¿Cómo aprendemos?: Los cuatro pilares con los que la educación puede potenciar los talentos de nuestro cerebro

by Stanislas Dehaene, Josefina D´Alessio (Translator)

AN ILLUMINATING DIVE INTO THE LATEST SCIENCE ON OUR BRAIN'S REMARKABLE LEARNING ABILITIES AND THE POTENTIAL OF THE MACHINES WE PROGRAM TO IMITATE THEM.

The human brain is an extraordinary machine. Its ability to process information and adapt to circumstances by reprogramming itself is unparalleled and it remains the best source of inspiration for recent developments in artificial intelligence. In 'How We Learn', Stanislas Dehaene decodes the brain's biological mechanisms, delving into the neuronal, synaptic, and molecular processes taking place. He explains why youth is such a sensitive period, during which brain plasticity is maximal, but assures us that our abilities continue into adulthood and that we can enhance our learning and memory at any age. We can all learn to learn by taking maximal advantage of the four pillars of the brain's learning algorithm: attention, active engagement, error feedback, and consolidation.

The exciting advancements in artificial intelligence of the last 20 years reveal just as much about out remarkable abilities as they do about the potential of machines. 'How We Learn' finds the boundary of computer science, neurobiology, and cognitive psychology to explain how learning really works and how to make the best use of the brain's learning algorithms in our schools and universities, as well as in everyday life.

©2020 Stanislas Dehaene (P02020 Penguin Audio

Genres: Nonfiction, Science, Psychology, Neuroscience, Education, Self Help, Brain

352 pages, Paperback

First published September 5, 2018

Reviews

[Atila Iamarino · Rating 5 out of 5]

Temos uma discussão sobre aprendizagem enorme no Brasil, mas pouco da nossa educação leva em conta experimentos controlados que comparam como as pessoas aprendem. Só este passo já é o que faz livros sobre como aprendemos como o ótimo How We Learn: The Surprising Truth About When, Where, and Why It Happens e este livro do Dehaene acrescentarem muito.

Mas Dehaene vai além. Ele aproveita o avanço da inteligência artificial até aqui para comparar como modelos de aprendizagem de máquina funcionam e onde falham, e o que nós fazemos de diferente. Isso é muito bom para avançar a computação e nosso entendimento do cérebro também.

Um dos maiores saltos que a computação ainda precisa fazer e nosso cérebro dá conta muito bem é de quão pouca informação precisamos para aprender. Nós extrapolamos muito com bem pouco. O Google Translate precisou de milhares de livros e documentos, anos e treino de pessoas para chegar em um modelo de tradução parecido com o que uma pessoa pode aprender por conta própria e muito menos leitura. Dehaene vai fazendo esse tipo de comparação, decomposição e explicação o longo do livro todo mostrando como derivamos informação, a importância de outras pessoas para podermos aprender e como funcionam modelos de aprendizagem com recompensa (como nossa curiosidade satisfeita). Tudo isso dando extrapolações de como o ensino pode aproveitar essas habilidades.

[Morgan Blackledge · Rating 5 out of 5]

I took cognitive science at the end of my psychology degree.

It was AMAZING.

The findings and insights of cognitive science are all about how people learn.

In other words, when you learn cognitive science, you learn how to learn.

In a crazy kind of Möbius strip, life imitating art type lived irony, you are literally applying the findings that you are learning, as you are learning them, and they help you learn the things your learning, which is essentially how to learn.

Much of the findings are counter intuitive, and slaughtered many of the sacred cows of liberal education.

Like learning styles for instance.

Not really a real thing.

But there’s more.

Common college truisms like, underlining, re-reading, long study sessions and cramming were all debunked as ridiculously inefficient and ineffective.

Other WAY less intuitive techniques were found to be WAY more helpful including, short study sessions distributed over longer periods of time, systematic self testing, plenty of sleep, exercise, healthy diet, no booze, location based mnemonics, and more.

It was an extremely difficult class, but I took a leap of faith and broke my old study habits in order to apply what I was learning in the class, and BAM!

I CRUSHED the exams.

KILLED the course.

I did significantly better on the difficult, highly technical material, and it took WAY less time.

The time I did spend studying was better quality, and more effortful, but none the less...less.

Less (but better) time spent got me better results.

All because I was studying the correct way, for the FIRST TIME!!!

It was UNREAL.

I remember talking to the instructor and reflecting that it was strange that learning how to learn wasn’t part of every curriculum at every stage of every educational path.

I reflected this same course should have been offered in 9th grade, because it was literally the blueprint for academic success.

I went on to reflect that, at minimum, the course should have come at the very beginning of my college career, not at the end of a masters degree.

Of course the instructor just smiled and said “you’re preaching to the choir dude”.

What was perhaps even more terrifying, was that I was working as an adjunct professor for undergraduates in psychology at that very time, and I had (until that point) received ZERO training in pedagogy, until that very course.

Thats right, I was a college instructor, and I was learning how to learn at close to the end of my career as a college instructor.

WTF!!!

And it gets worse.

The real reason no one ever offered you or I that education in education (call it a meta-education) was that they didn’t get it either.

Most educators, including school teachers and college professors don’t get any specific training in cognitive pedagogy.

I think this is the issue that Stanislas Dehaene is addressing in this book.

And he’s really brilliant, and a really good writer and educator, so...you guessed it...the book is really really good.

You’re not gonna get tons of study tips.

It’s not that kind of book (thank god).

It’s much more meta-level than that.

But it’s easy to apply Dehaene’s talking points to whatever field of study that you’re engaged in, at whatever level.

So whether you’re a teacher or a student, or just a human with the brain, I highly recommend that you get this book, and continue to pursue a basic education in cognitive science.

It will pay dividends in terms of better results and less anxiety regarding learning, for yourself and for who ever will listen to you (unfortunately this probably excludes your kids).

Great Book 📚

[Toni · Rating 5 out of 5]

Written in a very captivating manner, easy to follow and understand, How We Learn discusses one of the most important abilities that dstinguishes humans form the rest of the world- our ability to learn in a conscious way. Stanislas Dehaene is a neuroscientist who has written a number of books to help the general public understand better how our brains use information in order to learn and create.
The amount of information available to us nowadays is staggering. Our ability to select and process input in order to adapt to and enhance our environment as well as using feedback as the most important learning tool are the reasons why we keep surviving and advancing our knowledge. Stanislas Dehaene looks in detail at the biological processes that happen in our brain, and discusses the issue of neuroplasticity and learning at different ages.
As we live in a very technological society, it is not surprising that our interest in the science of learning is partly driven by our desire to develop 'smarter' machines and artificial intelligence. But it is not the only reason why we should keep trying to understand our learning process better. Our life is already calling for a life-long - 360° learner able to cope with the speed with which our world is changing. This is why this thought-provoking book is so important and timely.
Thank you to Edelweiss and Viking for the ARC provided in exchange for an honest opinion.

[Debjeet Das · Rating 5 out of 5]

HOW WE LEARN

• The brain draws its knowledge from its environment.
• Nerve cells posses a remarkable ability to constantly adjust their synapses to signal they perceive.
• Our 23 pair of chromosomes contain 3 billion pairs of the letters A,C,G,T -the molecules adenine,cytosine,guanine and thymine. Information is contained in bits and DNA contains 6 billion bits.
• There are 86 bn neurons and a thousand trillion connections
• The capacity of brain is on the order of 100 terabytes
• Animal that possessed even a rudimentary capacity to learn had a better chance of surviving than those with fixed behaviors & they were more likely pass their genome to next generation.
• Most of what we know know about the world was not given to us by our genes: we had to learn it from our environment or from those around us.
• From making fire & designing tools to agi,exploration etc is the story of humanity of constant self exploration.
• The brain has an extra ordinary ability to formulate hypothesis & select those that fit with our environment.
• The active verification of knowledge, rejection of simple heresay & the personal construct of meanings are essential filter to protect us.
• Mindset plays an important role in learning. Having a deeply entrenched view that anyone can progress is, itself , a source of progress.
• retrieval practice- it is one of the most effective educational strategies. The mere act of putting your memory to test makes it stronger.
• self awareness or meta memory is useful b/c it allows learner to focus harder on difficult items.
• to get information into long term memory ,it is essential to study the material ,then test yourself.Self testing is one of best strategy.
• space out learning- the rule of thumb is to review the info. at intervals of app. 20% of desired memory durations-for instance , rehearse after 2 mnths if you want memory to last about 10 mnths.
• to keep information in memory as long as possible,it is best to gradually increase the time intervals. Start with rehearsels everyday,then review info. after weeks, a month,then a year. This strategy guarantees optimal memory at all points.
• 4 pillars of learning-
1 Focussed attention- it allocates resource to info. it considers most essential. Attention makes relevant info. sensitive & it increases influence on brain.
attention system
• alerting-when to attend
• orienting-what to attend to
• exec. attention-how to process attended info.
• selective attention-select the relevant ,ignore irrelevant.

multitasking:-
• our brain can process only one piece of info. at a time .
• if 2 simple task given simultaneously . 1st one might get an attention & 2nd one might get delayed or forgotten altogether.
How can we multitask?
• Automization frees the conscious workspace by routinizing an activity, we can execute it unconsciously, without tying up brains central resources.



2 active engagement-learning requires an active generation of hypothesis with motivation and curiosity.
• motivation is an essential : we learn well only if we have clear goal and we fully commit to reaching it.
• to better digest new concepts, rephrase them into words or thoughts of your own.
• curiosity- fundamental driver of an organism: it pushes us to act, it encourages us to explore
• the degree of curiosity corelates with activity of nucleus accumbens and ventral tegmental areas, 2 essential regions of dopamine brain circuit.
• the more curious you are about something, the more likely you are to remember it.
• laughing or positivity seems to increase curiosity and enhance subsequent memory.
• curiosity occurs whenever our brain detect a gap b/w what we already know & what we would like to know- a potential learning area

• to maximise what we learn,we have to constantly enrich our env, with new objects that are stimulating enough to not be discouraging.
• as learning progresses ,the expected learning gain shrinks: the better we master a field , the more we reach the limits of what it can offer & less interested we are in it.
• to maintain curiosity, one must always provide brain with stimulants that match their intelligence.
• one can restore desire to learn by offering them stimulating problems carefully tailored to their current levels.
• the neuroscience of motivation is clear: desire to do action x must be associated with an expected reward,be it material or cognitive.



3 error feedback- whenever we are surprised because world violates our expectation , error signal spreads throughout our brain.They correct our mental models,eliminate inappropriate hypothesis ad stabilise the most accurate ones.
• Every error offers an opportunity to learn.
• the quality and accuracy of feedback we receive determines how quickly we learn.
• surprise is one of the fundamental drivers of the learning.No learning is possible without an error signal. organisms can only learn when events violate their expectations. No surprise, no learning.
• being immersed in fear free,stimulating env. can re-open synaptic plasticity.

4 daily rehearsel and nightly consolidation- our brain compiles what it has acquired and transfer it to long term memory, thus freeing neural resources for further learning.Repetition plays an important role in consolidation process.
• learning is more efficient when done at regular intervals, rather cramming an entire session in 1 day, we are better off spreading out the learning. The reason is simple: every night our brain consolidates what it has learned during the day.
• while we sleep,our brain remains active, it runs a specific algorithm that replays the important events it recorded during previous days & gradually transfer them into more efficient compartment of memory
• durng sleep,the neural circuits that we used during preceding day gets reactivated.
• experiments indicate that information you learned is better consolidated the next morning if you had slept while being exposed to another smell.
• If we smell an aroma while we take on new knowledge & then sleep next to source to that same odor, we will find it easier to recall info. at later date.
• practice during the day and sleep during night to reactivate and consolidate what we acquired.


• By constantly attending to probabilities and uncertainties, brain optimises its ability to learn.
• To learn is to progressively form an internal model of the outside world.
• At its core ,intelligence is a process that converts unstructured information to useful and actionable knowledge.
• Through learning ,raw data that strikes our senses turned into refined idea,abstract enough to be reused in a new context.
• Behind the scenes,our sensory areas ceaselessly compute with probabilities & only most likely model makes it to our consciousness.
• It is the brains projection that ultimately give meaning to flow of data that reaches us from our senses.In the absense of an internal model,raw sensory inputs would remain meaningless.
• Learning allows our brain to grasp a fragment of reality that it had previosly missed & to use it to build a new model of the world. It can be part of external reality as when we learn from history,science etc but also brain learns to map the reality internals to our bodies, as we learn to play musical instrument etc.
• 7 definition of what leaning means
1. learning is adjusting the parameters of mental models-it boils down to searching among internal models that best correspond to the external model

2 learning is exploiting a combinatorial explosion
• combinatorial explosion-the exponential increase that occurs when you combine even a small no of possibilities.
• In our brain ,there are 86 bn neurons, each with about 10k synaptic connections whose strength can vary.The space of mental representation that opens up is practically infinite.
• Human bran breaks down the problems of learning by creating a hierarchical,multilevel model
3 Learning is minimising errors
• Brain observe the errors & use them to adjust internal state in directions that they feel is best able to reduce error

4 learning is exploring space of possibilities- random exploration,random curosity & noisy firing plays an essential role in learning .
• characteristic of human species is relentless search for abstract rules, high level conclusion that are extracted from a specific situation & subsequently tested on a new observation

5 learning is optimising a reward function- The network can correct itself by calculating the diff. b/w its response and correct answer. This procedure is known as “supervised learning”.
• reinforcement learning
• act & self evaluate
• actor-critic combination is one of the most effective strategies of contemp. AI, when backed by hierarchical neural network , it works wonders.
• bootstrapping-a neural network can become a world champion simply by playing againsti itself.
• adversarial learning- consists of training 2 opponent system-one that learns to become an expert & other whose sole goal is to make 1st one fail.
• metacognition-the ability to know oneself,to self evaluate,to mentally stimulate what would happen if we acted this way or that way-plays a fundamental role in human learning.
• meta rules allows to search for a possible meaning among object around them, rather tha treating word as verb or adjective.

6 learning is restricting a search space- the more parameter the internal model has, the more difficult it is to find the best way to adjust it.
• curse of dimensionality- having too many free parameter can be curse while system learns ,it is unable to generalise a new situation.The ability to generalise is key to learning.
• one of the most effective interventions ,which can both accelerate learning and improve generalisation is to simplify the models.
• what is learned in one place can be reused everywhere else.
• the system has to tune only a single filter that it applies everywhere ,rather than a plethora of diff. connections for each locations.

7 learning is projecting a priori hypothesis
• exploitation of innate knowledge- convolutional neural network learns better & faster than other type of neural network b/c they donot learn everything.
• what I learn in 1 place can be generalised anywhere else.
• rather than learning from scratch ,it is more effective to rely on prior assumption that clearly describe the basic laws of domain that must be explored & integerate this into very architecture of system.
• to learn is to eliminate.

why our brain learn better than current machines?
1. unlike a computer, we posses the ability to question our beliefs & refocus our attention on those aspects of image that dont fit with our 1st impression
2. human learning is not just setting a pattern recognition filter, but also forming an abstract model of the world.
3. social learning- we learn a lot from our fellow human through language.
• what human do exceedingly well is to integerate new information with an existing network of knowledge.
4 the ability to discover a general laws that lie behind specific cases-
• the ability to make infinite use of finite means characterises human thought.
• humans can integerate piecewise knowledge into single formula & integerated pattern of activity happens in “Brocas area”.

How scientific reasoning work?
1 researchers apply a simple logic-they state several theories, unravel the web of ensuing predictions and eliminate the theories whose predictions are invalidated by experiments and observations
2 In each of our mind,ignorance is gradually erased as our brain successfully formulates increasingly accurate theories of the outside world through observations.

3 learning is reasoning like detective-going to hidden cause of phenomenon, in order to deduce the most plausible models that govern them.

3 bayesian approach -
• To learn is to be able to draw as many inference as possible from each observations
• tracing every observation to its most plausible cause allows us to return to foundation of logic.
• the bayesian theory allows us to travel from observation to causes.The theory explains how to update our belief after each observations.
• the brain should calculate an error signal : the diff. b/w what the model predicted and what has been observed. The bayesian algorithm indicates how to use this error signal to modify internal models of the world.
• when you eliminate the impossible,whatever remains ,however improbable, must be the truth.

4 our adult judgement combines 2 levels of insights-innate knowledge of species and our personal experiences.

5 learning is most effective when we have vast spaces of hypothesis, a set of mental model with myriad settings to choose from & sophisticated algorithm that adjust those settings a/c to data received.

self organisation
1 self organisation is ubiquitous in the developing brain.
2 grid cells are ‘GPS of brain’- it forms a network of triangle which grouped together to form a network.
3 hexagon being frequently produced from giraffe skin to beehives
• hexagon is common b/c it is a shape that fills a plane with equal size units and leaves no wasted space.It is most mechanically stable
• brain contains a neuronal map with hexagonal symmetry.

4 foundation of our core brain circuits arise through self organisations by bootstrapping themselves from database generated inside the system.

5 the knowledge of mathematical thing is almost innate in us. For people who are illiterate also know how to count and reckon.

neuroplasticity
• in addition to reinforcement of pre established pathways,new pathways are created by the progressive growth of terminal dendritic and axonal process.
• neurons, synapses and microcircuits that they form are material hardware of brain plasticity, they adjust each time we learn.
• as we learn, all elements can change: the presence ,number and strength of synapses and neurons.
• our brain contains about 100 trillions of synapses & synapses are genuine nanoprocessors of the brain.
• hubbs rule-when 2 neurons are activated at same time or in short successions,their connections strengthens
• emotional circuits of our brain considered most significant.
• when we learn, the explosion of new synapses forces the neurons to grow additional branches.
• to make & break a few million synapses/second , it requires a balanced diet,oxygenation & physical exercises.
• vitamin B1 which contains thiamine is imp. for brains health.
• the acquisition of novel skill doesnt require a radical rewriting of cortical circuits but a repurposing of existing org.

memory
1 episodic memory- records the episodes of our daily lives
• neurons in hippocampus memorize the context of each event
• they encode where,when,how and with whom things happened
• hippocampus is involved in all kinds of rapid learning.

2 semantic memory-memory donot stay in hippocampus forever.
At night,brain plays them back and moves them to a new location within cortex.There,they are transformed into permanant knowledge.

3 procedural memory- when we repeat the same activity over and over again .Neurons gets strengthened and information flows better in future.

[Gretchen Rubin]

A friend recommended Dehaene's work, and I found this book so interesting that I immediately got my hands on his other work.

[Sabin · Rating 4 out of 5]

This one’s a hidden gem. Ok, the first third is a bit underwhelming. In the beginning, the author talks about general mechanisms by which we encode information about the world and how machine learning computer algorithms cannot match the human brain because they do not employ these mechanisms. Yet. He does point out that they can't do this stuff yet. But the repetition is tedious. And we’re talking about machine learning. So even though I think I listened to an updated version of the book (2020), at the current rate of AI research some content is bound to be outdated in one or two years.

The beauty and importance of the book lies in its up-to-date descriptions and theories about how humans learn and what new information recent neural imaging studies bring to this topic. His conclusions focus on human learning and development. The book discusses language acquisition, mathematical reasoning, brain plasticity and the developmental stages of the brain. He debunks myths about learning while sleeping and discovery learning (letting children stumble upon truths by themselves, unaided) while at the same time championing a 4-pronged learning approach which consists of attention, active engagement, error correction and consolidation. And yes, flashcards do work. But if you find them tedious, then you’re going about learning in the wrong way.

A great resource for parents and educators. Could be worth a shot even for machine learning engineers who want to prove the author wrong or at least force him to revise the book on one account.

[Юра Мельник · Rating 5 out of 5]

Приділено дуже багато уваги когнітивним дослідженням із навчання у немовлят і дітей до 6 років. Ця книжка дуже корисна для батьків і педагогів

[Emre Sevinç · Rating 4 out of 5]

This book strikes a special chord with me: I started my formal cognitive science training and research activities almost 20 years ago, and reading the cognition-related developments in brain sciences that occurred in the last 20-25 years brings a unique type of excitement.

I haven't been involved with academic research for a long time, alas, but professionally I'm under daily pressure to learn new aspects of technology and apply them in various settings, and on top of that, I'm trying to teach some of the things I know to my two sons, while observing some of their developmental struggle with the complexities of our world. Therefore, reading this book was a delight, because it not only summarizes the state-of-the-art of learning and teaching, but also sets the evidence-based path for future learners and teachers, that is, us.

Even if you're not into the scientific aspects of developmental neuropsychology, or how some of the cutting-edge research in machine learning and artificial intelligence are inspired by the neurological mechanisms in the brain, you'll probably get something useful and practical out of this book because some sections will force you to think very consciously about the basic and critical mechanisms of attention, memory, engagement with a topic, giving and receiving feedback during learning something new, and other relevant aspects of your life.

Needless to say, the message of the book is even more important for actual teachers, trainers and young pupils, as well as the administrators responsible for shaping the future of education, and I strongly recommend reading this book with a critical perspective if you're professionally involved with such activities.

I take 1 star, and give it 4 stars, because of the author's over-enthusiasm and exaggerated analogies with modern, artificial deep-learning systems. This topic deserves more nuance, and subtlety, not TED-like simplification and sometimes outright misleading phrases. I'm sure Prof. Dehaene is very well aware of dangers of forcing such analogies, and I don't think he'd be keen on claiming strong correspondence between the intricacies of human minds/brains and over-hyped AI systems (yes, that attitude gets a little bit on my nerves, not only emotionally but also philosophically).

Long story short, if you want to learn about our best and current understanding of learning mechanisms that happen in the brain, especially starting from birth, and scientifically validated ways of learning and teaching better, you can't go wrong with this highly readable book.

[Oleksandr Golovatyi · Rating 5 out of 5]

Швидкочитання. Як читати втричі швидше. (promo)

Кращі нотатки з книги:

"Якщо ми не знаємо, як люди вчаться, то звідки можемо знати, як їх учити? (Л. Рафаел Райф, президент МТІ)"

"Чому еволюція взагалі вигадала навчання?"

"каркас палацу зведений за інструкціями архітектора (нашого геному), а деталі залишилися на розсуд головн��го інжерена (середовища). Якщо настільки скрупульозно прошити мозок неможливо, то щоб доповнити роботу генів, необхідне навчання"

"Більшість людських знань про світ не успадковані, а засвоєні із середовища, від інших особин."

"Освіта - головний акселиратор людського мозку. Без неї таланти нейронів ніколи не перетворяться на шліфовані діаманти."

"Уміння вчитися - чи не найважливіша передумова академічних досягнень."

"Навчання без роботи над помилками не реальне, але багато дітей втрачають упевненість і допитливість через те, що їх радше карають, ніж виправляють."

"Негативні емоції придушують навчальний потенціал, а безпечне середовище здатне заново відімкнути браму нейропластичності. Освітній прогрес неможливий без рівноцінної уваги до емоційної і когнітивної грані діяльності мозку - і сучасна нейронаука вважає їх обох ключовими інгредієнтами навчального коктейлю."

"увага - можливість вибирати і підсилювати релевантну інформацію"

"сон - алгоритм для синтезу свіжих знань"

"навчатися - це поступово формувати у мережі нейронів або на кремнієвій мікросхемі внутрішню модель зовнішнього світу"

""Чотири стовпи навчання"... регулють нашу здатність учитися. Перший з них - увага. Це комплекс нейронних мереж, що відбирають, підсилюють і поширюють сигнали, які ми вважаємо важливими, стократ збільшуючи їхній вплив на пам'ять. Другий стовп - активне залучення. Пасивний організм не вчиться, тому що навчання вимагає генерації гіпотез, умотивованості й допитливості. Третій стовп (і зворотний бік активного залучення) - корекція помилок. Щоразу, коли світ дивує, не виправдовуючи наших очікувань, мозок сигналізує про помилку. Ці сигнали виправляють ментальні моделі, виключають недоречні гіпотези і фіксують найбільш точні. І четвертий стовп - це консолідація. З часом наш мозок компілює засвоєні знання і навички й переводить їх у довготривалу пам'ять, вивільняючи ресурси для подальшого навчання. У процесі консолідації засадничу роль відіграє повторення. Навіть уві сні мозок не відпочиває, а на більшій швидкості відтворює свої попередні стани і перекодовує вивчене вдень."

"Кілька простих ідей щодо гри, цікавості, соціалізації, концентрації і сну здатні примножити найважливіший талант нашого мозку - навчання."

"навчатися - це формувати внутрішню модель зовнішнього світу" !

"Дослідження випадковостей , стохастична допитливість і "шуми" нервових імпульсів відіграють важливу роль у навчанні Homo sapiens."

"метакогніція - здатність розуміти й оцінювати себе, подумки бачити, до чого приведе той чи інший учинок."

"І в людей, і в машин навчання починається з апріорних гіпотез. Ми проектуємо їх на вхідні дані і таким чином добираємо найбільш слушні в поточних обставинах. Як писав Жан-П'єр Шанже в бестселері "Нейронна людина" (1983): "Учитися - це відсіювати""

"За оцінкою психолінгвіста Еммануеля Дюпу, за рік життя у французькій сім'ї дитина чує від 500 до 1000 годин усного мовлення."

"навчаючись, люди не просто налаштовують фільтр для розпізнавання шаблонів, а формують абстрактну модель світу. Наприклад, опановуючи читання, наш мозок утворює узагальнені уявлення про кожну літеру алфавіту, і завдяки їм ми згодом упізнаємо букви у різних варіаціях і можемо створювати нові"

"Машини пожирають інформацію, а люди - використовують. Наш вид бере максимум з мінімуму даних."

"Вербалізованість - невід'ємна характеристика усвідомлених знань. Тільки-но людина достатньо чітко розуміє якийсь феномен, ментальна формула резонує з мовою мислення, і ми можемо використовувати слова, щоб повідомити іншим."

"Навчатися - це вписувати нові знання в уже наявну схему."

"наш мозок має невичерпну здібність конструювати формули, і вони працюють як своєрідна ментальна мова. Він виражає поняття нескінченності завдяки абстрактним функціям заперечення і квантифікації, що існують у мові... Американський філософ Джеррі Фодор (1935-2017) теоретично обгрунтував цю властивість. Його гіпотеза розглядає людське мислення як сукупність символів, які поєднуються відповідно до законів мови мислення. Оскільки ця мова має рекурсивну природу, її можливості невичерпні."

"навчатися означає керувати ментальною ієрархією правил і старатися максимално швидко виснувати найбільш універсальні. Ті, які пояснюють значний обсяг спостережень."

"Перше метаправило, що сприяє засвоєнню лексики: серед припущень, сумісних з даними, завжди обирати найпростіше."

"Друге метаправило - трюїзм: мовець звертає увагу на об'єкт, про який говорить. Щойно немовлята схоплюють цю закономірність, вони здатні суттєво звузити діапазон пошуку значення... Дитині ж вистачить простежити за поглядом чи пальцем матері, аби зрозуміти, про що вона говорить. Цей неабиякий механізм засвоєння мови має назву спільна увага."

"Третє метаправило, яке допомагає дітям швидше вивчати слова, - принцип двосторонньої ексклюзивності. Простіше кажучи: один предмет - одна назва."

"Вчитися - це міркувати як умілий статистик і вибирати з-поміж альтернативних теорій ту, яка найповніше відповідає наявним даним, а отже, має найбільше шансів бути правильною"

"ймовірність - не що інше, як кількісне вираження нашої невпевненості"

"Статистик Едвін Томпсон Джейнс, ... книжку "Теорія імовірності: логіка науки""

"судження дорослої людини поєднує у собі два джерела: вроджені знання виду (правдоподібні гіпотези, успадковані у ході еволюції, - баєсівці називають їх апріорне) і власний досвід (висновки, яких ми доходимо протягом життя, - апостеріорне)."

"Усі знання людини стоять на двох китах: перший - апріорні припущення, які передують взаємодії із середовищем, другий - здатність упорядковувати їх відповідно до апостеріорної імовірності після опрацювання реальних даних."

"ще у IV ст. до н.е. Платон у "Державі" писав, що кожна душа з народження має два складні механізми: силу знань і орган, який відповідає за навчання"

"Об'єкти, числа, ймовірності, обличчя, мова ... це далеко не повний перелік вроджених здібностей немовлят."

"З дуже раннього віку немовлята мають інтуїтивні знання з арифметики, фізики і навіть психології. Щоб виявити їх, дослідники показують малюкам очікувані й несподівані сценки і замірюють тривалість погляду (здивована дитина дивиться довше)"

"Численні контрольовані експерименти цього типу не залишають сумнівів, що малюки від народження, не рахуючи, інтуїтивно розпізнають приблизну кількість і візуальної, і аудіальної інформації."

"на останніх місяцях вагітності мозок плода росте й розпізнає, хоч і неусвідомлено, деякі звукові послідовності й мелодії."

"За кілька місяців (шість для голосних, дванадцять для приголосних) мозок немовляти переглядає початкові гіпотези і залишає ті фонеми, які вживає його оточення."

"у 6 місяців немовлята вже виокремлюють найчастотніші в їхньому світі слова: мама, тато, ляля, каша, рука, їсти, пити тощо."

"лінгвіст Ноам Хомський мав рацію, постулюючи, що кожен представник людського виду народжується з "механізмом засвоєння мови" - спеціалізованою системою, яка автоматично спрацьовує в перші роки життя."

"дослідження показало, що у двомісячних дітей, які слухають речення рідною мовою, активуються ті самі ділянки мозку, що й дорослих."

"З першої хвилини життя слухова інформація активує слухову кору, візуальна - зорову, а дотик - ділянки, пов'язані з тактильними відчуттями. Нам не потрібно цього навчатися. Поділ кори головного мозку на окремі ділянки для кожного чуття забезпечують гени. Він властивий усім ссавцям."

"Мовні магістралі. Активніть почергово перетікає з ділянки в ділянку, тому що між ними існують зв'язки. На сьогодні ми частково розуміємо, які нейронні шляхи сполучають мовні ділянки у дорослих. Неврологи виявили, що дугоподібний пучок - велкий "кабель" з мільйонів нервових волокон - з'єднує ділянки скроневої і тімяної часток з лобовою, зокрема зоною Брока. Поява цього жмута - віха в еволюції мови. У лівій півкулі, яка у 96% правшів підповідає за обробку мови, дугоподібний пучок помітно більший, ніж у правій. Ця асиметрія властива виключно людям і не притаманна іншим приматам, навіть нашим найближчим родичам - шимпанзе. І знову ж таки ця анатомічна особливість вроджена, а не набута. Якщо ми проаналізуємо зв'язки у мозку новонародженого, то виявимо там не тільки дугоподібний пучок, а й усі основні нервові волокна, які сполучають кіркові і підкіркові структури."

"За розпізнавання облич у всіх індивідів відповідає одна і та ж ділянка плюс-мінус кілька міліметрів. Так формується один з найбільш спеціалізованих кіркових модулів, у якому до 98% нейронів реагують на лиця й ігнорують інші зображення"

"Німецький нейродослідних Андреас Нідер продемонстрував, що цей регіон (тім'яна частка) кори містить нейрони, спеціалізовані на конкретні числа. Деякі клітини реагують на один предмем, інші - на два, ще інші - на три і так далі... Це дозовляє припустити, що кількісні модулі - вроджені структури, що пізніше зазнають впливу середовища."

"На сьогодні штучний інтелект - практично синонім великих даних, тому що машини пожирають гігабайти інформації і лише після цього демонструють які-не-які результати. Людський мозок не вимагає стільки досвіду. Його головні вузли, модулі з базовими знаннями нашого виду розвиваються спонтанно, здебільшого (а то й виключно) завдяки внутрішнім стимуляції."

"Синапс - це місце зустрічі двох нейронів, точніше, аксона одного нейрона з дендритом іншого."

"Нейроні, синаспси і сформовані ними мікромережі - матеріальна основа пластичності мозку, яка змінюється під час будь-якого навчання."

"приписати пам'ять одному відділу мозку неможливо. Вона розподілена по більшості, а то й усіх мережах нейронів, що модифікують синапси у відповідь на регулярні патерни активності. Але не всі ділянки відіграють однакову роль. Попри нечітку термінологію, яка продовжує розвиватися, учені розрізняють мінімум чотири види пам'яті."

"Робоча пам'ять здатна кілька секунд утримувати ментальну репрезентацію в активній формі. Вона постає з бурхливих імпульсів у тім'яній і префронтальній корі, які у свою чергу підтримують діяльність нейронів у периферійний відділах. Робоча пам'ять згодиться, щоб тримати у голові номер телефону. Доки ви набираєте цифри на клавіатурі смартфону, нервові імпульси підсилюють одне одного, тим самим затримуючи інформацію у свідомості."

"інформація не затримується у робочій пам'яті довше, ніж на кілька секунд. Тільки-но ми відволікаємося, блок активних нейронів гасне. У буфері мозку зберігається тільки найсвіжіша і найактуальніша інформація."

"Епізодична пам'ять. Глибоко всередині півкуль головного мозку розташований гіпокамп, який фіксує плин нашого життя. Нейрони гіпокампа запам'ятовують життєві епізоди разом з контекстом. Вони записують де, коли, як і з ким трапилася ситуація, і за допомогою синаптичних змін зберігають цю інформацію на потім."

"Найновіші дослідження наштовхують на думку, що гіпокамп бере участь у всіх видах швидкого навчання. Якщо виучувана інформація унікальна - нехай це незвичайна подія чи цікаве наукове відкриття, - нейрони гіпокампа "вистрілюють" у специфічній послідовності."

"Семантична пам'ять. Спогади не затримуються у гіпокампі назавжди. Уночі мозок ще раз програє їх і переносить у нову локацію кори. Там вони перетворюються у постійні знання. Мозок витягує інформацію з пережитого досвіду, узагальнює її і додає у величезниу бібліотеку відомостей про світ... відбувся перехід епізодичної пам'яті у сематничну. Одинична ситуація трансформується у довговічні знання, а відповідний нейронний код переноситься з гіпокампа у кіркові мережі."

"Процедурна пам'ять - це стиснені несвідомі записи патернів повсякденної діяльності. Гіпокамп не має до них стосунку, тому що постійна практика направляє спогад на зберігання у комплекс підкіркових нейронних мереж, що називається базальні ядра."

"Гра на музичному інструменті, читання, жонглерство, таксування у мегаполісі - опанування цих занять веде до явного збільшення товщини і сили зв'язків кортикальних ділянок. Енергійно використовуючи магістралі мозку, ми розвиваємо їх."

"Синапси - втілення навчання, але не єдине джерело змін у мозку. Коли ми вчимося, стрімкий розвиток нових синапсів часто стимулює нервову клітину обростати додатковими гілками аксонів і дендритів."

"Харчування - важливий елемент навчання."

"Перші два роки життя дерева нейронів рясно буяють, утворюючи у мозку непрохідні хащі. Кількість синапсів у дворічного малюка чи не вдвічі більша, ніж у дорослого. Після цього починається етап підрізання дендритного покрову під впливом нейронної активносіт. Корисні синапси залишаються і примножуються, а непотрібні - зникають."

"Немовлята несвідомо збирають статистику почутих звуків, і мозок звикає до актуального розмежування фонем. Приблизно у дванадцять місяців процес дає вичерпну картину рідної мови (чи мов) і зупиняється. Набір фонем, які ми здатні розрізнити, практично кам'яніє, заморожуючи здатність до подального навчання."

"здібності до граматики повільно знижуються протягом усього дитинства і різко падають після сімнадцяти. Щоб устигнути вивчити мову, до того як вікно закриється, учені рекомендують починати до десятирічного віку."

"Раннє дитинство - ключовий етап розвитку поняття синтаксичного переміщення. Якщо протягом першого року життя мозок не здобув лінгвістичного досвіду, вікно нейропластичності для цього аспекту граматики зачиняється."

"навчання - це модифікація наявних мереж у мозку."

"Екзаптація - це еволюційна поява нового застосування старого механізму."

"Переробка нейронної мережі - це перетворення функцій засобами навчання і освіти, без впливу генетичних модифікацій."

"Задовго до спроб освоїти читання у дітей формується складна зорова система, що дозволяє розпізнавати і називати предмети, тварин і людей. Малюки впізнають об'єкт, незалежно від його розмірів, розміщення і ораєнтації у тривимірному простоір, і знають, як пов'язати його з відповідною назвою. Читання рециклює мережу, що присвоює імена зображенням. Під час навчання грамоти у ній виникає ділянка, яку ми з колегою Лораном Коеном прозвали "зоною візуальної форми слова". Тут сконцентровані всі знання про літери і їх послідовності. Справжнісінька поштова скринька мозку. Завдяки їй ми впізнаємо слово, байдуже на розмір, розміщення, шрифт і регістр - верхній чи нижній. У мозку всіх грамотних людей ця зона розміщена однаково (плюс-мінус кілька міліметрів) і відграє подвійну роль: спочатку ідентифікує рядок знайомих символів, а тоді, використовуючи прямий доступ до мовних ділянок, швидко перекладає їх у звуки і значення."

"Мозок неграмотних сильніше реагує на обличчя. Що вправніше ми читаємо, то менше активуються мережі розпізнавання лиць у лівій півкулі, де розташована ніша друкованих слів... Мозок наче звільняє простір для літер, і читання перебиває основну функцію цієї зони - розрізнення облич і об'єктів. Але ми не розучуємося впізнавати лиця, отже, не проганяємо це вміння геть. Як показали дослідження, разом з рівнем грамотності зростає чутливість до облич у правій півкулі. Мова виселяє цю функцію зі своїх володінь, і лиця знаходять прихисток на протилежному боці."

"Щойно малюки починають складати літери у слова, у лівій півкулі виникає зона "візуальної форми слова". Водночас у симетричній ділянці правової півкулі посилюється реакція на обличчя. Ефект настільки потужний, що комп'ютерний алгоритм безпомилково визначить, чи навчилася дитина читати, з реакції її мозку на обличчя."

"Як передбачає гіпотеза нейронного рециклювання, читання конкурує з попередніми функціями зорової кори, у цьому випадку - з розпізнаванням облич. Підвищення рівня грамотності (від повністю неписьменних людей до досвідчених читачів) тягне за собою сильнішу реакцію на письмо. Водночас активність, викликана обличчями, переміщуються з лівої півкулі у праву."

"Навчаючись читати, люди блокують розростання зони розпізнавання облич у лівій півкулі."

"Коли діти йдуть у перший клас і починають читати, літери проникають у неспеціалізовані ділянки і рециклюють їх. Спершу я думав, що читання загарбує відведені під обличчя ділянки, але це не так. Воно просто поселяється на пустирі по сусідству.. Експансія першого руйнує другий, і оскільки літери заселяють домінантну для мови ліву півкулю, обличчям не залишається нічого іншого, ніж переїхати на правий бік."

"щоб глибоко рециклювати зорову кору і стати читачами-експертами, необхідно максимально використати дитячий період нейропластичності."

"Нейровізуалізація демонструє, що репрезаентації математичних об'єктів заполоняють бокові потиличні ділянки обох півкуль, і після тренувань ці зони реагують на алгебраїчні вирази значно сильніше, ніж у нематематиків. А ще ми знову спостерігаємо конкуренцію з обличчями: цього разу чутливі до них шматки слабнуть в обох половинах мозку. Тобто якщо грамота просто виживає обличчя з лівої півкулі у праву, то інтенсивна робота з числами і рівняннми конфліктує з ними повсюди."

"Дитячий мозок - структурований і пластичний водночас. Немовлята народжуються з цілим арсеналом спеціалізованих ділянок мозку, які відібрані мільйонами років еволюції і запрограмовані генетично. Самоорганізація мозку дарує малюкам глибокі інтуїтивні знання у важливих сферах: чуття фізики, яка впливає на об'єкти і їхній рух, уміння орієнтуватися у просторі, розуміння чисел, імовірності і математики, тягу до інших людських істот і навіть лінгвістичний талант."

"якість зворотнього зв'язку від учителя - найефективніший важіль впливу на успішність учнів."

"Щоб отримати найкращі результати несвідомої нічної роботи, слід повторити урок чи перечитати завдання прямо перед сном."

"Оскільки цикл сну підлітків зрушений, не варто будити їх рано-вранці!"

"Пізнавши самих себе, ми маємо шанс використати потужні навчальні алгоритми мозку на повну. Діти тільки виграють від розуміння чотирьох стовпів навчання: уваги, активного залучення, корекції помилок і консолідації. Їх можна підсумувати чотирма слоганами: сконцентруйся, будь активним на уроці, вчися на власних помилках і тренуйся щодна - спи щоночі."

Таблиці для тренування швидкого читання (promo)

[Oleh Bilinkevych · Rating 4 out of 5]

Дисить дивним було рішення в автора починати книгу з історії розвитку штучного інтелекту, і ця частина тексту найнудніша. Тільки починаючи з 2 частини книга повертається у потрібне русло, власне, з цього місця можна було і починати книгу. Але , але. Тим не менш, це пізнавальний текст, хороший науково-популярний нонфік без води👌🏻

[Mircea Petcu · Rating 5 out of 5]

O definiție a învățării pe care doar un neurocercetător poate să o formuleze: a învăța înseamnă a forma un model intern al lumii exterioare.

John Locke, marele empirist englez, a spus că omul se naște cu mintea tabula rasa. Mintea e ca o foaie albă de hârtie pe care se imprimă prin experiență informațiile despre lumea exterioară. Departe de adevăr. Ne naștem cu o cunoaștere sofisticată a obiectelor, a numerelor, a oamenilor și a limbilor. "Nu putem învăța totul" Steven Pinker. Apelăm la întelpciunea oferită de milioanele de ani de evoluție prin selecție naturală.

Mi se pare plauzibilă ipoteza reciclării neuronale. Matematica reciclează circuitele pentru aproximarea numerelor, iar cititul reciclează circuitul vederii și al limbajul vorbit.

Recomand

[Virginia MD · Rating 4 out of 5]

A must read for teachers, parents, and students of all ages

I was eager to read the latest book by neuroscientist Stanislas Dehaene because I found “Reading in the Brain" and “Consciousness and the Brain" to be quite compelling. In his new book Dr. Dehaene explores the latest research about how our brains learn. More importantly, he explains how these principles can be applied in the real world of education.

Babies are incredible “learning machines” but they are not born as blank slates. Thus, Dehaene argues that learning is “never nature or nurture, but always both.” His discussion of how human learning differs from the so-called “deep learning” of current computers will be of special interest to anyone who wonders whether we are about to be surpassed by Artificial Intelligence (AI).

After describing the fascinating science of learning Dehaene shares what he calls “the four pillars of learning,” which are the key principles he feels must be used to improve our educational efforts. These four pillars are Attention, Active Engagement, Error Feedback, and Consolidation. Dehaene gives a clear explanation of each pillar so that they can be applied by educational leaders, parents, teachers, and even individuals who strive to be life-long learners.

I just posted an interview with Dr. Dehaene at brainsciencepodcast.com. You can listen on my website or in your favorite podcasting app.

[Peter Gelfan · Rating 5 out of 5]

This is an important and fascinating book for parents, teachers, education administrators and bureaucrats, and anyone who wants to learn anything, including children, teens, college students, people starting a new profession, and retirees hoping to master new skills. For much of his career, the author, a brain scientist, has researched learning and education over the full spectrum from brain imaging and function to educational techniques and results. The book covers the same span.

Dehaene punctures a number of long-standing myths about learning and education. He explains the marvelous acquisitiveness and plasticity of the baby brain, which is far more active than our pared down and more settled adult brains. Sleep turns out to be vital throughout life but especially for young people, because while we sleep, parts of our brains rev up to consolidate and streamline the takeaway from the events and newly acquired knowledge of the day. Near the end, the book outlines a number of thoroughly tested educational principles and techniques that any school, student, or parent can use to great advantage—and don’t worry, it’s nothing weird, the best teachers you ever had already used some of these methods.

The lively, clear prose is easy to read and free of academese. Fittingly, the author makes it easy for us to learn about learning.

[Thanh Tuyen NT · Rating 5 out of 5]

A book that brilliantly takes stock of the state of science in terms of learning both about the human brain and artificial intelligence.

In a very fluid way, the author makes a parallel between the human brain and the artificial intelligence and especially how the research of the former influenced the technical advances of the latter.

We learn about how our brain works and how the engineers in artificial intelligence tried to apply the latest discoveries in neuroscience, for example, error feedback or alternating phases of "sleep" and active learning. But today, our brain is still much more capable than computer's brain and the brain of our child can still do much more amazing things.

If you can read it in French, please do so, this is a must read not only for people who are interested in learning, parents, teachers, educators, tutors... but also for everybody else because each of us has to be a lifelong learner in order to thrive in our ever-changing world.

[Steve · Rating 3 out of 5]

Maybe 3.5 stars. Interesting stuff about how the learning process works within the brain, with some contrast/compare of computer learning algorithms. Some suggestions about how schools should restructure educational techniques in light of knowledge about how learning works.

I felt disappointed that there was very little discussion about how what is learned is actually encoded in the brain for future retrieval. The simple fact is that scientists just don’t know how this works, but I would have loved to hear some intelligent informed speculation. Not sure if this is the cautious scientist sticking to scientific facts, or if the mysteries of how knowledge is stored is still so unfathomable that it’s just not worth speculating about. But I guess I expect that a book about how the mind learns would addressthis situation, if at least to say that nobody has any real idea how information is encoded in the brain.

[Kitty · Rating 4 out of 5]

olen viimasel ajal õppimise ja aju kohta nii palju lugenud, et suur osa selle raamatu sisust ei tulnud otseselt uudisena. ikka needsamad asjad, mida meile räägivad Jaan Aru ja Grete Arro ja teised maailma juhtivad teadlased - efektiivseks õppimiseks tuleb ise avastada ja kaasa mõelda, vigade tegemine õpetab, testimisest on abi, aga hinnetest mitte, vahepeal peab magama ka.

põnev oli see osa, kus räägiti vastsündinu ajust ja et mis seal juba varuks valmis on, kui laps sünnib - selgub, et põhilised arusaamad objektidest, arvudest, tõenäosustest ja inimpsühholoogiast on kõik juba paigas, neid ei hakata nullist õppima.

mulle tegi rõõmu, et pärast seda, kui iga õppimisteemalise teadmiskillu juures olin rõõmustanud, et jaa, just seda kasutab Duolingo ka keeleõpetamisel, namecheckiti Duolingo raamatus kenasti ära. ma ei kujuta seda ise ette, et nad on teadusega kursis ja teevad õiget asja!

[Marko Kivimäe · Rating 5 out of 5]

Ütlen kohe alguses ära, et minu silmis on see kohustuslik kirjandus neile, kes teisi õpetavad. Ning ma ei räägi siin ainult õpetajatest, räägin tegelikult sisuliselt meist kõigist, eriti lapsevanematest. Kui tsiteerida raamatu enda lõppu: "Just nagu meditsiin põhineb bioloogial, peab haridus rajanema süstemaatilisel ja rangel teadusuuringute ökosüsteemil, mis toob kokku õpetajad, õpilased ja teadlased, kes üheskoos peatumata otsivad tõhusamaid, tõenduspõhiseid õppimise strateegiaid."

Autori kohta saab raamatu tagakaanelt lugeda: "Stanislas Dehaene on Saclay ajukuvamise keskuse kognitiivse neurokuvamise osakonna direktor ja Collège de France’i eksperimentaalse kognitiivse psühholoogia professor."

Õpime me kõik, raamatu alguses ongi kohe ka juttu väiksest aga tublist ümarussist, nematoodist, keda mõnes seltskonnas nimetatakse varb- ehk sireussiks. Uurimislaborites on ta viimasel ajal tihe külaline, kuna ussike on ühtepidi piisavalt lihtne, teistpidi suudab omastada, kohaneda ehk siis õppida, olles samal ajal väga hea partner teadlastele. Inimeste puhul võikski Homo sapiens asemel kasutada väljendit Homo docens - liik, mis ise ennast õpetab. Muidugi on omaette küsimus, kuidas õpetada ja õppida nii, et ka midagi külge jääks. Sest eriti just laste puhul:

"... kuigi vigade tagasisidestamine on hädavajalik, kaotavad paljud lapsed enesekindluse ja uudishimu, sest neid pigem karistatakse vigade eest, kui aidatakse neid parandada. Kogu maailma koolides tähendab vigade tagasisidestamine sageli karistamist ja stigmatiseerimist - ja selles raamatus on mul edaspidi päris palju öelda koolis pandavate hinnete rolli kohta selle probleemi põlistamisel. Negativsed emotsioonid purustavad aju õppimise potentsiaali, samal ajal võib hirmuvaba keskkond taasavada aju neuronaalse plastilisuse väravad. Hariduses ei toimu mingisugust progressi, kui samal ajal ei võeta arvesse aju emotsionaalset ja kognitiivset tahku - tänapäeva kognitiivses neuroteaduses peetakse mõlemat õppimise kokteili tähtsaimateks koostisosadeks."

Tänapäeval küll kasutatakse koolides juba ka kujundavat hindamist, kus kiidetakse sõnaliselt, samuti antakse soovitusi, et mida teha paremini. Pikemalt võib lugeda näiteks Õpetajate Lehest: https://opleht.ee/2017/11/kujundav-hi... . Või siis siitsamast raamatust, kus viimases kolmandikus on juttu õppimise neljast tugisambast (tähelepanu, aktiivne kaasatus, vigade tagasisidestamine ja konsolideerimine).

Kui veel rääkida lastest siis:

"Valge lehe eeldus on selgelt vale: lapsed sünnivad tähelepanuväärsete tuumteadmistega, küllusliku universaalsete eelduste kogumiga keskkonnast, millega nad hiljem kohtuvad. Nende ajuvõrgustikud on sünnihetkel hästi organiseeritud ja annavad tugeva intuitsiooni kõikvõimalikes valdkondades: objektide, inimeste, aja, ruumi, arvude suhtes. Laste statistilised oskused on tähelepanuväärsed - nad tegutsevad nagu tillukesed teadlased ja keerukas õppimisvõime lubab neil koondada oma teadmisi kõige kohasemateks maailmamudeliteks."

Või kui rääkida keskendumisvõimest, siis kohati kurdetakse, et tänapäeva lapsed ei suuda enam millelegi keskenduda ning on kleepunud nutiseadmete külge. Kui hetkeks mõtelda, mida just kirjutasin, siis on ju aru saada, et see on poolenisti väär. Ilmselgelt suudavad lapsed väga hästi keskenduda, kui kaovad võimalusel tundideks nutiseadmesse ära. Sama arvutimängudega. Seega keskendumisvõime on väga selgelt alles, märksa olulisem küsimus on, kuidas lastele anda arendavamat tegevust. Ehk:

"Õnneks on palju muid viise, kuidas kutsuda esile virgutussüsteemi mõju, toetudes samal ajal aju sotsiaalsele tajule. Õpetajad, kes suudavad õpilasi köita; raamatud, mis lugejaid kaasa haaravad, ning filmid ja näidendid, mis sunnivad publikut end unustama ja tekitavad neis reaalsena tunduvaid elamusi, pakuvad arvatavasti samavõrra võimsaid virgutussignaale, mis stimuleerivad aju plastilisust."

Kusjuures tasub ka endale meelde jätta, et me oleme väga kehvad rööprähklejad ning väga head keskendujad. Seega automaatseid, kätteõpituid tegevusi jõuame teha küll paralleelselt tähelepanu nõudvate tegevustega (näiteks praadida muna ja samal ajal kassi peale karjuda)... mingi piirini. Autoroolis pole väga mõistlik moblat vahtida, kuna ekraan on võimas tähelepanutõmbaja ning autojuhtimine võib jääda kiirelt tahaplaanile.

Raamat on huvitavalt üles ehitatud teemade ja käsitluse mõttes. Ta on läbivalt üsna teaduslik-tehniline, näiteks esimeses veerandis keskendub masinõppele ja suurandmetele, räägib, kuidas arvutit õpetatakse ning milles on masinad osavad, milles neid aga väikelapsed ületavad. Sekka on juttu ka inimestest, suurem osa tähelepanu kulub aga esimeses osas sellele, kuidas käib õppimine arvutimaailmas, samal ajal tuues paralleele inimesega. Edasi on rohkem juttu inimestest, samas on ikka suur osa raamatust täidetud n-ö "keerulisema" jutuga. Minu jaoks oli see meeldiv üllatus, kuna on hea kui vaadatakse maailma ka selle nurga alt, näidatakse, kuidas on masinaid õpetatud, võttes aluseks inimese. Mingil hetkel kaldub autor omakorda ajuehitusse, mis nõudis süvenemist, oli pingutav, väsinud peaga seda väga edukalt ei lugenud. Seega lugesingi võimalusel raamatut hoopis hommikuti, värske peaga.

Sirvisin ja meenutasin vahepeal ka Jaan Aru "Ajust ja arust. Unest, teadvusest, tehisintellektist ja muustki" raamatut. Stanislas Dehaene "Kuidas me õpime? Miks aju õpib paremini kui mistahes masin… esialgu" on mingis mõttes Jaan Aru raamatu järg edasijõudnutele, inimestele, kes tahavad nende teemadega minna edasi rohkem süvitsi sest see raamat on mitmes mõttes väga mahukas. Kui alguses on juttu masinõppest, keskel ajuehitusest, siis lõpuosa on rohkem pehmemale poole kaldu. Kuigi raamat on algusest lõpuni omavahel loogiliselt seotud, siis mingi piirini saab ka lugeda teda jupikaupa, keskendudes sobivale osale. Samas see pole õige, kuna nagu ka algusepoole viitasin: selleks, et olla hea õpetaja, on väga hea saada aru, kuidas aju töötab. Ning seda see raamat ka tutvustab. Kuigi eks kõige kannatumad võivad üldsegi võtta kohe ette peatüki, mis annab kolmteist näpunäidetlaste potentsiaali parimaks rakendamiseks.

Üks stiilinäide veel sellest, kuidas raamat on vahepeal üsna pingutav (vähemalt mu jaoks):

"Võrerakud on närvirakud, mis asuvad roti aju piirkonnas, mida nimetatakse entorinaalseks korteksiks. Edvard ja May-Britt Moser pälvisid 2014. aastal Nobeli auhinna nende rakkude tähelepanuväärsete geomeetriliste omaduste avastamise eest. Nad salvestasid esimestena entorinaalse korteksi neuroneid ajal, kui loom liikus ringi väga suures ruumis. Me juba teame, et lähedalasuvas piirkonnas, hipokampuses, käitusid närvirakud nagu "koharakud", st ergastusid ainult siis, kui loom asus selle ruumi konkreetses punktis. Moserite teedrajav avastus tuvastas võrerakkude reageerimise mitte ainult kindlale kohale, vaid tervele asukohtade hulgale. Enamgi veel, need privilegeeritud asukohad, mis panid raku ergastuma, olid paigutatud korrapäraselt: need moodustasid võrgustiku võrdkülgsetest kolmnurkadest, mis grupeerusid kuusnurkadeks nagu laigud kaelkirjaku nahal või basaltsambad vulkaanilistes kivimites! Millal iganes loom ringi liigub, isegi pimeduses, teatab võreraku ergastumine rotile asukoha tervet ruumi katvas kolmnurkade võrgustikus. Nobeli komitee nimetas seda süsteemi õigustatult aju GPS-iks: see moodustab ülimalt usaldusväärse neuronaalse koordinaatsüsteemi, mis kaardistab välise ruumi. Aga miks kasutavad neuraalsed kaardid kolmnurki ja kuusnurki, mitte nelinurki ja ristuvaid jooni, nagu tavalised kaardid? Descartes'ist peale on matemaatikud ja kartograafid toetunud kahele ristteljele, mida tuntakse kartesiaanlike koordinaatidena (x ja y, abstsiss ja ordinaat, pikkus ja laius). Miks eelistab roti aju tugineda kolm- ja kuusnurkade kombinatsioonile? Kõige tõenäolisemalt sellepärast, et võreraku-neuronid iseorganiseeruvad arengu kestel ning looduses annab selline iseorganiseerumine alates kaelkirjakunahast mesitarude ja vulkaaniliste sammasteni tihtipeale tulemuseks kuusnurgad.[...]"

Lühidalt: ma ei oskagi nii põhjalikku raamatut lühidalt kokku võtta, praegu andsin ka vaid väga põgusa ülevaate, umbes nagu üks pime kerjus, kes elevanti kirjeldab. Mul on väga hea meel, et see raamat on eesti keelde tõlgitud, müts maha Argo kirjastuse ees.

Raamatust saab juppi lugeda kirjastus Argo kodulehel: https://argokirjastus.ee/toode/kuidas...

[Oksana Grabovska · Rating 5 out of 5]

Найкращий нон-фікшн цього року. Автор не просто сухо розповідає про особливості навчання, а дає приклади, описує, як вчитися з розумом, а не зубрити. Запам'ятала, буду користуватися

[Nick Clark · Rating 5 out of 5]

Very clear and highly readable book on the nature of learning. A (surprisingly) clear exposition of ways in which the brain is organised in handling and structuring information, with actionable recommendations for how to learn, pedagogy and teaching. Dehaene's four pillars of the human learning algorithm are attention, active engagement, error feedback, and consolidation.

I really like him calling out the relationship between making errors and learning, and how our current school systems in most countries does not sufficiently accommodate for the variation in proficiency and curiosity across pupils, and the risk of stigmatisation caused by the current usage of grading in schools. The role of sleep as an active consolidator of knowledge was an eye-opener. Great read. Give it a try. There is something here for everyone.

[George Hipp · Rating 5 out of 5]

Well researched and written book about how humans have evolved to learn, parallels with ML (machine learning), importance of education for children and how everyone can use the tools our physical brain provides to learn effectively. This is not a book about how to cheat time to learn or shortcuts to learning, more of information about what we know about our brain and theories around how we can optimize learning by understanding how it works. The relationship between theories around learning and experiments with ML is also fascinating. Approachable book, good for anyone interested in learning and even more so for people who teach.

[Nadine · Rating 4 out of 5]

This book brings research on learning and neuroscience up to date and dispels a number of myths circulating in education and parenting circles.
Written clearly and in an engaging way.

[Eugene Kernes · Rating 5 out of 5]

Overview:
From birth, the brain is equipped with a lot knowledge and the capacity to learn. Nature, known as genetics, provides the infrastructure for the brain to learn, but the content of what is learned depends on nurture such as culture, and interactions with others. Evolution is a slow process of adaptation, but evolution provided the capacity to learn. Learning enables quick adaptation to unpredictable conditions. Learning is how the external world becomes represented in an internal model. Updating the model when needed. New experiences change how the brain organizes itself. Synapses constantly change, reflecting what is learned. Learning can be accelerated or inhibited depending on the context. Learners need focused attention, active engagement, error feedback, and cycles of consolidation. Learning is a discovery and updating process which depends on how a culture treats curiosity and opportunities to learn.

The Brain:
The brain is far more detailed than the blueprints to build it. It would be impossible to code all information into the brain, so learning needs to supplement genes. Even the most simple of life’s creatures that have a brain, learn by habituation and association. Habituation learning is adaption to stimulus. Association learning is predictions based on prior information discoveries.

Even with limitations such as blindness or other brain impairments, individuals are capable of developing normal capacities, and using them with great dexterity. Brain dynamic of recycling means to reorient functions without genetic modification. Learning and education recycle functions. The brain appears to need room for more complex thoughts. Some functions become impaired.

As some environment information is the same throughout generations, evolution makes them predictably. Alternately, evolution makes some parameters change rapidly to adjust to volatile environmental aspects.

Babies are born with considerable knowledge inherited by evolutionary process. Nature and nurture are not opposites. Each rely on the other. Learning takes place within innate constraints. Learning does not start from nothing as learning uses many prior assumptions. The a priori hypotheses are used in obtaining meaning, and seeing what works best given the environment. Even from an early age, humans are capable of computing many abstract ideas and can access abstract institutions which enable higher learning.

Memory is a reconstruction. Memory is based on contact between two neurons. The more the related neurons fire together, the more they are wired together. Memory vanishes without retesting of knowledge. Long-term memory is based on testing the material, rather than just studying it.

The brain needs more than just intellectual stimulation, it takes a appropriate nutrition, oxygenation, and physical exercise. Brain development requires exposing to various stimulus to make it flexible, otherwise the brain won’t develop the circuits. During childhood, the brain is overhauling its organization quickly, by either creating or eliminating synapses. This quick change also explains a large reason for childhood sensitivity periods.

Learning:
What is seen are the projections that the brain has made meaningful from the flow of data. Learning uses previously missed information to change the internal model. Knowing what to learn to update the model.

New observations update thoughts in a probabilistic manner. A gradual rejection of false hypotheses, and maintenance of more rigorous hypotheses. Considers a myriad of ways to express the internal model, then utilizing that which incorporates the most data of the external world. The best fit for the state of external world.

Sometimes learning can get stuck. No options to do better seem to exist. Changes seem counterproductive, as they increase errors. Although better outcomes are possible, they are too far to be understood.

Convolutional neural networks learn faster and better because they generalize information. What was learned can be applies elsewhere.

Humans learn from each other. Even a single experience, a single trial, can bring about new understandings. Trying to learn more and more abstract rules, so that as many observations fit into the rule. While the brain creates a lot of meaning from very little data, machines need a lot of data to make some meaning. For computers, learning is difficult because there are so much data and possibilities to explore. Hard to select what to focus on. Artificial systems have a hard time learning abstract concepts, are not data-efficient, lack social learning, and lack composition.

Pillars of Learning:
To extract as much information from the environment, evolution created functions that facilitated learning. Stability requires all four functions. The functions are attention, active engagement, error feedback, and consolidation.

Attention amplifies focus. Attention is how the brain selections information, amplifies it, channels it, and deepens its processing. Decides when, what, and how to attend to information. Paying attention, also means choosing what to ignore. Directing attention means to choose, filter, and select. Without attention, students cannot perceive the teachers lesson, therefor cannot learn. Attention can be misdirected, which inhibits learning.

Active engagement encourages curiosity and experimentation. Active exploration of the world. Passive organisms learn little or nothing. To learn, the brain needs to form hypothetical ideas of the outside world, and then then test them. Passive or distracted students do not benefit from lessons, because their brains are not updating their models of the world. Only by actively following the course is information learned. Teachers aid in pedagogical progression, to guide student learning. Students do not learn much without guidance. But do need a structured learning environment with strategies for active engagement.

Error feedback corrects predictions of the world. Learning from mistakes is a popular form of learning. Every error is an opportunity to learn. Error reduction through feedback. Feedback that explains how to improve. Discovering errors enables correcting errors. Quality and accuracy of feedback influence speed of learning. Without a surprise, there is no learning. Prediction error is needed to learn. Error feedback is not punishment. Many children are punished or stigmatized for errors, and learn not to be curious to reduce errors. Errors should be corrected rather than punished.

Consolidation makes learned behaviors automatic, and involves sleep. Consolidation frees up mental energy for other purposes. Automation reduces the mental strain of an activity, allowing the mental bandwidth to be used elsewhere. Sleep is not inactivity, or just waste disposal. Brain remains active during sleep. Sleep goes over what was learned during the day, and gradually transfers it into an efficient compartment in memory. Sleep quality and quantity depends on how much was learned, as the more learned means more sleep is needed. During sleep, new information is not absorbed. Sleep makes discoveries more abstract and general.

Caveats?
Most of the information is about childhood learning, because childhood is a time of major brain development. The focus on childhood learning leaves out implications for adult learning. What does learning mean for adults? Childhood learning implications might not relate well to adult learning.

Error punishment during school is a major inhibitor of learning for children. But even adults are punished for errors, and there can be a lot of social sanctions against learning. Non-childhood learning inhibitors are missing, but that does incentivize considering how cultures can facilitate or inhibit learning.

Artificial Intelligence or machine learning, is explained in the book, but as a contrast to human learning. Highlighting how humans learn by expressing the limitations of machines. Machine learning is a feature of the book, but is not prominent.

[Kris Muir · Rating 5 out of 5]

Dehaene is a well-respected neuroscientist and his expertise is obvious from his intricate explanation of the neuronal scaffolding of the brain. For my own personal teaching, I appreciated his framework around the 4 pillars of learning (attention, active engagement, error feedback, consolidation). I found his chapter on error feedback very interesting, particularly the research by Roediger on memory retention being stronger when students tested themselves (and studied less) vs the students that actually spent more overall time studying. I'm also very curious about the notion of the spacing effect and what an ideal timeline might look like for something you're trying to keep for a lifetime (e.g. knowledge of neuroscience!). Dehaene references intervals of 20% of how long you want to remember something, but he doesn't give any specifics in terms of timeline. If I want to learn X topic and still be able to remember it 5 years from now, how often do I need to schedule retrieval practice? Is it every week, then every month, then every 6 months, etc.?

On balance, I liked the Dehane book but I felt that it lacked strong practical approaches to applying neuroscience in teaching and learning.

Advice for learners (and maybe teachers):
-build in more low-stakes tests (as simple as brain dumps on a white sheet of paper)
-do the hard thinking involved in retrieval practice (set a timer and see how much you remember)
-leverage the spacing effect (plan ahead and schedule study sessions over many days instead of cramming)
-consider cascading your study sessions (20% reviewing old material, 40% new material)

Inspired by this book, I attended a wonderful workshop on the neuroscience of learning given by neuroscientist Kristi Rudenga at Notre Dame. Her main insights were:
-“learning changes the physical structure of the brain”
-“repetition strengthens synapses”
-“richer networks = stronger learning”
-“stress short-circuits the brain”

Happy Reading!

[Paige McLoughlin · Rating 5 out of 5]

I read this in the early part of 2020. I had been exploring concepts in machine learning and developments in Artificial intelligence and I have always had an interest in human cognition and learning. This book combined both topics and compares and contrasts the functioning down to some real nuts and bolts of both human and machine learning. Humans and machine learning via neural networks (which oddly enough are modeled somewhat on animal neural systems) have a lot of similarities. It seems logical categorical processing systems are great expert systems and good algorithms for expert and bureaucratic decision-making bodies developed in the early days of AI research are very alien to the way humans think. Do things precisely and efficiently way beyond human capacity but needed strict rules and had a hard time with novelty and ambiguity or not well-formed problems to tackle. The newer neural nets took a long time to get off the ground but can do amazing things at visual or audio tasks of distinguishing objects that are novel and not labeled in a clear-cut manner. It is also good at cluster messy data and fitting it into digestible graphical layouts and clusters. It also much like people require large amounts of training sets and use feedback in performance to "learn" and be evaluate performance before it is unleashed on novel data. The book gets down to the nuts and bolts of such systems and where they look like human learning and cognition and places where they diverge from it. Definitely will hit this one again in the near future.

[Adam Osth · Rating 3 out of 5]

It's hard for me to give books like this a fair review because I'm a cognitive scientist and I'm already familiar with a lot of the work described in the book. Still, I think one thing that frustrates me about these pop-science books is how little they take *theory* seriously. This is the kind of book where every chapter deals with a different aspect of learning without there being much of a unifying perspective.

Throughout the book he buys into some more "classical" views without providing much of a defense. He believes in innate knowledge, learning by rules, and some modular perspectives on brain function. Some of these views he defends (the modular view) but others he just kind of accepts without describing their weaknesses. For instance, while there is a lot of compelling evidence that suggests people use rule-like knowledge, there is still no theory about how people can come to flexibly acquire them. And there to this day is no theory on what "innate knowledge" people possess or how that was refined by evolution. It's important to at least knowledge these limitations when discussing these perspectives.

[Matt Hutson · Rating 5 out of 5]

How We Learn by Stanislas Dehaene is one of the best books about learning that I have read. I imagine, as a teacher myself, that this book is most helpful for those in the education industry. However, if you are an entreprenuer, business owner, or any conceivable field of work, if you want to become better at learning this book can help you to do that.

The core of the book is based on The Four Pillars of Learning:

1. Attention
2. Active Engagement
3. Error Feedback
4. Consolidation

These four pillars can be used both in child learning and adult learning.

As a creator of an online course about how to become a better reader having a framework like the one above turns out to be extremely useful. And you, the reader of this review, can use this framework in your own learning.

I would highly recommend this book to anyone who is looking to improve the way they learn and to understand how they can teach others how to learn better as well.

[Mihai Cosareanu · Rating 5 out of 5]

Amazing book about how we learn. It contains an overview about all the research available related to learning presented in a well summarized and structured way. This has become my go to recommendation for anyone who asks "how can I improve my learning?".

Before this I would have recommended the "How We Learn" course on Coursera, and indeed it contains one concept that wasn't touched in this book, the concept of Chunking (you can google to understand what it means). Apart from this concept that's missing, I totally loved the book and the insights that came from it

(tiny spoiler: newborns don't have a blank state, they already can assess objects, quantities and probabilities, I loved the research on this :) )

[Olenka Dobosh · Rating 4 out of 5]

Чудова книжка з гарними прикладами навчання мозку та штучного інтелекту. Було особливо цікаво читати в час сплеску популярності чату GPT.
Поставила рейтинг 4, частково за переклад, який як на мене трохи ускладнює читання тексту українською. Англомовна версія як на мене легша у сприйнятті. Також часом було відчуття, що певні частини в книжці зʼявляються не послідовно, і це створювало трохи хаотичне сприйняття прочитаного. Але загалом книжка точно варта уваги! Раджу прочитати і паралельно стежити за розвитком машинного навчання та штучного інтелекту.

[Brandon · Rating 5 out of 5]

I was expecting this book to be just another think piece about learning and education, but I ended up being really surprised by how good it is. Dehaene does for the most part an amazing job at weaving lessons from neuroscience, cognitive science, and even artificial intelligence together to create a compelling narrative about how we learn and how we can take advantage of our understanding of how the brain learns to improve education.

I'll start off with some quotes I want to remember:

"Take a new group of kindergartners and put them into the passive, receptive pedagogical mode. All you have to do is give them the object while saying, “Look, let me show you my toy. This is what it does . . .” and then play the music box, for instance. One might think that this would stimulate the children’s curiosity . . . but it has the opposite effect: exploration massively decreases following this kind of introduction. Children seem to make the (often correct) assumption that the teacher is trying to help them as much as possible, and that he has therefore introduced them to all the interesting functions of the device. In this context, there is no need to search: curiosity is inhibited."

I wonder how many parents have accidentally turned their kids off of their field of specialty by doing this exact thing. Reminder to self - NEVER DO THIS TO YOUR KIDS!!!!

Another surprising point I want to remember:

"The myth of learning styles: According to this idea, each student has his or her own preferred learning style—some are primarily visual learners, others auditory, yet others learn better from hands-on experience, and so on. Education should therefore be tailored to each student’s favorite mode of knowledge acquisition. This is also patently false: as amazing as it may seem, there is no research supporting the notion that children differ radically in their preferred learning modality."

I was always told that people have different learning styles. But Dehaene really tries to emphasize that almost all children have very similar cognitive circuits, and learning techniques that work for one will work for another. I think that he pushes his myth-busting a little too far though - does he remember being a kid? There are alsl sorts of personalities and brains out there, and even if in general what works for one kid should work for all, there is undeniably a huge amount of mental diversity out there, not just in ability but in inclinations.

The book is a whole divided into three parts. The first part, which explains how modern AI works and how it falls short of human cognition, was the worst in my opinion. I almost gave up on the book actually, because it was extremely boring to me. I happen to have taken a course in exactly the sort of AI he was talking about (Bayesian models of cognition), and Part 1 was basically a pop-sci version of the course I took. Plus, it went deep enough into AI techniques that I expect it will lose a lot of readers who aren't familiar with AI, while not going deep enough to interest readers who know AI well. But he still makes some interesting points.

First, he gives nine different definitions of learning:
- adjusting the parameters of a model
- exploiting a combinatorial explosion
- minimizing errors
- exploring the space of possibilities
- optimizing a reward function
- restricting a search space
- projecting a prior hypotheses
- inferring the grammar of a domain
- reasoning as a scientist

I was impressed by this list - I think it does a pretty good job giving a lot of different AI-inspired ways of thinking about learning - but it's also a very weird presentation of these concepts, and the different definitions have a huge amount of overlap.

The next list was also interesting, a list of functions that modern AI is lacking:
- learning abstract concepts
- data-efficient learning
- one-trial learning
- social learning (the ability to use cues from other agents to speed learning)
- systematicity and language of thought (the ability to learn general laws governing an example)
- composition

I found the next part, about neuroscience, to be much more interesting, maybe because I hardly know anything about neuroscience. What struck me the most is that apparently neuroscientists have identified many brain circuits that are devoted to specific tasks, some of them much more specific and powerful than I would have imagined. For example, I already knew that there are specific areas of the brain for processing faces and language. But I didn't know that there are grid cells in the brain that are arranged in hexagons that keep track of our location in 2D space, or that there is a line of neurons in our brain that we use as a number line to compare quantities.

Another theme that Dehaene really hammers home is that literally everything we learn has a physical representation in our brains, and that we have specialized circuits for virtually everything we do. He advocates for his theory of "neuronal recycling", where, in order to learn a new task, we repurpose the most relevant specialized neural circuit to learn the task. For example, humans evolved to speak and listen, but not to write and read. So when we learn to write and read, we end up repurposing parts of our language system for writing and reading (but at no cost to language). Interestingly, literate adults are much better at many mental tasks than illiterate adults - "not only are [illiterate adults] incapable of recognizing letters, but they also have difficulties recognizing shapes and distinguishing mirror images, paying attention to a part of a face, and memorizing and distinguishing spoken words". However, I'm a little bit suspicious of this line of research. I don't know how you can fairly compare literate and illiterate adults (maybe the papers talk about this more), and maybe with more effort it would be possible to identify tasks that the illiterate adults are better at. So when Dehaene writes, "The myth of the illiterate bard who effortlessly musters immense powers of memory is just that: a myth", I have to call him out on his bullshit. Those "illiterate" bards were trained to memorize epic poems from childhood and definitely had greatly enhanced "neural "circuits" that were probably at least as impressive as our literate circuits.

Dehaene also talks about interesting results about how math is represented in the brain. We apparently upcycle our primitive built-in math circuitry to learn arithmetic and continue to repurpose those circuits to understand more and more advanced math. Even professional mathematicians rely on those same circuits to think of abstract concepts. This is actually really interesting because it shows that most people probably think of concepts, even abstract mathematical concepts, in similar ways, because those concepts are tied into the same brain circuitry. "parity, negative numbers, fractions . . . all these concepts are demonstrably grounded in the representation of quantities that we inherit from evolution.36 Unlike a digital computer, we are unable to manipulate symbols in the abstract: we always grind them in concrete and often approximate quantities."

Just by examining the brain's responses to things like phonemes, letters, or numbers, you can tell how that person was raised. The brain of someone born in China and adopted to the US, who knows no Chinese, will still activate slightly more when exposed to Chinese phonemes. The brain of someone who learned to read as a child will respond to letters differently than someone who learned to read as an adult. The brain of someone taught to read musical notes as a child responds differently to sheet music than someone who learned to read music later in life. Etc.

The last and most interesting section of the book is finally on how we learn, or the four pillars of learning:
"Attention, active engagement, error feedback, and consolidation. Four slogans effectively summarize them: “Fully concentrate,” “participate in class,” “learn from your mistakes,” and “practice every day, take advantage of every night.” These are very simple messages that we should all heed."

The "four pillars" sounds like some vapid self-help catch-phrase (the "four agreements"), but they were actually really interesting to learn about.

Regarding attention, Dehaene talks about Posner's three types of attention: "Alerting, which indicates when to attend, and adapts our level of vigilance. Orienting, which signals what to attend to, and amplifies any object of interest. Executive attention, which decides how to process the attended information, selects the processes that are relevant to a given task, and controls their execution."

It is really interesting how each of these types of attention have been carefully studied, and we know pretty well how they work. "Engaging all three types of attention" sounds technical, but at its finest, attention translates into passion - so when Dehaene says that we understand how attention works, in a way, it means that we have some understanding of how passion works. And, unsurprisingly, passion is crucial for learning. "Alerting" and "orienting" can be encouraged, and "executive attention" which is basically concentration or self-control can be trained (apparently playing music from a young age helps a lot). Concentration is also linked very closely with fluid intelligence (it affects how well we can hold and manipulate objects in our working memory), and fluid intelligence is closely linked to IQ - which is Dehaene's explanation for why every year of schooling seems to raise IQ.

One other interesting aspect of attention of its social element - children seem to have a built-in 'pedagogical stance', where they recognize when an adult is trying to teach them something and then pay very close attention to the adult to intuit what they are trying to teach. "Parents and teachers, always keep this crucial fact in mind: your attitude and your gaze mean everything for a child. Getting a child’s attention through visual and verbal contact ensures that she shares your attention and increases the chance that she will retain the information you are trying to convey." This information is sad to learn in our age of Zoom education.

Activate engagement translates to curiosity - Dehaene uses terminology that is suspiciously similar to the terminology used in "curiosity-driven reinforcement learning agents" in this section, and I'm not sure which came first, AI curiosity or neuroscience curiosity. But it's interesting either way. According to Dehaene, curiosity is the difference between what we expect (what our mental model predicts) and what we end up observing. There is crucially a sweet spot for learning. If there isn't enough stimulation/surprise, then we grow bored and our brains stop learning. If there is too much, we become overwhelmed and our brains also stop learning. Interestingly, the way that a teacher presents the material can have a huge effect on how "curious" we perceive the subject to be: "take a new group of kindergartners and put them into the passive, receptive pedagogical mode. All you have to do is give them the object while saying, “Look, let me show you my toy. This is what it does . . .” and then play the music box, for instance. One might think that this would stimulate the children’s curiosity . . . but it has the opposite effect: exploration massively decreases following this kind of introduction. Children seem to make the (often correct) assumption that the teacher is trying to help them as much as possible, and that he has therefore introduced them to all the interesting functions of the device. In this context, there is no need to search: curiosity is inhibited." (yes I think it's worth quoting this twice).

Error feedback is also surprisingly interesting. The concept is obvious - we learn more when we have the opportunity to make mistakes, or even the opportunity to possibly make mistakes - and when we receive frequent, informative, and positive feedback of what we did wrong or right. But what is interesting is that Dehaene makes the case that we shouldn't have tests and grades for measuring performance - they are not constructive. Of course, tests are still a useful pedagogical tool, but only to motivate students and get them to participate in a form of error feedback. But not as a way to punish students for not knowing the material. One way to get around this would be to have more frequent tests where the student is allowed to retake the test, receiving feedback each time, until they get all the answers right. As someone who has taken about 1 million tests over the course of my life, that makes a lot of sense to me.

Consolidation is ALSO surprisingly interesting! Dehaene talks mainly about two things - spaced repetition learning and sleep. Spaced repetition learning is great, and I need to use Anki more to memorize and practice things. And sleep is even more important than I thought - Dehaene really believes that our brains are generative models, and that sleep is used to sample experiences from our generative model. He seems to think that sleep's main function is to help us learn, because during the night our brain can sample experiences from its generative model much more rapidly than we can experience things during the day, allowing us to learn and commit things to memory as we sleep.

[Iryna · Rating 5 out of 5]

Це було неймовірно!

Перша частина книги для всіх хто хоче дізнатися про порівняння людського мозку та ШІ. Спойлер: мозок виграє (поки?).

Друга частина для батьків чи людей, що щодня взаємодіють з дітьми від народження і до 17 років, щоб розуміти як розвивається мозок і як можна йому допомогти розвиватись ефективніше.

Третя частина для педагогів і батьків про 4 стовби навчання, які притаманні будь якій людині.

Тут не було "води" в тексті, викладення наукове та обгрунтоване дослідженнями.
Я отримала величезне задоволення і книга однозначно входить і в топ року, і в список must read non fiction.