Saved in:
Bibliographic Details
Main Authors: Lade, Ankit Hemant, Jasti, Sai Krishna, Kumar, Indar, Chadha, Aman
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.09169
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914548443774976
author Lade, Ankit Hemant
Jasti, Sai Krishna
Kumar, Indar
Chadha, Aman
author_facet Lade, Ankit Hemant
Jasti, Sai Krishna
Kumar, Indar
Chadha, Aman
contents A Mamba state-space model trained only for next-step prediction appears to recover Granger-causal structure through a simple readout $S = |W_{out} W_{in}|$, with early experiments suggesting the phenomenon generalized across architectures and benefited from interventional data at $p < 10^{-5}$. We package the protocol used to test that claim -- standardized synthetic generators (VAR/Lorenz/CauseMe-style), three intervention semantics ($do(X=c)$, soft-noise, random-forcing), edge-provenance cards on three real datasets, and size-matched control arms -- as a reusable falsification benchmark, and walk the claim through it in five stages. The method-level claim does not survive: (i) a plain linear bottleneck does as well or better; (ii) tuned Lasso beats the bottleneck on synthetic CauseMe-style benchmarks, and on Lorenz-96 (the only real benchmark with unambiguous ground truth) classical PCMCI and Granger lead a tight cluster in which the bottleneck trails; (iii) the headline intervention advantage is roughly 60% a sample-size confound, and the residual disappears under standard $do(X=c)$ interventions, surviving only under a non-standard random-forcing scheme; (iv) even that residual reproduces, with a larger effect, in classical bivariate Granger -- the effect is method-agnostic. What survives is a narrow characterization result; the benchmark is the lasting artifact, and each stage above is one of its control arms.
format Preprint
id arxiv_https___arxiv_org_abs_2605_09169
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Prediction Bottlenecks Don't Discover Causal Structure (But Here's What They Actually Do)
Lade, Ankit Hemant
Jasti, Sai Krishna
Kumar, Indar
Chadha, Aman
Machine Learning
Artificial Intelligence
A Mamba state-space model trained only for next-step prediction appears to recover Granger-causal structure through a simple readout $S = |W_{out} W_{in}|$, with early experiments suggesting the phenomenon generalized across architectures and benefited from interventional data at $p < 10^{-5}$. We package the protocol used to test that claim -- standardized synthetic generators (VAR/Lorenz/CauseMe-style), three intervention semantics ($do(X=c)$, soft-noise, random-forcing), edge-provenance cards on three real datasets, and size-matched control arms -- as a reusable falsification benchmark, and walk the claim through it in five stages. The method-level claim does not survive: (i) a plain linear bottleneck does as well or better; (ii) tuned Lasso beats the bottleneck on synthetic CauseMe-style benchmarks, and on Lorenz-96 (the only real benchmark with unambiguous ground truth) classical PCMCI and Granger lead a tight cluster in which the bottleneck trails; (iii) the headline intervention advantage is roughly 60% a sample-size confound, and the residual disappears under standard $do(X=c)$ interventions, surviving only under a non-standard random-forcing scheme; (iv) even that residual reproduces, with a larger effect, in classical bivariate Granger -- the effect is method-agnostic. What survives is a narrow characterization result; the benchmark is the lasting artifact, and each stage above is one of its control arms.
title Prediction Bottlenecks Don't Discover Causal Structure (But Here's What They Actually Do)
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2605.09169