Sensory processing and prediction

Differences in how people respond to sound, touch, light, taste, smell and movement are part of the diagnostic criteria for autism.^5,1 A meta-analysis of 55 questionnaire studies, covering 4,606 autistic people, found large differences from non-autistic people in over-responsivity (strong or negative reactions), under-responsivity (little or slow reaction) and sensory seeking, with wide variation between studies. Compared with groups with other developmental conditions, over-responsivity and seeking were higher in autism; compared with other clinical groups, only over-responsivity was. For seeking, the size of the difference from non-autistic people depended on age, IQ and whether people rated themselves.²

In one study of the records of 25,627 autistic children aged 4 or 8 in a US monitoring network (2006 to 2014), 74% had sensory features written down.¹ They were recorded less often for girls, for children from several racial and ethnic minority groups and for children whose mothers had fewer years of education, which the authors suggest may reflect differences in how often these features are noticed.¹

Adults report the same. In one small study, 94.4% of autistic adults scored at an extreme level on at least one sensory pattern of a self-report questionnaire, and the patterns differed a great deal from person to person.³ In two online surveys from one research group, autistic adults reported more over-responsivity than non-autistic adults across the senses.^4,5

Over-responsivity is not specific to autism. In a Finnish birth cohort of 4,424 eight-year-olds, parents reported it in 6.4% of children. It was most strongly linked to autism, and also to ADHD traits, motor difficulties and other developmental conditions, all reported by parents.⁶

Several theories treat the brain as a prediction machine that compares what it expects with what arrives. One proposal holds that autistic people rely less on prior expectations ("hypo-priors"), so perception is less shaped by past experience.⁷ The "aberrant precision" account suggests that sensory evidence is given too much weight compared with prior beliefs.⁸ A third, called HIPPEA, holds that prediction errors are given high weight that does not adjust to the situation. Its authors suggest that insistence on sameness would then be a way of getting a sense of predictive success in a world full of errors.⁹ Another group has argued that difficulty predicting makes events seem to happen without cause, which could be overwhelming.¹⁰ Not all theories are about prediction. Others propose too much excitation relative to inhibition in sensory and other brain systems, stronger low-level perception that runs more independently of higher-level processing, or an altered time window for binding signals from different senses (also proposed for dyslexia and schizophrenia).^60,61,62 This entry follows the prediction, filtering and arousal accounts.

Other proposals look at filtering and arousal. The "leaky thalamus" idea, that unimportant sensory signals pass through the thalamus and become distracting, grew out of work in mice.¹² In male mice, deleting one gene, Ptchd1, only in the thalamic reticular nucleus, which helps control what the thalamus passes on, weakened its activity and produced attention problems and hyperactivity. Deleting it throughout the body caused learning, aggression and movement problems as well, while basic sensory responses such as startle were normal.¹¹ Another proposal holds that the locus coeruleus, which sets the brain's arousal level, is more active at rest and less responsive to events in autistic children, which would change how attention is drawn to things.¹³

For anxiety, a 2010 review set out three possibilities: anxiety causes sensory over-responsivity, over-responsivity causes anxiety, or both share a cause or overlap in how they are diagnosed.¹⁴ A 2025 review proposes a chain from differences in infant sensory filtering to over-responsivity, autistic traits and anxiety.¹⁵ Neither review extends this to prediction differences or to social avoidance. The evidence on these links is in the last two sections.

A systematic review of 47 studies found differences in learning which events predict which, and in how responses fade with repetition.¹⁶ A review of studies testing the "weaker priors" idea found the results highly mixed: a slight majority found no difference in how priors are used, priors learned during an experiment were the ones more often found weaker, and many studies were too small to tell.¹⁷ A meta-analysis of 23 studies found a small-to-moderate effect in the predicted direction, but the studies varied widely for reasons it could not explain, and its authors called the support limited.¹⁸

Several direct tests did not find the predicted difference. Priors were intact for judging where a face is looking, in autistic adults without intellectual disability, and for reading shading as lit from above, in 18 autistic children matched to other children on age and ability.^19,20 In 34 cognitively able autistic children aged 6 to 14, learning sped up in a changing environment, and no difference from other children was detected.²¹ In one study of 26 autistic adults, reported in two papers, predictions were learned and updated as well as in non-autistic adults, although some brain measures in the same people differed.^22,23

Other tests found differences, of varying kinds. In one study of 24 autistic adults, they were less surprised than others when expectations were broken; a model suggested they treated the environment as more changeable, and in the autistic group pupil size tracked this modeled surprise.²⁴ In another, autistic participants gave less weight to the most recent stimuli and relied more on longer-term patterns.²⁵ A re-analysis of two earlier datasets found normal use of priors in 29 autistic young adults lifting objects, but less flexible learning in the eye movements of 17 autistic participants, and no sign that autistic people treat the world as constantly changing.²⁶

Brain-wave measures of surprise give no clearer picture. Four meta-analyses of the mismatch negativity, a response to an unexpected sound, drew on many of the same studies. They report no overall difference (with smaller responses in young children and to non-speech sounds), a smaller response to speech sounds, a smaller response in children but a larger one in adults in some designs, and no significant difference in oddball tasks.^27,28,29,30 One fMRI study, in a small all-male sample of 16 autistic men left after 12 of 28 were excluded for poor task performance, found that signals for other people's prediction errors in the anterior cingulate cortex differed and went with differences in social interaction.³¹

Sensory gating is the brain's damping of a repeated or expected signal. A meta-analysis of 7 studies (99 autistic participants) found weaker gating of the startle reflex in autistic children and adolescents, but not in adults.³² A 2026 review of 44 studies, whose participants were about 80% male with an average age of 15.6, pooled the measures with enough comparable studies. Startle gating was preserved, with consistent results. The brain response to the first of two clicks was smaller, and results for most measures varied widely.³³ Single studies found normal click gating in 12 autistic children without intellectual disability, and a difference in a later gating response but not the earliest one in 34 mostly male adolescents and young adults.^34,35

The thalamus relays most sensory signals to the cortex. Studies of how strongly it is connected to the cortex in autism disagree on both direction and location.³⁶ One study of children and adolescents found weaker thalamic connections with four of five cortical regions tested; connections with the temporal lobe looked stronger only in one type of analysis.³⁷ Two analyses of the same large shared dataset found stronger connectivity: with primary sensory networks in the first, which included only males, and with prefrontal, temporal and sensorimotor cortex in the second.^38,39 Two later studies, one with careful control of head movement and one small study comparing autism with ADHD, also found stronger connectivity with temporal or sensorimotor cortex.^36,12

In infants with an autistic older sibling, thalamic connections with emotion-related regions were stronger at 6 weeks, and those with prefrontal and motor regions weaker at 9 months. Infants with more early over-responsivity had stronger thalamic links with sensory areas and the basal ganglia, and weaker ones with higher-order cortex.⁴⁰

In brain scans by one research group, autistic youth responded more strongly than non-autistic youth to mildly unpleasant sounds, touch and images, in sensory cortex and in the amygdala. The size of these responses went with parent-rated over-responsivity, even after anxiety was taken into account.^41,42 When 19 autistic youth were split by over-responsivity, those with it showed less fading of the response with repetition, while those without it showed a pattern in which the orbitofrontal cortex appeared to damp the amygdala.⁴²

In the same sample, connectivity between the pulvinar, part of the thalamus, and the amygdala went with how severe over-responsivity was.⁴³ In a further study of 35 autistic and 35 non-autistic youth, the level of GABA, the brain's main inhibitory chemical messenger, which damps activity, in the thalamus did not differ between groups; within the autistic group, lower thalamic GABA went with more parent-rated over-responsivity. Among the 20 autistic participants with usable data, higher glutamate and glutamine, which drive activity, in somatosensory cortex also went with more over-responsivity.⁴⁴

Autistic youth also showed larger skin-conductance responses and higher resting heart rate, and heart-rate responses went with over-responsivity.⁴⁵ Youth with anxiety disorders showed similarly heightened brain responses to the same stimuli, while over-responsivity and anxiety had different physical signatures.⁴⁶

Pupil size is used as an indirect sign of activity in the locus coeruleus: baseline size for its resting (tonic) level, and quick changes for its responses to events (phasic).^47,48 A meta-analysis found one consistent difference in autism, a longer delay before the pupil responds, while evidence on baseline pupil size conflicted.⁴⁹

One small study reported larger baseline pupils in young autistic children.⁵⁰ One study of 24 autistic children found larger resting pupils and smaller responses to events, but these measures related more to ADHD traits than to autistic traits.⁴⁷ In a study of 139 autistic people aged about 6 to 30, three-quarters male, and 98 non-autistic people, baseline pupil size was slightly larger in the autistic group, but the difference was no longer significant once age and data quality were taken into account. After allowing for a related brain measure, pupil responses to the repeated standard tones were larger in the autistic group, not smaller.⁴⁸ Over the 14-minute task, baseline pupil size rose more at first and then fell more steeply in the autistic group. Its authors interpret these results as raised tonic activity, while noting that the study could not show the locus coeruleus was the source.⁴⁸

In adolescents from the same research group, a handgrip exercise used to raise arousal increased baseline pupil size in the autistic group and in a group with other mental health conditions. It did not sharpen how selectively the brain responded to sounds, but briefly increased responses to all of them.⁵¹ A meta-analysis of pupil responses found non-autistic people's pupils widened more for social than non-social images, while autistic people's appeared not to differ, with wide variation between studies.⁵²

Sensory differences and anxiety go together in autistic adults. In 246 autistic adults, over-responsivity, under-responsivity and sensory seeking each correlated with anxiety.⁵³ In 176 autistic adults from another country and research group, sensory sensitivity and anxiety were correlated, and statistically the link ran through intolerance of uncertainty.⁵⁴ Two small studies report similar links in autistic children aged 8 to 16 and in autistic preschoolers.^55,56 In one records study of autistic children, sensory features went with emotional and fear-related problems.¹

For social anxiety in particular, one study of 426 autistic adults found it correlated with sensory differences. In that study the link between sensory differences and repetitive behaviors ran through uncertainty and anxious arousal, not through social anxiety.⁵⁷

In 149 autistic toddlers followed for a year, over-responsivity predicted a later rise in anxiety, while anxiety did not predict later over-responsivity.⁵⁸ In a community sample of 917 children aged 2 to 5, not selected for autism, of whom 191 were assessed again at age 6, 43% of children with over-responsivity also had an impairing anxiety disorder, and preschool over-responsivity predicted anxiety at age 6, not the reverse.⁵⁹ In 161 infants with and without a family history of autism or ADHD, parent-rated over-responsivity at 14 months went with more fearfulness, a possible early sign of anxiety, at 24 months.⁶³

Autistic adults have also been asked how they see the link. This is their view, not a test of which comes first. They rated their over-responsivity as more a cause than an effect of their anxiety, and their sensory seeking as more an effect.⁵³ Over-responsivity and anxiety also have partly different physical signatures, which suggests they are related but distinct.⁴⁶

No study found as of October 2026 has tested whether sensory or prediction differences lead to social anxiety or social avoidance. That idea remains one of the possibilities set out in 2010, alongside anxiety driving sensory over-responsivity and both sharing a cause.¹⁴

Many autistic writers describe change as costly and predictability as a support. Devon Price writes that "Change takes us more time to adjust to than it does for others, and we rely upon certainty and predictability as anchors when life does not make sense."^c1 Price ties this partly to taking in new information in an effortful, detailed way, and partly to trauma: after sudden losses and rejections, the unfamiliar can register as a threat.^c1 In Unmasking Autism, Price sums up autism as processing "in a careful, systematic, bottom-up way", while non-autistic people entering a new place jump to reasonable conclusions: "They can wing it."^c2 This resembles the prediction theories above, whose direct tests are mixed. Some writers describe the autistic brain as slow to adapt; in several of the lab tests above, autistic participants learned and updated predictions as non-autistic participants did.

Sarah Nannery writes that "The ASD brain is notoriously slow to adapt to change", and uses timers to warn her autistic son before a transition.^c3 She describes a few small daily routines as "like a safety blanket" that carries him through an otherwise unpredictable day. When one broke at the end of a noisy, crowded school day, she argues, his distress was not oversensitivity or manipulation: "He is simply coping."^c3

Bianca Toeps places the stress slightly differently. She writes that not knowing what will happen stresses her, but that on reflection she gets anxious mainly when she risks losing control, for example over when she will eat; if she can decide when to take a break and when to go to bed, she will set off on a five-day trip with no plan at all.^c4 Because she cannot reliably predict how tired an activity will make her, she gave up detailed plans for a list of things she could do but did not have to, an idea she took from a friend.^c5

Autistic adults quoted in a clinicians' guide by Donna Henderson, Sarah Wayland and Jamell White give similar reasons for wanting sameness. One, Alix L., writes: "I don't freak out when my routine changes because I'm too rigid; I freak out because now I'm going into something without being able to prepare."^c9 Another, Peter B., eats the same lunch almost every day to save limited executive function for the things that matter.^c9 Orion Kelly adds that warning does not always help: when a day set aside for work turns, a day ahead, into a day at home with a sick child, the change is still "an immense trigger" for anxiety.^c10

Price describes a costlier way of seeking predictability. Some autistic people Price interviewed built rigid belief systems of their own: "They wanted to make their own worlds predictable, easy to make sense of, small."^c6 For change itself, Price recommends repeated, tolerable exposure to the unfamiliar, which builds "a useful internal database about how to handle unexpected events".^c1

Some writers use the image of an intense world. Megan Anna Neff writes that "it's as if your nervous system exists on the outside of your body", absorbing every sensory input, social interaction and piece of information.^c7 Toeps endorses the "intense world" theory of Kamila and Henry Markram, including their call for a calm, predictable environment for young autistic children, and rejects Uta Frith's warning that withdrawing stimulation could do harm.^c8 The research sections above do not test that theory.

Luke Beardon, writing for autistic adults, describes how sound lands, and separates sudden noises from background noise. An unexpected firework may be harder to bear than one at a festival, though both remain unpredictable, and for some people "The ticking of the clock could be just as noticeable to you as the person who is speaking to you."^c12 These descriptions resemble the prediction and gating ideas above but do not test them.

Huan Vuong, who now communicates by letterboard, describes a time before he had that tool, when a helicopter passed overhead: "Its sounds attack my ears and I'm helpless from the massacre on my brain."^c11 In that moment, he writes, he could find no way to communicate, and his mouth would not work when a speech therapist tried to make him talk.^c11