Anke Ehlers

The clinician whose findings gave the heartbeat-counting task its stakes, and the principal defender of reading its score as a continuous measure of a real ability.

She arrived in the wiki the way Schandry did — as a name that had been doing work on several pages without a page of her own, via a paper she is not first author on (the 2000 reanalysis). That is no longer the whole story: Ehlers (1993) is now in raw/, and it is the first time the wiki reads her first-hand and sole-authored — her own review of her own programme, at the moment before the sceptics arrived.

First-hand at last: the 1993 review

Ehlers (1993), “Interoception and panic disorder” is the source this page was written around the absence of. It supplies, in Ehlers’s own framing, the three characteristics of panic patients her programme identified — enhanced cardiac perception, an attentional bias toward physical threat, and rating bodily symptoms as dangerous (fear of fear) — and a fourth hypothesis she tested and rejected: panic patients are not generally more autonomically reactive (Roth et al. 1992), so the disorder is about perceiving and catastrophizing ordinary signals, not about a hot body.

What is most useful for this wiki is how much of the eventual sceptical case Ehlers had already conceded in 1993, as the position’s protagonist:

• the cardiac-perception effect is paradigm-specific (present on Schandry counting, absent on tone-pip matching and Katkin discrimination) — a claim about the panic-patient group difference, and not contradicted by Wiens et al. (2000), who find an individual-differences emotion-intensity effect on the Katkin task in healthy undergraduates: different question, and arguably the pattern an attentional/belief account of the panic effect predicts;
• only a subgroup of patients shows accurate perception;
• the design is correlational and “cannot determine cause–consequence relationships”;
• and — the striking one — she offers the deflationary reading herself: the group difference “may not reflect differences in the subjects’ general ability to perceive their heartbeats, but differences in the habit of attending to bodily cues.”

So the “publishing against herself” pattern below did not start in 2000. The 1993 review already contains the caveats; the reanalysis operationalized them. See is-the-heartbeat-counting-task-valid.

The finding that started the programme

Ehlers & Breuer (1992): panic disorder patients count their heartbeats more accurately than normal controls, infrequent panickers, and specific phobics. In a second study, panic and GAD patients both beat depressed patients. Replicated in Ehlers et al. (1995), where panic patients again had significantly lower error scores than controls.

This is the result that turned a psychophysiology curiosity into a clinical variable, and it fit the cognitive model of panic neatly: patients who attend to their bodies constantly should detect what is happening in them.

Replicated from outside her lineage. Until now every version of this finding the wiki held was Ehlers’s own or the reanalysis she contributed to. Zoellner & Craske (1999), an independent UCLA lab, reproduced it in an analogue sample — infrequent panickers counted more accurately than nonanxious controls (44.7% vs 61.7% error), and the difference survived covarying felt confidence and felt cardiac arousal. That the group difference travels beyond Marburg, and beyond diagnosed patients, is a genuine point for the pro-validity side of is-the-heartbeat-counting-task-valid — even as the same study’s within-subject anxiety effect hands the sceptics their artefact (see that page).

The finding with the teeth

Ehlers (1995): good heartbeat perception predicted poor treatment outcome and recurrence of panic attacks after initial remission, in a one-year prospective study.

This is the strongest clinical “more interoception is worse” result the wiki has, and it is prospective rather than cross-sectional, which is more than most rows on is-more-interoceptive-awareness-better can say. The wiki carries the published 1995 paper by citation only — it is not in raw/. But the 1993 review now gives its preliminary form first-hand: 17 remitted panic patients retested after a year, the 6 who relapsed having shown significantly better baseline heart-rate perception (unexplained by sex, remission length, or anxiety/depression). Ehlers flags it as needing replication in a larger sample — which the 1995 paper presumably supplied. So the finding’s origin and its author’s own hedging are now on the record here, even if the published effect sizes are not.

Van der Does et al. (2000) propose it is really about anxiety-sensitivity: accurate perceivers score higher on the ASI and on nothing else, so “good perceivers do worse” may be “people who believe their bodily sensations are dangerous do worse,” which is unsurprising and not about perception at all.

Her position on the measurement, stated fairly

Ehlers holds that the heartbeat-counting task is valid, with three arguments recorded in the 2000 reanalysis:

1. Test-retest reliability is adequate (Ehlers & Breuer 1992; Van der Does et al. 1997 — her opponents agree).
2. Almost all participants undercount, which is the logical outcome if people accurately perceive their heartbeats but miss a few.
3. Performance correlates with stroke volume (Schandry, Bestler & Montoya 1993) — so participants are demonstrably responding to real cardiac events.

And time estimation — the obvious alternative strategy — has been ruled out in several studies (Ehlers & Breuer 1992).

Given validity, dividing participants into accurate and inaccurate imposes an artificial boundary on a continuum, which is why every paper of hers after 1992 reports only % error. That is a principled position, not an oversight.

The inversion worth recording

Ehlers cites the stroke-volume finding as evidence of validity. This wiki carries the identical finding as the cardiodynamic confound that undermines the task.

Both readings are coherent, and they are not compatible:

• Ehlers: if counting performance tracks cardiodynamics, participants are responding to real cardiac events, so the task measures cardiac perception. Validity.
• This wiki (following Oldroyd et al. and Paulus & Stein 2010): if a louder heart produces a better score, the score indexes the signal’s amplitude rather than the perceiver’s skill, and any group difference where the groups differ in autonomic tone is confounded at the root.

The 2000 reanalysis contains the datum that adjudicates this, and neither side frames it that way: exercise raises accuracy above ~100 bpm and it decays to baseline by ~95 bpm, equally in patients and controls, with exactly one of 60 participants durably improving. Ehlers’s inference survives — the task does track real cardiac events. What does not survive is the construct: a score you can raise by making the heart beat harder is tracking the heart, not the perceiver. See is-the-heartbeat-counting-task-valid.

Publishing against herself

The 2000 reanalysis, which she co-authored, does the following to her own work:

• reduces her group difference to a claim about prevalence (17.1% of panic patients vs 7.9% of controls are accurate) and shows it is not specific to panic — no difference from social phobia, GAD or specific phobia
• overturns Ehlers & Breuer’s (1996) conclusion that good heartbeat perception is a stable individual characteristic: fewer than half of accurate patients were still accurate after treatment
• shows her distraction manipulation, which she had reported as having minimal effect on % error, pushes 35% of accurate perceivers into the inaccurate category when scored categorically — the two scores disagree about her own manipulation on her own data
• and relocates her most striking clinical finding from perception to anxiety-sensitivity

She contributed the data and signed the paper.

This wiki has now seen the same pattern twice on the same instrument — Schandry publishing the confound against his own task, Ehlers co-authoring the reanalysis against her own conclusions. Recorded as a fact about this literature’s character, and a favourable one: the field’s protagonists have repeatedly supplied the evidence against themselves, and the field has mostly not noticed. Post-2000 interoception research went on reporting continuous heartbeat-counting scores from unselected samples — Ehlers’s practice — without engaging the argument that the practice is unsound.

Elsewhere

Ehlers is at least as well known for the cognitive model of PTSD (Ehlers & Clark 2000) as for panic, and that work is entirely outside what this wiki has read. Nothing on this page should be taken as a summary of her career; it is a summary of her role in one measurement dispute. Her PTSD work would be the natural bridge to interoceptive-training-clinical’s trauma section, and the wiki has nothing on it.