{"id":"https://openalex.org/W7166863291","doi":"https://doi.org/10.18653/v1/2026.acl-long.304","title":"From \"Aha Moments\" to Controllable Thinking: Toward Meta-Cognitive Reasoning in LRMs via Decoupled Reasoning and Control","display_name":"From \"Aha Moments\" to Controllable Thinking: Toward Meta-Cognitive Reasoning in LRMs via Decoupled Reasoning and Control","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166863291","doi":"https://doi.org/10.18653/v1/2026.acl-long.304"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.304","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.304","pdf_url":"https://aclanthology.org/2026.acl-long.304.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.304.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139737317","display_name":"Rui Ha","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rui Ha","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139820060","display_name":"Rui Pu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rui Pu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139723260","display_name":"Chaozhuo Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chaozhuo Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139804288","display_name":"Li Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li Sun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139738040","display_name":"Sen Su","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sen Su","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.85216512,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"6698","last_page":"6710"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10906","display_name":"AI-based Problem Solving and Planning","score":0.2003999948501587,"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/T10906","display_name":"AI-based Problem Solving and Planning","score":0.2003999948501587,"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/T11010","display_name":"Logic, Reasoning, and Knowledge","score":0.08630000054836273,"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/T12805","display_name":"Cognitive Science and Mapping","score":0.07450000196695328,"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/control","display_name":"Control (management)","score":0.482699990272522},{"id":"https://openalex.org/keywords/control-theory","display_name":"Control theory (sociology)","score":0.4560999870300293},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.4115000069141388},{"id":"https://openalex.org/keywords/control-system","display_name":"Control system","score":0.3337000012397766},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.29159998893737793},{"id":"https://openalex.org/keywords/controllability","display_name":"Controllability","score":0.2818000018596649}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5508000254631042},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.482699990272522},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.4560999870300293},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.4115000069141388},{"id":"https://openalex.org/C17500928","wikidata":"https://www.wikidata.org/wiki/Q959968","display_name":"Control system","level":2,"score":0.3337000012397766},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3296999931335449},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2976999878883362},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.29159998893737793},{"id":"https://openalex.org/C48209547","wikidata":"https://www.wikidata.org/wiki/Q1331104","display_name":"Controllability","level":2,"score":0.2818000018596649},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.27639999985694885},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.2653000056743622},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2648000121116638},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.26019999384880066},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.25279998779296875}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.304","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.304","pdf_url":"https://aclanthology.org/2026.acl-long.304.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.304","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.304","pdf_url":"https://aclanthology.org/2026.acl-long.304.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/10","score":0.5970974564552307,"display_name":"Reduced inequalities"}],"awards":[],"funders":[{"id":"https://openalex.org/F4320335777","display_name":"National Key Research and Development Program of China","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166863291.pdf","grobid_xml":"https://content.openalex.org/works/W7166863291.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Reasoning":[1],"Models":[2],"(LRMs)":[3],"can":[4],"exhibit":[5],"step-by-step":[6],"reasoning,":[7],"reflection,":[8],"and":[9,35,54,89,111,141],"backtracking,":[10],"but":[11],"these":[12],"behaviors":[13],"are":[14],"often":[15],"unregulated,":[16],"leading":[17],"to":[18,49,57,73,106,126],"overthinking.As":[19],"a":[20,64,86],"result,":[21],"LRMs":[22],"continue":[23],"generating":[24],"redundant":[25],"reasoning":[26,52,66,70,133],"even":[27],"after":[28],"reaching":[29],"high-confidence":[30],"conclusions.This":[31],"increases":[32],"inference":[33],"cost":[34],"latency,":[36],"limiting":[37],"practical":[38],"deployment.The":[39],"root":[40],"cause":[41],"is":[42,101],"the":[43,51,108],"absence":[44],"of":[45,77],"an":[46],"intrinsic":[47],"mechanism":[48],"monitor":[50],"state":[53],"decide":[55],"when":[56],"continue,":[58],"backtrack,":[59],"or":[60],"stop.We":[61],"propose":[62],"MERA,":[63],"meta-cognitive":[65],"framework":[67],"that":[68,136],"decouples":[69],"from":[71],"control":[72,78,124,130],"enable":[74],"independent":[75],"optimization":[76],"strategies.MERA":[79],"constructs":[80],"high-quality":[81],"reasoning-control":[82,95],"supervision":[83],"data":[84],"via":[85],"takeoverbased":[87],"pipeline,":[88],"transforms":[90],"long-horizon":[91],"traces":[92],"into":[93],"structured":[94,109],"alternating":[96],"sequences":[97],"for":[98],"training.The":[99],"model":[100],"trained":[102],"with":[103,114,123],"supervised":[104],"fine-tuning":[105],"internalize":[107],"separation,":[110],"further":[112],"optimized":[113],"Control-Segment":[115],"Policy":[116],"Optimization":[117],"(CSPO),":[118],"which":[119],"combines":[120],"segment-wise":[121],"GRPO":[122],"masking":[125],"focus":[127],"learning":[128],"on":[129],"segments.Experiments":[131],"across":[132],"benchmarks":[134],"show":[135],"MERA":[137],"improves":[138],"both":[139],"efficiency":[140],"accuracy.":[142]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}