## Data There are three data sets that can be accessed in either Matlab or CSV format. 1. Sample of 1000 users who played at least 60 gameplays. The gameplay data is restricted to only the first 60 gameplays of those users. The sampling was constrained to sample an equal number of users across six age groups, 21-30, 31-40, 41-50, 51-60, 61-70, and 71-80. This data set is associated with the label: `sim6007`. 2. Sample of the 16 users with the most extensive history of practice among users in our Lumosity data sample. We included the practice history of these users that are part of the set [ 1:1000 2000:3000 5000:6000 9000:10000 13000:14000 ]. We omitted some of the practice history to speed up model simulations. This data set is associated with the label `sim7001`. 3. Sample of users in the 61-70 and 71-80 age range who played at least 1000 gameplays and the full history up to 1000 gameplays is included. This data set for the 61-70 and 71-80 users is associated with the label `sim7004` and `sim7005` respectively ## Data Format CSV files Each CSV file has the following fields: Field | Interpretation --------------------------- | -------------------------------------------- correct | T=correct response F=incorrect game_result_id | unique id for each gameplay movement_direction | L, R, U or D corresponding to the motion direction of the leaves pointing_direction | L, R, U or D corresponding to the pointing direction of the leaves response_direction | L, R, U or D corresponding to the response of the user response_time | Expressed in msec trial_num | Trial number within gameplay trial_type | Task Cue: P = pointing; M = moving user_id | ID of user (does not correspond to any lumosity id) where values 1..N are used for all N users in the particular data set accuracy | 1=correct; 0=incorrect uid | ID of user (does not map to any ID used by Lumosity) compatible | 0=no; 1=yes (the pointing and motion direction are the same) gamecount | number of Task Switching games played by user totalcount | total number of gameplays user actually played (each dataset might only contain a subset of gameplays agebin | 1='1..20'; 2='21..30'; 3='31..40'; 4='41..50'; 5='51..60'; 6='61..70'; 7='71..80' trialtypecount | position in run rtsum | cumulative time in run runlength | total run length of current run maxrtblock | total time in current run runlengthprev | run length of previous run rtsumprev | total time in previous run isswitch | 0=no; 1=yes, this is the first trial in a run movementd | field movement_direction in numeric format pointingd | field pointing_direction in numeric format task | 1=movement task cue; 2=pointing task cue choice | field response_direction in numeric format rlprev | same as field "runlengthprev" but where NaNs are replaced by value 20 (used for modeling purposes) newuid | same as user_id trialtypecount2 | same as trialtypecount but where values larger than 8 are mapped to 8 (used for modeling purposes) ## Matlab Data Files The Matlab files `data_sim6007.mat`, `data_sim7001.mat`, and `data_sim7004.mat`, `data_sim7005.mat` contain the data from the three user samples. Each data file contains the following Matlab variables: Variable | Interpretation --------------------------- | -------------------------------------------- modeldata_all | a N x 1 cell array containing tables of observed gameplay data (N is number of users). Each table contains columns as specified above for the CSV files heldoutdata_all | a N x 1 cell array containing tables of heldout gameplay data. Heldout data was not used in the model simulations to evaluate model performance d_all | a structure containing observed gameplay data used for DE-MCMC routine dt_all | a structure containing heldout gameplay data used for DE-MCMC routine (not used in simulations) ns | number of users modelp | structure containing parameters used to construct the data set and DE-MCMC parameters ## Matlab MCMC samples These data files contain the MCMC samples from the DE-MCMC simulations. Download and save into a folder `results` in the folder that contains the Matlab code