HW
Stanislas Dehaene

How We Learn

Education
Back to Categories
Education16 min read

How We Learn

by Stanislas Dehaene

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

Published: October 15, 2024
4.2 (201 ratings)

Book Summary

This is a comprehensive summary of How We Learn by Stanislas Dehaene. The book explores why brains learn better than any machine . . . for now.

what’s in it for me? discover the brain’s learning algorithms#

Introduction

ever wondered what really happens inside your brain when you’re learning something new? the concept of learning might seem straightforward, yet it’s a complex and dynamic process that often goes unnoticed in our daily lives.

few people are better equipped to shed light on this process than cognitive psychologist stanislas dehaene. in this chapter, we’ll be delving into the intersections of neuroscience, psychology, and technology to unveil how our brains function as highly efficient learning machines. along the way, we’ll learn about the differences between human and artificial intelligence as well as what we can do to boost our children’s mental growth. 

learning transforms perception into knowledge#

understanding “learning” involves appreciating how our brains construct and refine internal models of the world around us. these models are vast and varied, covering everything from the layout of our local environments to complex language structures and even the nuances of other people’s behaviors. these internal models aren’t static representations; they are dynamic, constantly updated frameworks that help us navigate and interpret everyday experiences.

when we learn, we essentially train our brains to simulate possible realities. for instance, consider the mental maps we carry of places like our neighborhoods or homes. these aren’t simple maps; they’re rich, interactive simulations that allow us to navigate these spaces mentally without physical presence. such capability extends to language, where our brains can distinguish familiar words from nonsensical ones, and to physical coordination, like the intricate balancing act involved in riding a bicycle.

the brain’s ability to dream represents another fascinating aspect of its modeling capabilities. dreams aren’t random images; they are intricate narratives crafted by the brain’s internal models. they allow us to experience realities that do not exist and to play out scenarios that have never happened. in a way, our ability to dream illustrates how the brain rehearses, explores, and refines its models continuously.

our brains also play a significant role in how we perceive the world when we are awake. every sensory input involves the brain in interpreting and predicting based on previous knowledge and context, transforming ambiguous visual information into a coherent perception. this process highlights a crucial aspect of learning: it’s not only about acquiring facts, but about enhancing our capacity to interpret and respond to our environments.

learning is also about making continual adjustments. when new information presents itself, or when an unexpected outcome occurs, our brains quickly incorporate this new data into existing models. this is evident in how we adapt to physical changes, like adjusting to distorted vision with prisms or glasses, and how these adjustments can recalibrate our perceptions even after the tools are removed. this adaptive quality allows us to navigate the world more effectively and to correct errors in our understanding and actions.

learning, in other words, is about more than just gathering information – it’s about forming, testing, and refining the mental models that allow us to understand and interact with the world more effectively. as we navigate through life, our ability to adapt our learning processes determines how well we can adjust to new challenges, understand others, and perceive the world around us.

human learning outpaces artificial intelligence in complexity and efficacy#

learning is a complex, multifaceted capability deeply embedded in the human experience and profoundly different from what even the most advanced artificial intelligence or ai systems can achieve today. while ai has made significant strides, particularly in areas like pattern recognition and data processing, it still lacks several fundamental qualities that define human learning.

human learning is incredibly efficient, for starters – especially in how it handles data. unlike ai, which requires vast amounts of information to develop basic competencies, humans can learn from remarkably few data points. for example, children can grasp a new language with only a fraction of the exposure that the most sophisticated language algorithms require, showcasing our brain’s ability to make significant leaps from minimal data.

another key area where human learning excels is in understanding abstract concepts. humans can recognize and understand objects and ideas in various forms and contexts, something that remains a challenge for ai. machine learning models, particularly deep learning networks, often misinterpret objects if even minor features are altered. this limitation highlights ai’s dependency on specific data patterns rather than an overarching understanding of concepts.

the flexibility of human cognition also stands in stark contrast to ai. humans can shift gears from rapid, intuitive judgments to slow, deliberate decision-making, crafting complex symbolic representations of the world that can be shared and discussed with others through language. this ability to engage in deep, symbolic thought and to communicate these ideas effectively remains uniquely human.

social learning is another domain where humans distinctly outperform ai. from an early age, humans learn from social cues, understanding and responding to the intentions and emotions of others. this capability allows for the nuanced transfer of knowledge through language, where a simple glance or gesture can enrich communication. here, too, ai struggles to replicate human cognition.

while ai continues to advance, it still lacks the depth and adaptability of human learning. from processing minimal data to engaging in complex social interactions, humans demonstrate a remarkable capacity to learn that machines have not yet matched. this comparison not only underscores the unique strengths of human cognition but also highlights the areas where ai may eventually grow to enhance its capabilities.

education transforms basic human potential into complex cognitive abilities#

education is a powerful tool that transforms basic human capabilities into highly sophisticated cognitive skills. it is fascinating that humans, who are not genetically pre-programmed for specific modern tasks, can learn complex skills like reading and mathematics. this capability underscores the remarkable adaptability of the human brain and the transformative power of education.

the impact of education on literacy is profound. studies of illiterate adults in various parts of the world show that lack of education not only affects the ability to recognize letters but also impairs the recognition of shapes, memory of spoken words, and the ability to distinguish mirror images. contrary to historical beliefs that reading might weaken memory by relying on external sources, education and literacy actually enhance cognitive functions. educated individuals typically display better memory capabilities than those who have never been schooled.

in mathematics, the influence of education is equally striking. many indigenous populations without formal education, such as many amazonian indian groups, have only rudimentary mathematical intuitions. these groups often lack a formal counting system and may only differentiate between vague quantities like “few” and “many.” their understanding of numbers and arithmetic is fundamentally different from that of educated societies. for instance, they might not recognize that there is a consistent interval between consecutive numbers, a concept that seems intuitive to trained minds.

education not only fills these gaps but also massively expands basic numerical intuitions. through learning, individuals come to understand that each number has a successor, and this recognition leads to a conceptual reorganization of the number line. this shift from an approximate to a precise understanding of numbers allows for the development of advanced mathematical thinking, a skill set that is crucial in the modern world and forms the basis for further scientific and technological advancements.

the differences observed between educated and uneducated individuals highlight the crucial role of education in cognitive development. it not only enhances innate abilities but also enables individuals to engage with and contribute to the world in more meaningful ways. 

human teaching is uniquely rooted in our understanding of minds#

human teaching is uniquely sophisticated compared to any known animal behavior due to our ability to understand and contemplate the minds of others. this capability allows educators to tailor their instructions thoughtfully, considering what their students do not yet know, thereby optimizing learning experiences. this deep understanding of another’s knowledge gaps and how they process information is what sets human pedagogy apart.

in the realm of teaching, humans continuously adjust their educational strategies based on a dynamic understanding of the learner’s mind. teachers choose specific examples and explanations with the awareness that their students are recognizing these efforts as targeted learning interventions. this “recursive awareness” – teachers know that students know that teachers understand their ignorance – enriches the educational process, making it distinctively human.

this complex interaction does not appear to be mirrored in the animal kingdom, despite instances that might superficially resemble teaching. for example, meerkats show an interesting behavior in which adults help their young handle prey like scorpions by initially removing the deadly stingers and gradually introducing them to more challenging tasks. while this behavior meets some criteria of teaching, including cost to the teacher and accelerated learning for the pupil, it lacks evidence of mutual recognition of knowledge states – what might be called a “theory of mind.”

the teaching seen in meerkats and other animals often seems pre-wired and specific to certain survival behaviors, lacking the generalized, adaptable approach found in human pedagogy. human teachers not only impart knowledge but also engage with students on a mental level, adapting their methods according to the perceived understanding and emotional state of the students. this process involves a continuous exchange of attention, respect, and mutual trust, elements absent in the more instinctual teaching behaviors observed in nature.

furthermore, effective human teaching transcends mere information transmission; it involves crafting lessons that resonate with and are adapted to individual students’ current knowledge and cognitive abilities. this bidirectional flow of understanding and adjustment is essential for profound learning experiences and is something that even advanced artificial teaching systems struggle to replicate.

optimal learning environments unleash children’s cognitive potential#

the human brain’s ability to develop and learn is profoundly influenced by its environment. unfortunately, many children do not achieve their full potential because they do not have optimal learning conditions at home or in school. comparing international educational outcomes reveals significant disparities, with some countries showing declines in academic performance, highlighting the urgent need for better educational strategies.

advancements in the science of learning provide actionable insights that can help reverse trends of declining educational achievement. these insights emphasize the importance of not underestimating children’s capabilities. even at birth, infants are equipped with core skills and a basic understanding of objects, numbers, and language. educational approaches should build on these innate abilities, linking new concepts to children’s existing knowledge to enhance learning.

the early years are critical due to the brain’s heightened plasticity. during this period, children’s brains undergo rapid changes that facilitate the absorption of new information, particularly in language acquisition. introducing a second language early in life can significantly shape cognitive development. moreover, this plasticity extends into adolescence, suggesting that language immersion can be beneficial well beyond the early years.

enriching a child’s environment is crucial. a stimulating environment that includes complex language use, puzzles, and games can significantly enhance cognitive development and prolong brain plasticity. conversations that challenge children to think about complex ideas are particularly beneficial.

the myth of individual learning styles has been debunked by brain imaging, which shows that the neural circuits involved in reading and mathematics are similar across individuals. while each child may have unique motivations and learning paces, the fundamental processes of brain learning are largely the same. educators should focus on understanding each child’s current knowledge level to tailor educational approaches effectively, but ensure foundational skills in language, literacy, and mathematics are solidly acquired.

attention is also critical in learning; students must be able to focus on the material without distractions. thus, educational environments should minimize potential distractions, ensuring that children’s attention is not divided. furthermore, learning is enhanced by making and correcting mistakes. effective learning involves recognizing and addressing errors promptly with constructive feedback, which is essential for adjusting mental models and deepening understanding.

leveraging these principles of learning can transform educational practices. by recognizing and nurturing children’s innate capabilities, providing rich and focused environments, and using strategies that foster attention and correct mistakes, educators and parents can significantly improve educational outcomes. embracing these principles can help all children reach their full potential, shaping a future where every child has the opportunity to succeed academically and cognitively.

final summary#

Conclusion

in this chapter to how we learn by stanislas dehaene, you’ve learned that human learning transcends artificial intelligence by efficiently processing minimal data to form complex concepts and social interactions. education enhances this innate potential, particularly when tailored to individual cognitive levels and supported by an enriched environment. effective teaching requires understanding students’ minds, making human pedagogy uniquely adaptable and deeply interactive.

okay, that’s it for this chapter. we hope you enjoyed it. if you can, please take the time to leave us a rating – we always appreciate your feedback. see you in the next chapter.