Plain-language answer
Predictive processing is an influential framework in cognitive neuroscience proposing that much of what brains do is continuously predict incoming sensory information, compare those predictions against actual input, and update using the mismatch (“prediction error”). Under this view, surprise is not a side effect of perception — it is close to the central signal that drives learning and attention. It is an active, still-debated research program, not an agreed-upon fact about how brains work, and this page marks it that way throughout.
Why it matters
If something like predictive processing is broadly right, it offers a unifying mechanistic story for why coincidences — events that violate a prediction — grab attention, feel significant, and trigger a search for an explanation: a large prediction error is, on this account, exactly the kind of signal the brain is built to prioritize and resolve. That would connect several of this site’s threads (surprise, pattern perception, causal narrative-building) to one underlying architecture, rather than treating them as a list of separate quirks.
What the framework claims
A widely cited target article lays out predictive processing as proposing that perception, action, and even the boundaries between them can be understood through hierarchical prediction: higher levels of the brain send down predictions about expected input, lower levels compare those predictions to what actually arrives, and only the error — the unpredicted part — propagates back up for further processing and potential belief updating.(Clark, 2013) On this account, a coincidence is, structurally, a large, hard-to-resolve prediction error: an event pattern the brain’s existing model did not anticipate and cannot immediately fold into routine expectations.
Why this counts as a live debate, not settled science
The article introducing this framework in its most-cited form was published specifically as a target article for open peer commentary, and it received dozens of published responses, several disputing central claims — about how much of cognition it can really explain, whether “prediction error minimization” is doing genuine explanatory work or redescribing existing findings in new vocabulary, and how the framework should be tested against rival accounts.(Clark, 2013) That structure (a strong proposal published alongside substantial, serious disagreement) is itself part of why this page treats predictive processing as a promising research program rather than an established mechanism.
Common misconception
“Neuroscience has shown the brain is a prediction machine, which is why coincidences feel so significant” overstates the current state of evidence. Predictive processing is a serious, actively developed framework with real explanatory ambitions — not yet a demonstrated, uncontested account of how perception and cognition work, and specific claims about its relevance to any one phenomenon (including coincidence-noticing) are extensions of the framework, not established findings in their own right.
Limits and open questions
Critics of the framework have raised concerns about falsifiability (whether predictive-processing explanations can be specific enough to be tested and potentially wrong), about how cleanly it scales from low-level sensory processing to complex reasoning and emotion, and about whether it is best understood as a single unified theory or a loose family of related but distinct proposals. This page will be updated as the debate develops rather than treated as settled.
Related
- Emotion, salience, and surprise covers the more empirically established side of why surprising events get prioritized in attention and memory, independent of any specific brain mechanism.
- Pattern perception and the clustering illusion covers the behavioral phenomenon a predictive-processing account would aim to explain mechanistically.
Key takeaways
- Predictive processing proposes that brains work mainly by predicting incoming information and learning from prediction error — a framework that would neatly explain why surprising coincidences draw outsized attention, if the framework itself holds up.
- It was introduced as, and remains, a debated research program with substantial published disagreement about its scope and testability — not an established consensus.
- Treat specific claims that “the brain does X because of predictive processing” as extensions of an active theory, not as settled neuroscience, until shown otherwise.