What does it mean that experimental ethics with yourself?

Do not experiment with behaviors that could cause serious harm.

Example: You have been running behavioral experiments successfully for weeks — adjusting your morning routine, testing different work intervals, trying new approaches to difficult conversations. The momentum feels good, and you start thinking bigger. You have struggled with anxiety for years, and you wonder whether you could design an experiment to address it directly. You sketch a protocol: stop your prescribed medication for two weeks, replace it with a combination of cold exposure, breathwork, and journaling, and measure whether your anxiety decreases. The hypothesis is clear, the measurement is defined, the time box is set. By every structural standard from this phase, it looks like a well-designed experiment. But it is not. It is a dangerous one. Stopping psychiatric medication without medical supervision can trigger withdrawal effects, rebound anxiety worse than the baseline, and in some cases neurological consequences that are not reversible on a two-week timescale. The experiment is structurally sound but ethically reckless — not because you lack the right to experiment on yourself, but because the potential downside is catastrophic, irreversible on any practical timeline, and falls in a domain where your self-assessment is unreliable. The ethical version of this experiment is not "do not investigate your anxiety." It is "investigate your anxiety with a professional who can help you design a safe protocol, monitor for adverse effects, and intervene if something goes wrong." The experiment still happens. The self-experimentation simply acquires a guardrail commensurate with the risk.

Try this: Review your experiment backlog or your list of behavioral experiments you have been considering. Select three experiments — one that feels clearly safe, one that feels somewhat uncertain, and one that feels ambitious or edgy. For each experiment, run it through the four-gate ethical screen. Gate one: reversibility. Ask yourself, "If this experiment goes badly, can I return to my previous state within a week?" Write down the specific mechanism of reversal. Gate two: impact radius. Ask, "Who besides me is affected by this experiment, and have I considered their interests?" List every person who would be impacted and how. Gate three: domain competence. Ask, "Do I have sufficient knowledge to anticipate the range of possible outcomes, including the bad ones?" Be honest about the boundaries of your expertise. Gate four: informed consent from your future self. Ask, "If I fully understood the worst realistic outcome, would I still choose to run this experiment?" Write your answer for each of the three experiments. Notice which experiments pass all four gates easily, which ones require modifications to pass, and which ones should not be self-experiments at all but rather collaborations with a professional. Adjust your experiment plans accordingly.