how are you going to ethically test whether you can manipulate people without… manipulating people.

That’s a great question. In the US, researchers are generally obliged (by their universities or their funders) to use an Institutional Review Board to review any proposed experiment involving human subjects. The IRB look for things like: causing physical or emotional harm to the subjects, taking advantage of vulnerable populations, using deception without consent, etc. The IRB might let you do something like manipulate a subject, if the subjects were informed that they might be manipulated or deceived. Yes, this might introduce an observer effect, but this type of review is generally accepted as being necessary for doing ethical research. However, I’m not familiar with the research in question or with the requirements of the Univ of Zurich where the researchers are from.