Module 8 · Training Adaptations: Habits & the Week-Three Cliff

The Week-Three Cliff: A Dopamine Event, Not a Character Flaw

By Darian Williams·July 25, 2026·7 min read·Lesson 8.4

Now the pattern you came here for, because you've lived it more times than you can count: you start something new, the first stretch goes great, you're energized, it's working, you think this time it's different, and then somewhere around week three the bottom drops out.

The energy's gone, the behavior feels like dragging a body uphill, and you quit, and you add it to the private file labeled "things I start and don't finish." I need you to hear the actual cause, because it is not the one you've been assuming. That drop is a dopamine event, and it is scheduled. Here's the mechanism. Your reward system doesn't fire for reward, it fires for surprise. When something is new, doing it produces a hit of phasic dopamine, a "novelty bonus", and that hit is a big part of what made the first two weeks feel good and pull you forward. But dopamine is a prediction-error signal: as an outcome becomes expected, the response to it declines toward zero. So as your new behavior becomes familiar, as it works, as it becomes predictable, as it does exactly what you knew it would, the novelty dopamine habituates away. The behavior didn't get harder. The reward for its novelty disappeared on schedule, and the behavior suddenly has to run without the chemical tailwind that was secretly carrying it. That's the cliff. It's not a willpower collapse; it's the predictable expiry of a starter bonus. Now add your brain, and it gets sharper and more forgiving at the same time. There's a well-developed theory that in ADHD the anticipatory dopamine response fails to transfer properly from the reward to the cues that predict it, meaning the "this is coming, keep going" signal that bridges a normal person through the boring middle is weaker for you. The direct consequence, named in the research: familiar, predicted systems lose their dopaminergic pull faster for ADHD brains, which is exactly why ADHD-friendly systems tend to need novelty rotated back in. So the cliff isn't just real for you, it may arrive harder and sooner. Read what that reframe does. The thing you've called a character flaw your whole life, "I get bored and quit, I can't stick with anything", is, at least in large part, a describable neurological event with a known mechanism and a known cure. You're not flaky. Your novelty bonus expired and nobody told you it was going to, so you read a scheduled chemistry change as a verdict on your character. It was never that. And because it's predictable, you can do the one thing that changes everything: put it on the calendar before it happens and meet it with a plan instead of a story about yourself. That plan is the next lesson.

The research

the mechanism, Schultz (1998/2016, dopamine reward-prediction-error coding): phasic dopamine fires to unexpected reward and, as outcomes become predicted, the response declines toward zero; a novel/surprising stimulus elicits a phasic response that habituates as it becomes familiar (the "novelty bonus", Bunzeck & Düzel 2006, Neuron; Kakade & Dayan 2002). The ADHD-specific bridge, Tripp, G., & Wickens, J. R. (2008, dopamine transfer deficit, Journal of Child Psychology and Psychiatry 49(7), 691-704): in ADHD the anticipatory dopamine response fails to transfer from reward to reward-predicting cues, so rewards must be immediate and salient to motivate and familiar/predicted systems lose their dopaminergic pull faster, the mechanistic basis for needing novelty rotation. (Elevated novelty-seeking is robustly observed in ADHD and shares DRD4 genetic correlates.) The adherence data give the cliff its real-world face: behavioral-treatment dropout is high and the documented barrier is executive function itself (forgetting to use skills, poor planning, losing materials), the same EF deficits that make systems necessary make them hard to sustain.

The honest caveat

two honest limits. First, "week three" is a memorable label, not a measured constant, given Lally's 18-254-day spread, your cliff might land at week two or week five; the pattern (novelty fades, the pull drops) is what's grounded, not a precise clock. Second, the dopamine accounts here are theoretical models with supporting but not definitive evidence: reward-prediction-error and habituation are general mechanisms, not ADHD-specific proof, and the dopamine-transfer deficit is one of several competing dopaminergic theories of ADHD reward, not the settled answer. So the honest claim is: the cliff is real, it has a plausible and well-reasoned mechanistic basis, and ADHD may steepen it, not "your dopamine transfer deficit is a proven fact." The design response in the next lesson works regardless of which dopamine theory ultimately wins, because it's built on the part everyone agrees on: novelty fades as things become predictable.

This article is drawn from the Attention Deficit course and describes population-level research, not personal medical advice. Talk to a qualified practitioner before changing anything.