You are correct.
If you want, have a read about the boy-girl paradox here: https://en.wikipedia.org/wiki/Boy_or_Girl_paradox
This is a very similar problem, where I tell you something along the lines of "Mr Smith has two children, and one of them is a boy". You then have to try and decode whether I intended to tell you:
- Either something like: I met Mr Smith's oldest child yesterday, who's a boy. This means I know the sample space of Mr Smith's children is {BG, BB} with equal probability.
- Or something like: Mr Smith was invited to an event that parents of boys only get invited to. So I know one of his children is a boy, and possibly both. So I know the sample space is {BG,BB,GB}, but with equal probability as I have no reason to suppose one combination over the other.
In the former the chance of two boys is a half, and in the later, a third.
Of course, the real problem is that natural language doesn't map nicely to the precise statements made by mathematicians (which must be partly why I never know what anyone is talking about). This is especially true about probability. When someone casually says "one of the children is a boy", they don't mean "there exists a child such that is a child of Mr Smith, and that child is a boy". But when a maths exam paper says that, it certainly is what they mean!
So in your example, statement 1) "at least one is a head", you are being told a very precise piece of information, which is weaker than saying "I saw the first toss and it was a head". You are really being told something like: "I saw a million tosses of two coins. I threw away all the examples of two tails. I randomly picked one of the remaining examples, and I am giving that example to you".
I hope this helps!