{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:RW5V6QND5ATB2JCF7NBTX45JGR","short_pith_number":"pith:RW5V6QND","schema_version":"1.0","canonical_sha256":"8dbb5f41a3e8261d2445fb433bf3a9344c2c88f27260824fbcbbf97bac86804f","source":{"kind":"arxiv","id":"2606.22430","version":1},"attestation_state":"computed","paper":{"title":"Words as Difference Makers: How Large Language Models Determine Causal Structure in Text","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Wolfgang Pietsch","submitted_at":"2026-06-21T10:40:47Z","abstract_excerpt":"Because large language models (LLMs) are impressively successful in predicting text, it appears that they must have access to a 'world model' representing causal and definitional structure. However, the dominant formalisms of modern causal inference -- Judea Pearl's interventionist approach and the Neyman-Rubin potential outcomes framework -- struggle to illuminate how LLMs learn causal structure. I resolve this puzzle by arguing that LLMs employ a specific inductive approach based on a difference-making logic -- sometimes called variational induction. I demonstrate how central aspects of this"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.22430","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-21T10:40:47Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"ac91f194805ed26d33b1ff75501d083ce15dcfbcf54bd9b14bd7d8df2593b2ac","abstract_canon_sha256":"d8aebc0c1945d0dfb4bb2b9e4949417bda346bc42b7d13f97e64f02168860a1b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T02:13:38.252253Z","signature_b64":"x4+IupbRNk+rrPbj7x3safFEUZrgr4NJ1y2TqWANi8W546SXuMJMvOu5tPHmvJc87QkGYBleZg+PrFhv2UQ6BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8dbb5f41a3e8261d2445fb433bf3a9344c2c88f27260824fbcbbf97bac86804f","last_reissued_at":"2026-06-23T02:13:38.251887Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T02:13:38.251887Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Words as Difference Makers: How Large Language Models Determine Causal Structure in Text","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Wolfgang Pietsch","submitted_at":"2026-06-21T10:40:47Z","abstract_excerpt":"Because large language models (LLMs) are impressively successful in predicting text, it appears that they must have access to a 'world model' representing causal and definitional structure. However, the dominant formalisms of modern causal inference -- Judea Pearl's interventionist approach and the Neyman-Rubin potential outcomes framework -- struggle to illuminate how LLMs learn causal structure. I resolve this puzzle by arguing that LLMs employ a specific inductive approach based on a difference-making logic -- sometimes called variational induction. I demonstrate how central aspects of this"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.22430","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2606.22430/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.22430","created_at":"2026-06-23T02:13:38.251951+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.22430v1","created_at":"2026-06-23T02:13:38.251951+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.22430","created_at":"2026-06-23T02:13:38.251951+00:00"},{"alias_kind":"pith_short_12","alias_value":"RW5V6QND5ATB","created_at":"2026-06-23T02:13:38.251951+00:00"},{"alias_kind":"pith_short_16","alias_value":"RW5V6QND5ATB2JCF","created_at":"2026-06-23T02:13:38.251951+00:00"},{"alias_kind":"pith_short_8","alias_value":"RW5V6QND","created_at":"2026-06-23T02:13:38.251951+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR","json":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR.json","graph_json":"https://pith.science/api/pith-number/RW5V6QND5ATB2JCF7NBTX45JGR/graph.json","events_json":"https://pith.science/api/pith-number/RW5V6QND5ATB2JCF7NBTX45JGR/events.json","paper":"https://pith.science/paper/RW5V6QND"},"agent_actions":{"view_html":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR","download_json":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR.json","view_paper":"https://pith.science/paper/RW5V6QND","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.22430&json=true","fetch_graph":"https://pith.science/api/pith-number/RW5V6QND5ATB2JCF7NBTX45JGR/graph.json","fetch_events":"https://pith.science/api/pith-number/RW5V6QND5ATB2JCF7NBTX45JGR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR/action/storage_attestation","attest_author":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR/action/author_attestation","sign_citation":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR/action/citation_signature","submit_replication":"https://pith.science/pith/RW5V6QND5ATB2JCF7NBTX45JGR/action/replication_record"}},"created_at":"2026-06-23T02:13:38.251951+00:00","updated_at":"2026-06-23T02:13:38.251951+00:00"}