{"id":"https://openalex.org/W7123645466","doi":"https://doi.org/10.48550/arxiv.2601.07226","title":"Lost in the Noise: How Reasoning Models Fail with Contextual Distractors","display_name":"Lost in the Noise: How Reasoning Models Fail with Contextual Distractors","publication_year":2026,"publication_date":"2026-01-12","ids":{"openalex":"https://openalex.org/W7123645466","doi":"https://doi.org/10.48550/arxiv.2601.07226"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2601.07226","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.07226","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2601.07226","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5122942539","display_name":"Seongyun Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Seongyun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122973916","display_name":"Yongrae Jo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jo, Yongrae","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122950568","display_name":"Minju Seo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Seo, Minju","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5043485792","display_name":"Moontae Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Moontae","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5087565126","display_name":"Minjoon Seo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Seo, Minjoon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.4641000032424927,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.4641000032424927,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.24770000576972961,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.15320000052452087,"subfield":{"id":"https://openalex.org/subfields/3311","display_name":"Safety Research"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.5852000117301941},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.49549999833106995},{"id":"https://openalex.org/keywords/adversarial-system","display_name":"Adversarial system","score":0.4814999997615814},{"id":"https://openalex.org/keywords/perception","display_name":"Perception","score":0.4602000117301941},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.43209999799728394},{"id":"https://openalex.org/keywords/workflow","display_name":"Workflow","score":0.4235999882221222},{"id":"https://openalex.org/keywords/identification","display_name":"Identification (biology)","score":0.392300009727478}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7605999708175659},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5852000117301941},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5145000219345093},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.49549999833106995},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.4814999997615814},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.4602000117301941},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.43209999799728394},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.42419999837875366},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.4235999882221222},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.392300009727478},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.32010000944137573},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.3197000026702881},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.3109000027179718},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.30640000104904175},{"id":"https://openalex.org/C157657479","wikidata":"https://www.wikidata.org/wiki/Q2367247","display_name":"Closed captioning","level":3,"score":0.2921999990940094},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.29019999504089355},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.2847000062465668},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.2824000120162964},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2784999907016754},{"id":"https://openalex.org/C2779585090","wikidata":"https://www.wikidata.org/wiki/Q3457762","display_name":"Resilience (materials science)","level":2,"score":0.2685999870300293}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2601.07226","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.07226","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2601.07226","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.07226","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advances":[1],"in":[2,52,84,132,166],"reasoning":[3],"models":[4,86,175],"and":[5,56,69,107,124,169],"agentic":[6,96],"AI":[7],"systems":[8],"have":[9],"led":[10],"to":[11,36,82,129,163],"an":[12,154],"increased":[13,159],"reliance":[14],"on":[15,178],"diverse":[16,60],"external":[17],"information.":[18],"However,":[19],"this":[20],"shift":[21],"introduces":[22],"input":[23],"contexts":[24],"that":[25,31,44,95,119,174],"are":[26],"inherently":[27],"noisy,":[28],"a":[29,41,76],"reality":[30],"current":[32],"sanitized":[33],"benchmarks":[34],"fail":[35,128],"capture.":[37],"We":[38,117],"introduce":[39],"NoisyBench,":[40],"comprehensive":[42],"benchmark":[43],"systematically":[45],"evaluates":[46],"model":[47],"robustness":[48],"across":[49],"11":[50],"datasets":[51],"RAG,":[53],"reasoning,":[54],"alignment,":[55],"tool-use":[57],"tasks":[58],"against":[59],"noise":[61],"types,":[62],"including":[63],"random":[64],"documents,":[65],"irrelevant":[66],"chat":[67],"histories,":[68],"hard":[70],"negative":[71],"distractors.":[72,91],"Our":[73],"evaluation":[74],"reveals":[75],"catastrophic":[77],"performance":[78,165],"drop":[79],"of":[80,146,189],"up":[81],"80%":[83],"state-of-the-art":[85],"when":[87],"faced":[88],"with":[89],"contextual":[90],"Crucially,":[92],"we":[93,152],"find":[94,118],"workflows":[97],"often":[98],"amplify":[99],"these":[100],"errors":[101],"by":[102,142],"over-trusting":[103],"noisy":[104,167],"tool":[105],"outputs,":[106],"distractors":[108],"can":[109],"trigger":[110],"emergent":[111],"misalignment":[112],"even":[113],"without":[114],"adversarial":[115],"intent.":[116],"prompting,":[120],"context":[121],"engineering,":[122],"SFT,":[123],"outcome-reward":[125],"only":[126],"RL":[127],"ensure":[130],"robustness;":[131],"contrast,":[133],"our":[134],"proposed":[135],"Rationale-Aware":[136],"Reward":[137],"(RARE)":[138],"significantly":[139],"strengthens":[140],"resilience":[141],"incentivizing":[143],"the":[144,186],"identification":[145],"helpful":[147],"information":[148],"within":[149],"noise.":[150],"Finally,":[151],"uncover":[153],"inverse":[155],"scaling":[156],"trend":[157],"where":[158],"test-time":[160],"computation":[161],"leads":[162],"worse":[164],"settings":[168],"demonstrate":[170],"via":[171],"attention":[172],"visualization":[173],"disproportionately":[176],"focus":[177],"distractor":[179],"tokens,":[180],"providing":[181],"vital":[182],"insights":[183],"for":[184],"building":[185],"next":[187],"generation":[188],"robust,":[190],"reasoning-capable":[191],"agents.":[192]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-01-14T00:00:00"}