0% READ
● A Visual Essay

The Length
of a Thought

A feeling runs its course in about ninety seconds. Rumination is the mechanism that keeps restarting the clock.

2026.02.09·13 min·Awareness
↓ Scroll

You have had the same thought forty times today. It only happened once.

There is a number that circulates in therapy rooms and meditation apps, and it is one of the few pieces of pop neuroscience I have found genuinely useful: ninety seconds. The claim is that when an emotion fires, the physiological part of it, the actual chemistry moving through your body, runs its course in about a minute and a half. After that, anything you still feel is being re-created.

It comes from Jill Bolte Taylor, a neuroanatomist who wrote about it after her own stroke.* Her version is blunt: ninety seconds is the duration of the flush. Past that, you are choosing to run the circuit again, even if the choosing does not feel like choosing.

I want to take that claim seriously, which means also taking it apart. The number is softer than it sounds. But the mechanism underneath it is real, and it is more interesting than the slogan: a feeling does not have a duration so much as a shape, and rumination is what happens when you keep redrawing it.

Not clinical advice. One person's understanding, drawn from years of reading and being a patient.

The shape of the thing

Start with the physical event. Something happens: a message, a memory, a tone of voice. Your amygdala reads it as significant and fires. Within a couple of seconds adrenaline is in your bloodstream and your heart rate is up. It peaks fast, somewhere around five seconds in, and then it starts to fall.

That fall is the part nobody describes, because nobody notices it. The curve is not a plateau. It is a wave: sharp rise, brief crest, long decay. Left completely alone, it is nearly flat again inside a minute and a half.

The figure below runs that arc. The dashed line is what a single untouched surge looks like. Press the button and watch it happen. Then, while it is still running, press think about it again and watch what the second press does to a wave that had already started coming down.

Fig. 01 · One wave, then several
A single surge is nearly gone by ninety seconds. Each re-trigger restarts it from wherever it had got to, so the peaks stack instead of resetting.

Turn on the automatic rumination toggle and stop touching it. Nothing new happens in the world. No second message arrives. The curve never comes down anyway, because the only input it needs is you, returning to it every fourteen seconds.

Why ninety

The number is not arbitrary, and it is not precise either. It is an estimate of how long the fast chemistry takes to clear.

Adrenaline has a short life in the blood, on the order of a couple of minutes. That is the sharp, hot, hands-shaking part of a feeling, and it does genuinely disappear on roughly that timescale if nothing renews it. So the ninety-second figure is a reasonable description of one component of an emotional response.

But it is only one component. Cortisol, released through the slower HPA axis, operates on a scale of tens of minutes to hours. Muscle tension outlasts both. And the cognitive layer, the part that decides what the event meant, has no half-life at all. It runs until you stop running it.

I should be straight about the provenance here. Taylor's ninety seconds is a clinician's observation and an introspective one, not a controlled measurement. You will not find a study that clocked emotions at ninety seconds, because that is not really what the number is claiming. It is a claim about the floor: this is how short it could be. Everything above the floor is built by something else.

"The ninety seconds is real. What is not real is the idea that you only get one."

The clock has a button

Here is the mechanism that makes the whole thing work, and it is the same one that makes anxiety possible in the first place: your alarm system cannot distinguish between an event and a sufficiently vivid representation of an event.

A memory of the argument, rehearsed in detail, is processed as an argument. A simulated future conversation is processed as a conversation. The amygdala does not check timestamps. It receives a pattern that matches threat, and it does the only thing it knows how to do, which is fire.

So when you return to the thought, you are not remembering the wave. You are causing a new one. And because the previous wave has not finished decaying, the new one starts from an elevated floor and stacks on top of it. Three returns in ninety seconds and you are at an intensity the original event never produced.

This is why the experience feels like escalation without cause. There is a cause. It is just that the cause is the previous lap.

A note has this shape too

If the curve in Fig. 01 looked familiar, it may be because you have seen it somewhere else entirely. It is an envelope: the shape a sound makes over time, independently of what the sound actually is.

Every synthesiser has one, usually in four stages: attack, decay, sustain, release. A struck piano string is almost all attack and decay, loud immediately and then falling away for seconds. A bowed violin has a long sustain, because energy keeps going in. The waveform decides the timbre. The envelope decides whether it reads as a pluck or a pad.

An emotional surge is a pluck. Sharp attack, a crest around five seconds, no sustain stage at all, then a long release. Nothing is holding the key down.

Which is why the synth vocabulary describes rumination better than the psychological vocabulary does. Play a note again before its release has finished and the envelope does not wait its turn. It jumps straight back to the attack stage from whatever level it had reached. Synth builders call this retriggering, and it is precisely what Fig. 01 does every time you press the button.

"A thought you keep returning to is a key you keep striking before the note has died."

There is a second piece of studio equipment that describes the loop even better: a delay line with the feedback turned up. A delay repeats the signal and feeds each repeat back into its own input. Below a certain amount of feedback the repeats get quieter and the tail dies out. Above it, every pass arrives at least as loud as the one before, and the delay never resolves. It rings until somebody reaches over and turns it down.

So the difference between a feeling that passes and a feeling that will not is not the size of the original signal. It is the feedback coefficient. I find that genuinely useful, because a coefficient is a thing you can turn down, and the original event is not.

(More on envelopes, and why shape matters more than waveform, in The Shape of a Sound.)

What rumination actually is

Rumination has a technical definition, and it is narrower than "thinking a lot". Susan Nolen-Hoeksema, who did most of the foundational work on it, defined it as repetitive, passive, self-focused attention to your distress and to its causes and consequences, without moving toward a solution.

The key word is passive. Rumination is not problem-solving, though it is extremely good at impersonating problem-solving. Problem-solving asks "what do I do". Rumination asks "why is this happening to me" and "what does this say about me", questions that are shaped like inquiry but have no terminating condition. You cannot finish them. You can only get tired.

That is what makes it so sticky. It presents itself as diligence. It feels like the responsible thing to do. Meanwhile every lap around the circuit is a fresh trigger going into a body that never got back to baseline.

Start the loop below and leave it running. Watch the lap time, not just the counter.

Fig. 02 · The circuit
Thought, meaning, body, proof, back to thought. The intensity climbs and the laps get shorter, because the route gets more familiar every time you take it.

The speed-up is the part I find most honest about that diagram. A well-worn loop is not just louder, it is faster to enter. After enough repetitions you skip the middle steps entirely: the thought arrives and the chest tightness arrives with it, already fused, no reasoning required. The circuit has been rehearsed into a reflex.

What the stopwatch says

If emotions really lasted ninety seconds, we would all be fine. So what do they actually last, when someone measures?

Philippe Verduyn and Saskia Lavrijsen asked people to report on emotional episodes and how long they ran.§ The answers spanned an enormous range. The briefest emotions, disgust and shame and irritation, were over in something like half an hour. Sadness, the longest by a wide margin, averaged on the order of five days.

Note the scale. Switch the chart to linear and everything below a day disappears into the axis.

Fig. 03 · How long emotions actually last
Approximate self-reported averages. The ranking is the finding, not the individual figures. The dashed line is ninety seconds, which is almost invisible at this scale.

At first glance this looks like a refutation of the ninety-second claim. It is not, and this is the part I think is genuinely elegant. The same researchers looked at what predicted the duration, and the two strongest factors were how important the person judged the triggering event to be, and how much they ruminated about it.

So the two findings fit together perfectly. Ninety seconds is the floor. Rumination is the elevator. Sadness is not intrinsically a five-day chemical; sadness is the emotion we are most inclined to keep returning to, and the returning is what makes it five days.

Brooding and pondering

One important complication, because "stop thinking about it" is both useless and slightly cruel as advice.

When researchers factor-analysed rumination, it split into two components. Brooding is the passive, comparative, why-me variety: why do I always do this, what is wrong with me. Reflective pondering is turning toward the problem to actually understand it. Only brooding reliably predicts worse outcomes. Reflection is roughly neutral and sometimes helpful.

The difference is not the amount of thinking. It is whether the thinking moves. Reflection has a direction and can arrive somewhere. Brooding revisits the same frame from the same angle and calls it progress.

Which gives you something better to ask than "am I thinking about this too much". Ask instead: has this lap produced anything the last one did not? If the answer is no, and it has been no for four laps, you are not analysing. You are re-triggering.

The body keeps paying

There is a cost to all this that is easy to miss because it is invisible from the inside. It has a name: the perseverative cognition hypothesis.

The argument is that most stressors are brief. The threat lasts a minute. But worry and rumination extend the physiological response far beyond the stressor, keeping heart rate elevated, cortisol circulating and recovery suppressed for hours after nothing is happening. The body does not experience one ninety-second event. It experiences a nine-hour one.

This reframes rumination as something more than an unpleasant mental habit. It is the mechanism by which a small event becomes a large physiological load. The thought is not just distressing. It is metabolically expensive.

Ninety seconds, for real

Reading about a duration is not the same as feeling one. Ninety seconds is genuinely short, and almost nobody has ever sat through one deliberately while a feeling was happening.

The technique is old and has a slightly embarrassing name: urge surfing. You do not fight the wave, argue with it or try to make it stop. You watch it, on the explicit understanding that it has a far side. The instruction is not "calm down". The instruction is "stay put and let the chemistry finish".

The timer below runs ninety real seconds. If you want to use it properly, wait until you actually feel something and come back.

Fig. 04 · Ninety real seconds
Surge, crest, settle. If you press the return button, the clock starts over. That is not a gimmick in the widget. That is the whole essay.

What actually interrupts a loop

Things that work, roughly in order of how reliably they have worked for me:

  • Name the mode, not the content. Not "is this true" but "this is brooding". Arguing with the content is another lap. Labelling the process steps outside it.
  • Change the sensory channel. Rumination is verbal and internal. Anything loudly non-verbal competes for the same resources: cold water, a walk with actual scenery, music with words you have to follow.
  • Write the thought down once. Externalising it seems to partially satisfy the thing that keeps re-raising it. The loop is often trying to make sure you do not forget. A note tells it you will not.
  • Postpone it, on purpose. A scheduled fifteen minutes later today, in a fixed place, and every intrusion before then gets deferred to the appointment rather than fought.** It sounds like a trick that could not possibly work. It works more often than it has any right to.
  • Say it out loud to someone. Rumination is remarkably hard to sustain in dialogue. Speaking it forces it into a linear, finishable shape, and someone else's nervous system does the rest.

And the one that does not work: trying not to have the thought. Suppression is a poor strategy, partly because monitoring for the thought requires holding it. You cannot check whether you have stopped thinking about something without thinking about it.

The number is soft, the shape is not

I do not think ninety seconds is a fact in the way a boiling point is a fact. It is an estimate of one component of a complicated response, popularised well past its evidence, and I would not defend the specific figure to anyone.

But I would defend the shape. Feelings arrive as waves rather than states. They have a rise and a crest and a decay, and the decay is real and is usually much shorter than it seems from the inside. What extends them is almost never the original event. It is the returning.

What I actually took from all this is smaller than a technique. It is a question I can ask mid-spiral, when the reasoning brain is barely available: is this the wave, or is this the fortieth wave? I am often surprised by the answer. The event was this morning. The feeling is from eleven seconds ago, and I made it.

If you're struggling, please talk to someone. A therapist, a doctor, a friend, a hotline. None of this replaces actual care. It's just a map of the terrain.

END · 13 MINReply by email
◆ Newsletter

New essays. New tracks. One email a month, max.

Reply to anything I send and it goes straight to me.

NEXT →

Borrowing Calm

What happens in your brain and body when anxiety takes over, and what actually helps.

Notes
  • *Jill Bolte Taylor, My Stroke of Insight (2006). The ninety-second figure is her clinical and introspective observation rather than a controlled measurement, and it has travelled a lot further than its evidence. I use it as a frame, not a finding.
  • Circulating adrenaline clears on the order of a couple of minutes; cortisol, released through the slower HPA axis, persists for tens of minutes to hours. Two different clocks running on the same event, which is why the "how long does a feeling last" question has no single answer.
  • Susan Nolen-Hoeksema's Response Styles Theory is the origin of most serious rumination research. Her core finding, repeated across decades: a ruminative response style predicts both the onset and the duration of depressive episodes.
  • §Verduyn & Lavrijsen (2015), "Which emotions last longest and why: the role of event importance and rumination," Motivation and Emotion. Durations are self-reported, so read them as a ranking rather than a stopwatch. The figures in Fig. 03 are rounded.
  • The brooding/reflection split comes from Treynor, Gonzalez & Nolen-Hoeksema (2003), who factor-analysed the standard rumination scale and found the two components behave very differently.
  • The perseverative cognition hypothesis, that worry and rumination prolong physiological stress activation well past the stressor itself, is Brosschot, Gerin & Thayer (2006). It is the strongest argument that rumination is a physical health issue and not only a mood one.
  • Urge surfing comes from Alan Marlatt's relapse-prevention work in the addiction literature, where the wave shape of a craving is very well documented. It was later absorbed into DBT and the mindfulness-based therapies.
  • **Scheduled worry time is a stimulus-control technique from Thomas Borkovec's work on generalised anxiety. The counterintuitive part is that you are not refusing the worry, only moving it, which seems to be why it does not backfire the way suppression does.