. (2011). Mechanisms of social cognition. Annual Review of Psychology 63. p. 287-313. https://doi.org/10.1146/annurev-psych-120710-100449
Holldobler, S. et al. (2011). An Abductive Model for Human Reasoning. Logical Formalizations of Commonsense Reasoning, Papers from the 2011 AAAI Spring Symposium, Technical Report SS-11-06, Stanford, California, USA, March 21-23, 2011.
Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
Kuroda, K. & Kameda, T. (2019) You watch my back, I'll watch yours: Emergence of collective risk monitoring through tacit coordination in human social foraging. Evolution and Human Behavior 40(5), p. 427-435. https://doi.org/10.1016/j.evolhumbehav.2019.05.004
Magnani, L. (2005) An Abductive Theory of Scientific Reasoning. Semiotica 153(1/4). p. 261-286. https://doi.org/10.1515/semi.2005.2005.153-1-4.261
Malafouris, L. (2020). How does thinking relate to tool making? Adaptive Behavior
29(2), p. 1–15*.* https://doi.org/10.1177/1059712320950539
Mercer, N. et al. (1999). Children's Talk and the Development of Reasoning in the Classroom. British Educational Research Journal 25(1), p. 95-111*.* http://dx.doi.org/10.1080/0141192990250107
Mercier, H. & Sperber, D. (2017) The Enigma of Reason. Harvard University Press.
Mesoudi, A. & Thornton A. (2018). What is cumulative cultural evolution? Proc. R. Soc. B 285(20180712). http://dx.doi.org/10.1098/rspb.2018.0712
Moll, H., & Tomasello, M. (2007). Cooperation and human cognition: the Vygotskian intelligence hypothesis. Philosophical transactions of the Royal Society of London. Series B, Biological sciences 362(1480), p. 639–648. https://doi.org/10.1098/rstb.2006.2000
Müller, U. & Carpendale, J. (2000). The role of social interaction in Piaget's theory: language for social cooperation and social cooperation for language. New Ideas in Psychology 18(2-3), p. 139-156. https://doi.org/10.1016/S0732-118X(00)00004-0.
Osiurak, F., et al. (2021). Technical reasoning is important for cumulative technological culture. *Nature Human Behavior ***5, p. 1643-1651. https://doi.org/10.1038/s41562-021-01159-9
Osvath, H. & Gardenfors, P. (2005). Oldowan culture and the evolution of anticipatory cognition. Lund University Cognitive Science, 122.
Packer, M. & Cole, M. (2019). Evolution and Ontogenesis: The Deontic Niche of Human Development. Human Development 62, p. 175–211. https://doi.org/10.1159/000500172
Tomasello, M., et al. (2005). Understanding and sharing intentions: the origins of cultural cognition. The Behavioral and Brain Sciences 28(5), p. 675–735. https://doi.org/10.1017/S0140525X05000129
Veissiere, S. et al, (2020). Thinking through other minds: A variational approach to cognition and culture. Behavioral and Brain Sciences 43(90), p. 1-75. **https://doi.org/10.1017/S0140525X19001213