Your Reaction Is Not Evidence

You're in a meeting where somebody else's project is being discussed. Your heart is going fast, your palms are damp, your breathing is shallow. The thought arrives by itself: I don't belong here, I'm not up to this level.

Now the simpler version: you drank a third coffee on an empty stomach.

The body produced the same set in both cases. The only thing telling them apart is the explanation, and the explanation gets picked from whatever was to hand.

The body doesn't report what it's feeling

In 1962 Stanley Schachter and Jerome Singer ran an experiment that's still foundational for the whole subject of emotion.

Participants were told the study was about a vitamin's effect on vision, and given an injection. In fact they received adrenaline or a placebo. Adrenaline produces a recognisable set: racing heart, trembling, quickened breathing. Some participants were warned about side effects, some weren't.

Each was then put in a room with a person who was actually an actor. In one version the actor behaved cheerfully and clowned around; in the other he was irritable and annoyed.

The result went like this. Those who'd been warned about the injection attributed their state to the injection and stayed more or less calm. Those who hadn't been warned attributed it to the setting: next to the cheerful actor they felt cheerful, next to the irritated one they felt irritated.

Same substance, same physiology, different emotions. Out of this came the two-factor theory: an emotion requires two things, bodily arousal and a label assigned to it. The body supplies the intensity, the explanation supplies the name.

In 1974 Dutton and Aron took this out of the laboratory. Men crossed a ravine either on a long, swaying suspension bridge or on a low stable one. On the far side a woman met them with a questionnaire and left her phone number in case of questions.

The men from the swaying bridge called noticeably more often.

The heart was pounding from the height, and a more agreeable explanation presented itself. Though it should be said honestly: that study has been criticised a great deal since, because the arousal level there was assumed rather than measured. The two-factor theory holds; that particular bridge is shaky in more ways than one.

What follows from this

Arousal is non-specific. A racing heart from fear, from excitement, from coffee, from too little sleep and from five flights of stairs are physiologically similar. The difference arises not in the body but in the caption you attach.

And the caption comes from context. From whatever is nearby. From whatever you happened to be thinking about. From the explanation you reach for most often.

It worked like this for me. I turned up to a strength session, and halfway through I felt frankly awful: heart pounding, nausea, the sense that something was about to happen. Two explanations arrived instantly: stress I'd apparently accumulated without noticing, and a heart problem, which I diagnosed myself with on the spot.

The real cause was duller. That day I'd drunk about two litres of coffee, and then added water on top during the session. Neither stress nor my heart had anything to do with it.

The interesting part: I didn't invent those versions deliberately. They arrived on their own and sounded convincing. The body produced a signal, and the head found it a story out of whatever was lying closest.

The link to the pillar article is direct: the detector went off, and that's all that's known. A detector going off doesn't report what's burning. It reports that a threshold was crossed.

The formula for this cluster: the intensity of the reaction is taken as the reliability of the conclusion. The harder it hits, the truer the explanation feels, although there's no connection whatsoever between force and truth.

Four traps in this cluster

1. Emotional reasoning. I feel it, so it's true

What it feels like. An emotion is processed as data of the same quality as sight or hearing. The brain doesn't check the source: if fear arrived, the danger is real. If shame arrived, there's something to be ashamed of.

Example. "I feel worthless, so I am worthless." There's no transition at all between the two halves of that sentence; the proof has been skipped, and skipped invisibly.

When it isn't a trap. When the emotion is a signal rather than a verdict. Unease around a person who later turned out to be dangerous was information. The difference is what happens next: an emotion says "look over here", not "the question is settled".

What helps. Put the missing word back in. Not "I'm worthless" but "I FEEL worthless". The difference looks cosmetic and it restores what was lost: the feeling stays a fact, and the conclusion drawn from it becomes a hypothesis you can test.

2. Tunnel vision. The vivid thing crowds out the rest

What it feels like. Attention is captured by whatever produced an emotional response, and the rest stops existing. Not ignored, but literally out of the frame, because the attention budget went where the anxiety is.

Example. In a project with forty tasks, one was done badly. You see that one. The other thirty-nine aren't disputed, they simply aren't in shot, and the conclusion "the project failed" feels like a description of reality.

When it isn't a trap. When one detail genuinely is decisive. One error in a load calculation cancels out a good bridge design. The question isn't whether you may focus on a problem, but who chose the focus: the consequences, or the strength of the emotion.

What helps. Widen the frame physically. Write out all forty tasks and look at the list with your eyes. It works for the same reason counting works in the mental filter: inside your head the frame doesn't widen, the budget is already allocated, so the picture has to be taken outside.

3. Minimising effort. If it's hard, I'm doing it wrong

What it feels like. Effort spent reads as a sign of error. If I were doing this right, it would be easy. Since it's hard, either the method is wrong or something is wrong with me.

Example. You got the work done, and it cost you everything. Someone else did something comparable and apparently didn't break a sweat. The conclusion arrives automatically: so I'm doing it wrong.

When it isn't a trap. When effort really does signal a poor method. If you're manually consolidating data that a built-in function handles, the difficulty is useful information and worth listening to.

What helps. Separate the two questions glued together here: how much resource did this cost, and was it done well. The second is judged by the result, not by the sensation during the process. And hold on to the obvious: the same action costs different people different amounts of resource, and that's a measure of their expenditure, not of their worth.

4. Reverse telepathy. It's obvious to me, so it's obvious to them

What it feels like. Your own state registers as plainly visible from outside. Hence the expectation that people close to you will work it out themselves: see that you're exhausted, understand that they hurt you, realise you need help.

Example. "He should have understood I was upset without me saying it." The hurt was obvious from the inside. From the outside there was a level face and the words "everything's fine".

When it isn't a trap. When the signal genuinely was sent and ignored. You said it plainly, you were heard, and nothing happened: that isn't about telepathy any more, that's about the person.

What helps. Say it in words. "I'm upset about this" requires no special abilities from the other person and, more to the point, doesn't put them in an exam they didn't know they were sitting. A clear request isn't weakness or defeat, it's the only interface that works.

What they have in common

Four traps, one substitution.

Emotional reasoning takes the strength of a feeling as proof of a conclusion. Tunnel vision takes the vividness of a detail as its importance. Minimising effort takes difficulty as a sign of error. Reverse telepathy takes internal obviousness for external obviousness.

In each case the intensity of an internal signal substitutes for a check. And the substitution happens before thinking switches on, so the conclusion doesn't feel like a conclusion. It feels like an observation.

Which explains why arguing with yourself mid-episode is useless. You're trying to refute the content, and the content didn't come from there: first the reaction arrived, then an explanation attached itself. A refuted explanation is simply replaced by another one, because the reaction hasn't gone anywhere.

What to do: separate the signal from the conclusion

Name the reaction as a reaction. Not "everything is terrible" but "I'm extremely anxious right now". That isn't denying the feeling, quite the opposite: the feeling is acknowledged in full, and it's the explanation that stays an open question.

Check the boring causes first. How much you slept. When you ate. How much coffee. What day of your cycle. Not because feelings reduce to physiology, but because the intensity often comes from there while the explanation gets glued to whatever caught your eye.

Take the picture outside. A list, a count, a written note. Anything that widens the frame works better than talking yourself down, because talking yourself down runs on the channel that's already occupied.

Wait, if waiting is possible. A reaction has a duration, and an hour later the same question looks different. The rule that important conclusions aren't drawn at the peak sounds banal right up until you start following it.

And the professional part, because this is my job. Every control system separates two things: the sensor reports that it tripped, and the decision is made by logic that has other inputs too. A sensor wired straight to an actuator gives you a system that stops the line at every sneeze. Such systems exist, they're called unstable, and the first thing done to them is inserting a check between signal and action.

The reaction is a sensor. The conclusion is a decision. There should be something in between.

All cognitive traps >

That was the fifth and final cluster. All the breakdowns live on the cognitive traps page.