# Workflow for "Cognitive flexibility improves memory for delayed intentions"
## Outline
- There were two experiments run for this project, one which was behavioral only (N=50), and one that was done also collecting fMRI data (N=28).
- For the outline, we are going to first extract data into summary arrays, and then analyze.
- I've also included the summaries already, so you can skip step 1 and go straight to the statistics in step 2 if you'd prefer.
- The first thing to do is download the zipped folders and put them together in a folder.
## Analysis Steps
### Step 1: Data Summary Creation (in MATLAB)
1. The first step is to extract relevant data for by-probe analyses from the behavioral data sample.
- Script: ~/Scripts/byProbe_behSample_preprocessing.m
- Output: ~/Data/Summaries/byProbe_BehSample.csv
- This is for the most part the raw behavioral data with pmCost on each probe, as well as an indicator whether or not the overall OG accuracy on the trial a probe belongs to was >50% (OG chance).
- Labels for the columns need to be added manually from: ~/Summaries/byProbe_beh_colLabels.xlsx
2. The next step is to do the same for the fMRI sample
- Script: ~/Scripts/byProbe_fMRIsample_preprocessing.m
- Output: ~/Data/Summaries/byProbe_fMRIsample.csv
- This is the same basic analysis process as above, but we also collect the difference in classifier evidence for the target-nonTarget categories, referred to as PM intention evidence. We have raw target classifier evidence as well.
- Labels for columns need to be manually added from: ~/Summaries/byProbe_fMRI_colLabels.xlsx
3. Next, we are going to use the same inclusion parameters to extract the by-trial measurements for the behavioral (and then fMRI) samples
- Script: ~/Scripts/byTrial_behSample_preprocessing.m
- Output: ~/Data/Summaries/byTrial_BehSample.csv
4. fMRI sample by-trial data summary extraction
- Script: ~/Scripts/byTrial_fMRIsample_preprocessing.m
- Output: ~/Data/Summaries/byTrial_fMRIsample.csv
### Step 2: Data Analysis (in R)
- Analyses here are shown for both samples combined. Scripts are included that separately analyze the behavioral and fMRI samples to replicate separate analyses. For example: suppl_byProbe_mriOnly would replicate the by-Probe analysis for only the fMRI sample.
1. Analyze the by-Probe data
- Script: ~/Scripts/stats\_byProbe_bothhSamples
- We use this to get:
<br /> **a.** OG RT, OG accuracy, and PM cost by trial type and difficulty
<br /> **b.** PM Accuracy by trial type
<br /> **c.** PM intention evidence by difficulty
<br /> **d.** False Alarm Rate
<br /> **e.** General probe count for each participant
2. Analyze the by-Trial data
- Script: ~/Scripts/stats\_byTrial_bothSamples
- We use this to get:
<br /> **a.** PM Cost Slope x Trial Type
<br /> **b.** PM Intention Evidence Slope x Trial Type
<br /> **c.** Average PM Intention Evidence x Trial Type
<br /> **d.** Bootstrap analyses relating PM Accuracy on each trial to PM cost slope and PM intention evidence independently
<br /> **e.** Correlation of PM cost slope and PM intention evidence
<br /> **f.** Bootstrap including both factors
<br /> **g.** Bootstrap for looking at the reliability of the different types of interactions from the full model
<br /> **h.** Looking at PM accuracy for trials that have end of trial costs indicative of proactive or reactive control
<br /> **i.** Model selection script
<br /> **j.** Partial regression analyses
3. By-trial model fitting scripts.
- Script: ~/Scripts/stats\_byProbe_trialFitsAIC
- These were separately analyzed just to make scripts less confusingly long/clunky
- We use this to get:
<br /> **a.** AICc and AIC for each trial polynomial fits
<br /> **b.** Akaike weights for each trial polynomial fits
