Brooks et al. (2005) — Somatotopic organisation of the human insula to painful heat
The wiki’s evidence that the dorsal posterior insula — Craig’s primary interoceptive cortex and the terminus of the lamina I → VMpo pathway — is somatotopically body-mapped has been carried entirely by cytoarchitecture and by one secondhand citation (Zeharia et al.’s 2012 “insula homunculus”, referenced on cortical-somatotopy and insular-cortex). This is that claim read first-hand, for a lamina I modality (pain), with A. D. (Bud) Craig as a co-author — the direct fMRI test of his own prediction that the VMpo→dpIns projection “should be arranged somatotopically.”
What it did
Fourteen healthy adults received moderately painful contact-thermode heat (adjusted per site to a pain rating of 5–6/10; mean 48.5–49.6 °C) to the right face, hand and foot, scanned with high-resolution 3T EPI (2 mm in-plane, 2 mm smoothing) targeted at the operculo-insular region. Three analyses were run: a mixed-effects group map, cross-subject frequency maps, and — the load-bearing one — single-subject analysis in acquired space comparing the peak-Z coordinate across the three body sites within five a-priori ROIs (anterior/mid/posterior insula, SII, inferior parietal lobule).
The result: a pain homunculus in the primary interoceptive cortex
Only one of the five regions carried a body map. In the contralateral (left) dorsal posterior insula, activation coordinates depended on where the pain was (multivariate ANOVA P = 0.012); the other four ROIs, SII included, did not. The layout is orderly: the face representation is anterior and lateral, the foot sits medially in the circular sulcus, the hand is lateral — a rostrocaudal (face→foot) axis crossed with a mediolateral one, running ≈45° to the mediolateral direction. Peak left-dpIns MNI coordinates were face (−40, −16, 11), hand (−40, −19, 14), foot (−35, −21, 11). These fall almost on top of the coordinates recovered by direct human depth-electrode stimulation (Ostrowsky et al. 2002) and recording (Frot & Mauguière 2003), the paper’s strongest external check.
This is the one place in the wiki where somatotopy and interoception genuinely meet with an actual measured body-part gradient rather than an asserted one: the interoceptive cortex is not a featureless “feeling” sheet but is laid out by body region, exactly as Craig’s labelled-line account requires. See insular-cortex and lamina-i-spinothalamocortical-pathway.
SII carries no pain map — so dpIns is the critical site
The negative half matters as much. SII responded bilaterally and non-somatotopically; the authors “did not observe nor can we find any evidence for the presence of an SII body map for pain.” Combined with their earlier finding that dpIns is the only region active contralateral to painful stimulation and unaffected by attention (Brooks et al. 2002), this makes dpIns, not SII, the critical operculo-insular substrate — and suggests it is the dpIns lesion, not the SII lesion, that produces the thermanesthesia and pain loss seen after operculo-insular strokes (Bowsher et al. 2004; Greenspan et al. 1999).
How it sits with the rest of the wiki
- It confirms a presumption the anatomy pages state as one. lamina-i-spinothalamocortical-pathway describes the VMpo→dpIns terminus as “presumably also somatotopically arranged (Craig, 1995)”; Brooks et al. supply the human functional evidence. cortical-somatotopy and insular-cortex both cite the Zeharia insula homunculus for the same point — this is an independent, earlier, modality-specific demonstration.
- It is not in tension with the lesion sceptics, but it sharpens the puzzle. Jones, Ward & Critchley (2010) report Starr’s patients rating noxious stimuli as more intense with no insular activation at all — pain awareness surviving insula destruction. Brooks shows a robust, somatotopically organized dpIns pain response in the intact brain. Both can be true (activation ≠ necessity), but together they say the dpIns is where pain is mapped while not being required for the crude awareness that something hurts — a dissociation the nociception page’s scope problem should hold onto.
- Nociception as an interoceptive channel. The paper is squarely inside Craig’s broad-inclusion frame — pain as a lamina I homeostatic afferent reaching the interoceptive insula — and predates the debate the nociception page records; it is evidence for the anatomy that frame rests on, not a position on the boundary.
Brakes
Small n with relaxed thresholds for weak (especially foot) activators; a mostly fixed stimulus order; a group-level effect too fine to survive standard group analysis (single-subject coordinates carry it); a left-hemisphere-only map from right-side-only stimulation; and a non-fibre-selective thermode. The somatotopy is real and replicated against invasive data, but it is a millimetre-scale effect demonstrated in a targeted, high-resolution, a-priori-ROI design — not a whole-brain result.