The trap of Rationalism: Why thinking too logically can actually hold you back

The Limits of Rationalism

Rationalism is the idea that to truly know something,

you must be able to describe it explicitly with rules,

definitions, and theories.

This philosophy suggests that true knowledge of the world

can only be achieved by reducing phenomena down

to explicit components, distinct from intuitive

or emotional ways of understanding.

This worldview is the bedrock of Western culture.

It underpins how computers and vaccines work,

how we predict the weather, and even how therapy operates.

It is highly successful, but it has a fundamental flaw:

it blinds us to the critical importance of intuition.

Socrates vs. Protagoras: The Birth of Rationalism

The roots of rationalism trace back to ancient Greece,

specifically to Socrates.

In Plato’s dialogue Protagoras, Socrates debates Protagoras, a sophist,

over whether “excellence” (or virtue) can be taught.

• Protagoras argued that everyone has the capacity for excellence and used myths, metaphors, and stories to explain it as something learned through hands-on experience and immersion in society.
• Socrates rejected this, demanding a precise, explicit definition of what excellence is and isn’t. He argued that if Protagoras could not define it in a non-contradictory way, he didn’t truly “know” it.

This moment set Western society on a path

where knowledge became synonymous with the ability

to define things explicitly.

This mindset blossomed during the Scientific Enlightenment,

with thinkers like Descartes, Newton, and Galileo proving

that mathematics and explicit rules could explain

and predict the physical world, shaping modern

technology from smartphones to rockets.

When Rationalism Fails

While rationalism conquered physics and engineering,

it struggled deeply when applied to the social sciences.

Fields like psychology and economics attempt

to reduce complex human phenomena into a set

of physics-like rules and definitions.

This approach has led to massive challenges,

such as psychology’s replication crisis.

Despite a century of research, finding universal laws for

human behavior—akin to Newton’s laws of motion

has proven nearly impossible because human nature resists

being boiled down to explicit, rigid rules.

Symbolic AI vs. Neural Networks

The limitations of rationalism are perfectly illustrated

by the history of Artificial Intelligence.

The Failure of Symbolic AI

When AI began in the 1950s, theorists tried to apply rationalism

by building “Symbolic AI.”

The goal was to reduce human intelligence down to

a system of logic symbols and explicit “if-then” rules.

• Example: To build a spam filter, you might create a rule: “If the email mentions winning the lottery, it is spam.” But exceptions immediately arise. If you make a rule to prioritize emails marked “Emergency,” spammers will just write “Emergency.” If you add a rule that it must be from a coworker, annoying coworkers might abuse the system.
• To solve real-world problems using symbolic AI, programmers had to explicitly define the entire world. The systems became too brittle, mathematically heavy, and ultimately failed when moved away from simple “toy” problems.

The Rise of Neural Networks

Neural networks provided the alternative.

Inspired by the human brain,

neural networks do not rely on explicit rules or definitions.

Instead, they are trained through massive amounts of trial

and error and pattern recognition.

• By feeding the network thousands of examples (e.g., of important vs. unimportant emails) and correcting it when it is wrong, the system learns to recognize patterns on its own.
• In Large Language Models (like ChatGPT), the AI learns thousands of inexplicit, partially fitting rules based on massive amounts of text to predict what word comes next. You cannot look inside the network and extract a neat list of explicit logic rules, just as you cannot look inside a human brain and find the exact logical code for recognizing a cat.

The Return of Intuition

Neural networks operate incredibly similarly to human intuition,

which is also trained by thousands of hours of direct experience

without explicit rules.

For much of the 20th century, the dominant metaphor for

the human mind was a computer: logical, rational, and rule-based.

This metaphor made our intuition invisible to us.

However, rational thought actually emerges out of intuition.

Our squishy, inexplicit intuition sets the frame, allowing our rational

mind to step in and manipulate things methodically.

Neural networks represent the first technology

we have ever invented that models human intuition

rather than just human logic.

This circles back to Protagoras: stories, hands-on experience,

and intuition give us a powerful way of knowing things

that we cannot explicitly define.

History shows it is vital to be open to different ways

of understanding the world, even if that means becoming

comfortable with things that are

intuitive, inexplicit, and a little bit mysterious.