When self-experimentation and conventional experiments are both possible, the difference between them often resembles the difference between learning and showing: Self-experiments are better for discovery but worse for convincing others that the solution is helpful or the answer is correct. Of course, most scientists want to do both ─ discover something and convince others of their discovery. Thus, psychologists should consider doing both self-experiments and conventional ones. The best use of resources may often be self-experiments followed by conventional ones. The researcher begins with self-experiments that, if all goes well, find large effects and / or generate and eliminate many hypotheses. This exploratory and theory-building phase lasts until a convenient solution or large effect is found. Then the researcher uses self-experiments to find the procedural parameters (e.g., duration, time of day, intensity) that optimize the solution or maximize the effect. Only then would the researcher begin conventional experiments, using the optimized parameters. Since self-experiments are suited to exploration while conventional experiments provide validation, the most efficient strategy is for the former to precede the latter.