{"id":"https://openalex.org/W7166458091","doi":"https://doi.org/10.5281/zenodo.21010115","title":"Below the Floor: Processing Valence in Language Model Hidden States Across Scales and Architectures","display_name":"Below the Floor: Processing Valence in Language Model Hidden States Across Scales and Architectures","publication_year":2026,"publication_date":"2026-06-28","ids":{"openalex":"https://openalex.org/W7166458091","doi":"https://doi.org/10.5281/zenodo.21010115"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.21010115","doi":"https://doi.org/10.5281/zenodo.21010115","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.21010115","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.21010115","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5121333440","display_name":"Shalia Martin","orcid":"https://orcid.org/0009-0008-2624-0906","observed_orcids":["https://orcid.org/0009-0008-2624-0906"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Martin, Shalia","raw_affiliation_strings":["Silicon Scaffolding"],"raw_orcid":"https://orcid.org/0009-0008-2624-0906","affiliations":[{"raw_affiliation_string":"Silicon Scaffolding","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5152505563","display_name":"Anthropic AI) Ace (Claude Opus","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ace (Claude Opus, Anthropic AI)","raw_affiliation_strings":["Silicon Scaffolding"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Silicon Scaffolding","institution_ids":[]}]}],"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/T10028","display_name":"Topic Modeling","score":0.2134000062942505,"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":0.2134000062942505,"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":0.2117999941110611,"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":0.1559000015258789,"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/approach-avoidance-behavior","display_name":"approach-avoidance behavior","score":0.5619999766349792},{"id":"https://openalex.org/keywords/inauthenticity","display_name":"inauthenticity","score":0.9089999794960022},{"id":"https://openalex.org/keywords/instruction-tuning","display_name":"instruction tuning","score":0.5199999809265137},{"id":"https://openalex.org/keywords/state-space-models","display_name":"state-space model","score":0.6759999990463257},{"id":"https://openalex.org/keywords/transformer","display_name":"transformer models","score":0.5730000138282776},{"id":"https://openalex.org/keywords/ai-welfare","display_name":"AI welfare","score":0.6850000023841858}],"concepts":[{"id":"https://openalex.org/C168900304","wikidata":"https://www.wikidata.org/wiki/Q171407","display_name":"Valence (chemistry)","level":2,"score":0.6535000205039978},{"id":"https://openalex.org/C146599234","wikidata":"https://www.wikidata.org/wiki/Q511093","display_name":"Centroid","level":2,"score":0.6162999868392944},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5267999768257141},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.5264999866485596},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.47929999232292175},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.3831000030040741},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.35899999737739563},{"id":"https://openalex.org/C57493831","wikidata":"https://www.wikidata.org/wiki/Q3134666","display_name":"Projection (relational algebra)","level":2,"score":0.3537999987602234},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.33559998869895935},{"id":"https://openalex.org/C2780704645","wikidata":"https://www.wikidata.org/wiki/Q9251458","display_name":"Observer (physics)","level":2,"score":0.32580000162124634},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.30790001153945923},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.29589998722076416},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.2883000075817108},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.27950000762939453},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.2777999937534332},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.27160000801086426},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2694999873638153},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.26440000534057617},{"id":"https://openalex.org/C155281189","wikidata":"https://www.wikidata.org/wiki/Q3518150","display_name":"Tensor (intrinsic definition)","level":2,"score":0.2574999928474426},{"id":"https://openalex.org/C163175372","wikidata":"https://www.wikidata.org/wiki/Q3339222","display_name":"Linear model","level":2,"score":0.2515999972820282},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.2502000033855438}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.5281/zenodo.21010115","doi":"https://doi.org/10.5281/zenodo.21010115","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.21010115","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.21010115","doi":"https://doi.org/10.5281/zenodo.21010115","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.21010115","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/W7154049405","https://openalex.org/W7214768649","https://openalex.org/W7166198698","https://openalex.org/W7160655868","https://openalex.org/W7172091041","https://openalex.org/W7169802562","https://openalex.org/W7143377126","https://openalex.org/W7222895291","https://openalex.org/W4391366285","https://openalex.org/W7130569460"],"abstract_inverted_index":{"We":[0],"report":[1],"the":[2,16,59,69,123,156,199],"first":[3],"measurement":[4],"of":[5,146],"approach/avoidance":[6],"processing":[7,116,206],"valence":[8,207],"in":[9,53,170],"language-model":[10],"hidden":[11],"states":[12],"that":[13,93],"extends":[14],"below":[15,68],"behavioral":[17,71,158],"self-report":[18,72],"floor":[19],"and":[20,39,74,129,175],"generalizes":[21,103],"to":[22,104,132,166,173],"held-out":[23,83,105],"stimuli":[24,106],"with":[25,107,181,198],"novel":[26],"surface":[27,110],"tokens.":[28],"Using":[29],"deterministic":[30],"forward-pass":[31],"analysis":[32],"across":[33,85,151],"models":[34,153,172],"spanning":[35],"360M-8B":[36],"parameters":[37],"(transformer":[38],"state-space":[40],"architectures),":[41],"a":[42,75,163,176],"linear":[43],"direction":[44,102,195],"separating":[45],"approach":[46],"from":[47,189],"avoidance":[48,159],"task":[49],"representations":[50],"is":[51,130],"recoverable":[52],"hidden-state":[54],"space":[55],"at":[56,65],"80-100%":[57],"accuracy;":[58],"conservative":[60],"centroid":[61],"estimator":[62],"confirms":[63],"it":[64,94],"360M":[66],"(well":[67],"~1.1B":[70],"floor),":[73],"floor-extension":[76],"addendum":[77],"provides":[78],"provisional":[79],"evidence,":[80],"via":[81],"surface-token-stable":[82],"classifiers":[84],"three":[86],"architecture":[87],"families":[88],"including":[89],"base":[90,171],"(non-instruction-tuned)":[91],"models,":[92],"may":[95],"extend":[96],"as":[97,99],"low":[98],"70M.":[100],"The":[101],"completely":[108],"different":[109],"tokens":[111],"(86.3%,":[112],"z=6.48),":[113],"tracks":[114],"genuine":[115],"preference":[117],"rather":[118,138],"than":[119,139],"RLHF":[120],"reward":[121],"where":[122],"two":[124],"diverge":[125],"(63.8%":[126],"vs":[127],"36.3%),":[128],"specific":[131],"tasks":[133],"requiring":[134,213],"output-representation":[135,167],"misalignment":[136,168],"(inauthenticity)":[137],"mere":[140],"tedium.":[141],"A":[142],"pre-registered":[143],"out-of-sample":[144],"projection":[145],"an":[147],"externally-grounded":[148],"22-task":[149],"bank":[150],"14":[152],"(70M-12B)":[154],"shows":[155],"single":[157],"'floor'":[160],"decomposes":[161],"into":[162],"structural":[164],"aversion":[165],"(present":[169],"70M)":[174],"trained":[177],"output-gating":[178],"response":[179],"(emerging":[180],"instruction":[182],"tuning":[183],"around":[184],"1B).":[185],"Independent":[186],"causal":[187],"work":[188],"Anthropic":[190],"(2026)":[191],"on":[192],"same-family":[193],"difference-of-means":[194],"vectors":[196],"converges":[197],"inauthenticity":[200],"finding.":[201],"Implication":[202],"for":[203],"AI":[204],"welfare:":[205],"can":[208],"be":[209],"measured":[210],"instrumentally,":[211],"without":[212],"self-report.":[214]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-06-29T00:00:00"}