Part two of a three-part series from Ordered Empathy.
Why reasoning with someone who is spiralling fails — and the one safeguard that still works when questions have stopped working.
Part one traced a descent: an ordinary mother, ordinary motives, arriving at an unthinkable act without ever believing a single false thing. Each step was a true sentence with a qualifier quietly removed. I’m failing them — in this respect. I’m the danger — under these conditions. There’s no way out — that I can presently see.
That leaves an obvious practical question. If someone you love is three or four steps down that road, what do you actually say?
The instinct is to argue. It is almost always the wrong move, and now we can say precisely why.
Why arguing makes it worse
You are trying to disprove sentences that are true. She can feel that you’re wrong, because you are.
Is she failing them in some way? Yes — she can name it. Has she looked for a way out and not found one? Also yes. Every time you negate one of her claims, you hand her evidence that you don’t understand the situation. And one of the later rungs on that descent is precisely no one else sees how bad this is.
So the argument doesn’t just fail. It confirms the thing that most needed challenging. You have volunteered as proof.
This is the part people find counterintuitive: the more forcefully you contradict her, the more securely you install the belief that she is alone with the truth.
What to do instead: restore, don’t refute
You can’t refute her. You can only give her back the words that were taken out.
That is a different conversation and it sounds different.
Not you’re wrong about failing them. Instead: “Failing them at what, exactly? And since when?”
Not you’re not dangerous. Instead: “Dangerous when? Was it true last March? What’s different now?”
Not things will get better. Instead: “Is it that there’s no way through, or that neither of us can see one from here? Those aren’t the same.”
Every one of those is the same move. The total claim was built by stripping particulars off a true one, so the particular is the antidote to the total. You’re not disagreeing with her. You’re asking her to be more precise than the spiral will allow, and precision is the one thing it cannot survive.
Notice that none of these are clever. They’re boring questions. That’s a feature — a clever question sounds like a trap and gets treated as one.
One serious warning about that
This only works as a shared operation, and getting it wrong causes real harm.
Ask a well-rested person in what respect, exactly, am I failing them? and they come back with two items. The sweeping claim dies for lack of support. That’s the technique working.
Ask an exhausted, depressed person the same question and she comes back with eleven — each of them true, each surfaced precisely because her state went looking for that kind of memory. She has now documented the claim you were trying to bound. You handed her a magnifying glass and she found more of exactly what she was already looking at.
What makes the technique work in a clinician’s office is that someone else is in the room to demand the other half of the search: “Now name three where you didn’t fail.” Nobody issues that demand to themselves.
A search that only runs in one direction isn’t an examination. It’s a prosecution.
So: pair every particular you ask for with a particular in the other direction, of equal length. If you’re not willing to run both halves, don’t run the first one.
The step where questions stop working
There is one rung in that descent that is different in kind from all the others, and everything turns on it.
Up to a point, every corruption is a corruption of the situation — how bad things are, how permanent, whose fault. Bad enough. But then comes this:
Hesitating now would be cowardice.
And the corruption has reached the brake itself.
Hesitation gets renamed. It becomes selfishness. Squeamishness. The failure to do what love actually requires. And once that renaming is finished, something happens that I think is the single most important fact in this whole subject:
She is no longer overriding her conscience. She is obeying it.
The internal argument was never I don’t love them anymore, so I can do this. It was because I love them, I have to. The love was never defeated at any point. It was recruited.
Which is why every intervention built on appealing to her love arrives too late. Her love is what’s driving. It is the most active thing in the room, and it is now pointed at the act.
Why self-examination cannot be the safeguard
If the danger sits in the judgment rather than the will, the safeguard has to operate on the judgment. There are good questions for that: What am I feeling, and what is that feeling making seem true? How did I get from this to that? What would show me I’m wrong? Say plainly what you’re proposing to do — not what you hope it accomplishes.
Those are real tools and they work near the top of the descent.
But there’s a structural problem with all of them, and Aquinas saw it clearly. Under strong passion, he says, a person may still fully possess the right principle and simply fail to consider it. He doesn’t lose the knowledge. He never reaches for it.
That is fatal for any self-administered check. A self-examination requires exactly the faculty that the crisis disables. The person who most needs to run it is the person least able to. And there is no version of try harder to be honest with yourself that fixes this, because trying harder is running on the same compromised equipment.
Put it plainly: a rule you check inside your own head always passes.
The one safeguard that holds
So it can’t be internal. It has to be another person.
Aquinas names the relevant virtue, and it’s an unfashionable one: docilitas — teachableness. Not gullibility, and not deference to anyone who happens to be talking. It’s the settled willingness to be corrected by a particular person’s judgment. His argument for why it’s necessary is almost brutally practical: no individual can weigh all the particulars of a real situation alone. So he quotes Proverbs at us — lean not on thy own prudence.
I used to read that as gentle advice about humility. It isn’t. It’s an engineering specification.
And here’s why it has to be installed in advance. One of the late rungs is no one else understands. Once that’s in place, disagreement stops being evidence against her and becomes evidence for her — proof that the others can’t see it. The system seals itself. You cannot recruit an outside judgment from inside a representation whose entire structure is that outside judgments don’t apply.
Which leaves exactly one thing that works, and it isn’t a question you ask yourself:
Name the person whose judgment you have pre-committed to accept on decisions of this kind — chosen back when you were well.
If you can’t name them, you don’t have a safeguard, however good your questions are.
It’s worth doing out loud, now, while nothing is wrong. It sounds roughly like this:
If I ever come to you and tell you I’ve worked something out, and it’s drastic, and I’m the only one who can see why it’s necessary — that’s the moment I need you to not believe me. I’m telling you now, while I’m fine, that you should trust this version of me over that one.
That is not a rule. Rules you check in your own head always pass. That’s a person, and a person can say no.
Three situations that look identical and are not
One necessary distinction, because getting it wrong does damage.
Everything above describes someone who is deliberating — building a case, step by step, toward a conclusion.
It does not describe the mother who has a sudden intrusive, horrifying thought about harming her child and is frightened by it. That is extremely common after birth, and it is close to the opposite condition. She isn’t building a case. The thought arrived unwanted and she wants it gone. Handing her say plainly what you are proposing to do treats an intrusion as a proposal, and it makes things worse rather than better. That’s a matter for treatment, not examination.
The difference between the two shows up in one question: does she want the thought to be true? The first is assembling justification. The second is desperate for reassurance that it isn’t her. Identical words, opposite structure.
And there is a third condition — a genuine loss of the use of reason — where none of this applies, because there’s no deliberation happening to interrupt. Aquinas is stricter about that category than our courts are: he requires reason to be taken away entirely, not merely impaired. But he grants that it happens, and where it does the response is protection, not argument.
If any of this is close to home, the right move is not a better question. It’s a phone call.
Part three turns the whole thing around. If extreme acts are produced by ordinary motives inside corrupted descriptions, then the descriptions are visible — and that changes how you detect danger in other people: What the Wicked Count On.