{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:53NWLEWVHEGZX4JSPQQ7YZIJLB","short_pith_number":"pith:53NWLEWV","schema_version":"1.0","canonical_sha256":"eedb6592d5390d9bf1327c21fc650958655b1fc690bc134df428390fd00aea1e","source":{"kind":"arxiv","id":"2602.22766","version":2},"attestation_state":"computed","paper":{"title":"Imagination Helps Visual Reasoning, But Not Yet in Latent Space","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chi Chen, Fanhu Zeng, Jinan Xu, Kaiyu Huang, Maosong Sun, Yanghao Li, You Li","submitted_at":"2026-02-26T08:56:23Z","abstract_excerpt":"Latent visual reasoning aims to mimic human's imagination process by meditating through hidden states of Multimodal Large Language Models. While recognized as a promising paradigm for visual reasoning, the underlying mechanisms driving its effectiveness remain unclear. Motivated to demystify the true source of its efficacy, we investigate the validity of latent reasoning using Causal Mediation Analysis. We model the process as a causal chain: the input as the treatment, the latent tokens as the mediator, and the final answer as the outcome. Our findings uncover two critical disconnections: (a)"},"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":"2602.22766","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-02-26T08:56:23Z","cross_cats_sorted":[],"title_canon_sha256":"c5f423cbb5b5dac768d9d1ef8e4ee00bfb01c020e393880da3e66517f8cf6108","abstract_canon_sha256":"4090fa21fd84e63308c959e0c9b6c9770ffec2914d340e58f8bd3713cc868420"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:15.114663Z","signature_b64":"9ZkkqyOLw5motDvTp99gHbodXtCODl+9DKgTFgS/w6JGaB/YPGrIs6l5C0TKmgbCK0CXa2zxhjH62WSHdLv5AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eedb6592d5390d9bf1327c21fc650958655b1fc690bc134df428390fd00aea1e","last_reissued_at":"2026-06-09T01:05:15.114172Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:15.114172Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Imagination Helps Visual Reasoning, But Not Yet in Latent Space","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chi Chen, Fanhu Zeng, Jinan Xu, Kaiyu Huang, Maosong Sun, Yanghao Li, You Li","submitted_at":"2026-02-26T08:56:23Z","abstract_excerpt":"Latent visual reasoning aims to mimic human's imagination process by meditating through hidden states of Multimodal Large Language Models. While recognized as a promising paradigm for visual reasoning, the underlying mechanisms driving its effectiveness remain unclear. Motivated to demystify the true source of its efficacy, we investigate the validity of latent reasoning using Causal Mediation Analysis. We model the process as a causal chain: the input as the treatment, the latent tokens as the mediator, and the final answer as the outcome. Our findings uncover two critical disconnections: (a)"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.22766","kind":"arxiv","version":2},"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/2602.22766/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":"2602.22766","created_at":"2026-06-09T01:05:15.114231+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.22766v2","created_at":"2026-06-09T01:05:15.114231+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.22766","created_at":"2026-06-09T01:05:15.114231+00:00"},{"alias_kind":"pith_short_12","alias_value":"53NWLEWVHEGZ","created_at":"2026-06-09T01:05:15.114231+00:00"},{"alias_kind":"pith_short_16","alias_value":"53NWLEWVHEGZX4JS","created_at":"2026-06-09T01:05:15.114231+00:00"},{"alias_kind":"pith_short_8","alias_value":"53NWLEWV","created_at":"2026-06-09T01:05:15.114231+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2605.18445","citing_title":"What's Holding Back Latent Visual Reasoning?","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18641","citing_title":"Leveraging Latent Visual Reasoning in Silence","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07106","citing_title":"Retrieve, Integrate, and Synthesize: Spatial-Semantic Grounded Latent Visual Reasoning","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2604.21027","citing_title":"HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering","ref_index":55,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB","json":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB.json","graph_json":"https://pith.science/api/pith-number/53NWLEWVHEGZX4JSPQQ7YZIJLB/graph.json","events_json":"https://pith.science/api/pith-number/53NWLEWVHEGZX4JSPQQ7YZIJLB/events.json","paper":"https://pith.science/paper/53NWLEWV"},"agent_actions":{"view_html":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB","download_json":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB.json","view_paper":"https://pith.science/paper/53NWLEWV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.22766&json=true","fetch_graph":"https://pith.science/api/pith-number/53NWLEWVHEGZX4JSPQQ7YZIJLB/graph.json","fetch_events":"https://pith.science/api/pith-number/53NWLEWVHEGZX4JSPQQ7YZIJLB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB/action/storage_attestation","attest_author":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB/action/author_attestation","sign_citation":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB/action/citation_signature","submit_replication":"https://pith.science/pith/53NWLEWVHEGZX4JSPQQ7YZIJLB/action/replication_record"}},"created_at":"2026-06-09T01:05:15.114231+00:00","updated_at":"2026-06-09T01:05:15.114231+00:00"}