>It’s not like the experimentor can issue a placebo to themselves.

TFA describes a protocol for doing just that. The author randomly selected between the treatment dose and placebo. They didn't reveal the choice until after the effect should be complete so they could record the results.