How can I control the amount of same random letters back to back? (N-back)

Hey,

I am designing an n-back experiment for my project (on psychtoolbox) and I have troubles controlling the letters that needed to be presented back to back. My projects demands that there should be certain amount of same letters. For instance, if there are 50 trials (1 trial = 1 letter) 5 times there should same letters back to back (avoiding the first 2 trials) and the letters should be random for each time the code is run.

So far what I have is this;

alphabet = {'A','B','C','D','E','F','G','H','I','J','K','L','M','N','O','P','R','S','T','U','V','Y','Z','W','Q'};
num_shuffled =  randi([1 25], [1 50])
alphabet_new = alphabet(num_shuffled);
for checked = 1:49
    if alphabet_new{checked} == alphabet_new{checked+1}
        new_num = randi([1 25], 1)
        alphabet_new{checked} = alphabet_new{new_num}
    elseif alphabet_new{checked} ~= alphabet_new{checked+1}
        alphabet_new{1,checked} = alphabet_new{1,checked}
    end
end
%This part is to control for 5 trials just to try it out. 
integ =[];
for cond = 1:5
    integ(1,cond)=randi([2 48]);
    alphabet_new(integ(cond)) = alphabet_new(integ(cond)+1);
end

This works, but not always. Sometimes I end up with more than 5 times or less than 5 times. So I am wondering If there is a more efficient way to do it?

I'd really appreciate any ideas!

Thanks.

Answer 1

[Karim]

Quite an interesting problem, I'm not familiar with a n-back experiment. But I guess one of the issues could be that your 'integ' indexes are not guarenteed to be unique values. Also you only loop ones over the inital shuffeling, so again it could be that after the shuffeling you still end up with repeated values.

Below I tried to modify the code so that it (hopfully) produces the output that you want.

Essentially it picks 'testsize' characters, and keeps on shuffeling them until it obtains a string with non repeated random characters. Afterwards if picks 'trial_value' locations to introduce the repeated character.

alphabet = ["A","B","C","D","E","F","G","H","I","J","K","L","M","N","O","P","Q" "R","S","T","U","V","X","Y","W","Z"];

TestSize = 50;
trial_value = 5;

% generate a string with "testsize" characters
num_shuffled = randi(numel(alphabet),1,TestSize);
% make sure there are no repeated values...
while true
    % test if integers are repeated
    change_me = [false diff(num_shuffled)==0];
    % if not end the loop
    if all(~change_me)
        break
    end
    % if so pick new integers
    num_shuffled(change_me) = randi(numel(alphabet),1,sum(change_me));
end

% create the new string
alphabet_new = alphabet(num_shuffled);

% now make sure that we have "trial_value" repeated characters (at unique locations)
trial_idx = sort(1 + randperm((TestSize-1),trial_value));
alphabet_new(trial_idx) = alphabet_new(trial_idx+1);

% print the string to the display
fprintf('Current test string is: %s \n',join(alphabet_new))

Current test string is: C V V N D E E M T U W I Z O R A Q E E D S E X O V R G B N E I B N F Q L B J S D H L L A M C Y A T B 

fprintf('Trial indexes are: %i %i %i %i %i \n',trial_idx)

Trial indexes are: 2 6 17 18 42 

fprintf('Trial characters are: %s \n',join(alphabet_new(trial_idx)))

Trial characters are: V E Q E L 

[Selen Soylu]

This works pretty smoothly, thank you very much. It also allows me to flexibly alter the code according to the target trials and from 1-back to 2-back! Thanks really