Two-factor model of relapse vulnerability
The clinical model that organizes Farb’s depression-imaging programme, and the reason the wiki’s “Week 11: Absence of Emotion” papers hang together. Its claim: a recovered depressive’s risk of relapse is not one thing but two dissociable things, which differ in what they are and — crucially — in whether treatment can touch them.
Name collision, flagged once. This is not the Schachter-Singer two-factor theory of emotion (arousal + cognitive label). Same phrase, unrelated claim. This page is about depression relapse.
The two factors
1. Static factor — sensory deactivation (a fixed trait). Under dysphoric mood, sensory and bodily representation is turned down — deactivation across somatosensory, motor, insular, and visual cortex. In vulnerable patients this deactivation is deeper, and it is treatment-invariant: present equally before and after prophylactic therapy, and correlated with past episode count and residual symptoms alike. It is the enduring vulnerability — and, discouragingly, the one neither MBCT nor WB-CT moved in Farb et al. (2022).
2. Dynamic factor — prefrontal cognitive reactivity (a treatment target). Under the same dysphoria, prefrontal cortex over-engages in cognitive elaboration (ruminative self-reference). Unlike the static factor, this responds to treatment: in patients who stay well, prefrontal reactivity falls over the course of therapy; in patients who relapse, it does not. It is the modifiable half of the model.
The two are not independent lesions. The paper’s PPI shows the relationship: under dysphoria, prefrontal elaboration actively inhibits the sensory representation — a positive-to-negative connectivity flip between lateral PFC and the somatosensory marker, whose magnitude predicts relapse. So the unifying claim is a single trade-off: over-reliance on cognitive elaboration to the detriment of sensory integration.
The evidence, across four papers
| paper | contribution to the model |
|---|---|
Farb et al. 2010 (sadness/MBSR; pending ingest, Week 11) | dysphoria deactivates right middle insula / somatosensory; deactivation tracks concurrent symptoms |
| farb-2011-relapse-prediction | first prospective link: elaborative (mPFC) reactivity predicts relapse, sensory (visual) reactivity protects — but one timepoint, so static vs dynamic cannot be separated |
Farb, Irving, Anderson & Segal 2015 (J Abnorm Psychol; cited, not in raw/) | the model stated as “a two-factor model of relapse/recurrence vulnerability” |
| farb-2022-relapse-biomarkers | RCT with pre/post scans separates the two factors by their treatment response: static somatosensory deactivation + dynamic left-DLPFC reactivity, combined C = 0.86 |
The 2022 paper is what makes it a genuine two-factor model rather than a single sensory-vs-elaborative axis: only a design with two scans, before and after an intervention, can show that one pole is fixed and the other moves.
Why it lives on an interoception wiki
The model is the clinical backbone of Farb’s dual-mode interoception framework. The static/sensory pole is perceptual-inference (stay with the sensation); the dynamic/prefrontal pole is active-inference (elaborate, regulate, problem-solve) gone maladaptive; decentering is the trainable shift from the second toward the first. The honest boundary the wiki keeps visible: the “sensory” pole measured is largely exteroceptive (visual in 2011; somatosensory/motor/visual in 2022), with the interoceptive insula present but not dominant — so the model motivates an interoceptive reading rather than proving one. See farb-2022-relapse-biomarkers for the full caveat and is-more-interoceptive-awareness-better for how the “which mode, not how much” verdict feeds the debate.
The unresolved clinical tension it creates
If the sensory-deactivation factor is the enduring trait vulnerability and no prophylactic therapy moved it, then existing treatments work only on the modifiable prefrontal factor — leaving the deeper vulnerability intact. Farb et al. (2022) speculate sensory deactivation might yield to skills consolidated over longer timescales, but their own data show it did not budge in eight weeks. This is the model’s sharpest open problem, and an argument (untested) for interventions that target sensory/bodily representation directly rather than cognitive reactivity — which is what the wiki’s body-awareness applied pages (interoceptive-training-clinical, mindfulness-interoceptive-training) implicitly propose.