I don’t know if you have a techical mindset, but think of this formally.
Let’s say we have individuals A, B, C, and D, where you are A. Maybe you can learn things about B, C, and D, but what you’re really interested in is the pairwise behavior: (A,B), (A,C), (A,D). Because B may behave differently with A than they may with C.
But B may also behave differently if D is present. So the behavior of B in the setting (A,B) is going to be different than in the setting (A, B, D). Imagine that D is the workplace manager, and you can see why.
However, professional and personal context will also play a role. Think of professional contexts a = in a work meeting, b = at work but in the cafeteria for lunch, c = in the parking lot on the way home. Think of personal contexts x = a loved one is terminally ill at home, y = their neighbors initiated a lawsuit against them, z = their sibling just had a child. In each individual’s case, they will react differently to those personal and professional contexts.
Finally, all of this is “noisy”, meaning each individual is working with limited information, and likely to misunderstand why a given person is acting the way they are. So imagine the setting (B, D) where B knows that D is going through a messy divorce. Compare it to (B, D) where B thinks that D is just annoying. Clearly this will change the behavior of B, and therefore of the interaction.
All of this may seem overwhelming, but in fact it’s fascinating. @Today recommended “just follow basic social norms” and that’s great advice. Cultivate a baseline way to act professionally, accept that you may never really know why someone acts the way they do, and take a detached but interested approach to the complexities of human interaction.