Try as we might to simplify the way we teach reading, reading isn’t simple. It’s made up of interconnected processes that our policies don’t necessarily reflect.

For many people, reading is a gateway, a skill that grants access to information and the understanding of that information. Reading is not only a matter of unlocking the text on the page, it actually has a far more lasting effect on language development, educational attainment and employability.

A new meta analysis

Every so often a meta analysis is published which draws out the common findings of a large number of studies. A recent meta-analysis of the neurobiology of reading has revealed new insights into the intricacies of reading behaviour using brain imaging techniques.

The team from the Max Plank Institute for Human Cognition and Brain Sciences led by Sabrina Turker analyzed findings from 163 experiments involving 3031 adults. None of the studies included participants with specific learning difficulties such as dyslexia and so the findings represent reading processes in people with typical development. The meta-analysis included studies of alphabetic reading processes in ten different languages that used whole brain imaging. The team examined five subprocesses of reading: letters, words, pseudowords, sentences and extended text.

Educational policy tends to talk about reading as a unitary skill: “the ability to read.” Among the most enduring models of reading, the 'Simple View of Reading' has stood the test of time, because it includes language processing and decoding skills and reading comprehension as its three key components. The actual complexity of reading tells a more interesting story of human experience.

Reading activates:

• classical left-hemisphere language areas,

• bilateral visual systems,

• auditory and phonological regions,

• fronto-parietal executive networks,

• and subcortical structures including the cerebellum.

Classical reading areas

The classical reading areas are located in the language regions in the left hemisphere of the brain. This view of reading suggests that specific reading operations of single word recognition, phonological processes and semantic processes are distinctive. The meta analysis confirmed the importance of these left hemisphere language areas in reading.

Attention network

Other brain areas were also activated, pointing to links between the visual word form area and the dorsal attention network (DAN). The DAN is associated with spatial attention and orienting. It’s easy to overlook the visual orienting of the eyes as they alight on individual words or glide across a line of connected text, but these movements demand focus and coordination between the tasks of sustaining attention, controlling eye gaze and also visually processing the words.

Executive network

When we pick up a text and decide to read it, we are engaging in a goal-directed activity, whether to satisfy our curiosity, to solve some kind of problem, or to simply read for pleasure. Our executive functions support goal-directed behaviour, for example through inhibitory control, enabling us to narrow our focus onto the reading task. Other executive functions such as task-switching support our capacity to respond in the moment and to pivot with changes to syntactic content in particular. Working memory is perhaps the most obvious of the three core executive functions and represents the capacity for holding and managing context and content, what phenomenologists refer to as ‘given’ and ‘new’.

Auditory and subcortical processes

The auditory areas are of course involved because, we are processing sound when we process language, but the intriguing part of this is how we allocate our resources to different units of sound. And related to this we need to include the cerebellum - also known as the ‘Little brain’, so called because it’s involved as a regulator of tasks and is implicated in timing precesses.

This broad distribution of processes explains something educators witness daily: some children can decode accurately but struggle to comprehend; others have strong comprehension but weak fluency; others read aloud confidently but falter in silent reading tasks. The brain is not activating “one skill” but a constellation of interconnected processes, each with distinct demands.

For leaders, this raises an important strategic question. When we support reading development, do we support all the networks required for reading — or only the ones most easily measured?

The bilaterally spread network

The classic approach to researching reading development has focused mainly on the left hemisphere, in the peri-Sylvian region where language processing takes place.The meta analysis showed that all the studies and all the subprocesses activated peri-Sylvian language regions, pre-central motor and auditory of the left hemisphere, right hemisphere regions involved in emotional processing as well consistent activation of bilateral areas of the cerebellum, associated with subcortical processes involving timing and regulation. The authors described this as a 'bilaterally spread network'.

There were also differences between the tasks with certain tasks having ‘high specificity’ for particular brain areas and the authors propose these formed a hierarchy: letter, words, pseudo words, sentences and text. It’s interesting that there were considerable contrasts between the activation areas.

Letter reading

This was very specific indeed with only a tiny area activated. In my opinion, reading an individual symbol with no intrinsic meaning would not have a biological basis - this task is a social construct.

Word reading

A large area of the left hemisphere is activated and showed a distributed pattern of activation. It is unsurprising that reading words engages orthographic (shape recognition) and lexical-semantic regions in the left hemisphere, forming a familiar and efficient route for recognising known words. Word reading naturally bootstraps onto word processing areas that have evolved over millions of years.

Pseudoword reading

The authors expected a more distinctive outcome for this subprocess, but commented that the close similarity between reading words and non-words was probably due to the participants processing the non words as syllables. Typically, reading non-words requires more effortful phonological assembly and relies on dorsal regions associated with phonological decoding.

Sentence processing

Sentence reading bootstraps onto the natural processing of phrases and sentences across a wide area of activation, which shows that sentence processing take place in distributed networks.

Extended text comprehension

The activation pattern for reading extended text was very specific indeed and located in a small area. This task does not translate easily into a speech processing pattern, but as it draws upon inference-making, working memory, coherence building, and discourse-level integration, the overview of the text involves synthesising a range of different types of information.

Summary

So taken together, we can see that this hierarchy of reading tasks is associated with language processes of the left hemisphere and that at the word and sentence levels, these are distributed networks, within the broader bilaterally spread network. Different tasks produce different activations and no doubt these layers, are exquisitely complex.

As reading is a network process, then our literacy strategies must be too. The challenge is to design systems that nurture every part of that network - from decoding to comprehension; from attention to executive function.The findings show clearly that reading is not one skill but a whole brain endeavour! The question is - to what extent are our systems reflecting this?

REFERENCE

S. Turker et al., (2025) The ‘reading’ brain: Meta-analytic insight into functional activation during

reading in adults, Neuroscience and Biobehavioral Reviews 173 (2025) 106166