One of your three blocks, and the one worth removing first — because the other two can't be worked on until you're actually in the conversation.
An earlier version of this lesson claimed you don't need openers and aren't searching for words. That was wrong — I over-read a single multiple-choice answer. You have three blocks, not one: the physical spike, not knowing how to start, and running dry mid-conversation. Lesson 2 and Lesson 3 handle the second and third. This lesson still stands on its own, but it is one third of the picture.
You told me the block is physical: a spike — heart, adrenaline — and you wait it out, and by the time it settles the moment has gone. That part is real, and it's worth attacking first for a practical reason. The other two blocks are about what you say, and you cannot practise saying anything until you've started. This one gates the others.
What you have here is a rule, and the rule is: wait until this settles, then act. This lesson is about why that rule works against you, and what to replace it with.
For decades, exposure therapy was taught as habituation: stay in the feared situation until the fear comes down, and that decline is the treatment working. Craske and colleagues took that apart. Their review of the evidence is blunt:
"The amount by which fear reduces by the end of an exposure trial or series of exposure trials is not predictive of the fear level expressed at follow-up assessment."
"Sustained arousal throughout extinction is associated with less fear at retest in animals and in humans." Craske, Treanor, Conway, Zbozinek & Vervliet 2014, Behaviour Research and Therapy 58, 10–23
Read that second sentence again. It is not merely that calming down first is unnecessary. Staying aroused is associated with the better outcome. Every time you wait for the spike to pass before approaching, you are optimising for the variable that doesn't predict anything, and discarding the one that does.
In clinical language, waiting-until-calm functions as a safety behaviour: something you do to make the situation feel survivable, which prevents you from learning that it was survivable anyway. It's not weakness. It's a strategy that works brilliantly in the short term and guarantees you'll need it again tomorrow.
If fear reduction isn't the mechanism, what is? Expectancy violation — the gap between what you predicted would happen and what did. An approach teaches you something only to the extent that it surprises you.
This is why the research on talking to strangers is so lopsided. Chicago commuters instructed to strike up a conversation had the most positive commute of any condition (d=0.63), with no measurable productivity cost. A separate, matched group of commuters predicted the precise opposite — that connection would be less pleasant than solitude. Epley & Schroeder 2014, JEP: General 143(5)
Your forecast about how the approach will go is not a neutral reading of the situation. It is a known, measurable, reliably-wrong instrument. The point of a rep is to collect evidence against it.
In a five-day stranger-talking study, on day one only 40% of participants believed the first person they approached would talk to them. 92% did.
Across the week — 1,336 conversations — 87% happened with the very first person approached. And here's the part that matters most: over the week, predicted rejections fell steadily while actual rejections never changed at all. They were near zero from day one. Nobody learned to be rejected less. They corrected an estimate that was already wrong. Sandstrom, Boothby & Cooney 2022, JESP 102 Full protocol and the 29-mission task list: The 29 Missions.
Two months of consistent training means you have several hundred recent experiences of elevated heart rate, adrenaline, and shortness of breath that you interpreted as fine — as readiness, even. Same physiology. Different label.
That isn't a motivational metaphor; it's the actual intervention. Jamieson and colleagues gave participants facing a stressor a short briefing: that increased arousal during stress is not harmful, that the body's stress response evolved to help you meet the demand, and that arousal aids performance. Nothing else changed — no relaxation, no breathing, no attempt to reduce anything.
The reappraisal group's cardiovascular profile shifted from threat toward challenge: cardiac output up (d=.76), peripheral resistance down (d=.81). They also showed less attentional bias toward negative information (d=.77). Jamieson, Nock & Mendes 2012, JEP: General 141(3)
Brooks tested the smallest possible dose of this: saying one sentence out loud before a high-pressure task. The script participants were given was exactly this —
Against "I am calm," people who said "I am excited" before a two-minute filmed persuasive speech were rated more persuasive (d=.68), more competent (d=.46), more confident (d=.71), and spoke for 167 seconds versus 132 (d=1.48). In karaoke, singing accuracy went from 52.98% ("I am anxious") to 80.52% ("I am excited"). Brooks 2014, JEP: General 143(3)
Self-reported anxiety did not differ between the "excited" and "anxious" conditions. The intervention did not make anyone feel calmer. It changed what they did while feeling exactly the same. Do not expect this to feel better. Expect it to work anyway. If you wait until it feels better, you're back to the rule we just deleted.
Reappraisal handles the interpretation. It doesn't handle speed — and your window closes in seconds. For that there's a separate, well-evidenced tool: implementation intentions, or if-then plans.
Across 94 independent tests, specifying if situation X, then I will do Y in advance produced a medium-to-large effect on goal attainment, d=.65. The mechanism is two-part: the specified cue becomes more perceptually accessible, and the response to it becomes automated — it fires without a decision. Gollwitzer & Sheeran 2006, Advances in Experimental Social Psychology 38
That paper has a section titled "Failing to Get Started," which is more or less the name of your problem. In one study it cites, only 37% of people holding an ordinary goal intention succeeded at a task with a short window. If-then plans are the documented fix for exactly that failure mode.
The critical design feature: the trigger must be something you notice, not something you decide. "When I feel ready" is not a trigger. "When I notice the spike" is.
Note what this does not say. It doesn't say approach only when it's a good opportunity. It doesn't say approach when you've thought of something to say. The spike is the trigger — which converts the thing that currently stops you into the thing that starts you.
You will sometimes fire it and get a mediocre interaction. That's fine and expected. This week is not about interaction quality.
Three spiked reps per day. A rep counts only if you spoke within roughly five seconds of noticing the spike. If you waited for it to settle first, it does not count — not as a failure, just as a different thing that isn't this.
Reps are deliberately trivial this week. Ask a barista what they'd order. Ask someone in the gym how many sets they have left. Ask a stranger for directions you don't need. Comment on the queue. Thirty seconds is a complete rep.
Zero preparation. Preparing an opener is a safety behaviour and will blunt the drill. The point is proving the window works when you have nothing ready.
| Reps | How many times you spoke within five seconds. Target 3/day. |
| Passes | How many times you noticed a spike and didn't fire. Not a punishment — the ratio is your actual baseline. |
| Disasters | Times something genuinely went badly — hostility, a complaint, real embarrassment. Record honestly. This is the expectancy-violation number, and it is the one that will surprise you. |
Nine boxes — three days at three reps. Tap as you go; it saves in this browser.
How calm you felt. How well it went. Whether they seemed pleased. All three will tempt you, and all three reinstate the rule we're removing — the first directly, the other two because your read on them is biased downward. People reliably underestimate how much a new acquaintance liked them, and the bias is largest for people who are shy about it. Boothby et al. 2018
From memory, no scrolling: what are the three numbers you're counting this week, and what is the one thing you must not count?
Count: reps (spoke within five seconds), passes (noticed the spike, didn't fire), disasters (genuinely went badly). Don't count: how calm you felt — nor how well it went, since that read is biased downward.
Spike → "I am excited" → speak in 5.
The spike is the starting gun, not the obstacle.
It is not supposed to feel better. Move anyway.
Screenshot that to your lock screen. You will need it at the moment you least want to read.
Craske, M. G., Treanor, M., Conway, C. C., Zbozinek, T., & Vervliet, B. (2014). "Maximizing exposure therapy: An inhibitory learning approach." Behaviour Research and Therapy, 58, 10–23. Free full text at PMC
Read the section on expectancy violation and the one on removing safety signals. It is a clinical paper, so skim the animal work — but it is the document that explains why the standard advice you've probably absorbed about "calming down first" is backwards. Twenty minutes, and it will do more for you than any amount of technique.