Lesion-symptom mapping
The wiki has leaned on lesion evidence since its first somatic-marker ingest — the VM patients are a lesion argument, the insula’s craving and alexithymia findings are lesion arguments, Bonaz et al. call the lesion material their best evidence — without a page describing what the design can and cannot license. Jones, Ward & Critchley (2010) organize their entire review by this method’s internal hierarchy (their Table 1 groups studies as case report / case series / lesion-symptom mapping / VLSM), which is the occasion for the page.
Why it is the strongest design here, and the most fragile
Nearly every claim in this wiki about where interoception happens rests on correlational evidence: this region activates when people count heartbeats, this region’s grey matter covaries with accuracy, these regions are functionally connected. None of it distinguishes a structure that performs the function from a structure that receives its output, funds its attention, or merely lights up alongside it.
Lesion evidence answers a different question — does the function survive without this? — and that is why the dissociations it produces are the sharpest things the wiki holds. Feeling one’s own heartbeat turns out to be at least two capacities, because bilateral insula/ACC damage can leave the somatosensory one and remove the viscerosensory one (via Bonaz et al.). Pain intensity and pain unpleasantness come apart, because insula-lesioned patients rate the first higher and the second normally (Starr et al., via Jones et al.). No imaging design produces facts of that shape.
The fragility is that the independent variable is not chosen by the experimenter and is not anatomically clean. Strokes follow arteries. The insula’s artery is the middle cerebral, whose territory includes the internal capsule and basal ganglia, so the typical “insula patient” has a large lesion of which the insula is one part. Tumours are better localized but are resected, and resection is followed by reorganization.
The worked cautionary case
The most useful thing on this page, because it shows the confound producing a wrong answer that stood for eight years.
Dronkers (1996), in Nature, localized apraxia of speech to left anterior insula by lesion overlay — a canonical finding, replicated by others, and one of the most cited claims about this structure. Hillis et al. (2004) re-examined it using diffusion- and perfusion-weighted imaging, which separates tissue that is infarcted from tissue that is merely underperfused. Apraxia of speech correlated with damage to Broca’s area, not the insula. Their conclusion, quoted by Jones et al.: the insula “is most commonly damaged because it is most vulnerable to disruption of the middle cerebral artery,” so the symptom and the insular damage “may be independent manifestations of large stroke.”
The finding was not fabricated or badly measured. It was produced by the sampling distribution of human brain damage, and it took a better imaging method to see that. Every lesion row in this wiki should be read with that possibility live.
The four designs, and what each buys
| design | what it can show | what it cannot |
|---|---|---|
| case report | that a deficit is possible after damage including region X; a striking dissociation in one person | anything about necessity, frequency, or which part of the lesion mattered |
| case series | convergence across a handful of patients; crude hemisphere or subregion contrasts | statistical inference; control for lesion size or aetiology |
| lesion-symptom mapping | the site most commonly damaged in symptomatic vs. matched asymptomatic patients | that the site is sufficient, or that the association is not driven by lesion extent |
| VLSM | voxel-level association without pre-specified regions | the same size confound, plus dependence on how many patients happen to have damage at each voxel |
Rorden & Karnath (2004) — “Using human brain lesions to infer function: a relic from a past era in the fMRI age?” — is the methodological reference Jones et al. cite, and the answer implied by their review is no: the lesion method is what the imaging age has been doing without.
Where this page bites in the wiki
- insular-cortex — the lesion section there is now sourced first-hand, and its anterior/posterior division of labour comes from this design.
- anosognosia — the posterior-insula localization is a VLSM result (Karnath et al. 2005) whose principal weakness is precisely the size confound this page names.
- does-somatic-feedback-guide-decisions — the somatic marker programme is a lesion argument throughout, and the debate’s largest correction (Everitt’s peripheral-manipulation experiment) turns on peripheral lesion and denervation groups (pure autonomic failure, spinal cord section, peripheral neuropathy) rather than cortical ones. Same method, different tissue; the same caveats about small n and adaptation apply, and Dunn et al. say so.
- craving — Naqvi et al.’s 19 insula-stroke patients are the wiki’s cleanest instance of the design doing something correlational evidence could not: removing a structure and watching one motivational feeling disappear.