Code and Data for "When Somatic Attention Creates Sensation: A New Class of Bodily Experience"

This repository contains the analysis code and processed/raw data accompanying our manuscript. It provides a fully reproducible pipeline for preparing, clustering, and post-processing sensation reports and shape features, as used in our study. Contents Raw data (../raw_data/): Participant sensation drawings (images, PNG format). Original descriptor selections (file names, German-language labels preserved). Demographics and experiment metadata (CSV). Masks (../masks/): Body-template masks used for pixel-based analyses. Scripts (scripts/): 00_Preprocessing.py — Cleans raw data, computes initial descriptor counts, prepares input tables. 01_Prep_Descriptors_for_Clusteranalysis.py — Extracts descriptor co-occurrence data for clustering. 02_Clusteranalysis_by_Descriptors.py — Performs hierarchical clustering on descriptor co-occurrence, outputs centroids and term–cluster assignments. 03_Jaccard.py — Computes Jaccard indices between sensations and descriptor clusters, derives absolute cluster assignments, and sorts/exports sensation-level results. 04_Referred_Sensations.py — Splits sensation images into “referred” vs. “local” categories based on pixel area threshold (20 px). 05–07_Shape_*.py — Detect contours, compute geometric shape features, and cluster shapes. 08_Postprocessing.py — Merges and consolidates all intermediate outputs into final point-, subject-, and sensation-level tables. All scripts are hands-off (no interactive input required), use relative paths, and produce English-only descriptor outputs. Results (../results/): Intermediate and final CSV tables at descriptor-, sensation-, point-, subject-, and shape-level. Cluster assignments and stability analyses. Organized masked images (copied into cluster-/category-specific folders). Requirements Python ≥ 3.10 Dependencies: pandas, numpy, matplotlib, scikit-learn, seaborn, scipy Usage To reproduce the analysis, run the scripts sequentially (00–08) from the scripts/ folder. Intermediate and final outputs will be generated under ../results/.