I was wondering if anyone has had experience with implementing a QUEST staircase, specifying a negative slope parameter (i.e. correct answer = increase in next trial intensity).
QUESThandler seems to dislike a negative slope value (beta), I have been using a workaround of subtracting n from the next calculated intensity value and doing so for threshold estimate once the staircase has completed. I just wondered if there was a better way of doing this?
I’ve had some success recently using a similar process. For example, if there is a maximum value that the staircase can reach I subtract the estimate from that using something like:
this_val = max - level. But I’ve also had success using the inverse, i.e.,
this_val = 1/level.
I guess the only thing I’d recommend not doing is to reverse code your correct answers, as tempting as the logic sounds!
Thank you for the advice! I am implementing it this way, my ‘level’ ranges from 1-0, so have been subtracting 1 from ‘level’ to feed my stimulus the next intensity (this_val)
Cheers for the heads up about being careful reversing the correct answer etc. I’ve saved a raw output and a ‘retro-transformed’ output for thresholds, just in case I’ve gone wrong with my transformations somewhere.