Isn't it A simply P(touch_low | touch_high)? P(touch_low | touch_high) = P(touch_low) * P(touch_high) = (0.2 * 2) * (0.2 * 2) = 0.16, no? This, of course assumes no skew or term structure and assumes delta ~ risk neutral probability.
B is much harder, since it's conditional on one hitting...