Learning to Read (or Do Anything) Means Building a Brain Network

Research on how the brain changes as we learn to read provides insight into learning, in general

What happens in the brain when you learn how to do a complex task, such as reading?

You didn’t evolve to read. There weren’t books grazing the plains of Africa. So, you don’t have a “reading” area of the brain in the same way you have, say, a “visual cortex.”

So, when you start learning to read, your brain works with what you have. It looks for parts that already do similar things, and uses them for reading.

And because reading involves many small steps, you end up drawing on multiple different parts. At minimum:

• Readers need to recognize the shapes of letters and combinations of letters. The brain draws on tissue in the left fusiform gyrus that we already use to recognize individual objects.

• Readers must know the sounds that make up words. The brain already processes the sounds of spoken words in parts of the auditory system. It will learn to apply these sounds to written words, too.

• Readers learn to associate the shapes of letters with the sounds they make. Readers draw on an area in the parietal lobe that already integrates signals coming from the visual and auditory systems.

• And of course, readers need to understand the meanings of the words and sentences they read. So, your brain draws on neurons in association areas of the temporal lobe that already help you understand the meanings of spoken words.

Notice that:

• The requirements of the task determine which parts of the brain get involved.

• Several brain regions work together to do the task.

As you read more often, these regions get used to working together. For example, when you read the word “go,” the neurons that recognize the letter “g” fire with the neurons that recognize the sound /g/, those that recognize the letter “o” fire with the neurons that recognize the sound /o/, and the whole group of them fire together.

When neurons repeatedly fire together, they develop stronger connections between them. There are several mechanisms (see the infographic below), but in general, new synapses form and existing ones strengthen.

Infographic by the National Institute for the Clinical Application of Behavioral Medicine, found on Michael Netzley’s blog.

The basic principle is, “neurons that fire together wire together.”

With stronger connections between neurons, information can travel between them faster. Thus, you can, for example, associate a letter with its sound more quickly. That’s one reason learning to read becomes faster and easier with practice.

As most people learn to read, neurons involved in reading fire more often with other neurons involved in reading, and less often with neurons not involved in reading. Their connection with neurons that participate in reading strengthen, while their connection with neurons that do not may weaken.

The result is that a “network” develops.

Here’s what it means for a “reading network to develop”:

1. At the level of individual neurons, the functions of neurons become more specialized: they read more and do other things less.

2. At the level of the whole brain, activity when you read becomes more “focal”: fewer neurons are involved. Only the ones that specialize in reading participate.

Instead of having a whole bunch of generalist neurons involved, you have a few specialists. It’s like a hospital hiring 10 doctors instead of 100 random people who know how to do CPR. Thus, your brain becomes more efficient at reading.

This process is called “interactive specialization.”

So, when you use fMRI to look at brain activity while people read, experienced readers have strong activation in small, highly demarcated groups of neurons. By contrast, people still learning how to read have weaker, more diffuse activation. Their reading network is still developing.

In the image above, shared by famous neuroscientist Stanislas Dehaene on X, one child’s brain activation before learning to read (at the end of preschool), during their first year of learning to read, and at the end of their second year of reading. During the first year, reading is still difficult, and is accompanied by widespread brain activation. By the end of the second year, when reading has become more automatic, activation has become focused in an area that specializes in recognizing letters and words, and a temporal lobe area processing meaning.

Meet the Reading Network

At its most oversimplified, the reading network would look like 3 blobs on the left hemisphere in the brain:

1. Frontal lobe: relates to the sequencing of sounds in words; 

2. Where temporal and parietal lobes meet: integrating auditory and visual information (word and letter sounds with word and letter symbols);

3. Temporal lobe, just above the cerebellum: visually recognizing letters and words. Part of this blob is called the “visual word form area,” which becomes specialized in recognizing words in the same way its counterpart on the right side specializes in recognizing faces.

These 3 areas were the first to be discovered and understood. They, and their connections between them, are needed for decoding: identifying the words on the page.

However, decoding is not enough. You also need to understand what words mean. Have you ever read an abstruse science or medical research article, and realized you “read" a word, but don’t really know what it means? You’ve experienced decoding without comprehension.

Thus, the complete reading network is larger than the 3 regions that enable decoding. It also includes brain regions that process the sounds and meaning of language. 

A Modern Model of the Brain’s Reading Network is a version of an image pulled from an unknown research article by Ms. Judy Araujo’s Reading Blog, edited by me for clarity and accessibility. It shows the Reading Network: Phonology regions, in orange triangles; Meaning regions, in green ovals, and Orthography, the red rectangle. These are supported by top-down attention (yellow ovals) and basic visual input from the visual cortex (blue diamond).

Instead of 3 regions, the modern schematic of the brain’s reading network above contains 3 functional categories. These groups include Phonology, Orthography, and Meaning, 2 of which include multiple regions. Other regions (shown in yellow and blue) support the reading network.

Orthography: Recognizing Letters

When we start reading alphabetic languages, we learn to recognize a letter whether it is in upper or lower case, handwritten or typed, in any font or style. We can tell a “b” from a “d” and an “m” from an “n,” and we can distinguish between letters and other symbols. Later, we learn to recognize meaningful clusters of letters, such as “igh” in English. Eventually, we can recognize whole words by their shape without having to read each letter. Our reading may or may not be accurate, as demonstrated in the internet-famous graphic below:

Version of an image I’ve often seen on Facebook, in which the middle letters of most words are rearranged. Yet, most people correctly read it as, “According to a researcher at Cambridge University, it doesn’t matter in what order the letters in a word are, the only importent [sic] thing is that the first and last letter be at the right place. The rest can be a total mess and you can still read it without problem. This is because the human mind does not read every letter by itself, but the word as a whole.” This image is analyzed (and critiqued) by Science Alert, who point out that no Cambridge researcher has made any such claim. Also, much of our ability to read these words comes from our powerful expectations of what word should come next — which are influenced by other parts of the reading network. However, these visual cues allow a quick check of these predictions. Most people would predict “Cambridge” is followed by “University,” but if the next word is “Street” instead, the first letter, last letter, shape, and length all contradict the prediction. Moreover, these cues are available within ~250 milliseconds, compared to the ~400 ms needed to read and understand a whole word.

Knowing the Sequence of Sounds in Words

You might wonder why the order of sounds in words matters when reading silently. The more finely we can break down and arrange the sounds of spoken words, the better an understanding of sound we can bring to the process of matching sounds and symbols. Starting around preschool age, we become aware, first that words contain syllables and rhymes, and later that they contain phonemes. Children learn to add, subtract, and re-order phonemes. For example, they can “say bat without the b.” Children who are better at these “phonological awareness” tasks are also better readers.

These regions may be highly active early in learning to read, when children sound out most words, and less so later, when they read automatically.

Associating Letters and Speech Sounds

Suppose you see an unfamiliar word, “zint.” Do you pronounce the “int” as in “pint” or as in “mint?” Decisions like these make even experienced readers pay attention to both the letters in a word and their associated sounds. This integration draws on a large area where parietal and temporal lobe meet, especially a specific ridge, the supramarginal gyrus. As with phonological regions, these areas may be more active in people learning to read than in skilled, automatic readers.

Understanding Word Meanings

When you understand a word, you know more than just the dictionary definition. You also know its connotations, the part of speech it has, and the ways it can be used in sentences; this is knowledge about language. You also know the contexts in which you’ve read and heard the word (is it used at school, at home, at work, or in a specific community?). That is knowledge about people. You also draw on your knowledge about the world. For example, most adults can read the phrase “car engine” and know what it does, but a car mechanic might picture the engine in detail, with each part in its correct place, and remember what each parts does and how to fix it.

The temporal lobe supports memory for both facts and personal experiences. It may also support imagery (picturing what you’re reading). Generally, the more abstract a concept, the further forward (“anterior”) the relevant brain tissue will be.

…

Of course, the regions in the “reading network” are not the only parts of the brain working when you read. 

For example, your visual system registers the way the letters look, without “knowing” that they’re letters. 

Your cerebellum is probably coordinating the timing of other regions’ activity.

An entire “attention network" helps you focus on reading. The posterior parietal lobe (top back of the brain) is especially important for guiding attention during reading. 

Your frontal eye fields, at the border of the frontal lobe and motor cortex, move your eyes and attention as you read.  

And of course, your brain stem coordinates the many unconscious processes that keep you alive and breathing.

None of these parts specialize in reading. They’re working most of the time, whatever you’re doing. So, they’re not part of the reading network. However, they often "light up” in fMRI studies.

Keep in mind…

Keep in mind that most people learn to read as children. Thus, in the background, our brains are still developing. If you were to watch brain development in slow motion, you would see waves of blooming, then pruning, move across the brain. That is, you would see synapses proliferate, then vanish, beginning with the sensory cortices and ending with the frontal lobe. 

This infographic from Harvard University Center for the Developing Child shows which areas “bloom” (increasing number of synaptic connections) and prune (lose connections) at what ages. Notice that during ages 5-8, when most children are learning to read, the sensory pathways are pruning, while language pathways reach their peak number of connections. Higher cognitive functions are still blooming. Children draw on all these pathways to read.

What happens if you have dyslexia?

People with dyslexia, as a group, have difficulty associating letters with speech sounds. Individuals may have other difficulties too, with attention or visual processing of letters. However, the most consistently replicated finding is difficulty with letter-sound association. This ability is called “phonological awareness”.

At first, children with dyslexia may read both slowly and incorrectly. With time, they learn to read correctly, but remain slow. With the same teaching and the same amount of practice other children get, they have more difficulty reading. Why?

People with dyslexia don’t develop the full reading network. They don’t develop groups of neurons that specialize in reading and exchange information super-fast.

Above: Image comparing dyslexic readers (left) to typical readers who are more efficient (right), from The Reading Well. Notice how efficient readers use a frontal region (leftmost), associated with speech sounds, a temporal lobe region associated with visual letter and word recognition, and a temporal-parietal region to integrate letter symbols and sounds. Dyslexic readers activate only their frontal region, much more than efficient readers do.

Yet, as shown in the picture above, they also don’t have the diffuse, all-over-the-place brain activity you see in typically developing kids just starting to learn to read.

Instead, people with dyslexia seem to rely on their frontal lobes — associated with conscious, deliberate, effortful thinking — to read.

Implications

Each Task We Do Often Enough Has its Own Network

The reading network isn’t the only network most humans develop.

There’s also a network for mathematical operations.

Most likely, any time we consistently do something for years, we develop a brain network for it. Expert musicians will develop a network selective for listening to or playing music. Expert chess players will have a “chess network.” And so on.

When Researching the Brain, The Task Matters

Researchers have to understand something about how a task works to predict what parts of the brain might be involved. Thus, it matters what the researchers ask participants to do.

Researchers sometimes don’t take experimental tasks seriously enough, and reporters even more so. To be complete, an explanation of a study in the news must clearly describe what participants perceived and did.

Researchers sometimes try to make the “reverse inference”: they infer how people do a task based on what brain regions they use. That doesn’t work, because many brain regions participate in multiple networks, doing multiple things. Neuroscientists call this faulty thinking the “reverse inference error,” yet they — and reporters — often fall prey to it.

The reading network is unusual because we understand so well what each region does during reading. That’s because we have done so much research on reading itself, and on brain activation during reading. 

It’s more common for researchers to know what brain regions participate in a task, but still be figuring out how each one contributes. How do researchers try to figure out a brain region’s function? That’s a topic for another post.

You may have noticed this post barely scratches the surface of what we know, and don’t know, about the reading network. How much does the reading network differ across languages? How does it change as children develop? Researchers have some answers to these and other questions. 

However, my point is not to explain how we learn to read, as interesting as it is. Instead, I’m using the reading network as an especially clear example of how the brain changes as we learn to do anything.

My inspiration came from a friend, who was reading about a study that talked about a “task related network" and a “default mode network." I started thinking about what a “task related network" means – something that these sorts of studies do not necessarily explain. Now, here we are.

What "task related” network most interests you? 

What questions do you have about the reading network and how it develops? Or, do you have questions about the way researchers represent it in images? (While researching this post and finding images to use, I was frustrated by how much the number and naming of regions varies across research studies and blog posts. 

You can share your questions, thoughts, and experiences by hitting the comment button below!