{"paper":{"title":"Do Fair Models Reason Fairly? Counterfactual Explanation Consistency for Procedural Fairness in Credit Decisions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Outcome-fair credit models can still apply different reasoning to similar individuals, a hidden procedural bias missed by standard metrics.","cross_cats":["cs.AI","cs.CE","cs.CY"],"primary_cat":"cs.LG","authors_text":"Gideon Popoola, John Sheppard","submitted_at":"2026-05-12T19:54:25Z","abstract_excerpt":"Machine learning algorithms in socially sensitive domains (e.g., credit decisions) often focus on equalizing predictive outcomes. However, satisfying these metrics does not guarantee that models use the same reasoning for different groups. We show that existing outcome-fair models can still apply fundamentally different reasoning to individuals, a ``hidden procedural bias'' missed by standard fairness metrics and algorithms. We propose Counterfactual Explanation Consistency (CEC), a framework that detects and mitigates this bias by aligning feature attributions between individuals and their co"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We show that existing outcome-fair models can still apply fundamentally different reasoning to individuals, a ``hidden procedural bias'' missed by standard fairness metrics and algorithms.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That nearest-neighbor counterfactuals and aligned integrated gradient attributions reliably capture and enforce procedural fairness without introducing artifacts from the generation method itself.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Outcome-fair credit models often exhibit hidden procedural bias through inconsistent reasoning across groups, which the CEC framework mitigates by enforcing consistent feature attributions via counterfactuals.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Outcome-fair credit models can still apply different reasoning to similar individuals, a hidden procedural bias missed by standard metrics.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c61f4223f3f2bc5544adab6a1f3dbc58653f93f80f98b3b9abea0b1b73e3eaa0"},"source":{"id":"2605.12701","kind":"arxiv","version":1},"verdict":{"id":"7832dd7f-fe60-4542-b456-616da47c0a29","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:46:43.747692Z","strongest_claim":"We show that existing outcome-fair models can still apply fundamentally different reasoning to individuals, a ``hidden procedural bias'' missed by standard fairness metrics and algorithms.","one_line_summary":"Outcome-fair credit models often exhibit hidden procedural bias through inconsistent reasoning across groups, which the CEC framework mitigates by enforcing consistent feature attributions via counterfactuals.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That nearest-neighbor counterfactuals and aligned integrated gradient attributions reliably capture and enforce procedural fairness without introducing artifacts from the generation method itself.","pith_extraction_headline":"Outcome-fair credit models can still apply different reasoning to similar individuals, a hidden procedural bias missed by standard metrics."},"references":{"count":299,"sample":[{"doi":"","year":null,"title":"Women in the American Political System: An Encyclopedia of Women as Voters, Candidates, and Office Holders , volume=","work_id":"9f1d1a74-e41e-4a7b-8f5b-c6a82d2818a0","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Home Mortgage Disclosure Act, and Community , year=","work_id":"670a24b9-7f65-4c5b-b9a8-cbcd24e09522","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2009,"title":"Causality , author=. 2009 , publisher=","work_id":"afebe501-3ef8-491b-998d-c996051cc590","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"International Journal of Management , volume=","work_id":"779390ec-7311-4932-b16b-6382e5308615","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Applied Soft Computing , volume=","work_id":"abdad045-9571-4baa-836a-f8de4c13d140","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":299,"snapshot_sha256":"6ef382d75938b8b9534e0d77b367682ba5b5c618db3a9f47de579b79a650bf92","internal_anchors":5},"formal_canon":{"evidence_count":2,"snapshot_sha256":"61275c580486d032009e64746eb07ccb2fd11f20f2f5a354f990c7f8dff0cd59"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}