Shame and guilt

The distinction was Helen Block Lewis's, in 1971, and it fits in one line: guilt is about the act, shame is about the self. I did something bad. I am bad.

Everything else follows from that sentence, including the part that surprises people — the two feelings do not lead to similar places with different intensities. They lead in opposite directions. Guilt points at the thing that happened and asks what can be done about it. Shame points at you and asks how to get out of the room.

They arrive in the same situations, often in the same minute, so the test is not the trigger. It is what your attention does next.

Guilt is specific and forward-facing. The thought has an object and a verb: I snapped at her. I forgot the thing I promised. It comes with tension and an urge to do something — apologise, fix, replace. It is unpleasant and it is workable, and it usually ends when the thing is addressed.

Shame is global and about exposure. The thought has no verb in it: I am a terrible friend. There is something wrong with me. The body goes with it — the heat, the wish to be smaller, the involuntary look away. And the urge is not to fix but to vanish: leave, change the subject, stop replying.

The reliable tell is the replay. Guilt replays the scene. Shame replays the audience. If what loops at midnight is what they must think of you rather than what you did, you are in shame, and the entry on rumination describes what that loop costs.

Tangney and colleagues spent two decades measuring shame-proneness and guilt-proneness separately, and the results are among the cleaner findings in emotion research.

Guilt-proneness looks good almost everywhere. It predicts apologising, making amends, empathy for the person harmed, and — in the longitudinal work summarised in the 2007 Annual Review — lower rates of later offending and substance problems. Guilt keeps the person and the deed apart, which leaves someone available to fix the deed.

Shame-proneness does not. It is associated with anger, defensiveness and the externalisation of blame. Stuewig and colleagues traced this directly: shame-proneness predicted blaming others, which in turn predicted aggression, across four separate samples including children and jailed adults. The intuition that shame is at least a sign of conscience does not survive the data.

Shame reliably produces hiding. Which is the practical problem. The person in shame is the least likely to apologise, precisely because apologising means staying in contact with the person who saw them. This is why "he never even said sorry" is so often the wrong inference from the right observation.

They are not different amounts of the same thing. Both feelings can be measured in the same person after the same event, and they predict different behaviour. Any advice that treats them as interchangeable — "a bit of shame is healthy, it keeps you honest" — is arguing against the evidence.

Almost everything people do to improve themselves runs on one of these two engines, and most of it runs on the wrong one.

A test result, an honest piece of feedback, a pattern you finally see in your own behaviour — each can be read as this is a thing I do, which is guilt-shaped and useful, or as this is what I am, which is shame-shaped and terminal. The second reading is not more honest. It is simply more final, and it happens to end the work.

This is the mechanism behind three other entries here. Facing disowned material, in the shadow self, only works when the material is survivable to look at — shame is exactly what makes it unsurvivable, so shame is what pushes it back into the dark. Self-compassion is not a softer alternative to self-improvement; it is the condition under which accurate self-assessment stays possible, and the research says self-compassionate people take more responsibility, not less. And a fixed reading of yourself is what the Growth Mindset test is about: the belief that the trait is the person is shame's premise, stated as a theory.

It also explains an ugly relationship pattern. Shame converts smoothly into anger, because anger restores size. That is the sequence behind many disproportionate reactions to small criticisms — and why the person who feels worst about a fight is sometimes the one who shouted.

The move is always the same: convert the sentence back to a verb.

• Write the finding as an action. Not "I'm a bad friend" but "I did not call her back for three weeks." One of those can be fixed by Thursday.
• Name the exposure. Shame needs to have been seen, even if only by you. Naming who you feel seen by usually shrinks it, because the answer is often one person whose opinion you have installed as a jury.
• Repair beats confessing. A specific apology for a specific thing settles guilt. A general confession about what you are like invites reassurance, which does not settle anything, and puts the labour on the person you harmed.
• Do not use shame as fuel. It is a poor engine with a known failure mode: it works for about a week and then produces avoidance of the whole area.
• Watch the direction of the words. "I feel bad" is ambiguous and the ambiguity is not harmless. One more notch of granularity usually decides the whole evening.

A boundary worth stating plainly. Shame is a normal emotion and everybody has it; that is not the concern. What is worth taking seriously is shame that has stopped being about anything — a steady background conviction of being defective, present when nothing has happened, or shame attached to something that was done to you rather than by you. Neither of those is a self-knowledge problem, and neither yields to better framing. That is work to do with a professional, and it is common enough work that asking for it is unremarkable.