I want to look at relationship between BOLD response and duration of behavioural response (measured during a visuomotor task). In my understanding, I could model this by two regressors; EV1 as fixed-height impulse functions and EV2 as variable-height impulse function with height equals to demeaned response duration. If I orthogonalize EV1 with EV2 and run the model, the parameter estimates for EV2 will reflect percentage signal change per unit reaction time. My question is:

1) Is this the correct way of obtaining percentage BOLD signal Vs response duration relationship?

2) I want to create a plot with % signal change in Y axis and response duration in X-axis. I think this will require different EVs for different response duration. Is there any other way of doing this?