Since you're already given the sample space, we can simply count how many outcomes satisfy the request, and divide that number by the cardinality of the sample space.
In other words, we're using the formula
[tex] P(E) = \dfrac{\text{Cases in favour of event }E}{\text{Number of possible cases}} [/tex]
So, the outcomes with a number higher than two are
3H, 4H, 5H, 6H, 3T, 4T, 5T, 6T
out of these outcomes, we can filter those with heads flipped:
3H, 4H, 5H, 6H
So, there are 4 cases out of 12, and the probability is thus
[tex] \dfrac{4}{12} = \dfrac{1}{3} \approx 0.33 [/tex]
