Why is there no matching of timing data when trying to match participant timing data (separate file) with note onset data (one file)?

Hello!

I want to analyze synchronization data. I have the timing of note onsets in seconds of a 30s long audio file in an xlsx file. I also have the timestamps of a participant's taps in relation to a stimuli in a separate xlsx file. The experiment is thus: the participant will listen to a 30s long audio stimulus where a C4 note is played at different rhythms. There will be 6 different audio files - a file will be randomly selected from the list for the participant to synchronize to by tapping along. This will happen 24 times as each stimuli has to be selected 4 times.

In my participant's data file there are several columns of interest to me. The most important being the column "keyTap.rt" as this is the column that contains the timestamps of each trial. Each row of this column will contain all the tap timestamps of one trial. Another column is the "Rhythmsequence" column, which designates what audio file was playing at the time of the trial. Hence, each row of this column tells me what audio file was playing at that particular trial. Rows of keyTap.rt and rows of Rhythmsequence have to match up. They also have to match up to the data in the note onsets file (RPTiming.xls), which has a column that called "Rhythmsequence" which says what audiofile and another column called "Notetiming" which has the timing of the note onset. So I have a multi-level matching task I have to perform prior to analysis.

So the problem is that I don't think I am matching the data properly.

As a result, I am unable to generate the circular histogram plots that I need.

Here is my code:

% Load participant data
participant_data = readtable('945386.xlsx');
keyTap_rt = participant_data.keyTap_rt;
Rhythmsequence = participant_data.Rhythmsequence;
trials_thisN = participant_data.trials_thisN;

% Display Rhythmsequence for debugging
disp('Rhythmsequence:');
disp(Rhythmsequence);

% Load note onset data
note_onset_data = readtable('RPTiming.xls');
note_onset_Rhythmsequence = note_onset_data.Rhythmsequence;
note_onset_Notetiming = note_onset_data.Notetiming;

% Define durations for different audio stimuli
file_duration_data = { ...
    '44_neutral_eighths.mp3', 33; ...
    '44_prosody_eighths.mp3', 33; ...
    '44_reverse_eighths.mp3', 33; ...
    '78_neutral_eighths.mp3', 30; ...
    '78_prosody_eighths.mp3', 30; ...
    '78_reverse_eighths.mp3', 30 ...
};

% Initialize duration_map
duration_map = containers.Map();

% Populate duration_map
for i = 1:size(file_duration_data, 1)
    filename = file_duration_data{i, 1};
    duration = file_duration_data{i, 2};
    duration_map(filename) = duration;
end

% Initialize matched data arrays
matched_keyTap_rt = [];
matched_Rhythmsequence = [];
matched_Notetiming = [];

% Store results for each trial
trial_results = struct('filename', {}, 'circular_mean', {}, 'circular_variance', {});

% Initialize a map to store results per audio file
audiofile_results = containers.Map('KeyType', 'char', 'ValueType', 'any');

% Define the list of relevant audio files
relevant_files = {'44_neutral_eighths.mp3', '44_prosody_eighths.mp3', '44_reverse_eighths.mp3', '78_neutral_eighths.mp3', '78_prosody_eighths.mp3', '78_reverse_eighths.mp3'};

% Iterate through each trial in the participant data from rows 7 to 30
for i = 7:30
    % Get the full filename from the Rhythmsequence
    full_filename = Rhythmsequence{i};
    
    % Check if full_filename or keyTap_rt is empty
    if isempty(full_filename) || isempty(keyTap_rt{i})
        warning('Empty full_filename or keyTap_rt at index %d', i);
        continue;
    end
    
    % Debugging statement to check the full_filename
    disp('Full filename:');
    disp(full_filename);
    
    % Remove 'resources/' prefix if it exists
    filename_without_prefix = strrep(full_filename, 'resources/', '');

    % Debugging statement to check the filename without the prefix
    disp('Filename without prefix:');
    disp(filename_without_prefix);

    % Check if the filename without prefix is in the list of relevant files
    if ~ismember(filename_without_prefix, relevant_files)
        continue; % Skip this trial if the file is not relevant
    end

    % Check if the filename without prefix is present in the duration map
    if isKey(duration_map, filename_without_prefix)
        % Retrieve duration from duration_map
        duration_of_audio_stimulus = duration_map(filename_without_prefix);
        % Debugging statement to check the duration
        disp('Duration of audio stimulus:');
        disp(duration_of_audio_stimulus);
    else
        warning('Duration not found for audio file: %s', filename_without_prefix);
        continue; % Skip this trial if duration is not found
    end

    % Find the index of the matching Rhythmsequence in the note onset data
    matching_idx = find(strcmp(note_onset_Rhythmsequence, full_filename));
    
    % Match keyTap.rt with corresponding Notetiming based on Rhythmsequence
    for j = 1:numel(matching_idx)
        % Extract note onset timings for the current Rhythmsequence
        note_onset_times = note_onset_Notetiming{matching_idx(j)};
        
        % Convert note onset timings to numeric array
        note_onset_times = str2double(strsplit(note_onset_times, ','));
        
        % Check if the trial index is within the range of available keyTap_rt values
        if i <= numel(keyTap_rt)
            % Iterate over each note onset and find the closest tapping event
            for k = 1:numel(note_onset_times)
                % Find the index of the closest tapping timestamp to the current note onset timestamp
                [~, closest_idx] = min(abs(keyTap_rt{i} - note_onset_times(k)));
                
                % Match participant's tapping timestamp with note onset timestamp
                matched_keyTap_rt = [matched_keyTap_rt; keyTap_rt{i}(closest_idx)];
                matched_Rhythmsequence = [matched_Rhythmsequence; full_filename];
                
                % For the first note, set the timestamp to 0.0
                if k == 1
                    matched_Notetiming = [matched_Notetiming; 0.0];
                else
                    matched_Notetiming = [matched_Notetiming; note_onset_times(k)];
                end
            end

            % Convert timestamps to angles for current trial
            angles = mod(2*pi * matched_keyTap_rt / duration_of_audio_stimulus, 2*pi);

            % Calculate circular mean and variance for current trial
            trial_circular_mean = circ_mean(angles);
            trial_circular_variance = circ_var(angles);

            % Perform Rayleigh test for circular mean
            p_mean = circ_rtest(angles);

            % Perform Rayleigh test for circular variance
            p_variance = circ_rtest(angles);

            % Store results for current trial
            trial_results(end+1).filename = full_filename;
            trial_results(end).circular_mean = trial_circular_mean;
            trial_results(end).circular_variance = trial_circular_variance;

            % Retrieve or initialize results for the current audio file
            if isKey(audiofile_results, filename_without_prefix)
                results = audiofile_results(filename_without_prefix);
            else
                results = struct('means', [], 'variances', []);
            end

            % Append current trial results to the audio file results
            results.means(end+1) = trial_circular_mean;
            results.variances(end+1) = trial_circular_variance;

            % Update the audio file results in the map
            audiofile_results(filename_without_prefix) = results;
        end
    end
end

Full filename:

resources/44_prosody_eighths.mp3

Index of matching Rhythmsequence:

2

Index of matching Rhythmsequence:

No matching Rhythmsequence found for 44_neutral_eighths.mp3

Full filename:

resources/44_prosody_eighths.mp3

Index of matching Rhythmsequence:

2

Index of matching Rhythmsequence:

No matching Rhythmsequence found for 44_prosody_eighths.mp3

Full filename:

resources/44_prosody_eighths.mp3

Index of matching Rhythmsequence:

2

Index of matching Rhythmsequence:

No matching Rhythmsequence found for 44_reverse_eighths.mp3

Full filename:

resources/44_prosody_eighths.mp3

Index of matching Rhythmsequence:

2

Index of matching Rhythmsequence:

No matching Rhythmsequence found for 78_neutral_eighths.mp3

Full filename:

resources/44_prosody_eighths.mp3

Index of matching Rhythmsequence:

2

Index of matching Rhythmsequence:

No matching Rhythmsequence found for 78_prosody_eighths.mp3

Full filename:

resources/44_prosody_eighths.mp3

Index of matching Rhythmsequence:

2

Index of matching Rhythmsequence:

No matching Rhythmsequence found for 78_reverse_eighths.mp3

>>

So what I expect and actually did manage to get is an overall comparison of synchronization data across 24 trials, as well as the comparison of trials of specific audio files. I end up getting something like this

Summary for all trials combined: Circular Mean: 2.109612 radians Circular Variance: 0.058917

Audio File: 78_neutral_eighths.mp3 Trial 1 - Circular Mean: 2.918635 radians, Circular Variance: 0.059712 Trial 2 - Circular Mean: 2.918385 radians, Circular Variance: 0.060746 Trial 3 - Circular Mean: 2.918424 radians, Circular Variance: 0.060585 Trial 4 - Circular Mean: 2.918635 radians, Circular Variance: 0.059712

Trial Number: 24 Filename: resources/44_prosody_eighths.mp3 Circular Mean: 2.109612 radians Circular Variance: 0.058917

And so I want this on a circular histogram for each audio file (I didn't include the plotting section), showing the tapping data in different colors (e.g. for audio file n we have tapping data for trial n1 in green, trial n2, in blue, etc.)

The plots are unfortunately not generating and I suspect it's cause there is an issue with the index? Is there a way to fix this?

[Image Analyst]

If you have any more questions, then attach your data files with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

[Andrea]

Hello! I have read through the link, thanks so much! I also included the files. I still find myself hitting a wall - I think I might be having a problem with properly extracting the data from the participant file because I end up generating a variable called matched_keyTap_rt which is filled with just commas rather than the timestamps.

Answer 1

[Divyajyoti Nayak]

Hey Andrea,

Like you mentioned, the issue is in fact in the data extraction from the participant data file. In the data file, the rows of the ‘keyMaprt’ column contains an “array” of numbers enclosed in square brackets but when this is read by MATLAB, it will be considered as an array of characters and not numeric data. That’s the reason the ‘matched_keyTap_rt’ variable gets populated with commas and square brackets.

To fix this issue, before finding the index of closest tapping timestep you should convert the character array to an array of doubles. You can make use of the ‘erase’ and ‘split’ functions to do so. Here are the documentation links for them:

https://www.mathworks.com/help/matlab/ref/erase.html?searchHighlight=erase&s_tid=srchtitle_support_results_1_erase

https://www.mathworks.com/help/matlab/ref/split.html?searchHighlight=split&s_tid=srchtitle_support_results_1_split

Here’s your code that I edited to fix the issue:

if i <= numel(keyTap_rt)
            % Iterate over each note onset and find the closest tapping event
            for k = 1:numel(note_onset_times)
                %Remove square brackets from character array
                keyTap_rt_arr = erase(keyTap_rt{i}, {'[',']'}); 
                %Split the array with comma delimiter
                keyTap_rt_arr = split(keyTap_rt_arr,',');
                %Convert the string array into double array
                keyTap_rt_arr = str2double(keyTap_rt_arr);
                
                % Find the index of the closest tapping timestamp to the current note onset timestamp
                [~, closest_idx] = min(abs(keyTap_rt_arr - note_onset_times(k)));
            
                % Match participant's tapping timestamp with note onset timestamp
                
                matched_keyTap_rt = [matched_keyTap_rt; keyTap_rt_arr(closest_idx)];
                matched_Rhythmsequence = [matched_Rhythmsequence; full_filename];
                
                % For the first note, set the timestamp to 0.0
                if k == 1
                    matched_Notetiming = [matched_Notetiming; 0.0];
                else
                    matched_Notetiming = [matched_Notetiming; note_onset_times(k)];
                end
            end
end