Evidence for Cognitive Placebo and Nocebo Effects in Healthy Individuals

This repository contains data and analyses for the paper "Evidence for Cognitive Placebo and Nocebo Effects in Healthy Individuals".

If you want to use this data/analysis in a research publication, please cite our paper.

Turi, Z., Bjørkedal, E., Gunkel, L., Antal, A., Paulus, W. & Mittner, M. (2018). Evidence for Cognitive Placebo and Nocebo Effects in Healthy Individuals. Scientific Reports, 8(1):17443. doi: 10.1038/s41598-018-35124-w.

@article{turi2018evidence,
  title={Evidence for Cognitive Placebo and Nocebo Effects in Healthy Individuals},
  author={Turi, Zsolt and Bj{\o}rkedal, Espen and Gunkel, Luisa and Antal, Andrea and Paulus, Walter and Mittner, Matthias},
  journal={Scientific Reports},
  volume={8},
  number={1},
  pages={17443},
  year={2018},
  publisher={Nature Publishing Group}
}

Requirements

Analysis are coded in R and stan. Quite a lot R-packages and the Stan are required.

This repository uses the ProjectTemplate directory layout. When you install ProjectTemplate and call load.project() within this directory, all dependencies should be automatically installed.

Data

Raw data is located in data/raw and is provided in .csv format.

The .R scripts located in data load the raw files into R workspace under the name of the R-file (without the .R extension).

The data is structured as follows

Subjectively reported expected and perceived cognitive performance

stored in variable subj.outcomes

> summary(subj.outcomes)

  participant     group      expected          perceived             effect             type            subj   
 1      : 4   n_cond :16   Length:64          Length:64          nocebo :32   conditioning:32   n_cntrl_1 : 1  
 2      : 4   p_cond :16   Class :character   Class :character   placebo:32   control     :32   n_cntrl_10: 1  
 3      : 4   n_cntrl:16   Mode  :character   Mode  :character                                  n_cntrl_11: 1  
 4      : 4   p_cntrl:16                                                                        n_cntrl_12: 1  
 5      : 4                                                                                     n_cntrl_13: 1  
 6      : 4                                                                                     n_cntrl_14: 1  
 (Other):40  

Variables are coded as follows:

• participant - participant identification number from 1 till 16 in each group
• group - group identification;
  • n_cond: nocebo group
  • p_cond: placebo group
  • n_cntrl: nocebo control group
  • p_cntrl: placebo control group
• expected - expected direction of change in cognitive performance
  • decline means that subjects expected to get worse,
  • neutral means no expected change,
  • improve means subjects expected to have better performance
• perceived - perceived direction of change in cognitive performance
  • decline means that subjects perceived to be worse,
  • neutral means no perceived change,
  • improve means subjects perceived to be better
• effect - whether the group belongs to nocebo or placebo effect
  • nocebo: n_cond and n_cntrl
  • placebo: p_cond and p_cntrl
• type - type of manipulation
  • conditioning: n_cond and p_cond
  • control: n_cntrl and p_cntrl
• subj - 64 unique, group-independent participant identification number

Data from the reward-based learning task (learning phase)

data from the five different groups are stored in the variable learn

> summary(learn)

  participant        group      symbl_pair symbl_position    accuracy      reaction_time        trial       
 1      : 2400   n_cntrl:7680   AB:12800   Min.   :1.0    Min.   :0.0000   Min.   :0.0333   Min.   :  1.00  
 2      : 2400   n_cond :7680   CD:12800   1st Qu.:1.0    1st Qu.:1.0000   1st Qu.:0.6830   1st Qu.: 60.75  
 3      : 2400   nhg    :7680   EF:12800   Median :1.5    Median :1.0000   Median :0.8331   Median :120.50  
 4      : 2400   p_cntrl:7680              Mean   :1.5    Mean   :0.7861   Mean   :0.8665   Mean   :120.50  
 5      : 2400   p_cond :7680              3rd Qu.:2.0    3rd Qu.:1.0000   3rd Qu.:1.0330   3rd Qu.:180.25  
 6      : 2400                             Max.   :2.0    Max.   :1.0000   Max.   :1.6662   Max.   :240.00  
 (Other):24000                                                             NA's   :343                      
     reward       day               subj      
 Min.   :0.0000   1:19200   n_cntrl_1 :  480  
 1st Qu.:0.0000   2:19200   n_cntrl_10:  480  
 Median :1.0000             n_cntrl_11:  480  
 Mean   :0.6165             n_cntrl_12:  480  
 3rd Qu.:1.0000             n_cntrl_13:  480  
 Max.   :1.0000             n_cntrl_14:  480  
                            (Other)   :35520  

Variables are coded as follows:

• participant - participant identification number between 1 and 16 in each group
• group - group identification;
  • n_cond: nocebo group
  • p_cond: placebo group
  • n_cntrl: nocebo control group
  • p_cntrl: placebo control group
• symbol_pair - pair type (AB,CD,EF) with (80/20, 70/30 or 60/40 % reward contingency)
• accuracy - accuracy: 1 correct, 0 incorrect
• reaction_time - reaction time in s
• trial - trial number between 1 and 240
• reward - 1: reward was received, 0: no reward
• day - 1: first day, 2: second day
• subj - 64 unique, group-independent participant identification number

Analyses

All analyses are located in src/. To run the scripts, you need to have the ProjecTemplate package and various other packages installed.

The first two lines in each file

library(ProjectTemplate)
load.project()

convert the raw data into a more convenient format by

1. running the data/<dataset>.R file
2. running the preprocessing scripts in munge
3. loading the convenience functions in lib