It is 9:47am. Your one-on-one with your manager starts at ten. You have already played through the opening sentence eleven times in your head, in three slightly different versions. The body is reading prepared, but the breathing is reading hunted. You are not preparing anymore. You are rehearsing.
Rehearsing conversations before they happen is a different cognitive shape from the post-event replay people more often write about. The version that follows an awkward exchange is the one most articles treat as the canonical case, the one you also see when people replay conversations in their head for years after the fact. The version that runs in the twenty minutes before a meeting, while you are reheating coffee and pretending to read your inbox, is its own loop. Researchers call it pre-event rumination or anticipatory processing. Some of it is useful preparation. Some of it is the cognitive shape that the Clark-Wells model of social anxiety identifies as a maintaining factor. The piece below names both, draws the line between them, and lays out a runnable move-set for the minutes before a hard conversation starts.
Why your brain rehearses something that has not happened
The brain rehearses upcoming conversations because that is, partly, what brains are for. Mental time travel into the near future is one of the prefrontal cortex’s standard moves, and the default mode network treats imagined social scenes as if they were happening, lighting up many of the same regions involved in real interaction. When you rehearse, you are not idling. You are running a simulation. Stanford researchers using brain-machine interfaces showed this directly: mental rehearsal moves neural activity patterns into the same configuration the brain uses for the real action. That is part of why deliberate, bounded rehearsal genuinely helps. Athletes use it. Musicians use it. People preparing for stage talks use it. It is not a malfunction.
The cognitive model of social anxiety, originally laid out by David Clark and Adrian Wells in the mid-1990s and refined since, draws a careful line inside that broad category. When the simulation is about anticipated poor performance, it stops being preparation and becomes anticipatory processing, a self-focused loop that scans for threat and tries to manage the outcome in advance. The model treats this loop as one of the central maintenance factors that keeps social anxiety running, alongside in-situ self-focused attention and post-event rumination. The line between useful preparation and anticipatory processing is not whether rehearsal happens at all. The line is what the rehearsal is for.
It helps to see what the empirical record looks like. A 2024 systematic review and meta-analysis from Donohue and colleagues at the University of Sydney pooled 26 studies covering 1,524 participants and found that psychological treatment produces a large within-group effect on pre-event rumination, g = 0.86. CBT alone was even higher. More usefully, the analysis found that interventions which specifically target the rumination outperform those that hope it dissolves as a side effect of treating social anxiety in general. The cognitive shape, when named and worked on, responds. The same review draws on Wong and Moulds and others who have run randomized work on “banning” pre-event rumination using detached mindfulness, with measurable reductions in frequency, uncontrollability, and distress over four days. The takeaway worth carrying is not the numbers themselves but their implication. Anticipatory processing is treatable, and the National Social Anxiety Center summary of the Clark-Wells model captures the mechanism cleanly: anticipated negative performance appraisals mediate the loop between social anxiety and the pre-event spin.
There is one corollary worth holding. The same machinery, pointed at sleep instead of an upcoming meeting, becomes the shape covered in how to stop replaying conversations at night. The rumination network does not really care which direction it is pointed. It just needs an open lane.
The thesis to carry into the next section is small. Rehearsal becomes a problem not because it happens, but because it does not stop.
Useful preparation vs anticipatory rumination
The practical question is not whether you are rehearsing. It is whether the rehearsal is doing what rehearsal is supposed to do. Three signals tell you which side of the line you are on. Walk them slowly.
The first signal is whether you can name the one move you actually want to make. Useful preparation has a target. You know that you want to ask for the new project, or admit you missed the deadline, or tell your sister you cannot host Thanksgiving. The move is a single declarative line, not a paragraph. When you have it, the rehearsal converges. You say the line in your head, hear it, adjust the wording once, and you are done. Anticipatory rumination has no such target. The rehearsal generates new variations of the opening sentence without converging on one. You are not getting closer to a line you want to say. You are running endless drafts of a line you are afraid to say.
The second signal is whether you stop rehearsing once you have rehearsed it. Useful preparation has an off-switch. When the line is set, the brain releases the loop, often by switching to something else: refilling the water bottle, opening a tab, noticing the time. Anticipatory rumination has no off-switch. You finish a pass, and the brain starts the next pass automatically. A clean tell that you have crossed over is described by practitioner-facing pieces: your anxiety actually rises while you are rehearsing, instead of dropping. Useful rehearsal lowers anticipatory anxiety because it converts uncertainty into a concrete plan. Anticipatory rumination raises it because each pass elaborates on what could go wrong.
The third signal is whether the imagined other person’s responses are getting darker the more you rehearse. Useful preparation lets the other person stay neutral. You picture them roughly the way they actually behave, with the rough range of replies they have actually used in your past conversations. Anticipatory rumination starts at the realistic baseline and drifts steadily worse. By the fourth pass they are angrier. By the seventh pass they are dismissing you. By the eleventh pass they are saying something cutting that they have never said in any real interaction you have had with them. This is what flashforward imagery research describes: a vivid first-person mental scene of being rejected by the audience, which raises anxiety and increases avoidance instead of preparing you for the actual exchange.
There is a cleaner frame for what the looping version is doing. Mental rehearsal that runs without resolving is what the Salkovskis cognitive account treats as a safety behavior, a covert action taken to prevent a feared outcome that, paradoxically, keeps the fear alive. It tells the brain that the conversation needs eleven passes to survive, which is precisely the message that makes it feel like an eleven-pass conversation. The International OCD Foundation describes the same phenomenon at higher intensities as a mental compulsion: an action you feel compelled to perform mentally that offers no real relief. Both frames make the same practical point. The relief you are reaching for inside the rehearsal is the thing the rehearsal is preventing you from getting.
One useful sanity check before moving on. The honest line is not that all rehearsal is harmful. Brief, goal-directed rehearsal is helpful, sometimes very. The line is duration and intent. If you have used up most of the time between deciding to have the conversation and actually having it on mental rehearsal that does not converge on a single move, you are inside the loop the cognitive model is talking about.
What to do in the twenty minutes before the conversation
The move-set below is not three tips. It is one approach with four steps, each of which takes a small amount of time, and the whole thing fits inside the window between deciding to have the conversation and walking into it. The aim is not to script the conversation. The aim is to enter prepared but unscripted.
Start by naming the one move you actually want to make. Not the conversation. The move. “I want to ask for ownership of the new project.” “I want to tell my manager I missed the deadline because the spec was unclear.” “I want to say I cannot host Thanksgiving and I am still available for the day-of cleanup.” The move is one sentence, declarative, in your own voice. Write it down on the back of an envelope or on a sticky note. The act of writing forces the loop to converge. The brain stops generating drafts because there is a final draft to look at.
Next, write the first line out once. Just the opening. Not the response to the response to the response. The first line is the thing you have direct control over. Everything after that is going to depend on what the other person says, and you cannot rehearse that without making them say it. Amy Gallo’s Harvard Business Review piece on mentally preparing for a difficult conversation is clear on the framing: the goal is to stay calm and clear, not to memorize a script. The line is a calibration tool, not a performance.
Third, say the first line out loud, once or twice, with your actual voice in the actual room. This step is the one most people skip, and it is the one that matters. Verbalizing the line breaks the closed loop of mental rehearsal because it forces the body to commit to a single version. The version your throat makes is always different from the version your inner monologue makes. The inner one is smoother and more confident; the throat one is the one that will actually leave your mouth. Hearing it out loud is also a useful surprise: most of the time, it sounds fine. The version your head was workshopping was a version no one was ever going to deliver.
Fourth, decide what you will do if the other person says the thing you are most afraid of, and then stop. This is the move that does the most work, and it is borrowed from Gabriele Oettingen’s mental contrasting research and the WOOP protocol she developed with Peter Gollwitzer. The structure is a single if-then plan: “If they say the project is going to a more senior person, I will ask what experience they would need to see from me before considering me next time.” One sentence. Pick the response you are most afraid of and write your one move for it. The mental-contrasting work is a useful corrective to the common impulse, which is to visualize the whole conversation going perfectly. Pure positive visualization, on its own, slackens readiness. Mental contrasting paired with a single implementation intention outperforms either component alone in a 2021 meta-analysis. You are not rehearsing the conversation. You are choosing one move for one feared response.
The fact that you cannot rehearse the rest is not a failure of preparation. It is the point. A conversation is a live exchange between two people, and the value you add by being there is the value you bring in the moments you cannot anticipate. The four steps above leave room for that. Name the move, write the first line, say it out loud, set one if-then. Then close the loop and walk in.
One small note for the harder cases. If you find that you cannot stop running additional passes after writing the if-then, the move is not more rehearsal. The move is to externalize the worry once and stop. Speak it to a friend in two sentences. Write the worst-case sentence into a notes file and close it. The brain releases the loop more easily when the worry has a place to live outside your head. If even that does not slow it down, the rehearsal is doing more than preparing you, and the next section is where to look.
When the rehearsal is doing more than getting you ready
Most pre-conversation rehearsal is ordinary. It happens for ten or twenty minutes before a meeting, converges on a move, and lets you walk in. The pattern that deserves more attention is the one that runs longer, harder, and across days, and that you cannot enter the conversation without.
A few specific signals are worth noticing. If the rehearsal of a single upcoming conversation has been running for hours, with the same opening line cycling in different variations, and the anxiety is climbing instead of dropping, you have moved from preparation into anticipatory processing. If you have been rehearsing the same upcoming conversation for several days, especially if you are also avoiding small adjacent conversations (declining a coffee, putting off a quick email), the rehearsal is functioning as the safety behavior the Salkovskis frame describes. If you find that you cannot enter the conversation at all without the loop running first, the loop has become a precondition, which is the shape mental compulsions take. None of these are diagnoses, and the cringe of recognizing yourself in one of them is normal. They are signals, and signals deserve to be taken seriously rather than argued with.
The practical good news inside this picture is that the cognitive shape responds to treatment. The Donohue 2024 figures are the strongest single argument: a g = 0.86 within-group effect on pre-event rumination across the treatment literature, with the largest effects in CBT and the largest gains coming from interventions that name and target the rumination directly. A therapist familiar with the cognitive model of social anxiety will recognize the loop quickly. The work is often less about willpower than about learning to notice the rumination before expecting yourself to stop it. That noticing-first move is the same move the IOCDF describes for mental compulsions, and it is the move most pre-event rumination interventions teach in some form. The first skill is not interruption. It is recognition without self-criticism.
If the loop is also pulling sleep into its orbit, the night-specific pattern in how to stop replaying conversations at night covers what to do when the rehearsal arrives near bedtime. If it is mostly running about the conversations you already had, the post-event side of the same loop sits with the broader habit of replaying conversations in your head. The three shapes share machinery. They just point it in different directions.
The sentence to leave with is small. Some rehearsal is preparation, and some is the cognitive shape the model is talking about, and the difference is not the rehearsal itself but whether it converges, releases, and lets you walk in. Aim for the first one. Notice when it has tipped into the second one. Get help if it has tipped without releasing for a long time. The conversation will go better than the loop suggested, and almost always better than the worst pass you ran on it.
