Altered use of stimulus history in face processing in congenital prosopagnosia

Overview

This repository contains the analysis workspace for the congenital prosopagnosia (CP) face-discrimination study described in the manuscript Altered use of stimulus history in face processing in congenital prosopagnosia.

The pipeline examines how face discrimination is shaped by:

• diagnostic group (TD vs CP)
• face race (Own-Race vs Other-Race)
• regression-to-the-mean condition (biasp vs biasm)
• age
• recent and accumulated stimulus history

The repository includes:

• Weibull psychometric fitting
• d-prime / signal-detection analysis
• trial-level GLMM analysis
• bootstrap follow-up for peak-age estimates
• SPSS repeated-measures ANOVA follow-up
• histogram and range-based descriptive plots
• RTM correlation analyses
• updating-model analysis

The manuscript contains the full theoretical and methodological interpretation. The README is intended only as a compact guide to the repository structure and execution workflow.

Manuscript Contact: Bat-sheva Hadad - bhadad@edu.haifa.ac.il

Repository / Analysis Contact: Eitan Gelfand - eitan.gelfand@gmail.com

---

Quick Start

1. Restore the R environment

This project uses renv for R package version control.

renv::restore()

2. Install Python dependencies

The d-prime workflow is implemented in Python.

pip install -r requirements.txt

3. Font setup

Shared plotting helpers are located in R/. Most figure scripts use:

source("R/setup_fonts.R")
setup_fonts()

The helper selects a Times-like serif font when available and otherwise falls back to a system serif font.

4. Execution order

Some analyses are independent, but the main derived-metric workflow is:

1. Run the Weibull fitting script.
2. Run the d-prime calculation script.
3. Run downstream model-based and descriptive analyses that use those derived outputs.

---

File Organization

Main directories

• weibull_analysis/ - subject-level psychometric fitting and Weibull-derived metrics
• d-prime/ - workflow for d-prime, criterion, and related summary outputs
• glmm_analysis/ - trial-level GLMM analysis and bootstrap follow-up
• SPSS/ - SPSS repeated-measures ANOVA syntax and short documentation
• rtm_corr/ - correlation analyses linking RTM-related measures
• hisotgrams/ - grouped summary plots for d-prime and PSE
• range_analysis/ - accuracy-by-range descriptive plots
• trail_updating_analysis/ - updating-model analysis of recent vs accumulated history
• R/ - shared helpers for paths, fonts, and plotting theme
• data/ - input and derived CSV files used by the analysis scripts
• output/ - generated figures and summary outputs

Shared R helpers

• R/paths.R - centralized project paths
• R/theme_pub.R - shared publication plotting theme
• R/setup_fonts.R - font setup for figures

---

Main Scripts

Derived metrics

• weibull_analysis/Weibull_CP_experiment.Rmd

  • Fits Weibull functions by subject and condition.
  • Generates psychometric parameters and prediction outputs used in downstream summaries.
  • Main outputs:
    • data/weibull_trial_predictions.csv
    • data/weibull_metric_results.csv
    • output/plots/Weibull_groups_by_race.png
    • output/plots/Weibull_groups_by_bias.png

• d-prime/dprime_calculation.py

  • Computes d-prime, criterion, and related signal-detection outputs.
  • Writes derived CSV files used by downstream scripts.
  • Main outputs:
    • data/dprime_results_with_range.csv
    • data/dprime_results.csv
    • output/plots/Dprime_summary.png
    • output/plots/CR_summary.png
    • output/plots/Bprime_summary.png

Main model-based analyses

• glmm_analysis/GLMM_CP.R

  • Main trial-level GLMM of accuracy.
  • Tests effects of Group, Race, Regression condition, and age.
  • Main visualization output:
    • output/plots/glmm_cp_.png

• glmm_analysis/GLMM_CP_bootstrap.R

  • Bootstrap follow-up to the main GLMM.
  • Evaluates stability of peak-age estimates from the modeled trajectories.
  • Main outputs:
    • bootstrap summary CSV/RDS files written to output/

• trail_updating_analysis/last_trail_updating_analysis.R

  • Compares updating models based on the long-term average (t-inf) and previous trial (t-1).
  • Main outputs:
    • output/updating_beta_results.csv
    • output/last_trail_updating_analysis.png

Follow-up and descriptive analyses

• SPSS/IBM SPSS Statistics - Code.sps

  • SPSS repeated-measures ANOVA syntax.
  • Use after generating the d-prime and Weibull/PSE-derived metrics.
  • See SPSS/SPSS_repeated_measures_ANOVA.md for a short explanation of the model and factors.

• hisotgrams/dprime_histograms.R

  • Subject-level d-prime grouped summaries by Group, Race, and Regression condition.
  • Main visualization output:
    • output/plots/dprime_histogram.png

• hisotgrams/dprime_young_sub_hist.R

  • Younger-subgroup version of the d-prime grouped summary.
  • Main visualization output:
    • output/plots/dprime_histogram_young.png

• hisotgrams/pse_histograms.R

  • Grouped PSE summaries based on the Weibull outputs.
  • Main visualization output:
    • output/plots/pse_histogram.png

• hisotgrams/dprime_cfmt_group.R

  • d-prime grouped summary split by CFMT subgroup.
  • Main visualization output:
    • output/plots/dprime_cfmt_group.png

• range_analysis/accXrange_CP_bias.R

  • Accuracy as a function of morph range, grouped by Race within Bias condition.
  • Main visualization output:
    • output/plots/acc_range_CP_bias.png

• range_analysis/accXrange_CP_race.R

  • Accuracy as a function of morph range, grouped by Bias condition within Race.
  • Main visualization output:
    • output/plots/acc_range_CP_race.png

• rtm_corr/rtm_correlation_cp.R

  • Correlates RTM magnitude with Bias- accuracy.
  • Main visualization output:
    • output/plots/Correlation_biasM_vs_magnitude_CP.png

• rtm_corr/dprime_rtm_corr.R

  • Correlates RTM magnitude with d-prime.
  • Main visualization output:
    • output/plots/Correlation_dprime_biasM_vs_magnitude_CP.png

---

Workflow

Stage 1: Generate derived metrics

Run:

• weibull_analysis/Weibull_CP_experiment.Rmd
• d-prime/dprime_calculation.py

These scripts generate the derived psychometric and signal-detection measures used later in the pipeline.

Stage 2: Run the main inferential models

Run:

• glmm_analysis/GLMM_CP.R
• glmm_analysis/GLMM_CP_bootstrap.R
• trail_updating_analysis/last_trail_updating_analysis.R

Dependency note:

• glmm_analysis/GLMM_CP_bootstrap.R is a follow-up to the main GLMM and should be run after glmm_analysis/GLMM_CP.R

Stage 3: Run descriptive and follow-up plots

Run as needed:

• files in SPSS/ repeated-measures ANOVA
• scripts in hisotgrams/
• scripts in range_analysis/
• scripts in rtm_corr/

Dependency note:

• The SPSS analysis should be run only after Stage 1, because it uses the derived d-prime and Weibull/PSE metrics.
• SPSS/IBM SPSS Statistics - Code.sps contains the syntax, and SPSS/SPSS_repeated_measures_ANOVA.md gives the short explanation.

These scripts are mostly parallel and can be run according to the figure or metric of interest.

---

Data Inputs

The scripts expect CSV inputs in data/. The main files used across the repository are:

• data/full_data_cp.csv - main trial-level accuracy dataset
• data/updating_data_cp.csv - updating-analysis dataset
• data/dprime_results.csv - d-prime summaries generated by the Python workflow
• data/dprime_results_with_range.csv - d-prime values retaining the Range dimension
• data/weibull_metric_results.csv - subject-level Weibull outputs

Additional intermediate files are generated by the scripts themselves.

Subject exclusions

• In the Weibull analysis, one CP subject: (829) was excluded because below-chance performance did not allow a stable Weibull fit within a reasonable range; accordingly, two matched control (1704, 1700) subjects were also excluded.

Main parameters

Parameter	Typical values	Role in the analysis
Subject	participant ID	Repeated-measures unit; random intercept / grouping unit across analyses
Age	continuous years	Main age-related predictor in the GLMM and age-based follow-up analyses
Group	TD, CP	Between-subject diagnostic grouping
ExperimentName	Caucasian, Asian	Face-race condition; interpreted as Own-Race vs Other-Race
Regression	biasp, biasm, sometimes Null	RTM-related condition; biasp and biasm are the main discrimination conditions, while Null is used in the d-prime workflow for false-alarm estimation
Range	morph-distance levels	Stimulus-difficulty / perceptual-distance variable
ACC	0, 1	Trial-level accuracy outcome
CFMTgroup	typically L, H	TD ability subgrouping used in selected follow-up summaries
New_tinf	numeric	Updating-model term reflecting accumulated / long-term stimulus history
New_t1	numeric	Updating-model term reflecting recent-history contribution from the previous trial

Main visualization outputs

The repository generates publication-oriented figures in output/plots/ and output/. Main examples include:

• GLMM age-trajectory figure: output/plots/glmm_cp_.png
• Weibull summary figures: output/plots/Weibull_groups_by_race.png, output/plots/Weibull_groups_by_bias.png
• Histogram summaries: output/plots/dprime_histogram.png, output/plots/dprime_histogram_young.png, output/plots/pse_histogram.png, output/plots/dprime_cfmt_group.png
• Range-based descriptive figures: output/plots/acc_range_CP_bias.png, output/plots/acc_range_CP_race.png
• Correlation figures: output/plots/Correlation_biasM_vs_magnitude_CP.png, output/plots/Correlation_dprime_biasM_vs_magnitude_CP.png
• Updating-model figure: output/last_trail_updating_analysis.png

---

Reproducibility

R

renv::snapshot()  # Capture current state
renv::restore()   # Restore to locked state

Repository R version stored in renv.lock:

• R 4.5.2

Python

pip install -r requirements.txt

Local Python version in this environment:

• Python 3.14.3

Key Python packages are pinned in requirements.txt.

The repository uses local path helpers in R/paths.R, so scripts can be run either from the project root or from their own subdirectories.

---

For More Context

The analytical methods, theoretical background, hypotheses, and full discussion of results are presented in the associated manuscript. This repository is intended to support reproducibility and transparency of the quantitative analyses.

To understand the research questions and scientific context, please consult the authors.

---

License

This repository is licensed under the MIT License.

---

Last Updated: April 2026