{"id":"https://openalex.org/W7164002273","doi":"https://doi.org/10.48550/arxiv.2606.08719","title":"Thinking Without Images: Internalizing Visual Manipulation with On-Policy Self-Distillation","display_name":"Thinking Without Images: Internalizing Visual Manipulation with On-Policy Self-Distillation","publication_year":2026,"publication_date":"2026-06-07","ids":{"openalex":"https://openalex.org/W7164002273","doi":"https://doi.org/10.48550/arxiv.2606.08719"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.08719","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.08719","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.08719","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114174972","display_name":"Yishuo Cai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cai, Yishuo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138261077","display_name":"Jiahui Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Jiahui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138261745","display_name":"Yuanxin Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Yuanxin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102543115","display_name":"Haobo Deng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deng, Haobo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138208771","display_name":"Linli Yao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yao, Linli","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102802347","display_name":"Yuhao Zheng","orcid":"https://orcid.org/0009-0008-1216-1684"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Yuhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138275279","display_name":"Kun Ouyang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ouyang, Kun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137632453","display_name":"Zhimo Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zhimo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138215221","display_name":"Ziyue Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Ziyue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138239588","display_name":"Xu Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Xu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138286323","display_name":"Haoli Bai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bai, Haoli","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138226123","display_name":"Xiaohui Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Xiaohui","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":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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.3946000039577484,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.3946000039577484,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.1251000016927719,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.11810000240802765,"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/semantic-reasoner","display_name":"Semantic reasoner","score":0.808899998664856},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6974999904632568},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.5934000015258789},{"id":"https://openalex.org/keywords/visual-reasoning","display_name":"Visual reasoning","score":0.5590000152587891},{"id":"https://openalex.org/keywords/outcome","display_name":"Outcome (game theory)","score":0.5254999995231628},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.4372999966144562},{"id":"https://openalex.org/keywords/zoom","display_name":"Zoom","score":0.41929998993873596},{"id":"https://openalex.org/keywords/visual-perception","display_name":"Visual perception","score":0.38359999656677246},{"id":"https://openalex.org/keywords/eye-tracking","display_name":"Eye tracking","score":0.3237999975681305}],"concepts":[{"id":"https://openalex.org/C9616225","wikidata":"https://www.wikidata.org/wiki/Q3929429","display_name":"Semantic reasoner","level":2,"score":0.808899998664856},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6974999904632568},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.5934000015258789},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.590499997138977},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.5595999956130981},{"id":"https://openalex.org/C2777508537","wikidata":"https://www.wikidata.org/wiki/Q7936620","display_name":"Visual reasoning","level":2,"score":0.5590000152587891},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5523999929428101},{"id":"https://openalex.org/C148220186","wikidata":"https://www.wikidata.org/wiki/Q7111912","display_name":"Outcome (game theory)","level":2,"score":0.5254999995231628},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.45249998569488525},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.4372999966144562},{"id":"https://openalex.org/C124913957","wikidata":"https://www.wikidata.org/wiki/Q1232548","display_name":"Zoom","level":3,"score":0.41929998993873596},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.39590001106262207},{"id":"https://openalex.org/C178253425","wikidata":"https://www.wikidata.org/wiki/Q162668","display_name":"Visual perception","level":3,"score":0.38359999656677246},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.35030001401901245},{"id":"https://openalex.org/C56461940","wikidata":"https://www.wikidata.org/wiki/Q970687","display_name":"Eye tracking","level":2,"score":0.3237999975681305},{"id":"https://openalex.org/C111370547","wikidata":"https://www.wikidata.org/wiki/Q7451120","display_name":"Sensory cue","level":2,"score":0.32100000977516174},{"id":"https://openalex.org/C90329073","wikidata":"https://www.wikidata.org/wiki/Q914232","display_name":"Ask price","level":2,"score":0.3197999894618988},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.2957000136375427},{"id":"https://openalex.org/C158495155","wikidata":"https://www.wikidata.org/wiki/Q2369151","display_name":"Visual search","level":2,"score":0.2870999872684479},{"id":"https://openalex.org/C193611912","wikidata":"https://www.wikidata.org/wiki/Q4677596","display_name":"Active vision","level":2,"score":0.28380000591278076},{"id":"https://openalex.org/C2777055276","wikidata":"https://www.wikidata.org/wiki/Q7936580","display_name":"Visual approach","level":2,"score":0.27300000190734863},{"id":"https://openalex.org/C2986089797","wikidata":"https://www.wikidata.org/wiki/Q6501338","display_name":"Visual attention","level":3,"score":0.272599995136261},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.26600000262260437},{"id":"https://openalex.org/C103057564","wikidata":"https://www.wikidata.org/wiki/Q4751139","display_name":"Analytic reasoning","level":3,"score":0.26440000534057617},{"id":"https://openalex.org/C98184364","wikidata":"https://www.wikidata.org/wiki/Q1780131","display_name":"Argument (complex analysis)","level":2,"score":0.2635999917984009},{"id":"https://openalex.org/C2778251979","wikidata":"https://www.wikidata.org/wiki/Q7936617","display_name":"Visual processing","level":3,"score":0.25870001316070557},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2554999887943268},{"id":"https://openalex.org/C2780103172","wikidata":"https://www.wikidata.org/wiki/Q1309721","display_name":"Visual Objects","level":3,"score":0.25529998540878296},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.2549999952316284},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.25440001487731934},{"id":"https://openalex.org/C2780203653","wikidata":"https://www.wikidata.org/wiki/Q17009571","display_name":"Dual process theory (moral psychology)","level":3,"score":0.2524000108242035},{"id":"https://openalex.org/C552651612","wikidata":"https://www.wikidata.org/wiki/Q7328862","display_name":"Visual thinking","level":2,"score":0.2513999938964844}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.08719","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.08719","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.08719","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.08719","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"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":{"''Thinking":[0,88,135,192],"with":[1,42,89,96,136,191,193],"Images''":[2,90,137,194],"has":[3],"emerged":[4],"as":[5],"an":[6,98,122,163],"effective":[7],"paradigm":[8],"for":[9],"fine-grained":[10],"visual":[11,65,76,109],"reasoning:":[12],"by":[13],"explicitly":[14],"zooming":[15],"into":[16],"relevant":[17],"regions":[18],"and":[19,46,106,151],"reasoning":[20,85,157],"over":[21],"crops,":[22],"models":[23,184],"can":[24,67,91],"access":[25],"local":[26],"evidence":[27,145],"that":[28,101,175],"is":[29],"difficult":[30],"to":[31,72,104],"recover":[32],"from":[33,57,148],"a":[34,128,134],"single":[35],"global":[36],"image.":[37],"However,":[38],"this":[39,79],"benefit":[40],"comes":[41],"redundant":[43],"tool":[44],"invocations":[45],"longer":[47],"inference":[48,188],"traces.":[49],"Moreover,":[50],"when":[51],"such":[52],"behaviors":[53],"are":[54],"learned":[55],"mainly":[56],"outcome":[58],"reward,":[59],"the":[60,84,131,153,178],"resulting":[61],"intermediate":[62],"crops":[63],"or":[64,70,166],"cues":[66,110],"be":[68,92],"noisy":[69],"fail":[71],"faithfully":[73],"capture":[74],"task-relevant":[75],"evidence.":[77],"In":[78],"work,":[80],"we":[81],"ask":[82],"whether":[83],"benefits":[86],"of":[87,133],"internalized":[93],"through":[94],"Thinking":[95],"Imagination:":[97],"internal":[99],"process":[100],"decides":[102],"where":[103],"look":[105],"imagines":[107],"what":[108],"closer":[111],"inspection":[112],"would":[113],"reveal":[114],"without":[115],"actually":[116],"invoking":[117],"tools.":[118],"We":[119],"propose":[120],"Imagine-OPD,":[121],"on-policy":[123],"self-distillation":[124],"framework":[125],"in":[126],"which":[127],"teacher":[129,165],"plays":[130],"role":[132],"reasoner":[138],"during":[139],"training:":[140],"it":[141],"receives":[142],"privileged":[143],"zoomed":[144],"views":[146],"derived":[147],"annotated":[149],"regions,":[150],"supervises":[152],"model's":[154],"own":[155],"imagination":[156,168],"trajectories.":[158],"Imagine-OPD":[159,176],"does":[160],"not":[161],"require":[162],"external":[164],"high-quality":[167],"demonstrations.":[169],"Experiments":[170],"on":[171],"vision-centric":[172],"benchmarks":[173],"show":[174],"achieves":[177],"best":[179],"average":[180],"performance":[181],"among":[182],"compared":[183,190],"while":[185],"significantly":[186],"reducing":[187],"overhead":[189],"methods.":[195]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-10T00:00:00"}