Trying to measure a psychoacoustic threshold and can't wrap your head around how to do a 3 down 1 up staircase? This function handles the decisions about whether (and by how much) to increment or decrement your signal on the next trial. The rules (e.g., 3 down, 1 up) are an input variable, allowing you to use any arbitrary rule scheme for your adaptive track.
You provide the function with all the subject’s answers up to that point (correct and incorrect), step sizes, number of reversals you want, and the adaptive rules. The function then outputs the change to apply to the signal (if any) and the reversal indices once the adaptive track is finished.
>> [change,reversals] = adaptive_track_engine(answers,steps,reversals,rules)
Based on an adaptive track with steps of 6 dB initial and 2 dB final, 3 reversals for the first stage, 5 for the second, and a 3 down 1 up method, given the listener's answer history so far:
>> answers = [0 1 1 1 1 1 1 1 1 1];
>> [change, reversals] = adaptive_track_engine(answers,[6 2],[3 5],[3 1])
change = -6
reversals = 
'change' will tell you if and by how much you need to change your signal on the next trial.
‘reversals’ will be empty until total reversals are exceeded, at which stage it will contain a logical index of the points in the track when reversals occurred.
Later on in the adaptive track:
>> answers = [0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 1 1 0 0 1 1 1 0];
>> [change, reversals’] = adaptive_track_engine(answers,[6 2],[3 5],[3 1])
change = 2
reversals = 
With this function, the code required to implement an entire adaptive track is simple:
[answers, reversals] = deal(); %clear out the reversals and answers
levels = 50; %this is the starting level
play_signal(sum(levels)); %play your stimulus at sum(levels) level
answers = [answers;get_answer()]; %get your subject response
[levels(end+1),reversals] = adaptive_track_engine(answers,[6 2],[3 4],[3 1]);
levels = cumsum(levels); %this is the vector of levels that were used
Optional use of 'method': The function will assume that your 'steps' variable is in decibels and will return positive or negative decibels. Alternatively, you can call the method as 'scalar':
>> [change,reversals] = ADAPTIVE_TRACK_ENGINE(answers,[1.414 1.225],[3 5],[3 1],'scalar')
in which case the code will return 1.414 for increments or its inverse 0.707 for decrements during the first stage, 1.225 or 0.817 and so on.
This submission includes an example
(interleaved_adaptive_track_example.m), showing how to use the core 'adaptive_track_engine' to run a set of interleaved adaptive tracks.