No. 221 of 500 No. 46 in Mind, Brain & Philosophy

How We Learn

Why Brains Learn Better Than Any Machine . . . for Now

By Stanislas Dehaene

Print length
352 pages
Publisher
Penguin Books
Reading level
Intermediate
ISBN-13
978-0525559900
Published
February 2, 2021
Written for
Teachers, parents, learning designers and AI readers curious how human learning differs from machine learning.

Synopsis

Stanislas Dehaene, the cognitive neuroscientist who directs France's NeuroSpin brain-imaging center, asks what learning is at the level of neurons and why the human brain still does it far more efficiently than artificial neural networks. He defines learning as building an internal model of the world, then contrasts deep learning systems, which need enormous labeled datasets, with infants who arrive equipped with intuitions about objects, numbers, faces and language and who generalize from a few examples. Drawing on brain-imaging research, including his own studies of reading and arithmetic, he argues that innate structure and experience work together, and that the brain's early organization makes fast learning possible.

The second half sets out four pillars of learning: attention, active engagement, error feedback and consolidation, with sleep playing a central part in the last. Dehaene turns each pillar into guidance for classrooms and self-study, such as frequent low-stakes testing, feedback that treats errors as information rather than failure, and spaced review. Readers gain an evidence-based account of why some teaching methods outperform others and a clear view of what current AI systems still lack. First published in French in 2018 and in English in 2020, the book is dense in places, and some reviewers found the translated prose demanding.

Scorecard

Editorial score

How it scored

Editorial score: 73.8 / 1005 weighted criteria

  • Influence30%

    21.0/30
  • Authority25%

    22.5/25
  • Practical value20%

    14.0/20
  • Currency15%

    9.8/15
  • Readability10%

    6.5/10
  • Editorial score

    73.8/100

How the rank was computed

Rank No. 221 of 500 comes from a composite of 55.7: observable signals 22.5 of 55 points plus editorial review 33.2 of 45 points.

All 11 inputs
InputPoints
Observable signals22.5/55
Reach (Amazon ratings volume)10.8/16.5
Consensus across best-of lists2.8/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted7.4/8.3
Amazon sales rank1.5/5.5
Recency0.0/2.8
Editorial review33.2/45
Author authority10.1/11.3
Influence9.5/13.5
Practical value6.3/9.0
Readability2.9/4.5
Currency4.4/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

A rigorous neuroscientist's account of why infants outlearn algorithms, demanding but rewarding for teachers and anyone comparing brains with machines.

Strengths

  • Written by a leading cognitive neuroscientist
  • Bridges neuroscience, education and machine learning
  • Turns research into concrete teaching principles

Keep in mind

Dense, academic prose demands close attention, and its machine learning comparisons were written before large language models reshaped the field.

Key takeaways

  1. Learning means building an internal model of the world, and brains do this with far fewer examples than today's deep learning systems require.
  2. Infants are not blank slates: they arrive with core intuitions about objects, numbers, faces and language that make rapid later learning possible.
  3. Four pillars govern learning efficiency: attention, active engagement, error feedback and consolidation, and teaching that respects all four produces better results.
  4. Frequent low-stakes testing and feedback that treats mistakes as information strengthen memory more than rereading or grades delivered as punishment.

Notable facts

About the author

Stanislas Dehaene

Stanislas Dehaene holds the chair of Experimental Cognitive Psychology at the Collège de France and directs the NeuroSpin brain-imaging center at CEA Saclay. Trained in mathematics at the École Normale Supérieure, he shared the 2014 Brain Prize, was elected a Fellow of the Royal Society in 2023, and has chaired France's Scientific Council for Education since 2018. His books include Reading in the Brain and Consciousness and the Brain.

Questions

What is How We Learn by Stanislas Dehaene about?
The book explains how the brain learns and why it still outperforms artificial neural networks. Dehaene describes infants' built-in knowledge of objects, numbers and language, shows how brain plasticity works, and identifies four pillars of effective learning: attention, active engagement, error feedback and consolidation during sleep. He then applies these findings to teaching, schooling and self-study at any age.
Is How We Learn worth reading in 2026?
Yes for educators and anyone interested in the science of learning, because its neuroscience and four-pillar framework remain well supported. Readers should note that its comparisons with artificial intelligence were written before large language models, so its account of machine learning's limits is partly dated. The prose is dense, and the book rewards slow, attentive reading more than a quick skim.
Does How We Learn argue that brains will always beat machines?
No. The subtitle's 'for now' signals Dehaene's position: today's machines learn far less efficiently than children, needing vast data where a toddler needs a few examples, but he treats this as a gap engineers could narrow by borrowing principles from the brain, such as built-in structure, curiosity and sleep-like consolidation. The book is a comparison, not a claim that machines will never close the gap.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.