N-back function, same letter repeats more than 4 times

function [trial, trial_letters] = my_N_back(trial_count,hit_count,N)
alphabet = ["B","C","D","F","G","H","J","K","L","M","N","P","Q" "R","S","T","V","Y","W","Z"];

false_alarm = zeros(1,trial_count - hit_count);
hit = ones(1,hit_count);

no_hit = 1; %determine where there shall be no hits (first two indices)

max_no_hit_N = max(N,no_hit);
trial = [false_alarm(1:max_no_hit_N),Shuffle([false_alarm(max_no_hit_N+1:end), hit])];

%When false_alarm(1:max_no_hit_N) is called, [0 0 ] is the ans, so the
%first two index, no hits and this is determined by the max_not_hit_N. 

trial_letters = strings(1,trial_count);
shuffle_alph = Shuffle(alphabet);
trial_letters(1:max_no_hit_N) = shuffle_alph(1:max_no_hit_N); %determine the first two. 

last_letter = '';
count_last_letter = 0;
for i= N+1:length(trial)
    if trial(i) == 0
        rand_letter = alphabet(randi(numel(alphabet),1,1));
        while rand_letter == last_letter
            rand_letter = alphabet(randi(numel(alphabet)));
        end
        last_letter = rand_letter;
        count_last_letter = 1;
        trial_letters(i) = rand_letter;
    else
        if trial_letters(i-N) == last_letter
            count_last_letter = count_last_letter + 1;
            if count_last_letter >= 3
                rand_letter = alphabet(randi(numel(alphabet),1,1));
                while rand_letter == last_letter
                    rand_letter = alphabet(randi(numel(alphabet)));
                end
                last_letter = rand_letter;
                count_last_letter = 1;
                trial_letters(i) = rand_letter;
            else
                trial_letters(i) = trial_letters(i-N);
            end
        else
            last_letter = trial_letters(i-N);
            count_last_letter = 1;
            trial_letters(i) = trial_letters(i-N);
        end
    end
end
end

>This function is supposed to generate a string array and a binary vector when given the input of trial numbers, hit ratio and the N-back specification.

i.e. With the input of 10 as trial numbers, 5 as hit trials and N as 1: the output could be [A B B B C D D E E E ] the binary vector trial should be

[ 0 0 1 1 0 0 1 0 1 1].

But, the trials and trial_letters do not match. The trial vector has ones, but the trial_letters indicate no hit trial. The function needs to be used with N = 2 and N = 3 as well, and same issue happens with that too. Mostly the mid one, so there are ones in the trial output, where there are not hits in the string array.

[Voss]

Please provide the values of the inputs (trial_count,hit_count,N) for which the function gives different outputs (trial, trial_letters) than you would like (at least sometimes), and please describe what outputs you would like it to give in that case and why.

It will also help if you describe what you want this function to do and how it's supposed to do that.

Also please provide code for the missing function Shuffle, if it is different than the following:

function x = Shuffle(x)
x = x(randperm(numel(x)));
end

[Voss]

I have no idea if the following is a good change to the code or not, since I don't know what the code's supposed to be doing, but last_letter is only updated when trial(i) == 0. Should it perhaps be updated each time through the loop? i.e.:

last_letter = "";
for i= N+1:length(trial)
    if trial(i) == 0
        rand_letter = alphabet(randi(numel(alphabet),1,1));
        while rand_letter == last_letter
            rand_letter = alphabet(randi(numel(alphabet)));
        end
        trial_letters(i) = rand_letter;
    else
        trial_letters(i) = trial_letters(i-N);
    end
    last_letter = trial_letters(i);
end

[Selen Soylu]

I apologize for the lack of info, I updated my question now, and The Shuffle function was a in-built function for me, but it randomizes the sequence. Could be the last letter as well, sometimes it generates that too, ´with different number of trials. Its almost 90% percent correct but there is a shift that I cant figure out.

[Voss]

"The trial vector has ones, but the trial_letters indicate no hit trial."

Please indicate how you know that.

[Selen Soylu]

I manually checked and compared

[Voss]

Please give an example of correct output and how you know it's correct.

Answer 1

[Tridib]

Hi @Selen Soylu,

The code does not work as expected because it relies on “last_letter” and “count_last_letter” to manage repetitions which does not align with the core logic of the N-back task. In an N-back task, whether a letter repeats should depend directly on the value of trial(i).

1. For hit trials (trial(i) == 1), “trial_letters(i)” should be set equal to “trial_letters(i-N)”.

2. For non-hit trials (trial(i) == 0), a new letter should be chosen that is different from “trial_letters(i-N)” to avoid accidental matches. Also, avoid repeating the same letter too many times in a row.

3. Remove “last_letter” and “count_last_letter” and replace the current for loop with the following code that checks trial(i) directly and assigns letters accordingly.

    % filling letters based on the trial binary vector
    for i = max_no_hit_N + 1 : trial_count
        if trial(i) == 1
            % Hit: should repeat the letter from N-back
            trial_letters(i) = trial_letters(i - N);
        else
            % Non-hit: pick a new random letter different from i-N
            rand_letter = alphabet(randi(numel(alphabet), 1, 1));
            while any(i > N & rand_letter == trial_letters(i - N)) || ...
                  (i >= 2 && rand_letter == trial_letters(i - 1)) % avoiding consecutive repeats
                rand_letter = alphabet(randi(numel(alphabet), 1, 1));
            end
            trial_letters(i) = rand_letter;
        end
    end
    end

Hope this helps!