A Computational Test of Visual Similarity Between the Indus Valley Script and Western and Eastern Bronze Age Writing Systems: A pHash-based permutation test of candidate glyph pairs

This study evaluates whether apparent visual resemblances between selected Indus Valley glyphs and glyphs from two geographically distinct Bronze-Age writing traditions — a “Western” group (Proto-Cuneiform and Proto-Elamite) and an “Eastern” group (Oracle-Bone Chinese and Classical Yi) — are more likely than chance to occur across the two corpora. Rather than attempting linguistic claims or decipherment, we treat each glyph as a visual object and ask a narrowly defined, testable question: are a curated set of hypothesized Indus ↔ other-script pairs systematically more visually similar than random cross-script pairings drawn from the same glyph pools? To answer that question we use perceptual image hashing (pHash) to produce compact fingerprints (64-bit) of standardized glyph images, then measure similarity with Hamming distance and evaluate significance using a nonparametric permutation test (10,000 random re-pairings). We further run a complexity-matched control (quantile bins by stroke length) to reduce the risk that effects are driven by simple/complex sign imbalances rather than by visual structure.