{"id":"https://openalex.org/W7104365267","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.1335","title":"Dissecting Persona-Driven Reasoning in Language Models via Activation Patching","display_name":"Dissecting Persona-Driven Reasoning in Language Models via Activation Patching","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W7104365267","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.1335"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.1335","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1335","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1335.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":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-emnlp.1335.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Ansh Poonia","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ansh Poonia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Maeghal Jain","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Maeghal Jain","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":2,"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":0.50160608,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"24553","last_page":"24566"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T14074","display_name":"Persona Design and Applications","score":0.9437999725341797,"subfield":{"id":"https://openalex.org/subfields/1709","display_name":"Human-Computer Interaction"},"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/T14074","display_name":"Persona Design and Applications","score":0.9437999725341797,"subfield":{"id":"https://openalex.org/subfields/1709","display_name":"Human-Computer Interaction"},"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/T10028","display_name":"Topic Modeling","score":0.0031999999191612005,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.002899999963119626,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.41100001335144043},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.34299999475479126},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.25870001316070557},{"id":"https://openalex.org/keywords/automated-reasoning","display_name":"Automated reasoning","score":0.257999986410141},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.2549999952316284}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6118999719619751},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5331000089645386},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4668000042438507},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.41100001335144043},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.34299999475479126},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.25870001316070557},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.257999986410141},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.25589999556541443},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.2549999952316284},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.2526000142097473}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.1335","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1335","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1335.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":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.1335","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.1335","pdf_url":"https://aclanthology.org/2025.findings-emnlp.1335.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":"Findings of the Association for Computational Linguistics: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[{"display_name":"Reduced inequalities","score":0.4335351288318634,"id":"https://metadata.un.org/sdg/10"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7104365267.pdf","grobid_xml":"https://content.openalex.org/works/W7104365267.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"exhibit":[4],"remarkable":[5],"versatility":[6],"in":[7],"adopting":[8],"diverse":[9],"personas.In":[10],"this":[11],"study,":[12],"we":[13,29,93],"examine":[14],"how":[15,36],"assigning":[16],"a":[17,20,31],"persona":[18,72],"influences":[19],"model's":[21,91],"reasoning":[22],"on":[23],"an":[24],"objective":[25],"task.Using":[26],"activation":[27],"patching,":[28],"take":[30],"first":[32],"step":[33],"toward":[34],"understanding":[35],"key":[37],"components":[38],"of":[39,61],"the":[40,48,58,62,82,90],"model":[41],"encode":[42],"persona-specific":[43],"information.Our":[44],"findings":[45],"reveal":[46],"that":[47,98],"early":[49],"Multi-Layer":[50],"Perceptron":[51],"(MLP)":[52],"layers":[53,70,87],"attend":[54,100],"not":[55],"only":[56],"to":[57,88,101],"syntactic":[59],"structure":[60],"input":[63],"but":[64],"also":[65],"process":[66],"its":[67],"semantic":[68],"content.These":[69],"transform":[71],"tokens":[73],"into":[74],"richer":[75],"representations,":[76],"which":[77],"are":[78],"then":[79],"used":[80],"by":[81],"middle":[83],"Multi-Head":[84],"Attention":[85],"(MHA)":[86],"shape":[89],"output.Additionally,":[92],"identify":[94],"specific":[95],"attention":[96],"heads":[97],"disproportionately":[99],"racial":[102],"and":[103],"color-based":[104],"identities.":[105],"1":[106]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-11-08T00:00:00"}