{
  "schema_version": "1.0",
  "id": "s13:https://arxiv.org/abs/2607.21574v1",
  "slug": "2607-21574v1-0m2upel",
  "url": "https://feed7.dev/p/2607-21574v1-0m2upel",
  "title": "Surprisal Theory is Tautological (without Rational Grounding)",
  "why_included": "The paper argues that unconstrained surprisal can fit any non-negative processing-difficulty pattern, so corpus fit alone cannot make claims about human language processing falsifiable.",
  "summary": "The paper shows that, under mild conditions, **any non-negative difficulty measure** can be represented as an affine function of surprisal under some language model. Without constraining that model, surprisal theory therefore makes no falsifiable prediction.",
  "practical_implication": "Builders evaluating language models against human behavior should not treat better corpus likelihood as evidence of better cognitive fidelity. The proposed remedy is to derive the model from independent assumptions such as memory limits or processing goals.",
  "agent_context": "The paper shows that, under mild conditions, **any non-negative difficulty measure** can be represented as an affine function of surprisal under some language model. Without constraining that model, surprisal theory therefore makes no falsifiable prediction.\n\nBuilders evaluating language models against human behavior should not treat better corpus likelihood as evidence of better cognitive fidelity. The proposed remedy is to derive the model from independent assumptions such as memory limits or processing goals.\n\nThis is a theoretical argument, not a new agent evaluation or empirical benchmark. Its practical force depends on whether researchers can specify independently motivated comprehender models that produce testable predictions.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2607.21574v1",
    "published_at": "2026-07-23T17:54:37.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "benchmark",
  "domains": [
    "research"
  ],
  "topics": [
    "benchmark-integrity"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "This is a theoretical argument, not a new agent evaluation or empirical benchmark. Its practical force depends on whether researchers can specify independently motivated comprehender models that produce testable predictions."
  ],
  "lifecycle": "Current",
  "published_at": "2026-07-23T17:54:37.000Z",
  "modified_at": "2026-07-23T17:54:37.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2607-21574v1-0m2upel",
    "json": "https://feed7.dev/p/2607-21574v1-0m2upel.json",
    "markdown": "https://feed7.dev/p/2607-21574v1-0m2upel.md"
  }
}