Bidirectional Sleep Experiment
Test a sleep change, remove it, and compare both directions
- Difficulty
- Easy
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 96%
The bidirectional sleep experiment adds a reversal phase to an ordinary self-test. Walker illustrated it with evening light: first observe normal sleep with lights on, then dim most lights for one hour before bed for seven days, and finally return to the prior lighting pattern. Evidence is stronger when sleep or sleepiness improves during the intervention and worsens again after it is removed. The design does not prove medical causation, but it is more informative than noticing improvement in only one direction. Keep the intervention bounded, track the same outcomes throughout, and avoid changing other major variables. Use it for low-risk habits and environmental cues, not medications or symptoms that need clinical assessment.
Origin
Extracted from The Diary of a CEO
Core principles
- 01A one-way improvement can be coincidence or placebo
- 02Removal provides a second direction of evidence
- 03Change one main variable at a time
- 04Track observable outcomes before interpreting them
How to run it
- 1
Capture the baseline
Record the current routine and a few outcomes, such as perceived sleepiness, bedtime, awakenings, and morning function. Do not improve the routine yet.
Pro tip Use the same short questions each day.
- 2
Define one intervention
Choose one low-risk change with a clear start and finish. Walker's example was dimming most lights for the hour before bed.
Pro tip Write the intervention precisely enough that another person could repeat it.
Watch out Do not self-test prescription dosing or deliberately restrict sleep.
- 3
Run the intervention
Apply the same change consistently for seven days while recording the same outcomes. Avoid adding other major sleep changes during the test.
Pro tip Record compliance as well as outcomes.
- 4
Reverse the change
Return to the previous routine and keep measuring. Look for whether any observed benefit weakens when the intervention is removed.
Pro tip Keep the reversal period long enough to observe ordinary nights.
Watch out Stop if returning to the baseline would be unsafe.
- 5
Compare both directions
Ask whether outcomes improved with the change and moved back when it stopped. Treat mixed results as uncertainty rather than forcing a conclusion.
Pro tip Retain only changes that are both helpful and sustainable.
Watch out A personal pattern is not proof that the intervention treats a medical condition.
In the wild
A sleeper records three baseline nights, then dims most household lights for the final hour before bed for seven nights. They return to their former lighting pattern and continue the same log, comparing sleepiness and continuity in both directions.
→ The sleeper gets stronger personal evidence than a one-direction trial would provide.
Common mistakes
Changing several variables
Simultaneous changes make it hard to know which one tracked with the result.
Skipping the reversal
Without removing the intervention, an improvement could reflect unrelated changes or expectancy.
Claiming medical proof
A personal A-B reversal is useful evidence for a habit decision, not a clinical trial or diagnosis.
Is it for you?
Best for
People testing a low-risk environmental or behavioral sleep change such as dimming evening lights.
Not ideal for
It is not appropriate for testing prescription changes, dangerous sleep restriction, or untreated medical symptoms.
From the transcript
“Once that 7-day period has finished, go back to doing what you were doing before”
“I'm trying to teach you bidirectionality in the experiment.”
From the episode
No.1 Sleep Expert: Magnesium Isn’t Helping You Sleep! This Sleep Habit Increases Heart Disease 57%