Thinking about thinking: How the mind (or AI) works -Governance Now

Thinking about thinking: How the mind (or AI) works

In ‘The Laws of Thought’, Tom Griffiths writes the exciting story of a quest that has given us thinking machines

GN Bureau | May 18, 2026

#Artificial Intelligence #Science #Mathematics

(Photo: Courtesy CrisNYCa via https://commons.wikimedia.org/wiki/File:Mus%C3%A9e_Rodin_1.jpg)

Tom Griffiths is one of those scientists working at the cutting edge of cognitive science and AI. He is a professor of psychology and computer science at Princeton University, and directs the Computational Cognitive Science Lab and the Princeton Laboratory for AI. His first book for general readership ‘Algorithms to Live By: The Computer Science of Human Decisions’ (co-authored with Brian Christian) applied computer science to daily life. His new work, ‘The Laws of Thought: The Quest for a Mathematical Theory of the Mind’ (HarperCollins) is a more ambitious, historical, and philosophical exploration. It shifts the focus from “how to act” to “how we work”. It presents a narrative history of cognitive science and a deep dive into the three competing frameworks that try to explain human intelligence.

Griffiths’ central question is, “Is there a ‘physics’ of the mind?’ We all understand the external world, with its “Laws of Nature”. Concepts such as gravity, force, entropy were formalized by scientists like Newton and Maxwell. But are there laws governing the internal world, that is, something like ‘Laws of Thought’? For 300 years, a quest has been going on to find a mathematical language for the mind, from Enlightenment philosophers to modern Large Language Models (LLMs). This book tells the story of that quest to formalize human thinking using mathematical frameworks. It is the story of how mathematics has been used to understand human cognition, from historical logic to modern AI systems.

The core of the book is organized around the three major mathematical frameworks that have attempted to "solve" the mind. Griffiths argues that the history of cognitive science is essentially a struggle between these three schools:

(1) Logic (Rules and Symbols): This was the first great hope. Starting with George Boole (who wrote the original The Laws of Thought in 1854), this school argues that thinking is like algebra. If you can define the rules (IF this, THEN that), you can simulate thought. Griffiths explores why this "Good Old Fashioned AI" (GOFAI) eventually hit a wall: the world is too messy for a finite list of rules.

(2) Neural Networks (The Connectionists): This school looks at the architecture of the brain rather than the rules of logic. It focuses on how simple units (neurons) connected in massive webs can "learn" patterns. Griffiths traces this from early "Perceptrons" to the modern "deep learning" revolution led by figures like Geoffrey Hinton.

(3) Probability and Statistics (The Bayesians): This is Griffiths’ own primary field of expertise. This framework suggests the mind is not a calculator or a simple web; it is a prediction engine. It uses "Bayesian inference" to update its beliefs about the world based on incomplete and noisy data.

A major theme of the book is why AI, despite its incredible power, still feels "different" from human thought.

Narrow vs. General: Griffiths points out that modern AI is often "super-specialized": it can beat a grandmaster at chess but can’t figure out how to fold a towel or navigate a social nuance.

The Data Gap: Humans learn from very little data (a child sees two cats and knows what a cat is). AI requires billions of parameters and massive datasets. Griffiths argues that the "Law of Thought" we haven't fully mastered yet is how humans use Prior Knowledge so efficiently.

Not a dry textbook, ‘The Laws of Thought’ uses the biographies of scientists to ground the math. We meet Gottfried Wilhelm Leibniz, the dreamer who wanted a "universal characteristic", a language where two philosophers in an argument could simply say, "Let us calculate!" to find the truth. Then there is George Boole, the schoolteacher who realized logic could be expressed as 0s and 1s, laying the foundation for the digital age. Then came the cognitive revolutionaries such as Alan Turing, Noam Chomsky, and Herbert Simon, who moved us away from Behaviourism (just looking at what people do) to Cognitivism (modeling what is happening inside the head).

Griffiths suggests that by trying to build ‘artificial’ intelligence, we have learned more about ‘natural’ intelligence. Every time a computer fails at a task that a toddler finds easy, it reveals a hidden ‘Law of Thought’ that we previously took for granted. He argues that we are currently in a ‘synthesis’ phase, where the most promising models of the mind combine symbols, neural networks, and probability.

In the final chapters, Griffiths moves into the ethical and social. As AI begins to make decisions about our lives, he argues that we must understand the mathematics of thought. If we don't know the "laws" the machine is following, we cannot hold it accountable, nor can we understand why it might develop biases that differ from our own.

Comments

Other News

Gandhian movement after Gandhi failed: Ram Guha

The Gandhian movement in India, after Gandhi’s death, failed, says Ramachandra Guha, and blames Vinoba Bhave for it. But the movement found new, fertile grounds around the world and succeeded in bringing about a change in South Africa, the US and elsewhere. It is high time it returned home. Of cour

When religion, politics and Bollywood come together

Tavleen Singh, noted journalist and political commentator, has turned to fiction, with her debut novel ‘The Actor’ (HaperCollins India). Set in the near future, it has a superstar, Amit Singh, turning to politics, when the political landscape is dominated by prime minister Guru Gaynendraj

Cabinet approves setting up of Integrated Transport & Logistics Authority

The union cabinet, chaired by prime minister Narendra Modi, on Tuesday approved the setting up of a Special Purpose Vehicle (SPV), called Integrated Transport & Logistics Authority (ITLA), to strengthen research, planning, appraisal, monitoring and impact assessment within the broad domains of Transp

Why foundational literacy and numeracy is key to school education

Foundational Literacy and Numeracy (FLN) are often described as the building blocks of education. Yet, their significance goes well beyond a child’s ability to read a sentence, write a response or solve a basic mathematical problem. These foundational skills shape how children understand the world,

When police data becomes the algorithm

For decades, policing has followed a familiar sequence: a crime occurs, the police respond, an investigation produces records and those records become part of the institutional memory of the criminal justice system. Artificial intelligence is beginning to alter that sequence. Instead of merely recording

Before India voted, the framers worried about the referee

“We are making a Constitution forever and not only for today.” Shibban Lal Saksena said this in the Constituent Assembly on June 15, 1949. He was warning that a ruling party might one day choose a loyalist to run India’s elections. Seventy-seven years later, the Supreme Court is wrestli