{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SFLR6BF2NL7HU26SLMLTN6DRX5","short_pith_number":"pith:SFLR6BF2","schema_version":"1.0","canonical_sha256":"91571f04ba6afe7a6bd25b1736f871bf5404d09ba0cb777f159690d8be3b273f","source":{"kind":"arxiv","id":"2606.01914","version":1},"attestation_state":"computed","paper":{"title":"Mechanistic Diagnostics of Spatial Lexical Bias in Multimodal Large Language Model Spatial Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Akiko Aizawa, Chuang Ma, Fei Cheng, Qianying Liu, Sadao Kurohashi, Shuyuan Zheng, Sudong Cai, Tomoyuki Obuchi, Wang Yang","submitted_at":"2026-06-01T08:49:47Z","abstract_excerpt":"Multimodal large language models (MLLMs) remain unreliable on spatial multiple-choice questions, and their failures are often attributed to poorly attended visual information. In this work, we identify a complementary failure mode, spatial lexical bias: adding a spatial relation word to the answer options can attract the model's decision and make the newly added option likely to be selected. Using nine open-weight MLLMs, we show that this phenomenon is widely observed. In particular, models can answer a binary spatial question correctly, yet consistently select an incorrect third spatial optio"},"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":"2606.01914","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-01T08:49:47Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"a86556de9a18b096a389f1ac71dace25de67b0aa9f24442df0dee635eb827d9f","abstract_canon_sha256":"c7489d165577fd9eee256b54da1cbc3009983ce76c601821f0be06e45dbbef38"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:05:00.366051Z","signature_b64":"K9SYSsswG05nFa/5Oe8F66jcmeAqKc6htOQtCukvXXzA1aacdg5hP5Dkf+k/MwxqNv0bCec3EjCl5UdZaIZ3Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"91571f04ba6afe7a6bd25b1736f871bf5404d09ba0cb777f159690d8be3b273f","last_reissued_at":"2026-06-02T02:05:00.365605Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:05:00.365605Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mechanistic Diagnostics of Spatial Lexical Bias in Multimodal Large Language Model Spatial Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.CL","authors_text":"Akiko Aizawa, Chuang Ma, Fei Cheng, Qianying Liu, Sadao Kurohashi, Shuyuan Zheng, Sudong Cai, Tomoyuki Obuchi, Wang Yang","submitted_at":"2026-06-01T08:49:47Z","abstract_excerpt":"Multimodal large language models (MLLMs) remain unreliable on spatial multiple-choice questions, and their failures are often attributed to poorly attended visual information. In this work, we identify a complementary failure mode, spatial lexical bias: adding a spatial relation word to the answer options can attract the model's decision and make the newly added option likely to be selected. Using nine open-weight MLLMs, we show that this phenomenon is widely observed. In particular, models can answer a binary spatial question correctly, yet consistently select an incorrect third spatial optio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01914","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/2606.01914/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":"2606.01914","created_at":"2026-06-02T02:05:00.365664+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01914v1","created_at":"2026-06-02T02:05:00.365664+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01914","created_at":"2026-06-02T02:05:00.365664+00:00"},{"alias_kind":"pith_short_12","alias_value":"SFLR6BF2NL7H","created_at":"2026-06-02T02:05:00.365664+00:00"},{"alias_kind":"pith_short_16","alias_value":"SFLR6BF2NL7HU26S","created_at":"2026-06-02T02:05:00.365664+00:00"},{"alias_kind":"pith_short_8","alias_value":"SFLR6BF2","created_at":"2026-06-02T02:05:00.365664+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/SFLR6BF2NL7HU26SLMLTN6DRX5","json":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5.json","graph_json":"https://pith.science/api/pith-number/SFLR6BF2NL7HU26SLMLTN6DRX5/graph.json","events_json":"https://pith.science/api/pith-number/SFLR6BF2NL7HU26SLMLTN6DRX5/events.json","paper":"https://pith.science/paper/SFLR6BF2"},"agent_actions":{"view_html":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5","download_json":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5.json","view_paper":"https://pith.science/paper/SFLR6BF2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01914&json=true","fetch_graph":"https://pith.science/api/pith-number/SFLR6BF2NL7HU26SLMLTN6DRX5/graph.json","fetch_events":"https://pith.science/api/pith-number/SFLR6BF2NL7HU26SLMLTN6DRX5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5/action/storage_attestation","attest_author":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5/action/author_attestation","sign_citation":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5/action/citation_signature","submit_replication":"https://pith.science/pith/SFLR6BF2NL7HU26SLMLTN6DRX5/action/replication_record"}},"created_at":"2026-06-02T02:05:00.365664+00:00","updated_at":"2026-06-02T02:05:00.365664+00:00"}