AEPsych uses GPs and active learning for adaptive experimentation, but there are older adaptive psychophysics methods, including the 1-up-N-down staircase (https://en.wikipedia.org/wiki/Psychophysics#Staircase_procedures). This method is still useful for finding thresholds along single dimensions. The algorithm has 3 parameters: N, step-up size, and step-down size. After the participant makes the correct response N times, the stimulus parameter is increased by step-up size. If they make an incorrect response, the parameter is reduced by step-down size. This algorithm can be implemented using an AEPsyh generator class. All parameters should be configurable via .ini files.
AEPsych uses GPs and active learning for adaptive experimentation, but there are older adaptive psychophysics methods, including the 1-up-N-down staircase (https://en.wikipedia.org/wiki/Psychophysics#Staircase_procedures). This method is still useful for finding thresholds along single dimensions. The algorithm has 3 parameters: N, step-up size, and step-down size. After the participant makes the correct response N times, the stimulus parameter is increased by step-up size. If they make an incorrect response, the parameter is reduced by step-down size. This algorithm can be implemented using an AEPsyh generator class. All parameters should be configurable via .ini files.