# Analysis plan: is the anterior insula a selective meeting point of time perception, interoception and vestibular processing, and of depressive symptom terms, in the neuroimaging literature?

Registered before any association between a term and the insula was computed on unshuffled data.

## Hypotheses under test

A theory of depressed experience proposes that a single anterior-insula system carries time perception, interoception and graviception (vestibular/gravitational processing) and mediates the felt heaviness, low mood, low energy and slowing of depression. A necessary (not sufficient) condition is that the published literature locates all three functions, and depression-symptom terms, in the anterior insula more selectively than arbitrary topics, given that the insula is reported in a very wide range of studies.

## Data

NeuroQuery v1 (Dockès et al. 2020; BSD-3): 13,459 full-text articles with reported MNI coordinates and body-text term counts (vocabulary of 7,547 terms). Harvard-Oxford cortical atlas, maximum-probability image thresholded at 25%, 2 mm (FSL; insula = label 2). Before registering we looked only at term coverage (numbers of studies per term at body-count thresholds 1, 2, 3 and 5), the atlas label positions, and a run of the full code with term usage randomly permuted across studies.

## Measures (`code/convergence.py`)

- Regions: insula voxels with MNI y >= 0 (anterior) and y < 0 (posterior).
- A study activates a region if any reported coordinate lies within 10 mm of a region voxel.
- A study uses a term if the term occurs at least twice in its body text (K = 2); terms combined where listed: interoception = "interoception" or "interoceptive"; sadness = "sad" or "sadness".
- Association: two-proportion z comparing the region's activation rate in studies using the term with that in all other studies with coordinates.
- Random-term null: all vocabulary terms used by 30-1000 studies at the same K, excluding the target terms.

## Tests

- **H1 (primary).** Conjunction: the minimum anterior-insula z over time perception, interoception and vestibular exceeds the minimum over random triples of null terms in at least 95% of 10,000 draws (one-sided p < 0.05).
- **H2.** For each of the three functions, anterior-minus-posterior z, located among the same difference for null terms (descriptive percentiles; the theory predicts anterior over posterior).
- **H3.** The mean anterior-insula z of fatigue, psychomotor, sadness and mood exceeds random 4-term means in at least 95% of 10,000 draws.
- **Sensitivity.** K = 1 and K = 3.

## Limits stated in advance

Term co-occurrence is not task content: articles mention terms in passing, and "vestibular" is a proxy for graviception (the vocabulary has no "gravity" or "graviception"); there is no term for heaviness. Coordinate reporting is uneven and the 10 mm kernel is coarse relative to insular subdivisions. A positive result shows literature-level co-localisation, not that one system mediates these functions or symptoms.