
The loop underneath every compulsion
Cue, routine, relief, and then a cost that feeds the next cue. Draw your own version on one sheet, find the link that is weakest for you, and break that one rather than trying to break all four.

Cue, routine, relief, and then a cost that feeds the next cue. Draw your own version on one sheet, find the link that is weakest for you, and break that one rather than trying to break all four.

Four steps, in order: stop the session now, tell one person, change the one condition that let it happen, and restart today rather than Monday. The shame is what produces the next one.

Log five things every time it happens for a week, then turn the log into one page: your hours, your places, your feelings and the two doors worth closing first.

The first week has a shape that repeats across behaviours: an easy start, a hard middle, a crash, and boredom that arrives late and matters most. Plus the substances where stopping suddenly is dangerous.
Four links: a cue, the behaviour, the relief, and the cost that feeds the next cue. Every compulsion runs on that circuit, whatever it is aimed at, and you do not have to break all four. Draw your own version on one sheet of paper, find the link that is weakest for you, and put your effort there.
For most people the weakest link is the gap between the cue and the behaviour, which is a matter of seconds, and seconds can be bought.
Read each against your own behaviour rather than in the abstract.
Something starts it. Almost nothing starts from nowhere. A time, a place, an object, a feeling, a message, or, most commonly of all, a transition: getting into bed, sitting down after work, ending a call, the moment the house goes quiet.
The US National Institute on Drug Abuse describes cues in a person's environment triggering strong cravings "even if the drug itself is not available", and says the learned response "can last a long time, even in people who haven't used drugs in many years". Its summary is three words: "The brain remembers."
The behaviour itself, and it is usually longer than the headline. It is not just "I bought something". It is opening the app, browsing without intending to buy, finding the thing, the twenty minutes of justifying, then the checkout. Almost every compulsion has a run-up, and the run-up is where it is still interruptible.
Write yours out in steps. People are often surprised there are six or seven of them.
Something improves, briefly. Not necessarily pleasure. Usually the removal of something unpleasant: restlessness, boredom, dread, loneliness, a feeling of being unfinished.
This is where the loop gets its power, and it is also where people misunderstand themselves. NIDA notes that scientists now think dopamine "has more to do with getting us to repeat pleasurable activities (reinforcement) than with producing pleasure directly". Wanting and liking come apart. You can be pulled hard toward something you stopped enjoying two years ago, and that is expected rather than evidence you are broken.
Then the bill. Money, sleep, time, a lie you had to tell, skin, a relationship, an hour you needed.
And the cost produces feelings, mainly shame and dread, which are themselves cues for the behaviour, because the behaviour is the fastest known way of making those feelings quieter. That is the closed part of the circuit. It is why "feel worse about it" is not a strategy, and why people who are hardest on themselves often have the tightest loop.
Twenty minutes, one sheet of paper. Do not do this from memory alone if you can help it, do it from a week of notes, which is what [writing your own trigger map](/guides/other/writing-your-own-trigger-map) produces.
Not all four are equally breakable, and knowing that saves you months.
**The cue is worth removing where you can, and often you cannot.** Some cues are physically removable: the object, the app, the route past the shop, the chair. Others are times of day, feelings, or other people, and those are not going anywhere. Remove what is removable and stop trying to remove the rest.
**The gap between cue and routine is the best target.** This is where friction goes. Log out, delete, move the object, hand the card to someone, put someone else's fingerprint on the second factor. You are not building a wall, you are buying seconds, and seconds are usually what is missing. [Friction, the one move that works for everything](/guides/other/friction-the-one-move-that-works-for-everything) has the full ladder.
**Inside the routine, the early steps are interruptible and the late ones are not.** Step two of seven can be stopped. Step six almost never can. So the intervention belongs at the start of the run-up, which means noticing earlier rather than resisting harder.
**Relief is the hardest link and it is not nothing.** You cannot decide to stop finding relief in it. What you can do is arrange other reliable sources of the same relief, so the behaviour stops being the only route. That is the subject of [replacing it with something real](/guides/other/replacing-it-with-something-real).
**The cost-to-cue arrow is the one people never think to cut, and it may be the highest-value cut of all.** You cut it by refusing to punish yourself after an episode: telling one person, making one small specific change, and going to bed. That sounds soft and it is mechanical. Shame is a cue. Removing a cue is a legitimate intervention.
Because it is being applied at the worst point in the circuit.
NIDA calls impaired self-control "the hallmark of addiction". A plan that requires more self-control at the peak of the urge is a plan that depends on the one resource the pattern has already eaten into. Every move above works somewhere else in the loop, and most of them are set up while you are calm, which is when your judgement is actually available.
Two things worth saying, because confident diagrams are easy to oversell.
**The neuroscience is drug research.** Everything quoted above was studied in substance use. Applying it to checking, buying, watching, picking or scrolling is an extension the sources themselves do not make. Researchers have [warned that "recent publications have suggested that nearly all daily life activities might lead to a genuine addiction"](https://pmc.ncbi.nlm.nih.gov/articles/PMC4627665/), and that caution is fair. Cue-driven learning as a general mechanism is uncontroversial. Anyone telling you precisely what your brain does when you refresh a page is going past the evidence.
**A loop is a description, not a cause.** Some compulsions sit on top of anxiety, depression, trauma, pain or a genuinely bad situation. Breaking the loop helps and it does not treat what is underneath. If the loop keeps rebuilding itself no matter how much friction you add, that is a reason to see a GP rather than a reason to add a sixth blocker.
Renovyn lets you define your own behaviour and log the loop as it runs, which makes the pattern visible over weeks rather than nights. It is a tracking tool, not treatment.
It is the same shape, and the difference is what sits in the relief box and how hard the arrow from cost back to cue pulls. A habit produces a small reward and no meaningful cost, so the loop is loose and easy to break. A compulsion produces relief from something unpleasant and a cost that regenerates the cue, which is why it holds.
The gap between cue and behaviour, in almost every case, because it is the cheapest to change and it works while you are calm. Move to the others once you know from your own log that the gap is not where yours is weakest.
Look at transitions rather than at feelings. People who find no cue usually have transition-driven episodes: sitting down, getting into bed, finishing something, arriving somewhere. Those are invisible until you go looking for them specifically.
Not on its own. It tells you where to put the effort, which is the difference between six months of general resolve and one specific change that reduces the count. Insight without a change is a better description of the same evening.
Renovyn is the app we wished existed in our worst weeks. Check-ins, protection, community, and a crisis button for 3am. Or if you just want the next piece in your inbox, we can do that too.
One email. No noise. Just work worth reading.