First up, I realise that the Palamedes toolbox is distinct from PsychToolbox, but many users will use them together – and there is no more appropriate forum (that I’m aware of) to ask this question. So I hope that either this post will be allowed here, or otherwise someone can direct me to a more appropriate site. I have scoured Kingdom and Prins’ book (Psychophysics) for the answer to the question below but cannot find it.
We use the Palamedes toolbox’ implementation of the Psi procedure (Kontsevich and Tyler 1999) as a standard for psychophysical studies (having moved on from the QUEST toolbox bundled with PsychToolbox). This involves setting up a procedure using PAL_AMPM_setupPM(), and then updating after each trial using PAL_AMPM_updatePM(response), where response is a logical variable indicating whether the user chose the ‘correct’ response or not. Most of the time, the dependent variable (for which we are trying to find threshold) works on a scale in which a higher value indicates an ‘easier to see’ stimulus (i.e. higher stimulus intensity). But, I wonder whether the Psi object assumes this to be the case? What if our dependent variable were back-to-front, i.e. a higher number indicated an ‘easier to see’ stimulus, as is the case with cycles per degree?
One of the parameters set when initialising a Psi object is the priorBetaRange, which, if left to default settings, is free to vary from -1 to +1. Beta is equivalent to the slope of the psychometric function. If I understand correctly, a negative slope would indicate a situation in which a higher ‘intensity’ value corresponds to a lower chance of seeing the stimulus. Since for any given scale of measurement, the ‘direction’ of the slope is known (e.g. cycles per degree is negative and contrast-as-a-percentage is positive), then I wonder whether when initialising Psi objects in Palamedes is it better to override the default priorBetaRange so that it is constrained only to work in that presumed direction? Or is there some other way to indicate the response direction to Palamedes?