{"id":"https://openalex.org/W7214036154","doi":"https://doi.org/10.5281/zenodo.22904980","title":"When Deliberation Hurts: Inverse Test-Time Scaling, Unfaithful Traces, and the Case Against a Unified System-2 in LLM Reasoning","display_name":"When Deliberation Hurts: Inverse Test-Time Scaling, Unfaithful Traces, and the Case Against a Unified System-2 in LLM Reasoning","publication_year":2026,"publication_date":"2026-09-23","ids":{"openalex":"https://openalex.org/W7214036154","doi":"https://doi.org/10.5281/zenodo.22904980"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.22904980","doi":"https://doi.org/10.5281/zenodo.22904980","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.22904980","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":"Preprint"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.22904980","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5147720734","display_name":"Pranay M. Mahendrakar","orcid":"https://orcid.org/0009-0003-7224-029X","observed_orcids":["https://orcid.org/0009-0003-7224-029X"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Mahendrakar, Pranay","raw_affiliation_strings":["SONYTECH"],"raw_orcid":"https://orcid.org/0009-0003-7224-029X","affiliations":[{"raw_affiliation_string":"SONYTECH","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5147720734"],"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/T10028","display_name":"Topic Modeling","score":0.2842000126838684,"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/T10028","display_name":"Topic Modeling","score":0.2842000126838684,"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":0.1754000037908554,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.10939999669790268,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/llm-reasoning","display_name":"LLM reasoning","score":0.7440000176429749},{"id":"https://openalex.org/keywords/dual-process-theory","display_name":"dual-process theory","score":0.7760000228881836},{"id":"https://openalex.org/keywords/system-1-and-system-2","display_name":"System 1 and System 2","score":0.8840000033378601},{"id":"https://openalex.org/keywords/chain-of-thought","display_name":"chain-of-thought","score":0.6919999718666077},{"id":"https://openalex.org/keywords/test-time-compute-scaling","display_name":"test-time compute scaling","score":0.7630000114440918},{"id":"https://openalex.org/keywords/reasoning-faithfulness","display_name":"reasoning faithfulness","score":0.5730000138282776},{"id":"https://openalex.org/keywords/hallucinations","display_name":"hallucinations","score":0.6890000104904175},{"id":"https://openalex.org/keywords/factual-accuracy","display_name":"factual accuracy","score":0.4620000123977661},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.5960000157356262},{"id":"https://openalex.org/keywords/mechanistic-interpretability","display_name":"mechanistic interpretability","score":0.5830000042915344}],"concepts":[{"id":"https://openalex.org/C2778023277","wikidata":"https://www.wikidata.org/wiki/Q321703","display_name":"Premise","level":2,"score":0.6754000186920166},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5467000007629395},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.5339999794960022},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5210000276565552},{"id":"https://openalex.org/C166088908","wikidata":"https://www.wikidata.org/wiki/Q308495","display_name":"Abductive reasoning","level":2,"score":0.44429999589920044},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.4381999969482422},{"id":"https://openalex.org/C2776946740","wikidata":"https://www.wikidata.org/wiki/Q358652","display_name":"Deliberation","level":3,"score":0.4122999906539917},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.4090000092983246},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.4083999991416931},{"id":"https://openalex.org/C36964233","wikidata":"https://www.wikidata.org/wiki/Q7920942","display_name":"Verbal reasoning","level":3,"score":0.40299999713897705},{"id":"https://openalex.org/C89288958","wikidata":"https://www.wikidata.org/wiki/Q7301504","display_name":"Reasoning system","level":2,"score":0.4000000059604645},{"id":"https://openalex.org/C115086926","wikidata":"https://www.wikidata.org/wiki/Q17004651","display_name":"Causal reasoning","level":3,"score":0.39590001106262207},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.38429999351501465},{"id":"https://openalex.org/C97364631","wikidata":"https://www.wikidata.org/wiki/Q484284","display_name":"Deductive reasoning","level":2,"score":0.3765999972820282},{"id":"https://openalex.org/C103057564","wikidata":"https://www.wikidata.org/wiki/Q4751139","display_name":"Analytic reasoning","level":3,"score":0.3709000051021576},{"id":"https://openalex.org/C79585631","wikidata":"https://www.wikidata.org/wiki/Q431498","display_name":"Confirmation bias","level":2,"score":0.3700000047683716},{"id":"https://openalex.org/C86827895","wikidata":"https://www.wikidata.org/wiki/Q7098582","display_name":"Opportunistic reasoning","level":4,"score":0.3637999892234802},{"id":"https://openalex.org/C37335422","wikidata":"https://www.wikidata.org/wiki/Q6888134","display_name":"Model-based reasoning","level":3,"score":0.353300005197525},{"id":"https://openalex.org/C183521366","wikidata":"https://www.wikidata.org/wiki/Q7256422","display_name":"Psychology of reasoning","level":4,"score":0.34950000047683716},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.34130001068115234},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.33009999990463257},{"id":"https://openalex.org/C108650721","wikidata":"https://www.wikidata.org/wiki/Q1783253","display_name":"Counterfactual thinking","level":2,"score":0.31790000200271606},{"id":"https://openalex.org/C111472728","wikidata":"https://www.wikidata.org/wiki/Q9471","display_name":"Epistemology","level":1,"score":0.299699991941452},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.2822999954223633},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.27480000257492065},{"id":"https://openalex.org/C83725634","wikidata":"https://www.wikidata.org/wiki/Q7268699","display_name":"Qualitative reasoning","level":2,"score":0.2736000120639801},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2667999863624573},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.2605000138282776},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2574000060558319},{"id":"https://openalex.org/C7149132","wikidata":"https://www.wikidata.org/wiki/Q1377840","display_name":"Forgetting","level":2,"score":0.25529998540878296},{"id":"https://openalex.org/C2777267654","wikidata":"https://www.wikidata.org/wiki/Q3519023","display_name":"Test (biology)","level":2,"score":0.25279998779296875}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.5281/zenodo.22904980","doi":"https://doi.org/10.5281/zenodo.22904980","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.22904980","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":"Preprint"},{"id":"doi:10.5281/zenodo.22904981","doi":"https://doi.org/10.5281/zenodo.22904981","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.22904981","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":"JournalArticle"},{"id":"doi:10.5281/zenodo.23022324","doi":"https://doi.org/10.5281/zenodo.23022324","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.23022324","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":"Preprint"}],"best_oa_location":{"id":"doi:10.5281/zenodo.22904980","doi":"https://doi.org/10.5281/zenodo.22904980","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.22904980","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"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":"Preprint"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W6966787974","https://openalex.org/W4417429064","https://openalex.org/W7221029564","https://openalex.org/W7222962320","https://openalex.org/W4416887915","https://openalex.org/W4415057168","https://openalex.org/W4414588372","https://openalex.org/W7081972478","https://openalex.org/W7119493921","https://openalex.org/W7166849418"],"abstract_inverted_index":{"(c)":[0],"2026":[1],"Pranay":[2],"Mahendrakar.":[3],"Licensed":[4],"under":[5,312,395],"CC":[6],"BY":[7],"4.0.":[8],"The":[9,322,379,460],"dominant":[10],"frame":[11,57],"for":[12,386,464,469],"large":[13],"reasoning":[14,53,98,199,226,282,377],"models":[15,175,249],"borrows":[16],"a":[17,22,28,97,160,165,231,257,285,296,336,340,361,371],"label":[18],"from":[19,67,430],"dual-process":[20],"psychology:":[21],"fast,":[23],"intuitive":[24],"System":[25,31,68,71],"1":[26,69],"and":[27,44,130,137,149,180,197,264,273,308,316,350,374,383,413,421,447,468],"slower,":[29],"deliberate":[30,224],"2,":[32],"with":[33,120,343,392,411,441],"longer":[34,281],"chains":[35],"of":[36,41,52,85,168,230,277,298,370,437],"thought":[37],"read":[38],"as":[39,58,73,339],"more":[40,48,181,192,366,369,375],"the":[42,56,59,65,74,158,169,235,269,299,355,396,418,431,438,465],"latter":[43],"therefore,":[45],"on":[46,108,187,201,301,348],"average,":[47],"reliable.":[49],"A":[50],"survey":[51],"LLMs":[54],"states":[55],"field's":[60],"own":[61],"premise":[62,90],"--":[63],"refining":[64],"transition":[66],"to":[70,76,112,177,354],"2":[72],"route":[75],"human-level":[77],"intelligence":[78],"(Li":[79],"et":[80,134,154,194,210,245,305,319],"al.,":[81,135,155,195,211,246,306,320],"2025).":[82,156,212,247,275],"Four":[83],"bodies":[84],"published":[86],"evidence":[87],"complicate":[88],"that":[89,171,220,280,325],"rather":[91,241],"than":[92,242,359],"confirming":[93],"it.":[94,442,474],"First,":[95],"extending":[96],"model's":[99,356],"chain":[100,159],"length":[101],"does":[102,142],"not":[103,143,164,191,251],"merely":[104],"show":[105],"diminishing":[106],"returns;":[107],"tasks":[109,132,189,317],"built":[110],"specifically":[111],"test":[113],"it,":[114],"it":[115],"produces":[116],"monotonically":[117],"worse":[118],"accuracy,":[119],"five":[121],"distinct":[122,373],"failure":[123],"modes":[124],"identified":[125],"across":[126],"counting,":[127],"regression,":[128],"deduction":[129],"safety-relevant":[131],"(Gema":[133],"2025),":[136],"separately,":[138],"test-time":[139],"compute":[140],"scaling":[141,294],"consistently":[144],"improve":[145],"closed-book":[146],"factual":[147],"accuracy":[148,349],"often":[150],"increases":[151],"hallucination":[152],"(Zhao":[153],"Second,":[157],"model":[161,236,314,337],"verbalizes":[162],"is":[163,227,310,428,462],"reliable":[166],"readout":[167],"computation":[170],"produced":[172],"its":[173,404,458],"answer:":[174],"trained":[176],"be":[178],"larger":[179],"capable":[182],"produce":[183],"less":[184],"faithful":[185],"chains-of-thought":[186],"most":[188],"studied,":[190],"(Lanham":[193,318],"2023),":[196],"unfaithful":[198],"appears":[200],"ordinary,":[202],"non-adversarial":[203],"prompts":[204],"without":[205],"any":[206],"injected":[207],"bias":[208],"(Arcuschin":[209],"Third,":[213],"at":[214,295],"least":[215],"one":[216],"mechanistic":[217],"study":[218,271],"reports":[219],"what":[221],"looks":[222],"like":[223],"System-2":[225],"frequently":[228],"downstream":[229],"System-1-like":[232],"snap":[233],"judgment":[234],"then":[237],"spends":[238],"tokens":[239],"rationalizing":[240],"revising":[243],"(Dang":[244],"Fourth,":[248],"do":[250],"reliably":[252],"know":[253],"how":[254],"much":[255,335],"deliberation":[256],"given":[258],"problem":[259],"needs,":[260],"overthinking":[261],"easy":[262],"problems":[263],"underthinking":[265],"hard":[266],"ones":[267],"in":[268,364,425,450,473],"same":[270],"(Su":[272],"Healey,":[274],"None":[276],"this":[278,387,426,451],"shows":[279],"never":[283],"helps;":[284],"compute-optimal":[286],"allocation":[287],"strategy":[288],"can":[289],"match":[290],"or":[291,406,433,455],"exceed":[292],"brute-force":[293],"fraction":[297],"cost":[300],"math":[302],"benchmarks":[303],"(Snell":[304],"2024),":[307],"faithfulness":[309],"recoverable":[311],"some":[313],"sizes":[315],"2023).":[321],"paper":[323,388,427,452],"argues":[324],"these":[326],"results":[327],"are":[328],"jointly":[329],"better":[330],"explained":[331],"by":[332,360,457],"treating":[333],"\"how":[334],"deliberates\"":[338],"token-length":[341],"variable":[342],"heterogeneous,":[344],"sometimes":[345],"negative,":[346],"effects":[347],"an":[351],"unreliable":[352],"relationship":[353],"actual":[357],"computation,":[358],"two-system":[362],"story":[363],"which":[365],"chain-of-thought":[367],"means":[368],"qualitatively":[372],"trustworthy":[376],"mode.":[378],"literature":[380],"search,":[381],"drafting":[382],"citation":[384,400],"verification":[385],"were":[389],"carried":[390],"out":[391],"AI":[393],"assistance":[394],"author's":[397],"direction.":[398],"Every":[399],"was":[401,445,453],"machine-verified":[402],"against":[403,417],"arXiv":[405],"DataCite":[407],"record":[408,419],"before":[409],"inclusion,":[410],"title":[412],"author":[414,461],"list":[415],"checked":[416],"returned,":[420],"every":[422],"quantitative":[423],"claim":[424],"taken":[429],"abstract":[432],"stated":[434],"headline":[435],"result":[436],"source":[439],"credited":[440],"No":[443],"experiment":[444],"run":[446],"no":[448],"number":[449],"measured":[454],"recomputed":[456],"author.":[459],"responsible":[463],"final":[466],"text":[467],"all":[470],"claims":[471],"made":[472]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-23T00:00:00"}
