The Geometry of Forgetting
Sambartha Ray Barman, Andrey Starenky, Sophia Bodnar, Nikhil Narasimhan, Ashwin Gopinath · 2026
Read The Geometry of Forgetting by Sambartha Ray Barman, Andrey Starenky, Sophia Bodnar, Nikhil Narasimhan, Ashwin Gopinath (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-06222.txt.
Description
Barman and four colleagues argue that forgetting comes from geometry, not biology. High-dimensional embedding spaces under noise, clashes, and decay mirror human memory patterns with no special engineering. Power-law loss, with an exponent near 0.46 close to the human value, comes from rival memories interfering: without rivals the exponent collapses fiftyfold. Models pack spread into about 16 dimensions, leaving them open to clashes. Even false memories need no design: similarity over untouched embeddings matches the false-alarm rate, showing memory quirks as features of meaning-based retrieval.
Catalog record
| CREATOR | Sambartha Ray Barman, Andrey Starenky, Sophia Bodnar, Nikhil Narasimhan, Ashwin Gopinath |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49664 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.06222