The issue with the button presses is actually how the task is designed. The random motion is always presented for 2 sec and a participant is allowed to give a response only after the motion pattern disappears. If you check the event sequence, there is the RDM event for 2000ms followed by the Response event. The RDM event contains only RDM motion element, so no key presses are registered there. In contrast, the next event contains the Button element, but no RDM generator. The dependent variable in this task is a participant's detection threshold, which is based on response's correctness, not RT, so the latter measure is ignored. In the same time, presenting RDM for the fixed time ensures that the same amount of evidence is presented in every trial.