1. confidence_dud

Data & scripts for the study "The presence of irrelevant alternatives paradoxically increases confidence in perceptual decisions" (Comay, Della Bella, Lamberti, Sigman, Solovey & Barttfeld; 2023). See the publication here: https://tinyurl.com/confidence-dud
Source code
github.com/nicolascomay/confidence_dud
Contributors
nicolascomay
Latest commits
Pages

confidence_dud

This repository contains all the experimental data and scripts used for the study "The presence of irrelevant alternatives paradoxically increases confidence in perceptual decisions" (Comay, Della Bella, Lamberti, Sigman, Solovey & Barttfeld; 2023). Publication available here. Preprint available here.

Usage

In order to run the scripts you'll have to download the experimental data. All scripts are in R language, and in order to be runned you will need to install 3 libraries: "tidyverse" and "DirichletReg". To do that, simply type in the R console the command "install.packages()" with the name of the library as a parameter.

In the 'scripts' folder there are 6 sub-folders. First 5 contains the source html/js code of the experiments, the .jzip archives to directly upload the experiments to JATOS and scripts that reproduce the main figures of the paper. In sub-folder 'modeling' you'll find the code that performs the model fitting and simulations of the 4 models proposed for this study: the max model, the diff model, the contrast model and the average residual model. In order to perform model fitting simply run the wrapper.R file.

How is data organized

In folder "data" you will find the data for the 5 experiments. Rows are trials, and the columns are organized as follows:

Experiments 1, 2 & 4

  • n3SquareOrCircle: indicates if the 3rd alternative was a square (1), a circle (2) or was not present (0).
  • BiggerCircleOrSquare: indicates if the biggest figure was a circle (1) or a square (2)
  • PosBig: this variable is the same as the former, 1 means that the biggest figure was a circle and 2 that was a square. Was used mostly for debugging.
  • StimVal: indicates the sizes of the second biggest stimulus, in proportions to the biggest figure. Is used to indicate the levels of task difficulty.
  • StimVal3: indicates the sizes of the third biggest stimulus, in proportions to the second biggest figure.
  • Nalternativas: indicates the number of alternatives present.
  • RT_type1: indicates the response time (in ms) of the type 1 task (the perceptual decision).
  • Response: indicates which response made the subject: a circle (1), a square (2) or the third alternative (3).
  • Correct: boolean indicator for correct (True) and incorrect (False) responses.
  • Confidence: reported confidence in a scale from 0 to 100.
  • RT_confidence: indicates the response time (in ms) of the type 2 task (confidence report).
  • N_sujeto: indicates the subject's number.
  • binary_correct: a binary variable for correct (1) and incorrect (0) responses.
  • Trial_number: trial number from 1 to 120 in experiment 1 and from to 480 in experiment 2.
  • Angle: random angle for stimuli arrangement.
  • Step_Angle: a rotation of +/-120 degrees of stimuli arrangement. Is expressed in radians (+/- 2.0943...)
  • Area1: size of stimulus 1 (the final size will depend on the screen size).
  • Area2: size of stimulus 2 (the final size will depend on the screen size).
  • Area3: size of stimulus 3 (the final size will depend on the screen size).
  • Mobile: binary indicator, a 1 if the subject performed the experiment on a mobile and a 0 if did it on a computer
  • Code: an alphanumeric code that identifies the subject.
  • Gender: a variable for gender identification of the participant. f=female; m=male; NoBinario=non binary gender identifications.
  • Age: subject's age.

Note: data from experiment 2 contains more columns because it had more alternatives. For example, there are variables called n4SquareOrCircle and n5SquareOrCircle, indicating if the fourth and fifth alternative was a square (1) or a circle (2) respectively (and 0 if it was not shown). The same is for the other variables, with a 3, 4 or 5 indicating to which alternative refers.

Experiments 3 & 5

  • distance_ratio: the distance of the target to the mean of the closest dot cloud. It can take 3 values: 3, 2.5 and 2.1. A value of 2 means that the target is in the middle of the 2 closest clouds.
  • xtarget: x position of target on the screen.
  • ytarget: y position of target on the screen.
  • sd: standard deviation of clouds' dots (60 as in Li & Ma, 2020).
  • nsamples: number of dots on each cloud (375 as in Li & Ma, 2020).
  • color_nube1: index for the color of the cloud 1. 0 = red; 1 = green; 2 = blue.
  • color_nube2: index for the color of the cloud 2. 0 = red; 1 = green; 2 = blue.
  • color_nube3: index for the color of the cloud 3. 0 = red; 1 = green; 2 = blue.
  • color_buttons1: index for the color of the button 1. Is always 0 as button 1 was always red.
  • color_buttons2: index for the color of the button 2. Is always 1 as button 2 was always green.
  • color_buttons3: index for the color of the button 3. Is always 2 as button 3 was always blue.
  • correct_color: the color of the correct cloud.
  • correct_cloud: the number of the correct cloud.

The rest of the variables are the same as in experiment 1, 2 & 4.

Contact

For any doubts or suggestions you can send an e-mail to the corresponding author at nicocomay@gmail.com