Are we born with brains that are blank slates waiting only to be filled with experience? And if so, why is it so difficult to simulate that same experience in a computer? Why can a sophisticated AI be easily tricked into thinking that a cat is guacamole or that a Granny Smith apple is an iPod? How have human brains managed to adapt to the diversity of environments across the globe, from the Stone Age to the Space Age?

In How We Learn: Why Brains Learn Better Than Any Machine … for Now, French neuroscientist Stanislas Dehaene shows that the “blank slate” hypothesis does not withstand scrutiny. Our brains are equipped, from birth, with specialized neural circuits that shape a child’s intuitions about physics, arithmetic, geometry, and probability.

Dehaene cites a large body of research (including his own) that shows how young infants’ brains make sense of the world using this “core knowledge”. This research has forced a re-appraisal of older psychological models that have become popular wisdom:

Incidentally, these results overturn several tenets of a central theory of child development, that of the great Swiss psychologist Jean Piaget (1896–1980). Piaget thought that young infants were not endowed with “object permanence”—the fact that objects continue to exist when they are no longer seen—until the end of the first year of life.

He also thought that the abstract concept of number was beyond children’s grasp for the first few years of life, and that they slowly learned it by progressively abstracting away from the more concrete measures of size, length, and density.

In reality, the opposite is true. Concepts of objects and numbers are fundamental features of our thoughts; they are part of the “core knowledge” with which we come into the world, and when combined, they enable us to formulate more complex thoughts. [p. 58]

Brains vs. Machines

Like a budding scientist, an infant’s brain forms predictions about the world and then adapts its wiring based on whether the predictions match reality. This neuroplasticity is powerful, but it is highest during early sensitive periods, and dependent on innate cognitive skills—a number sense, the ability to detect faces, to acquire languages, and so on.

Dehaene contrasts this with the current generation of artificial networks, which are closer to a “blank slate” on which the characteristics of a training dataset are gradually inscribed. As a result of this naive approach, AIs lack the human brain’s multifaceted reasoning skills—”for now”, as the book’s subtitle puts it.

Hinting at what tomorrow may bring, Dehaene cites examples of artificial neural networks modeled after human cognition. For instance, a 2008 algorithm for predicting the optimal structural representation of various datasets “rediscovered” human knowledge representations like the tree of life.

One may quibble with the author’s assertion that human brains learn “better” than machines, given that the latter have already excelled in specific domains. Just ask Go player Lee Sedol, who retired from professional play and called AI “an entity that cannot be defeated”.

A Theory of Learning

But the focus of the book are humans, not machines. The author’s goal is nothing less than a theory of learning that teachers, parents, and students can apply in practice. He suggests a model based on “four pillars of learning”: attention, active engagement, error feedback, and consolidation.

From these four pillars he develops concrete recommendations for educators. Many of them may match the instincts of progressive teachers: anxiety and stress prevent learning; students need to be engaged as individuals; sleep is essential. However, Dehaene is highly critical of unassisted discovery learning and the popular idea of learning styles.

Dehaene does not take a one-sided view in the “nature versus nurture” debate; instead, he explains the complex interplay between our biology and the world we inhabit. Biological evolution has placed constraints on us, but it has also allowed us to reason about the workings of our own brains, and to journey to other planets.

The Verdict

As a whole, I highly recommend Dehaene’s book. As befits the subject, he is a great (occasionally witty) teacher who uses straightforward explanations and illustrations in order to systematically advance the reader’s understanding of a complex subject.

While I found the author’s level of certitude—when he talks about animal intelligence or learning styles, for example—a bit off-putting at times, that too can be the mark of a teacher in love with their subject. It is a passion he successfully passes on to the reader.