Emotional reasoning
Anxious people read their own anxiety as evidence that a situation is dangerous, so the feeling of threat becomes a reason to believe in the threat.
Anxious people read their own anxiety as evidence that a situation is dangerous, so the feeling of threat becomes a reason to believe in the threat.
The claim is about the direction of an inference. Fear normally follows danger, and the reasoning under test runs the other way, from the feeling back to the world. Arntz, Rauner and van den Hout called it ex-consequentia reasoning, inferring the premise from the consequence. Its clinical interest is the loop it implies. Anxiety indicates threat, the inferred threat sustains the anxiety, and the person never finds out that the situation was safe. The task built to test it separates two sources of information, what a scene objectively contains and what the person feels, then asks which one the danger rating follows.
Repeats disagree, or the effect holds in some conditions and not others. The other 248.
Mixed The danger-inflating effect turns up again in anxious and fearful samples, and the result that separated patients from everybody else does not.
Engelhard, Macklin, McNally, van den Hout and Arntz gave the scenarios to Vietnam combat veterans. Veterans without PTSD inferred danger from the objective information, while veterans with PTSD also inferred it from anxiety and from intrusive memories, and both effects were largely mediated by perceived uncontrollability. The 2022 systematic review records this as the one included study that used Arntz's scenarios without modification.
Verwoerd and colleagues carried the design to disgust, twice. The first study compared 31 people above the clinical range on contamination fear with 27 selected as low, on scripts that varied objective threat of contamination and the actor's disgust. The second compared 35 people high on a measure of fear of vomiting with 38 low, on scripts that were all objectively safe and that varied disgust and anxiety separately.
High scorers in both inferred risk of illness from the disgust, and in the second study disgust rather than anxiety drove the effect. The control-group result is the part that has not held. Muris, Merckelbach and van Spauwen found a general emotional reasoning effect in 101 primary school children, with anxiety sensitivity and trait anxiety predicting how far a child leaned on anxiety information in the safety scripts.
Morren, Muris and Kindt found the effect in 508 children aged 7 to 13, and Morren and colleagues found it on two occasions ten months apart in 122 non-clinical children. Berle and Moulds gave an adapted task to 70 students split into 28 low and 42 high dysphoric on the Beck Depression Inventory. Both groups showed emotional reasoning on the dysphoria scripts, and the groups did not differ.
On the panic script carried over from Arntz neither group showed the effect at all, which the authors report as a contrast with the 1995 result rather than as a tie.
The safe route that feels wrongillustration
A cyclist takes a lit, busy road home, feels frightened anyway, and concludes from the fear that the road must be dangerous. What the street actually contains points one way, and the feeling is allowed to settle the question instead.
Dread before a meetingillustration
Someone feels a wave of dread before speaking at a meeting and takes it as a sign the talk will go badly. Nothing about the slides or the audience has changed. The forecast is read off the nerves rather than off the situation.
The child at the sleepoverillustration
A child feels uneasy at a friend's house and decides the house is unsafe rather than that she is homesick. Because she goes home, she never finds out the evening would have been fine, and the next invitation feels worse.
Cases marked as illustrations describe nobody in particular. They are there to make the pattern recognisable, not to report that a particular thing happened.
The 1995 experiment gave people short scripts and asked how dangerous the situation was. Each script crossed two factors: objective safety against objective danger, and an anxious response against a calm one. Gangemi and colleagues quote the panic script from page 919. The reader is in a crowded lift, and the ending either has a passenger falling into their arms as a welcome chance, or has the lift stuck between floors with two people fainting, each version running with or without the line that the reader suddenly becomes anxious.
A danger rating that rises with the anxious ending counts as emotional reasoning. The abstract reports 52 spider phobics, 41 panic patients, 38 social phobics, 31 other anxiety patients and 24 normal controls. Patients' ratings followed the anxiety information as well as the objective information, and the controls' ratings followed the objective information alone. The abstract adds that the effect was neither situation-specific nor confined to panic patients, so patients did it on scripts unrelated to their own diagnosis.
1 primary research8 replication and reanalysis2 commentary and critique
The abstract opens by saying the idea had already been suggested, that neurotic patients draw invalid conclusions from their own emotional response. Gangemi, Dahò and Mancini trace the description to Beck, Emery and Greenberg's 1985 book on anxiety disorders and quote it from page 198; that book was not read here.
The claim, and the replication this entry reads its figures from.
The evidence base is small and comes from one tradition. The 2022 systematic review searched PubMed, PsycInfo, Scopus and The Cochrane Library and included nine articles, the earliest from 1995 and the latest from 2016. Five were run in the Netherlands, three in Australia and one in Italy. Seven were case-control studies scoring 7 to 9 out of 12 on the review's quality tool, and one of the nine justified its sample size.
No meta-analysis, multi-site replication or preregistered replication of the 1995 experiment turned up in searches run on 21 September 2026 of Europe PMC and Crossref, and of the reference lists of the 2022 review and of Gangemi, Dahò and Mancini. The tasks rate imagined scenarios rather than felt emotion. The review says the authors of the included studies generally grant that, and Verwoerd and colleagues state it about their own work.
The 1995 paper was not obtained on that date, and the figures here come from its PubMed abstract. That matters, because the review's table describes the study as 133 people, 119 of them patients, with 9 spider phobics, where the abstract gives 52. The review's other group sizes of 41, 38, 31 and 24 match the abstract exactly, so the disagreement sits on the spider phobia group alone.
Gangemi and colleagues refer to a first experiment, so the paper may report more than one, but a single mistyped figure fits the pattern better, and this entry cannot settle it from the abstract.
Each of these is named in this entry's own prose as something it gets taken for. Where the two came out differently, that is worth knowing before you quote either.
The finding is often read as a line between anxious people and everybody else. Non-clinical children show the effect whatever their anxiety level, and dysphoric and non-dysphoric students showed it alike, so symptoms track the size of the tendency rather than its presence. The name also invites confusion with the affect heuristic, entry no. 12, which comes from the judgement-and-decision-making literature and is tested against risk and benefit ratings and against probability weighting.
Both describe a feeling standing in for evidence, and their evidence bases do not meet. No test of Arntz's task on the scale of the Many Labs 2 attempt that reversed one affect heuristic prediction appeared in the Europe PMC and Crossref searches of 21 September 2026. The depression case is also not the anxiety case. In a separate study Berle and Moulds compared 27 currently depressed people with 25 never depressed and 26 previously depressed, and none of the differences in emotional reasoning reached significance. That is not the same as no differences at all, because the previously depressed did engage in more non-self-referent emotional reasoning than the never depressed.
Every claim on this page was held against these sources on 2026-09-21. Nothing here is written from memory. Written and checked by Krishna Chagti; how entries are written and corrected. Found a mistake? Report it, with the sentence and a source.