2-Forced-Choice Reverse Correlation Task

Script Author: Katja Borchert, Ph.D. (katjab@millisecond.com), Millisecond

Created: January 20, 2020

Last Modified: January 06, 2023 by K. Borchert (katjab@millisecond.com), Millisecond

Script Copyright © Millisecond Software, LLC

Background

This script implements a template procedure to present stimuli for a 2 Forced-Choice (2FC) Reverse Correlation Task. The template script investigates people's mental representation of a 'biracial' face but can be easily adapted to investigate a different target concept.

The Reverse Correlation Paradigm is a data-driven method that helps generate visual proxies of people's mental representations of social categories. It is based on Signal Detection Theory, but instead of a-priori deciding what is the the signal and what is the noise, the Reverse procedure attempts to find what the participant perceives as signal and what as noise (see Brinkman, Todorov & Dotsch, 2017, for an in depth discussion).

2 FC Correlation procedure: For a 2FC Reverse Correlation procedure, a large amount of image pairs are prepared (see step 1 below). Each image pair is based on the same base image but each image pair has been created with a slightly different noise pattern (noise pattern = overlay of random black/white pixel noise). The individual images of a pair have been created with the same noise pattern, but in one case, the random noise pattern was the inverse of the other image.

Participants are presented all image pairs and have to decide which of the two images of an image pair best represents the target category (e.g. biracial face). The generated data can then be analyzed to generate the proxy visual image of the mental represention of the target category.

3 Steps:

(1) Stimuli Preparation: The stimuli for a Reversed Correlation procedure can be generated in 'R' (free software program). Follow instructions provided at: https://www.rondotsch.nl/rcicr/

Example: the stimuli for this script could be generated by using the following R code:

# Load reverse correlation package ("rcicr") library(rcisr)

# assign the base image to 'base_face_files' base_face_files <- list('base'='biracial base image.jpeg')

#generate the stimuli: R will generate a folder of stimuli generateStimuli2IFC(base_face_files, n_trials = 200)

(2) Stimuli Presentation This script provides a procedure to present the generated stimuli in a 2FC Reverse Correlation Procedure.

By default this script runs 200 images. The images are stored in item.stims_orig and item.stims_inverse and can easily be exchanged for new ones. The number of trials automatically adapts to the the number of images stored under item.stims_orig (Note: the script works best with an even number of trials).

(3) Analyze Data

Prepare Inquisit Data (raw data files) for analysis: - merge your RAW data files (if you want to look at the data across participants) - convert iqdat data files to .cvs files

Data Analyses is done in 'R'. Follow instructions provided at: https://www.rondotsch.nl/rcicr/

This is the R-code that could be used to generate the cis image based on data collected with this Inquisit script: (Note: the code is based on the demo provided at: https://www.rondotsch.nl/rcicr/)

# Load reverse correlation package ("rcicr") library(rcicr)

# Base image name used during stimulus generation baseimage <- 'base'

# File containing the contrast parameters (this file was created during stimulus generation) rdata <- 'rcic_seed_1_time_may_18_2020_13_41.rdata'

# Load response data # Note: the datafile should be a *.csv file # Load Inquisit data file in Excel and save as *.csv file # IMPORTANT: use the name of your data file instead of 'datafile.csv' responsedata <- read.csv('datafile.csv')

# Batch generate classification images by subject cis <- batchGenerateCI2IFC(responsedata, 'subject', 'index', 'responseCode', baseimage, rdata)

# Batch generate classification images by trait (here: targetCategory) cis <- batchGenerateCI2IFC(responsedata, 'targetCategory', 'index', 'responseCode', baseimage, rdata)

###Result: a folder 'cis' will be generated by R that contains the generated image file(s)

References

L. Brinkman, A. Todorov & R. Dotsch (2017) Visualising mental representations: A primer on noise-based reverse correlation in social psychology, European Review of Social Psychology, 28:1, 333-361, DOI: 10.1080/10463283.2017.1381469

https://www.rondotsch.nl/rcicr/

https://cran.r-project.org/web/packages/rcicr/rcicr.pdf

Duration

8 minutes

Description

This script presents 200 (default) pairs of images to participant. For each pair, participants have to decide which image best represents a 'biracial' face.

Procedure

This script runs 1 blocks of 200 trials (default: the number of trials run adapts to the number of images).
Each trial presents 1 pair of face images, consisting of one "original" and one "inverse" image (Brinkman, 2017, p.335).
Half the trials present the original image on the left side; the other half of trials present the original images
on the right side.
Images are on screen until one of the images is selected (= 2 forced choice trial).

Stimuli

provided by Millisecond.

New Stimuli can be generated in R. Follow instructions at:
https://www.rondotsch.nl/rcicr/

Instructions

provided by Millisecond - can be edited under section Editable Instructions

Summary Data

File Name: reversecorrelation_2fc_summary*.iqdat

Data Fields

Raw Data

File Name: reversecorrelation_2fc_raw*.iqdat

Data Fields

Parameters

The procedure can be adjusted by setting the following parameters.