Prediction error in autism is one of the most useful ideas in current neuroscience for understanding why a small, unexpected change can feel so much larger than it looks from the outside. For many autistic adults, a moved item at the store or a meeting pushed by fifteen minutes produces a response that is real, physical, and slow to fade, and that has often been labeled rigidity or overreaction. This framework offers a calmer and more accurate account of what your nervous system is actually doing.
I'll call him Daniel, and I've changed the details that would identify him. Daniel is a late-identified autistic adult in his mid-forties who came to see me after what he described as "a meltdown over nothing." His usual grocery store had rearranged its aisles. He stood in the new layout for a long time, got home late and, in his words, "wrecked," and spent the rest of the evening unable to talk to his wife or his kids. What brought him in was the shame afterward, and his wife's quiet confusion about why something so small had cost him so much. Daniel is the kind of person I work with every week: capable, self-aware, and carrying decades of explanations for himself that were never quite right.
In this blog post, I want to explain prediction error in plain language, why it tends to land differently in autistic brains, and why a mismatch can cost far more energy than the trigger seems to justify. I'll also look at how this plays out inside neurodiverse relationships, what it means for adults identified later in life, and the small adjustments that tend to help. The goal is not to fix anything. It is to give you a clearer name for something you have probably been navigating, without language, for most of your life.
3 Key Takeaways
• The brain is constantly predicting what will happen next. A prediction error is the signal it generates when reality does not match the prediction, and every brain produces these signals all day long.
• For many autistic adults, prediction errors tend to be louder, more frequent, and more expensive to process. This is a large part of why small surprises can feel disproportionately big and why recovery takes longer than the event seems to warrant.
• Understanding prediction error does not make surprises stop. It shifts the question from "what is wrong with me" to "what is my brain doing, and what does it need."
What Prediction Error Actually Is in Predictive Coding
Most of us picture the brain as something that reacts to the world. A large body of neuroscience over the past two decades tells a different story. The brain is better described as a forecaster than a reactor. At every moment it is generating predictions about what is about to happen, based on everything similar it has experienced before, and comparing those predictions against the sensory information coming in (Friston, 2010; Barrett, 2017).
What the sensory input signal means
Think about the jolt when you walk down a dark staircase and expect one more step than there is. Nothing dangerous happened, but your whole body registered it. That jolt is prediction error. When prediction and reality line up, the brain stays quiet. When they do not, the brain generates a signal that says, in effect, "something here does not match my map, pay attention." That signal is prediction error. It is not a mistake in the everyday sense. It is information.
What the brain does with the signal
Once a prediction error arrives, the brain compares what it predicted with the actual sensory input, then decides whether to ignore the mismatch as noise, update its model, or act on the world to make reality match the expectation. Bergenfield and Sweezy (2026) describe this process in Wired to Feel, and note that in most brains, most of the time, it happens quickly and quietly.
Why the brain bothers predicting at all
Prediction is efficient. If your brain had to treat every incoming signal as brand new, you would be exhausted before breakfast (Friston, 2010). A small amount of prediction error is also how we learn. Nervous systems work best with enough uncertainty to keep learning and not so much that the system shifts into protection.
How Prediction Error Tends to Land Differently in Autism Spectrum Disorder
Why This Happens
Several research groups have converged on related ideas. Pellicano and Burr (2012) proposed that autistic perception may give current sensory data more weight than past experience, which can make expectations harder to update flexibly, and they described this as the world becoming "too real." Van de Cruys and colleagues (2014) and Lawson, Rees, and Friston (2014) suggested that the autistic brain may assign unusually high weight to prediction errors, so that mismatches other brains would file as noise get treated as important. Sinha and colleagues (2014) proposed that difficulty predicting what comes next, especially in fast-moving social situations, may sit close to the center of the autistic experience, and Balsters and colleagues (2017) found that the brain's signaling of social prediction errors looked different in autistic participants.
These are still theories, but they point in a consistent direction. Bergenfield and Sweezy (2026) connect this to what they call sensory surplus. Where a non-autistic brain filters out a great deal of incoming information before it reaches conscious attention, an autistic brain often receives much more of it. More sound, more light, more texture, more social signal. Sensory differences are extremely common in autism; about 90% of children on the autism spectrum show sensory abnormalities, often as heightened sensitivity to stimuli other people can more easily ignore (Leekam et al., 2007; see also Ben-Sasson et al., 2009). More incoming data means more places where reality can fail to match the prediction, and so more prediction errors, which helps explain why reactions can be more pronounced when expectations and outcomes differ.
What This Can Look Like
• A small change in routine, such as a different route, a moved object, or a meeting time pushed by fifteen minutes, producing a response that feels far out of proportion to the change. Part of what makes these moments hard is the reduced ability to update quickly from new sensory information once the change has already happened, which can lead to rigidity in thinking and inflexible expectations.
• Unexpected sensory stimuli, such as a sound, smell, or texture, cutting through whatever you were doing and taking minutes or hours to recover from.
None of these are signs of fragility. They are signs of a system processing more and getting hit by more mismatches than the systems around it.
What Helps
Having the name. Daniel's grocery store evening made no sense to him as a personality trait. As a nervous system event, it made complete sense: a reliable prediction failed inside a system that treats failed predictions as urgent. Naming it that way did not change what happened. It changed what it meant.
Why Small Surprises Cost So Much
Why This Happens
Every prediction error costs something to process. In an autistic system with sensory surplus, those decisions are happening far more often than in a non-autistic system, and each one draws on a finite budget. That is one reason a strong need for predictability can be so critical, because an autistic individual may resist change or hold tightly to sameness when it lowers anxiety.
There is also a threat dimension. Porges (2011) describes how the autonomic nervous system continuously evaluates safety beneath conscious awareness, and shifts the body toward mobilization or shutdown when safety cannot be confirmed. When the extent of mismatch stays high across the day, the nervous system treats the absence of predictability as an ongoing problem rather than a small isolated event. A day full of unresolved prediction errors is, from the nervous system's point of view, a day in which safety keeps failing to arrive.
What This Can Look Like
The cumulative cost, and the outcome of repeated mismatches, can include exhaustion, irritability, sensory shutdown, difficulty speaking, and more pronounced distress, or the evening collapse that follows a day which looked, from the outside, completely manageable. Over longer timescales it feeds autistic burnout, which Raymaker and colleagues (2020) described as chronic exhaustion, loss of skills, and reduced tolerance to stimulus following sustained demands without adequate recovery. The visible event is never the whole story. The invisible story is how many mismatches the brain processed to get through it.
What Helps
Accurate accounting. If you know a new environment or a change in plans spends more budget than it appears to, you can plan recovery instead of being ambushed by the need for it.
Prediction Error in Neurodiverse Relationships
Prediction error rarely stays inside one person. In a neurodiverse couple, where partners may have very different sensory and predictive systems, the same change of plans can run through two nervous systems and cost them wildly different amounts.
Why This Happens
A last-minute switch that costs one partner a moment of mental flexibility can cost the other a substantial share of the day's budget, because the same event is being processed by a different system. In close relationships, that can also affect estimating what another person is likely to do or mean next, especially when more than one possibility seems plausible. This fits with one predictive impairment hypothesis: difficulties in social interactions and communication may reflect trouble estimating conditional probabilities from ambiguous social cues, which can shape everyday interactions as much as sensory or routine changes. When neither partner has language for this, the gap fills with interpretation. The non-autistic partner may read the reaction as controlling or dramatic. The autistic partner may read the request as careless. Both are guessing, and both guesses tend to be wrong.
What This Can Look Like
Daniel's wife, whom I'll call Renee, told me in a joint session that she had learned to "brace" before telling him about any change to the weekend. She was not afraid of him. She was afraid of the shutdown that followed. Daniel heard this and said, "I thought you'd stopped telling me things because you didn't care what I thought." Both had built a story about the other to explain a pattern that was really about prediction error.
What Helps
Language that names the mechanism instead of the character. Some phrases that clients have found useful:
• "That kind of change costs me more than it looks like it should. I'm not upset with you. I need a few minutes to update."
• "Can you give me a heads-up before we change the plan, even if it's small? It helps my brain get ready."
Gottman and Silver (1999) describe how couples who repair quickly after small ruptures fare far better over time. Naming prediction error is a form of repair. It gives both partners a shared, non-blaming account of what just happened. That kind of language can support repair by helping both partners focus on the mechanism rather than intent.
When Prediction Error Meets Late Identification
Why This Happens
Late-identified adults have typically spent decades building explanations for themselves out of whatever language was available, and often that language came from other people: too rigid, too dramatic, too sensitive, too much. Some differences in anticipatory or regulatory patterns can be visible long before a person is formally diagnosed, even if they are only recognized in adulthood. Many learned to mask, hiding the responses that drew criticism, at real cost to their energy and sense of self (Price, 2022). The prediction error framework arrives as something different: neither an insult nor an excuse, but an accurate description.
What This Can Look Like
You realize that the responses you have been apologizing for, such as the freezes at small changes and the after-effects that lingered for days, were never personality flaws. They were a nervous system telling the truth about how much it was processing. Bergenfield, who was identified at fifty, writes openly about how this recognition reshaped her relationship to her own responses (Bergenfield & Sweezy, 2026). The past does not change. Its meaning does.
What Helps
It tends to soften shame. When a difficult evening is the predictable cost of a high-prediction-error day, you no longer have to explain it to yourself as personal failure.
Working With the Predictive System, Not Against It
The instinct is to try to override these patterns. This rarely works, because the patterns are not choices. They are how the system manages its own load. What tends to help looks more like cooperating. Bergenfield and Sweezy (2026) draw on internal family systems, an approach that treats the different responses inside a person as protective parts with understandable jobs, to do this work with curiosity rather than judgment. When the nervous system learns that prediction errors will be met with care rather than pressure, it tends to stay in its working range more often, and shutdown, meltdown, and withdrawal get triggered less.
This also reframes the routines and rituals that so often get pathologized. A familiar route, meal, or piece of clothing is not rigidity. It is bandwidth management: the brain trying to create predictability in the face of chaotic sensory input, conserve focus for the parts of the day that require it, and reduce anxiety. This can look like a strong insistence on sameness, and repetitive behaviors can serve the same function when the system is trying to regain stability after too many mismatches.
What This Can Look Like in Practice
• Building in predictability where it is free. Same morning routine, same workspace, same lunch, so the budget is there for the parts of the day that need it, and that kind of consistency leads to less unnecessary updating and preserves energy for genuine changes.
• Warning the brain before changes. Even a few minutes of preparation before a known disruption gives the predictive system time to update before the change arrives, and supports temporal preparation so the system can adjust its timing more smoothly.
• Recovering on purpose. New environments, social events, and travel days all spend more budget than they look like they should. Schedule the recovery.
• Reducing avoidable sensory load. Clothes that do not fight your skin. Lights that do not fight your eyes. Every small reduction frees up budget for the prediction errors you cannot avoid.
• Naming what happened, without shame. "That was a lot of prediction error for my system" lands very differently than "I overreacted." The first is information. The second is shame.
Neural Correlates of Brain Function
Predictive Coding Regions
Researchers are also looking for neural markers and neural correlates in specific brain regions that may help explain why predictive coding differs across people on the spectrum. These regions are thought to be involved in how the brain generates and updates predictions, and how it processes mismatches between expectation and reality.
Implications for Support
Understanding which brain regions are involved in predictive coding can help guide support strategies and interventions. By targeting these neural pathways, therapies may be developed to help autistic individuals manage prediction errors more effectively and reduce the associated costs.
Hyperplasticity and Brain Development
The Role of Hyperplasticity in Autism
Hyperplasticity refers to the brain's heightened ability to change and adapt, especially during development. In autism, this increased plasticity may contribute to the sensory surplus and the way prediction errors are processed. While hyperplasticity can offer advantages in learning and adaptation, it may also make the brain more sensitive to mismatches and unexpected changes, further amplifying prediction error responses.
Integration: What to Take From This
If you take one thing from this post, let it be this. The idea of prediction error offers a unifying way to explain sensory strain, social strain, and the cost of change in autism. When small things feel big, your brain is not malfunctioning. It is doing exactly what every brain does: predicting the world, comparing reality to the prediction, and generating a signal when the two do not match. The difference for many autistic adults is not that the process is broken. It is that more of the world produces mismatches, the mismatches are felt more intensely, and the cost of processing them is real.
You probably need more predictability than the average person, and that is not a failing but a response to real processing limitations and challenges. You probably need more recovery time, and that is not laziness. It is the visible cost of invisible work.
The work is not to become a brain that generates fewer prediction errors. That brain is not available to you, and pretending it is has been part of what wore you out. The work is to build a life with room for the brain you actually have: the one that notices more, costs more to run, and also, in many cases, sees and feels and connects more than the systems around it.
You do not need a quieter brain. You need a life sized for the one you have.
If something in this post resonated and you would like to talk it through, whether for yourself as an autistic or ADHD adult or as a partner trying to understand someone you love, I offer a no-pressure conversation to see whether this kind of work might be a fit.
Book a FREE 20-minute “Clarity & Connection” Zoom session
Book a FREE 20-minute “Clarity & Connection” Zoom session
A closing note. I write these posts for the adults and couples I sit with every week, and for the many more still looking for language. If this one gave you a clearer name for something you have lived, that is enough for today. I'm Barbara (Blaze) Lazarony, MA, LMFT #151788. I work with neurodiverse couples and neurodivergent adults, helping people understand why small surprises can feel so big.
Client stories referenced in this post are composites, with identifying details changed to protect privacy.
Frequently Asked Questions
Is prediction error a problem to fix?
No. Prediction error is how every brain learns and updates. Prediction error itself is normal, but in autism the issue is often the precision weighted prediction errors assigned to mismatches, which can make ordinary surprises feel unusually important. The difference for many autistic adults is not that prediction errors are happening; it is that they happen more often, land louder, and cost more to process. The work is not to eliminate them but to build a life that respects how much processing the system is actually doing.
Does this mean autism is a brain disorder?
Bergenfield and Sweezy (2026) explicitly reframe it the other way. They describe autism as a condition of sensory surplus rather than deficit, meaning a brain that receives and processes more rather than less. Researchers are also looking for neural markers and neural correlates in specific brain regions that may help explain why predictive coding differs across people on the spectrum. Prediction error is part of what that surplus costs. The framing matters, because "disorder" tends to push toward fixing, while "different processing" tends to push toward accommodation and respect.
Why do small changes affect me so much when I can handle big stresses?
Big stresses are usually expected, so your predictive system has time to build a model for them. Small changes, such as a moved object, a different route, or an unexpected sound, can mean the brain had predicted one pattern and then encountered a violation of that pattern, which is why even a minor change can trigger a strong response, an example of how small unexpected events can consume more energy than larger but anticipated ones. The size of the trigger is not what determines the cost. The amount of mismatch, and whether the system had time to update for it, is what determines the cost.
How does this relate to sensory overload and meltdowns?
Meltdowns and shutdowns are often what happens when the cumulative prediction-error budget runs out. When enough unexpected stimuli accumulate, the likely outcome is overload because the system cannot keep pace with continual mismatch. Each mismatch may have been manageable on its own, but together they exceed what the system can process. It makes more sense to build in recovery before the budget runs out than to push through a system that is already over capacity, and professional support may help reduce avoidable overload and recover more consistently.
References
Balsters, J. H., Apps, M. A. J., Bolis, D., Lehner, R., Gallagher, L., & Wenderoth, N. (2017). Disrupted prediction errors index social deficits in autism spectrum disorder (often cited as evidence suggesting altered social prediction signaling in autism). Brain, 140(1), 235–246. https://doi.org/10.1093/brain/aww287
Barrett, L. F. (2017). How emotions are made: The secret life of the brain. Houghton Mifflin Harcourt.
Ben-Sasson, A., Hen, L., Fluss, R., Cermak, S. A., Engel-Yeger, B., & Gal, E. (2009). A meta-analysis of sensory modulation symptoms in individuals with autism spectrum disorders. Journal of Autism and Developmental Disorders, 39(1), 1–11. https://doi.org/10.1007/s10803-008-0593-3
Bergenfield, S., & Sweezy, M. (2026). Wired to feel: Autism as a condition of sensory surplus. New Harbinger Publications.
Friston, K. (2010). The free-energy principle: A unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138. https://doi.org/10.1038/nrn2787
Gottman, J. M., & Silver, N. (1999). The seven principles for making marriage work. Crown Publishers.
Lawson, R. P., Rees, G., & Friston, K. J. (2014). An aberrant precision account of autism. Frontiers in Human Neuroscience, 8, 302. https://doi.org/10.3389/fnhum.2014.00302
Leekam, S. R., Nieto, C., Libby, S. J., Wing, L., & Gould, J. (2007). Describing the sensory abnormalities of children and adults with autism. Journal of Autism and Developmental Disorders, 37(5), 894–910. https://doi.org/10.1007/s10803-006-0218-7
Pellicano, E., & Burr, D. (2012). When the world becomes "too real": A Bayesian explanation of autistic perception. Trends in Cognitive Sciences, 16(10), 504–510. https://doi.org/10.1016/j.tics.2012.08.009
Porges, S. W. (2011). The polyvagal theory: Neurophysiological foundations of emotions, attachment, communication, and self-regulation. W. W. Norton.
Price, D. (2022). Unmasking autism: Discovering the new faces of neurodiversity. Harmony Books.
Raymaker, D. M., Teo, A. R., Steckler, N. A., Lentz, B., Scharer, M., Delos Santos, A., Kapp, S. K., Hunter, M., Joyce, A., & Nicolaidis, C. (2020). "Having all of your internal resources exhausted beyond measure and being left with no clean-up crew": Defining autistic burnout. Autism in Adulthood, 2(2), 132–143. https://doi.org/10.1089/aut.2019.0079
Sinha, P., Kjelgaard, M. M., Gandhi, T. K., Tsourides, K., Cardinaux, A. L., Pantazis, D., Diamond, S. P., & Held, R. M. (2014). Autism as a disorder of prediction. Proceedings of the National Academy of Sciences, 111(42), 15220–15225. https://doi.org/10.1073/pnas.1416797111
Van de Cruys, S., Evers, K., Van der Hallen, R., Van Eylen, L., Boets, B., de-Wit, L., & Wagemans, J. (2014). Precise minds in uncertain worlds: Predictive coding in autism. Psychological Review, 121(4), 649–675. https://doi.org/10.1037/a0037665
