In the context of post-conviction polygraph screening, a question about lying to a probation officer (or a therapist) would be a probable-lie "control" question. In that context, the relevant questions are about direct violation of the specific terms of one's probation or parole.

For examples of probable-lie "control" questions used in post-conviction polygraph screening, see:

https://antipolygraph.org/cgi-bin/forums/YaBB.pl?num=1456979483

Alright, then it sounds to me like the easiest way to recognize a control question is by assuming it is a question that literally everyone would lie on. Because recognizing a control question sounds difficult to me so I'm trying to figure them out as best I can. For example the question: "Have you ever lied to your probation officer?" I cannot tell if this would be a control question or not, because it starts off sounding like a control question, but it is relevant to somebody with a probation officer, and it is also illegal to lie to a probation officer I think. But I cannot seriously imagine a probation officer sending you to jail just because of being caught lying on this question.

In CQT polygraphy, reactions to "control" questions are compared with reactions to adjacent relevant questions. Reactions are not averaged. Failure to show a reaction to a probable-lie "control" question may indeed result in a subject's failure to pass. Polygraph operators do not assume that some people will tell the truth on some control questions, and no allowance is made for this possibility.

You can read about polygraph chart-scoring methodology in the following National Center for Credibility Assessment document:

https://antipolygraph.org/documents/ncca-numerical-scoring-2017-08.pdf

I'm wondering about how scoring works on probable-lie polygraphs. If I were to not be able to tell if a question was a control question, I know I'm supposed to "err on the side of caution and assume it is a relevant question". But what actually happens scoring-wise if I treated it as a relevant question? Would I likely be called out for being deceptive if the adjacent relevant question were one that I actually needed to lie on? Or does the probable-lie polygraph also take into account your average/mode measurements on all control questions and compare that to the relevant questions? It doesn't seem quite right to me if this isn't the case because it's very reasonable to assume that people will tell the truth on some control questions since you know, it is called a probable-lie test.