Schwarzlose et al. (2023) — Sensory over-responsivity: a feature of childhood psychiatric illness
Not an interoception study. SOR is over-responsivity to exteroceptive stimuli — sounds, textures, smells — so this paper is filed the way the wiki files zeharia-2019-precuneus-homunculus and barr-2022-psychiatric-comorbidity: as a boundary/parallel case, useful precisely because it runs the interoception literature’s own transdiagnostic-marker logic in the exteroceptive domain, in a sample (N=11,210, ABCD) an order of magnitude larger than any interoceptive study here. Read it against Nord, Lawson & Dalgleish (2021) — the interoceptive counterpart — and the two together frame the question this wiki keeps circling: is there a sensory (as opposed to affective or cognitive) transdiagnostic pathway into psychopathology, and does it run through the same salience machinery on both sides of the skin?
What it did
Parent-reported SOR (one Short-SRS item) was used to sort 11,210 ABCD children into no (82%), mild (12%) and severe (6%) SOR groups. Linear mixed-effects models (family nested within site) then tested SOR group against demographics, autistic traits (social-communication impairment SCI, restricted/repetitive behaviour RRB), and dimensional psychiatric symptoms (CBCL depression, anxiety, OCD, ADHD, ODD, CD; PQ-BC prodromal psychosis), both concurrently (Y1) and one year later (Y2). The design’s discipline is an incremental covariate approach: significant SOR–symptom effects were re-tested with SCI added, then RRB added, with FDR correction at each step — so a surviving effect is one that is not reducible to subthreshold autistic traits.
The neural arm used resting-state fMRI: exploratory FC screening in the Y0 dataset (7,760 children) identified 21 network-pair candidates, preregistered (OSF) for independent replication in the Y2 dataset (5,117 children); 17 replicated, 15 survived stringent post-hoc covariates. Cortex was parcellated with the Gordon scheme into 13 functional networks; connectivity was network-averaged BOLD correlation.
The three-accounts framing
The paper is built around adjudicating three explanations of why SOR co-occurs with psychopathology — a cleaner version of the cause-or-consequence problem that interoceptive-psychopathology keeps hitting:
| account | claim | prediction |
|---|---|---|
| dysregulation | SOR is a behavioural readout of transdiagnostic emotion dysregulation | SOR adds no information beyond existing symptoms; associations are uniform across disorders |
| autism | SOR–symptom links are driven by subthreshold autistic traits | SOR adds nothing beyond autistic traits |
| sensory-specific | SOR is a neurobiological sensory-processing difference and an independent risk factor | SOR predicts symptoms beyond both symptoms and autistic traits, and has a sensory neural signature |
The data land on the sensory-specific account: associations are differential not uniform (positive for anxiety/depression/OCD/ADHD, negative for conduct disorder), survive autistic-trait control, predict future anxiety prospectively, and carry a sensory-network FC signature. The dysregulation and autism accounts are both rejected as sufficient.
Where it lands in the wiki
On sensory-over-responsivity (created here) — the construct’s home page, holding the prevalence, the three accounts, and the FC profile.
On interoception-exteroception-boundary. SOR is the exteroceptive mirror of the interoceptive over-responsivity the wiki carries (panic’s cardiac hypervigilance, anxiety sensitivity). That the same nodes — salience, ventral attention, amygdala — show altered connectivity for exteroceptive over-responsivity as the interoceptive literature implicates for internal over-attention is the kind of cross-boundary convergence that the boundary page argues should make us cautious about treating “interoceptive” and “exteroceptive” salience as separate systems.
On salience-network. The salience network here does its Barrett/Menon job in a developmental clinical population: enhanced sensorimotor↔salience coupling is read as the network assigning behavioural relevance to innocuous sensory input that should have been ignored — over-precision on exteroceptive prediction error, the exteroceptive analogue of the interoceptive precision failures on computational-psychiatry.
On interoceptive-psychopathology and nord-2021-mid-insula-transdiagnostic. A transdiagnostic sensory marker with a prospective-risk claim and a candidate neural signature — exactly the shape of the interoceptive mid-insula story, in the exteroceptive domain and with a far larger sample. The parallel is instructive and a caution: if an exteroceptive sensory feature predicts the same broad psychopathology transdiagnostically, then “interoceptive dysfunction across disorders” may be one instance of a more general sensory-precision dysfunction, not an interoception-specific mechanism.
Brakes
The whole edifice rests on a single parent-report item, never validated against real SOR instruments — the authors concede this first among limitations. The FC findings are correlational, the sensorimotor-hand emphasis may be an artefact of the item’s wording (or of auditory-SOR dropout under scanner noise), and the design cannot separate cause from consequence any better than the interoceptive literature it parallels — a prospective association is not a causal one. And it is, throughout, about the outside of the body: its value to this wiki is comparative, not confirmatory.