How Reasoning Emerges from Intuition
If every step that of reasoning (System 2) performs is a call to intuition (System 1) then there’s no separation between reasoning and intuition. Therefore, reasoning is just a meta-cognitive intuition of one’s intuition.
Left unanswered is the question of how human reasoning emerges from an intuition machine.
In the usual formulation of the dual-process cognitive model of Daniel Kahneman, System 1 and System 2 are delineated as an intuitive cognitive system and a rational cognitive system. My conjecture that I’ve been pursuing is that rational thought is a consequence of intuitive thought. That is, System 1 simulates System 2. In this article, I am going to explain how rationality arises from intuition.
First, we need to clarify Kahneman’s System 1 and System 2. Kahneman’s original depiction can be found in this table:
What is missing here is an explanation of how System 2 arises from System 1. Furthermore, how does System 1 simulate (or act as a virtual machine) to System 2. To discover this transition, we need to identify the essential intentional stance that differentiates homo sapiens from primates. What is uniquely distinctively human in our behavior?
Tomasello proposes that it is the motivation of humans for “shared intentionality”. Humans are uniquely inclined toward cooperative and coordinated behavior. This said, other social animals also have the cognitive ability to coordinate their actions in solving a task. This social behavior exists in living things as simple as bacteria and bees. Humans just happen to have larger brains and thus require more sophisticated cognition to achieve group coordination.
The utility of human reasoning is for the purpose of group coordination. One can argue, as Yuval Noah Harari has done, that it is the homo sapiens advanced ability for cooperation that has led to the civilization that we live in today. Harari writes in “Sapiens” why humans dominate the earth:
“Ants and bees can also work together in huge numbers, but they do so in a very rigid manner and only with close relatives. Wolves and chimpanzees cooperate far more flexibly than ants, but they can do so only with small numbers of other individuals that they know intimately. Sapiens can cooperate in extremely flexible ways with countless numbers of strangers. That’s why Sapiens rule the world, whereas ants eat our leftovers and chimps are locked up in zoos and research laboratories.”
Harari argues that it is in language and myth-making that humans are able to organize themselves. What is missing in Harrari’s formulation is “how are homo sapiens capable of language and thus myth-making?”. What we need is a compelling narrative of how a brain that is of System 1 transitions to a system capable of language and eventually myths.
The motivation for language is obvious. The purpose of language is to enable cooperative behavior. The basic elements of language, that is, the operation of giving things names. In Semiotics there are 3 kinds of signs: Icons, Index and Symbols. It is symbols where we assign names to things. The primary purpose of this operation is to support the coordination of behavior among multiple agents.
Coordination is still possible with just icons and indexes. This is how animals without language coordinate. Mammals and birds are capable of a ‘theory of mind’. They have consciousness and an ability to place themselves in the perspective of other animals. They are able to do this through sensing and interpreting the body language of another animal. In fact, it should be obvious that animals intrinsically use indexical information to infer the presence and movement of another animal. This indexical information also includes the sound an animal might make.
Vocal communication is an enhancement in that it allows an agent to express their intentions to another. The technological advance of vocal communication allows an animal to receive signaling information without actually seeing or feeling the other animal. This is signaling at a distance. It’s a level of indirection leads to more advanced coordination methods.
Social animals like dogs, birds, elephants, dolphins, whales, seals and primates all are able to use vocal communication to transmit information and thus coordinate their intentions over distances. Communication at a distance is an extremely useful tool. One can further argue that social animals have a proto-language where specific vocalizations have semantic meanings. A sound that is associated with semantics is, in fact, a symbol. Animals are therefore capable of symbolic communication.
More complex language is invented to support more complex signaling patterns. As a consequence, this leads to more sophisticated coordination behavior. Coordination requires shared knowledge to effectively work. Communication is clearly less effective if every utterance needs an explanation. To achieve this, common utterances are learned by a group.
But how are shared utterances learned, furthermore how are shared utterances taught and conveyed? At the most basic level, it begins with mimicry. This is taught through repetition of actions. Clearly, many animals are capable of this kind of learning. But there’s a limit to the information that can be conveyed by teaching by example. Teaching by example requires the student to see or to experience what is taught. John Vervaeke calls this perspectival and participatory knowing respectively. How can a student learn a new concept without seeing or participating in an experience?
Procedural knowing, the ability to learn how to perform a sequence of tasks is driven by the cognitive necessity to navigate. Many animals are capable of navigating vast distances. It’s a consequence of the machinery we have in the hippocampus. This is an essential skill that’s required for navigation and for tool creation. Vervaeke calls this procedural knowing.
It’s intriguing to note that bees are also able to communicate to their hive the navigation instructions. This is known as the waggle dance. A honey bee has just 1 million neurons, that 1/100,000th the number of neurons in a human brain.
But the human mind procedural knowing was forged by the evolutionary pressure to create tools. Homo sapien and her ancestors were gifted with great dexterity in their hands. This served as a catalyst for crafting tools and thus developed further our capability for refining procedural cognition. That is, homo sapiens learn how to optimize how they craft their tools. This optimization process in its most basic form is known as trial and error.
Navigation or procedural knowing is a stepping stone to our ability for reasoning. Navigation requires understanding the sequence of spatial movements required to get to a place. Reasoning requires understanding the sequence of explanations that lead to a conclusion. Complex language allows information about behavior to be conveyed not just by example but also via a more complex explanation.
The ability to reason about an explanation and also communicate an explanation requires the formulation of causal models in an agent's mind. To perform the latter is analogous to explaining how one can travel to another place. An explainer takes the perspective of the traveler. But rather than explaining directions and landmarks, an explainer substitutes causality directions and propositions. Vervaeke calls this propositional knowing.
The invention of a symbol (a vocalization) of causality patterns leads to an ability for more complex coordination. The ability to reason and thus explain symbols to another agent allows information to bridge the experiential gap. A map provides a traveler a means to navigate the territory that he’s never seen. By analogy, an explanation provides a listener with a way to understand the causal relationships that he’s never experienced. An agent thus can mimic a complex utterance through procedural knowing. An agent can understand the purpose of an utterance and through propositional knowing.
This explains why there is a synergistic coupling between language and reason. The purpose of reasoning is to explain to oneself the motivations for one’s goal. The other purpose of reasoning is to explain to others why a kind of coordinated behavior is useful. The purpose of language is also to tell stories and ultimately share one’s experiences.
I would, however, caution the reader that the cognitive ability for reasoning is not the same as the capability for rational thought. Reasoning is, in fact, a System 1 capability. Reasoning is our intuitive capacity to understand and formulate explanations. Not all explanations are rational or even correct. The only important goal is that explanations lead to a shared understanding between communicating parties. Reasoning is required for conversational cognition and in fact is enhanced through conversation.
If we are to assume the brain to be a consequence of many parallel thinking processes then it should be unsurprising that certain processes come to our attention without our awareness of its origins. It also should not be surprising that we become aware of the relevance of conscious thoughts also through the same intuitive process. It would be grossly inefficient if a mind was constantly in reflection of what it was contemplating.
Like consciousness that is a reflective cognitive ability, reasoning is also reflective in that it layers an explanation on top of the intuitions that we uncover. That layering of explanation happens after the thought and not the thought itself. It’s the mind introspecting on the validity of its thoughts.
The brain moves forward using intuition. Reasoning is the intuition we use to discover an explanation of our intuitions. If we can discover the chain of explanations, then we describe this as deliberate thought. The mistake of many cognitive models is that they substitute the explanation with the actual process. Consciousness and reason are a meta-cognitive capability to explain our intuitions. Our brains are all System 1.
Rationality relates to systematic reasoning that emerges as a consequence of cultural evolution. Rationality is a thinking tool, a technology that our intuitive brains can make use of. It is a set of procedures and rules that we employ to improve our reasoning. It is however not a cognitive ability that is intrinsic in human brains. The mistake of many AI practitioners is to believe that rationality is an intrinsic human ability. The reason for this delusion is a cognitive bias that assumes humans are distinct from animals because of our rationality. This kind of cognitive model has both a cultural and an anthropocentric bias.
Further Reading
