Martin P. Paulus
The senior author of the roadmap and, more importantly for the wiki, the source of the sentence its interoceptive-psychopathology page is built around. Paulus & Stein (2010), “Interoception in anxiety and depression,” proposed that mood and anxiety disorders arise from failures to appropriately anticipate changes in interoceptive states — and it is Paulus & Stein that the roadmap cites (ref 97) when it describes the pathological case as interoceptive input decoupling from prediction, presenting as “noisy afferent interoceptive inputs.” His framework is the clinical hinge between the predictive-coding theory the wiki tracks through Seth and Barrett and the actual disorders.
Anticipation as the locus of dysfunction
Paulus’s distinctive move is to locate psychopathology not in interoceptive sensing but in interoceptive prediction — the anticipatory limb. On this view an anxious brain is not one that feels its body too much or too little but one whose forecasts of upcoming bodily states are miscalibrated, so that ordinary fluctuations generate outsized prediction error. This is why he sits naturally at the head of a document trying to marry hierarchical Bayesian models to clinical categories: his 2010 claim is that clause of the computational theory, stated before the formal apparatus was imported.
Reach across disorders
His empirical programme spans the transdiagnostic list the roadmap assembles: interoceptive dysfunction in drug addiction (with Naqvi and Bechara, on the insula’s role in urges and decision-making), in anorexia nervosa (with Khalsa and Feusner, on altered caloric-anticipation interoception), and in anxiety and depression. That breadth is why interoceptive dysfunction reads as a dimension cutting across diagnoses rather than a feature of one.
The formal apparatus arrives (2021)
Petzschner et al. (2021) is where his 2010 thesis gets the machinery it was stated without. He co-authors the review that puts active inference an approach to interoceptive psychopathology (Paulus, Feinstein & Khalsa 2019) into a landscape alongside its rivals, and that proposes the translational pipeline (Paulus et al. 2016, “A roadmap for the development of applied computational psychiatry”) as the standard computational biomarkers must meet.
One consequence the wiki should hold against his framework: the same review names forecasting — modelling how internal states evolve in time — as the least-formalized arc of the sensory-control-loop. “Failure to appropriately anticipate interoceptive states” is a claim about a forecast, and the forecast has not been written down. His thesis is currently better supported as a clinical description than as a computational one.
The instrument, and where it came from (2012–2023)
The complaint above is about formalization. Measurement is a separate matter, and on measurement Paulus’s thesis has done better than the wiki previously recorded — through a paradigm lineage that starts with him and that he is not usually credited for here.
Paulus et al. (2012) introduced the inspiratory breathing load paradigm: cue an upcoming aversive breathing restriction, deliver it on some trials and not others, and ask what was experienced or what had been anticipated. Kruschwitz et al. (2019) turned the gap between those two ratings into interoceptive prediction accuracy, and Walter et al. (2020) put it in the scanner. That lineage is the operational form of “failure to appropriately anticipate interoceptive states” — his 2010 sentence with numbers attached.
Wilzok et al. (2023) is the current end of the line, and Paulus is in the acknowledgements (“additional advice provided by Dr. Martin Paulus”) rather than the author list — his junior LIBR colleagues Khalsa and Adamic carry it, with Croy’s Dresden group. It fixes the two defects of his original design (retrospective anticipation ratings, a single load level) and extends it to a second channel, pinprick pain.
What this changes and what it does not. It changes the standing of his thesis as an empirical programme: interoceptive anticipation is now a graded, reliable, per-trial, pre-stimulus quantity measurable in two channels — which it was not when the wiki last assessed him. It does not touch the forecasting complaint at all, and the reason is worth stating precisely. What every paradigm in this lineage measures is the anticipation of an experimenter-imposed event signalled by an arbitrary visual cue. The body never evolves under its own dynamics. So thirteen years of instrument development have produced a good measure of cue-outcome expectancy in an interoceptive modality, and still no measure of the thing his clinical claim is about — a forecast of how a bodily state will unfold.
The other unfinished business is that none of it has been run in a patient. The paradigms exist to serve interoceptive-psychopathology and have so far been validated only in healthy young volunteers.
Placement
The clinical-theoretical anchor of the LIBR cluster (sahib-khalsa, Feinstein) and the person whose 2010 thesis the roadmap’s psychopathology section elaborates. His disclosures on the roadmap are substantial (Warren Foundation and multiple NIH/NIDA grants; a LIBR affiliation and UpToDate royalties) — worth recording plainly, alongside the note that the roadmap is a Warren-funded, LIBR-organized document on which he is senior author.