Appearance
10.5 — Reading a State Against Reading a Character
A colleague is short with you in a meeting.
State reading: he is under pressure about something.
Character reading: he is a rude person.
The same observation, two conclusions, and they are not equally likely to be correct. State readings are usually right. Character readings are usually premature, and the slide from one to the other happens so fast that most people never notice they made it.
Why the slide happens
The fundamental attribution error, from 5.8: for other people's behaviour you overweight the person and underweight the situation.
The reason is informational rather than moral. You can see their behaviour and you cannot see their circumstances. From outside, the behaviour is all the data there is, so the behaviour becomes the person.
For your own behaviour you have the full context, which is why you were short with someone because of the morning you had, and he was short with you because he is like that.
And there is a second driver: a character conclusion is more useful, if true. A state tells you about today; a trait predicts every future interaction. Your mind is reaching for the more valuable conclusion, and paying for it with accuracy.
The three levels
Keep these apart and most reading errors disappear.
State — what is happening right now. Tired, angry, distracted, in pain, elated. Changes hourly. Frequently readable. Usually accurate.
Situation — what is going on in their life at the moment. A sick parent, a bad review, a new baby, money trouble, a divorce running quietly for eight months. Changes over months. Not readable, and knowable by asking.
Character — stable dispositions. Changes over years. Requires many observations across varied situations (10.11), and it is the level everybody jumps to first.
The middle layer is where the answer usually is
This is the practical core of the page.
Most puzzling behaviour is explained by a situation you do not know about.
The colleague who has become unreliable — a parent with dementia. The friend who stopped replying — a depressive episode. The manager who has become impossible — being managed out themselves. The relative who was cold at the wedding — an illness they have told nobody about.
None of these are visible, all of them are common, and every one of them produces behaviour that reads as character.
The habit that follows: when someone's behaviour changes or puzzles you, the first hypothesis is a situation you cannot see, and the second is character. Not the other way round.
And the way to test it is to ask. Not interrogate — one plain sentence. "You've seemed different the last few weeks. Everything all right?" People tell you a surprising amount when asked directly and without an agenda, and the answer is frequently something you would never have guessed.
When a state reading is enough
For most interactions, you do not need character at all, and pretending you do is what causes the trouble.
To handle this conversation, you need the state. Are they angry, upset, distracted, in a hurry? That is readable and sufficient (4.9).
To decide whether to raise something now, you need the state.
To help, you need the state and the situation.
Character only matters for a small number of decisions: whether to trust someone with something important, whether to enter a long-term arrangement, whether to rely on them, whether to be close to them.
Those decisions deserve real evidence. Everything else does not need the judgement at all, and making it anyway is where the error enters.
When character reading is warranted
It is not always wrong. The conditions are specific.
A pattern, not an instance. Three or more occasions.
Across different situations. Someone who is short when under pressure has a state. Someone who is short under pressure, when relaxed, at home, on holiday and at a party has a disposition.
Over time. Weeks or months, not one afternoon.
Where the situation does not explain it. If you know their circumstances and the behaviour persists beyond what those would produce.
And in the behaviour that costs them something. This is the strongest single indicator. What someone does when being decent is expensive tells you far more than what they do when it is free. Anybody is generous when it costs nothing.
One special case with an asymmetry worth naming: where safety is involved, a single instance is enough. Someone who has hit you once has told you something about their character, and the pattern rule does not apply. 13.6 is explicit about this.
The reverse error
The opposite mistake is less discussed and it does real damage: explaining away a genuine pattern as a state, indefinitely.
He's just stressed. She's been under a lot of pressure. That's not really him.
Said once about one incident, this is generous and usually correct. Said for three years about a recurring behaviour, it is how people stay in situations they should have left.
The distinguishing question: how long has this state been going on? A state that has lasted two years is not a state. At that point, whatever the original cause, it is how this person behaves, and you are dealing with it rather than with a temporary condition.
And note what the state explanation does to you. It is a reappraisal (4.6) and it is the correct tool for a one-off. Applied to a pattern, it is the mechanism that keeps people in harmful situations for years, and 13.8 covers what to do about that.
Reading yourself the same way
The error runs inward too, in the opposite direction.
For your own failures you use situations — I was tired, it was a difficult week — and for your own worst moments you sometimes jump to character: I'm a bad person, I'm someone who ruins things.
Both need the same correction as with other people. The situation is usually part of it, and one instance is not a character (4.8).
The useful discipline: apply the standard you would apply to a friend. From 9.6: you would look for their situation before concluding about their character. Do it for yourself, and — the harder half — do it for the person who has annoyed you.
What to do with this
Insert one question between the observation and the conclusion: what would make a reasonable person behave this way? Four seconds. It generates the situational hypotheses your mind skipped.
Then ask, if it matters. One plain sentence, no agenda.
Require three instances across different situations before concluding anything about character — with the safety exception, which requires one.
Notice how long you have been calling something a state. Two years is not a state.
And apply it in your own favour and in theirs equally. The generosity that comes easily for yourself and hard for others is the exact asymmetry this page is about, and correcting it in both directions is what accurate reading actually means.
Next: 10.6 covers the one thing everybody believes they can do from these signals and almost nobody can — detect a lie.