function randomStimsMC( subNo, objname, objtype )
% Natasha Parikh
% 12/16/14
%
% inputs:
% subNo if odd, experimental; if even, control group
% objname cell of all stimulus filenames
% objtype cell of image types: 0 = neutral, 1 = negative
%
% This function serves to randomizes all the stimuli and creates stimuli
% orders for each of the 3 sessions. It does this as a Monte Carlo
% simulation: create random permutations of the stimuli until we meet the
% criteria specified (no more than 4 negative pictures in a row). File
% outputs are named: "stimuliorder1/2/3_subNo.txt", respectively.
% define variables of split of each day's stims
neg1 = 90;
neut1 = 30;
neg2 = 30;
neut2 = 10;
neg3 = 22;
neut3 = 8;
s1orderList = strcat('stimorder1_',num2str(subNo),'.txt');
s2orderList = strcat('stimorder2_',num2str(subNo),'.txt');
s3orderList = strcat('stimorder3_',num2str(subNo),'.txt');
% Randomize order of negative and neutral pics separately
negTrials = find(objtype); % get indices of negative trials
neutTrials = find(~objtype); % get indices of neutral trials
negTrials = Shuffle(negTrials);
neutTrials = Shuffle(neutTrials);
% Masterlists!! DO NOT CHANGE negTrials OR neutTrials THROUGHOUT SCRIPT!!
%% SESSION 1
% Arrange so we have shuffled pairs of neg and neutral pics
% Combine all images for session 1
% Do not change this so we can refer to it for session 2
allS1trials = [neutTrials(1:neut1); negTrials(1:neg1)];
% Shuffle the odd numbered items so that paired items stay together
odds = 1:2:length(allS1trials);
odds = Shuffle(odds);
% pseudo randomization: if there are more than 4 negative pictures
% randomize everything again
while 1 % run an unlimited number of times until conditions are met
% extract the object types of the current trials we have
currentTypes = objtype(allS1trials(odds));
% find groupings of negative and neutral pictures
% groups will be something like [3 4 1 2 1 6], where each number
% indicates the number of repeating items (3 0's, 4 1's, etc.)
% Specifically, I am subtracting each pair of items from each other
% (diff), so if consecutive items are the same, they will come up with
% a diff of 0. Then, I use the find function to find the indices of
% non-zero items, suggesting that a switch of type occurs there.
% Lastly, I subtract out the indices of the switches, so if a switch
% occurs at 1 and then at 5, we know there are 4 of the same item in
% the middle (that's what the outer diff function does). In order to do
% this last step, we have to put a 0 at the front and the length of the
% list at the end to ensure we have proper subtractions from all sides.
groups = diff([0; find(diff(currentTypes)); numel(currentTypes)]);
% find the first instance where there are more than 4 of the same
% type in a row. moreThan4 will give us an index in groups
moreThan4 = find(groups > 4, 1);
% if there aren't any groups larger than 4, we're good!
if isempty(moreThan4)
break % quit the while loop
else % reshuffle everything and check again
odds = Shuffle(odds);
end
end
% initialize final lists for session 1
finalS1list = cell(length(allS1trials), 1);
finalS1type = zeros(length(allS1trials), 1);
% open ASCII file for writing
datafilepointer1 = fopen(s1orderList,'wt');
% change indices to names of files and a list of picture types
for i = 1:2:length(allS1trials)
finalS1list(i) = objname(allS1trials(odds((i+1)/2)));
finalS1list(i+1) = objname(allS1trials(odds((i+1)/2)+1));
finalS1type(i:i+1) = objtype(allS1trials(odds((i+1)/2)));
% write to file twice, one for odd image, one for its pair
fprintf(datafilepointer1, '%i %s %i\n', ...
i, ...
char(finalS1list(i)), ...
finalS1type(i));
fprintf(datafilepointer1, '%i %s %i\n', ...
i+1, ...
char(finalS1list(i+1)), ...
finalS1type(i+1));
end
fclose(datafilepointer1);
%% SESSION 2
% initialize finalS2list to the right size
total2 = neg2 + neut2;
finalS2list = cell(total2, 1);
% Separate stimlist creation for control and experimental group
if mod(subNo, 2) % only goes in here if odd = experimental group
% extract all the Picture A's from day 1
possibleS2 = finalS1list(1:2:length(allS1trials));
possibleS2type = finalS1type(1:2:length(allS1trials));
% figure out indices for which ones are negative/neutral
neutS2trials = find(~possibleS2type);
negS2trials = find(possibleS2type);
% Shuffle their indices
neutS2trials = Shuffle(neutS2trials);
negS2trials = Shuffle(negS2trials);
% combine the correct number of negative and neutral pics
allS2trials = [neutS2trials(1:neut2); negS2trials(1:neg2)];
allS2trials = Shuffle(allS2trials); % Shuffle them
finalS2list(:, 1) = possibleS2(allS2trials); % Convert to file names
else % even = control group
% Extract new images and combine
allS2trials = [neutTrials((neut1+1):(neut1+neut2)); negTrials((neg1+1):(neg1+neg2))];
allS2trials = Shuffle(allS2trials); % Shuffle them
finalS2list(:, 1) = objname(allS2trials); % Convert to file names
end
% pseudo randomization: no more than 4 neg in a row
% implementation is pretty identical to pseudorandomization for session 1
% pics above, so commenting here will mainly highlight differences
while 1
% to extract the pic types, I find the 3rd letter of the name "neG" or
% "neU", convert it to it's ASCII code, and store them in a matrix
% This is because the diff function only works for numbers
currentTypes = cell2mat(cellfun(@(x) double(x(3)), finalS2list, 'UniformOutput', false));
groups = diff([0; find(diff(currentTypes)); numel(currentTypes)]);
% find the first instance in where there are more than 4 in a row
moreThan4 = find(groups > 4, 1);
% if there aren't any groups larger than 4, we're good!
if isempty(moreThan4)
break % quit the while loop
else
finalS2list = Shuffle(finalS2list);
end
end
% need suppression and reappraisal equal amounts for neg, neut
regTypeNeg = [zeros(neg2/2, 1); ones(neg2/2, 1)];
regTypeNeg = Shuffle(regTypeNeg);
regTypeNeut = [zeros(neut2/2, 1); ones(neut2/2, 1)];
regTypeNeut = Shuffle(regTypeNeut);
% Initialize final regulation list
regType = zeros(total2, 1);
% open different output files
datafilepointer2 = fopen(s2orderList,'wt'); % open ASCII file for writing
% counters for regulation type used
negCount = 1;
neutCount = 1;
for i = 1:total2
stimName = char(finalS2list(i));
% add the appropriate regulation type depending upon neg or neut
if strcmp(stimName(1:3), 'neg')
regType(i) = regTypeNeg(negCount);
negCount = negCount + 1;
else
regType(i) = regTypeNeut(neutCount);
neutCount = neutCount + 1;
end
% print to file
fprintf(datafilepointer2, '%i %s %i\n', ...
i, ...
char(finalS2list(i)), ...
regType(i));
end
fclose(datafilepointer2);
%% SESSION 3
% Extract all Picture B's (file names)
oldS3 = finalS1list(2:2:length(finalS1list));
oldType = zeros(length(oldS3), 1); % denote them as old pictures
% Get the distractor images (indices from stimuli list)
newS3index = [neutTrials((neut1+neut2+1):(neut1+neut2+neut3)); negTrials((neg1+neg2+1):(neg1+neg2+neg3))];
newS3 = objname(newS3index); % convert to file names
newType = ones(length(newS3), 1); % denote them as new pictures
% Combine old and new pics
allS3list = [oldS3; newS3];
allS3type = [oldType; newType];
% since we both want allS3list and allS3type permuted in the same way (so
% we can still determine which are old and which are new, work on a list of
% indices 's3index' instead:
s3index = 1:length(allS3type);
s3index = Shuffle(s3index);
% pseudorandomization: no more than 4 neg in a row
% again, look above at implementation for sessions 1/2 for better comments
% session 3 has more stimuli, so I do this one slightly differently to
% ensure it doesn't take over 2 hours to converge.
while 1
% gather image names in the new order, convert to ASCII, and group
s3names = allS3list(s3index);
currentTypes = cell2mat(cellfun(@(x) double(x(3)), s3names, 'UniformOutput', false));
groups = diff([0; find(diff(currentTypes)); numel(currentTypes)]);
% find the first instance in where there are more than 4 in a row
moreThan4 = find(groups > 4, 1);
% if there aren't any groups larger than 4, we're good!
if isempty(moreThan4)
break % quit the while loop
end
% find the index right before the first instance of more than 4 items
% in a row (if it's the first index, we don't have anything to sum, so
% set okayTill to 1
if moreThan4 == 1
okayTill = 1;
else
okayTill = sum(groups(1:moreThan4-1));
end
% are there any neutral items left in the list? If not, there's no
% point reshuffling the remaining items because they're all negative
neutralLeft = find(currentTypes(okayTill+1:end) == 117, 1);
if isempty(neutralLeft)
% if only negative items are left, reshuffle the entire list
% because there's no way we can satisfy our constraints
s3index = Shuffle(s3index);
else
% otherwise, shuffle the remaining items, leaving 4 of the
% negative pictures of the offending large group
if moreThan4 == 1
s3index(5:end) = Shuffle(s3index(5:end));
else
s3index(okayTill+5:end) = Shuffle(s3index(okayTill+5:end));
end
end
end
% apply the new ordering to both the list of names and the types (old/new)
allS3list = allS3list(s3index);
allS3type = allS3type(s3index);
% write to file
datafilepointer3 = fopen(s3orderList,'wt'); % open ASCII file for writing
for trial = 1:length(allS3type)
fprintf(datafilepointer3, '%i %s %i\n', ...
trial, ...
char(allS3list(trial)), ...
allS3type(trial));
end
fclose('all');
end