Self-Experimentation

Self-experimentation is the administration of a substance to oneself, outside a protocol, an ethics approval, or clinical supervision, in order to learn what it does. Structurally it is a single-subject, unblinded, uncontrolled observation in which the experimenter, the subject, the outcome assessor and the person with an emotional stake in the result are all the same individual. Every design safeguard that distinguishes a trial from an impression is absent simultaneously, which is why the resulting data are weak in a specific and predictable way rather than merely small.

The historical cases people invoke are real. Barry Marshall swallowed a culture of Helicobacter pylori in 1984 and developed gastritis, and Werner Forssmann passed a catheter into his own heart in 1929. Both are also survivorship: the self-experiments that ended in serious injury or produced a confidently wrong conclusion are not the ones that get retold, and neither case established its claim until controlled work followed.

The interpretive limit is that a single unblinded person cannot separate a drug effect from expectation, natural fluctuation of the condition, regression to the mean after starting at a low point, concurrent changes in sleep or training, or the placebo response. Adding a compound to a stack multiplies the attribution problem rather than resolving it, and stopping or restarting without blinding does not fix it.

The characteristic error is treating a personal result as strong evidence for oneself even if weak for others. The biases listed above operate hardest on the person inside the experiment, and any dose-finding done this way is uncalibrated by definition, which is where self-experimentation stops being an epistemic problem and becomes a safety one.