Learning About Learning:
A Not-So-New Science of Reading
How Modern Neuroscience Validates Marie Clay’s Predictive Reading Brain
I’m always fascinated by new discoveries that deepen our understanding of the most complex phenomenon in the universe: the human brain.
One of the most significant neuroscience developments of this century has been the revelation that our brains are prediction machines. In other words, we now know that our brains are constantly guessing what will happen next and using our senses to update those guesses, rather than just passively reacting to the world.
It has been affirming to realise that Dame Marie Clay’s view of reading - as a highly complex task involving skills such as prediction, monitoring and self correction - proves her to be, not an outdated theorist, but a cognitive visionary who anticipated modern neuroscience.
In recent years New Zealand has seen the growth of a strict, “bottom-up” narrative dominate the conversation about reading instruction (Bottom-up refers to a method of decoding text that starts solely at the level of the letter or group of letters). Critics of “balanced literacy” have aggressively dismantled Clay’s work dismissing her “literacy processing theory” and the so-called “Three-Cueing system” as unscientific guesswork that misleads young readers. However, as the pendulum swings toward an exclusively linear, phonics-first Structured Literacy model, educational history in Aotearoa risks throwing the cognitive baby out with the ideological bathwater.
What if Clay wasn’t wrong? What if she was simply decades ahead of her time?
Long before functional MRIs could map neural pathways, and long before cognitive neuroscience adopted the framework of the “predictive brain,” Clay intuitively understood that the human mind is not a passive, tape-recorder-style data scanner. Now, in 2026, neuroscience views the brain as a sophisticated prediction machine; an organ that actively deploys top-down expectations to synthesise, accelerate, and make sense of bottom-up sensory data. When viewed through this contemporary lens, Clay’s emphasis on meaning, structure, and active problem-solving transforms from a dated educational trend into an effective, practical application of predictive processing. With the benefit of 21st century neurobiological research, is it time to recognise Clay as an accidental neuroscientist who cracked the code on how the active brain learns to read?
The Brain as a Prediction Machine
To understand why Clay’s work was so prophetic, we have to look at a massive shift currently happening in neuroscience.
For decades, scientists believed the brain processed the world like a camera. The eyes took in raw data - like letters on a page - and sent it up to the brain to be processed from scratch. This is a “bottom-up,” linear model and is in line with phonics-first reading models.
But recent neuroscientific findings have completely overturned this view.
Led by neuroscientists like Andy Clark and Karl Friston, the consensus has shifted to a framework known as predictive processing. Neuroscience now proves that the brain is not a passive receiver of data. It is a proactive, top-down prediction machine. In the drive for efficiency and to save metabolic energy, the brain continuously builds internal models of the world. It uses your past experiences to predict what you are about to see, hear, or read before the sensory data even arrives. Instead of processing every single line and dot of a letter from scratch, the reading brain projects its expectations downward. It then only focuses on the “prediction errors” - the differences between what it expected and what is actually on the page.
When Clay stated that reading is a “message-getting, problem-solving activity,” she was describing active inference. She realised that a child who brings their oral language, knowledge of grammar (structure and syntax), and life experience/prior knowledge (meaning) to the page is not guessing in a vacuum. They are using the brain’s natural architecture to make decoding faster, smoother, and more meaningful. They are deploying cognitive “priors.” In neuroscience terms a prior is an educated estimate - or an informed guess, based on past experience. It is an unconscious belief about what you are likely to see, hear, or feel before your senses actually send data to your brain.
Clay didn’t have the neuroimaging tools to see neural feedback loops, but she correctly identified their behavioural output. She knew that efficient reading requires a mind that anticipates text, rather than one that just reacts to it.
Self-Correction as Neural Error-Minimisation
If you want to see the predictive reading brain in action, you only need to watch a child self-correct. In traditional or Structured Literacy reading models, an error is simply a failure - a broken link in the decoding chain. But Marie Clay saw errors, particularly self-corrections, as goldmines of psychological data. Through her development of Running Records, Clay proved that when a child misreads a word and immediately fixes it without adult prompting, they are demonstrating a high form of reading competence. Indeed, Reading Recovery teachers do not promote students to higher instructional levels of text until they have a sufficient self-correction rate.
Modern neuroscience completely backs Clay up. In predictive processing, the brain only learns and adapts through the management of prediction errors.
When a child reads a sentence like “The family walked into the horse,” and suddenly stops, hesitates, and says, “No, house,” a correcting neural action has occurred. The brain’s top-down prediction (driven by semantic meaning and syntactic structure) temporarily overrode the bottom-up visual data. However, the reading brain does not operate in a vacuum. It is a hierarchy that is constantly monitoring and checking its own outputs across multiple layers of information simultaneously. The moment the child’s eye registers that the visual data (the letters ouse) does not match the predicted word (horse), or if the child realises the first attempt doesn’t make sense or sound right - a massive prediction error signal is triggered. The brain doesn’t collapse; it adapts. It immediately uses this visual or aural sensory feedback to update its internal model, leading to an instantaneous self-correction.
Clay famously noted that efficient readers do not just use one source of information; they cross-check meaning, structure, and visual cues against one another. If a prediction satisfies meaning but violates the visual print, the system flags it. If it satisfies the visual print but violates oral grammar, again the system flags it.
This is exactly what neuroscientists refer to as hierarchical predictive coding. The brain is constantly running a real-time, multi-layered quality control check. By teaching children to self-monitor and search for cues to verify their reading, Clay wasn’t teaching them to make empty guesses. She was fine-tuning the brain’s internal error-detection software, training young readers to become highly efficient, self-regulating prediction engines.
Restoring the Visionary
To reduce Marie Clay’s legacy to an outdated “guessing game” suggests a level of misunderstanding about both her work and the very architecture of the human mind. Clay was never advocating for reckless guessing; she was advocating for an active, thinking, self-monitoring brain. Decades before advanced neuroimaging could track neurobiological pathways, she looked through the lens of human behaviour and accurately mapped the reading brain as a dynamic prediction engine. As we watch the current educational pendulum in Aotearoa swing toward rigid, linear models of literacy instruction that risk reducing reading to a mechanical, passive chore, the insights of predictive brain processing offer a vital course correction. They remind us that visual decoding of words using a solely phonics or Structured Literacy approach is not the final destination of reading, but a necessary sensory aspect that serves a higher, predictive purpose. By honouring Clay’s methods through the framework of modern neuroscience, we will be on the way to settling the reading wars’ debate. It is time to restore Dame Marie Clay to her rightful place in educational history - not as a relic of the past, but as a brilliant cognitive visionary who was, all along, decades ahead of her time.
Clay, M. M. (1991). Becoming literate: The construction of inner control. Heinemann.
Clay, M. M. (2019). An Observation Survey of Early Literacy Achievement (4th ed.). Heinemann.
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioural and Brain Sciences, 36(3), 181-204.
Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press.
Doyle, M. A. (2013). Marie Clay’s theoretical perspectives and powerful messages for teachers. Journal of Reading Recovery, 13(1), 5-15.
Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138.
Peters, L. A., Gawthorpe, A., Goldenberg, G., & Wass, S. V. (2026). Knowing what’s coming: The neuroscience of why predictability and routines are more important for younger children, and for children with additional needs. [Preprint]. Institute for the Science of Early Years and Youth (ISEY).
Steffen, P. R., Hedges, D., & Matheson, R. (2022). The brain is adaptive not triune: How the brain responds to threat, challenge, and change. Frontiers in Psychiatry, 13, 802606.