New research from Nagoya University finds affect labelling reduces anxiety in adults with higher autistic traits
A cross-sectional study by Fujii and Hirai at Nagoya University, published in Scientific Reports in 2026, surveyed 505 Japanese adults aged 20 to 39 using four self-report measures: the Autism-Spectrum Quotient, the Intolerance of Uncertainty Scale, the Affect Labelling Questionnaire, and the State-Trait Anxiety Inventory. The authors tested two competing serial mediation models against the data, both connecting autistic traits to anxiety through intolerance of uncertainty and affect labelling, but in opposite causal sequences.
The first model — the Emotion Regulation deficit Model — treats affect labelling difficulty as the upstream problem: autistic traits produce affect-labelling deficits, which produce elevated intolerance of uncertainty, which produces anxiety. This is the familiar clinical reading. The second model — the Cognitive-Motivational Model — flips the order: autistic traits produce elevated intolerance of uncertainty, which paradoxically motivates the active use of affect labelling as a coping strategy, which then reduces anxiety. Both models fit the data identically. The authors selected the Cognitive-Motivational reading on theoretical grounds, not statistical superiority, and described what they found as a “novel adaptive pathway”: higher autistic traits → higher uncertainty intolerance → higher affect labelling → lower anxiety.
The substantive finding is that affect labelling functions as an emotion regulation strategy in this sample, and the trait usually framed as pure pathology — intolerance of uncertainty — appears to do motivational work, pushing people toward a coping strategy that actually reduces anxiety. The authors describe a “dual role” of uncertainty intolerance as both risk factor and motivational driver, and frame the resulting pattern as a “conflict”: high autistic-trait individuals struggle with the affect-labelling work yet are driven toward it as a way of managing what their architecture finds difficult.
This is closer to the kind of reframe The Neurodiversity Directory has been working toward across the corpus than most autism-and-anxiety research manages. It does not treat autistic traits as a pure deficit to be remediated. It identifies a coping mechanism that genuinely works. And it acknowledges that what looks like pathology from one angle may be doing adaptive work from another. The paper deserves engagement rather than dismissal — and that engagement involves following the move the authors make a little further than they themselves do.
The two models the paper tests reveal something the clinical frame still cannot quite name
What’s interesting about the Fujii and Hirai paper is not just which model they chose, but the fact that both models fit the data identically. Statistical equivalence between two opposite causal sequences is not a failure of the research; it is the data telling the researchers something the clinical frame is poorly equipped to hear.
The data are cross-sectional, which the authors acknowledge limits any directional inference. But the deeper issue is that the bidirectional fit reflects something real about the relationships being measured. Affect labelling and intolerance of uncertainty are not arranged in a simple linear chain. They co-construct each other across time in the lives of the people being studied. High uncertainty intolerance makes affect labelling harder and makes it more necessary. Affect labelling, when it works, reduces uncertainty intolerance and requires baseline access to the very cognitive resources uncertainty intolerance disrupts. The relationship is dynamic and recursive in ways that any cross-sectional model will compress into apparent linear sequence.
The reframe the paper makes — uncertainty intolerance as motivational driver, not just risk factor — is real and matters. But it does not go far enough. The frame still positions autistic architecture as the source of the difficulty, with affect labelling as a strategy for managing what the architecture produces. What the data are also consistent with, and what the paper cannot quite say, is a further reframe: that elevated intolerance of uncertainty in autistic individuals may itself be a sensible adaptive response to environments that systematically violate prediction, rather than a trait-level pathology requiring compensatory coping. And that reduced affect labelling capacity may be downstream of chronic masking — the accumulated suppression of internal-state awareness that develops in people who have spent decades managing their own affective expression to survive environments not designed around their architecture.
The authors note explicitly that the internal world is uncertain by its nature, and that this matters more for individuals with elevated uncertainty intolerance. The framing is reasonable as far as it goes. But the internal world is uncertain for everyone. What may be different in autistic experience is not the inherent uncertainty of internal states but the additional layer of having to translate those states into a neurotypicality-oriented representational system — neurotypical emotional vocabulary, neurotypical social-emotional cues, etc — that was not designed around original, authentic, autistic phenomenology. Affect labelling deficits, in this reading, may not be deficits in labelling affect. They may be deficits in mapping affect onto a vocabulary built for someone else’s interior life.
What the paper points at but does not follow through on — affect labelling as scaffolded coping
The most interesting moment in the paper is also the moment its argument breaks open and is left undeveloped. The authors note, briefly, that emotion education and interpersonal affect labelling — having one’s feelings labelled by others — may be particularly important for autistic adults who struggle with internal labelling. The recommendation appears almost as an aside. It deserves to be the centre of gravity.
Affect labelling, as a coping strategy, does not happen in a vacuum. It requires baseline conditions: a felt-safety that allows internal states to surface in the first place, a vocabulary that can carry the states being labelled, time and attentional capacity to do the labelling work, environments where the labelled states will be received rather than dismissed or pathologised, and ideally relationships with other people who can help name what is internally present but not yet verbalisable. Autistic adults reporting that affect labelling works for them — which the Fujii and Hirai data suggest is genuine — are typically reporting it works under particular conditions: when they have access to supportive scaffolding, when the environment is not actively hostile, and when their interlocutors have the patience to wait for the words and the receptivity to take the labelled states seriously.
This is what the paper points at without quite saying: the coping strategy that works requires environments designed to support it. Affect labelling is not a property of the individual exercised in isolation. It is a relational and environmental capacity that the individual brings to encounters with others, and that thrives or fails based on whether those encounters scaffold or undermine the labelling work. The clinical intervention implied by the paper’s findings is not “teach autistic adults to label their feelings better.” It is closer to “design environments and train interlocutors so the labelling work that genuinely helps becomes possible.”
This connects directly to what the Directory has been working toward across our neurodiversity news coverage. The masking-is-not-an-exhaustion-symptom work argued that masking functions as situational interface labour producing alexithymia-like phenomenology over time as a downstream effect of sustained internal-state suppression. The cognitive co-regulation piece raised the question of where coping support comes from when human relational scaffolding is absent — and what gets substituted in its place. The accommodation-as-alibi frame named the structural pattern: the support that would actually help is positioned as a special adjustment for specific cases, rather than as standard practice for human cognitive variation.
Affect labelling as scaffolded coping is the same structural pattern applied to emotion regulation. The coping mechanism works. The conditions for it to work are absent in most environments autistic adults navigate. The absence of those conditions is then framed as the autistic person’s deficit in regulating emotion, rather than as a structural failure of the environments those adults are operating in.
Why the absence of scaffolding is structural, not personal
The clinical pathway implied by the Fujii and Hirai paper, if taken at face value, points toward individual-level intervention: teach affect labelling skills, perhaps via metacognitive therapy, cognitive-behavioural therapy, or emotion-focused therapy adapted for autistic adults, and the anxiety should reduce. This is the standard direction the field defaults to when it identifies a coping mechanism that appears to work.
The structural pathway points elsewhere. If affect labelling works under supportive conditions and fails or becomes too costly under unsupportive ones, the question becomes: why are unsupportive conditions the norm, and what would change if they were not? Autistic adults frequently describe the experience of being asked to identify and articulate their emotional states in environments that have already demonstrated they will not receive those states well — workplaces where vulnerability is professionally risky, and families where emotional honesty was historically met with dismissal, medical settings where labelled distress becomes ammunition for diagnostic re-interpretation, and social contexts where the labelling work itself reads as oversharing, overreaction, entitlement, selfishness, etc.
The conditions for affect labelling to function as a coping mechanism are rare. Their rarity is a structural property of the environments most autistic adults inhabit, not a property of the autistic adults themselves.
This matters for what the field does next. A research programme that takes the affect-labelling finding seriously without examining the conditions under which affect labelling works will produce a wave of individual-skill interventions that work in clinical contexts and fail in real life — the familiar pattern across decades of behavioural intervention research. A research programme that takes seriously what the Fujii and Hirai paper points at without saying — that the coping mechanism that helps requires scaffolding to function — produces a different question entirely. It asks what environments would have to look like, what training interlocutors would need, what relational practices would have to become standard, for the labelling work to become reliably possible. That research programme does not yet exist at scale. The Directory’s editorial position is that it should.
The paper is genuinely interesting work and points at something the field has been slow to recognise. The further reframe — affect labelling as relational scaffolding rather than individual skill, the absence of scaffolding as structural rather than personal — is the move the authors gesture toward and leave for others to complete. That completion is part of what this editorial work across 2026 and beyond works on.
Citations
Fujii, A., & Hirai, M. (2026) — Autism related traits and anxiety in the general population are linked through intolerance of uncertainty and affect labeling
