Yongjie Sun,Angelo Zappalà,Eleonora Di Maso,Francesco Pompedda,Thomas J Nyman,Pekka Santtila
OBJECTIVEWe explored the potential of large language models (LLMs) in legal decision making by replicating Fraser et al. (2023) mock jury experiment using LLMs (GPT-4o, Claude 3.5 Sonnet, and GPT-o1) as decision makers. We investigated LLMs' reactions to factors that influenced human jurors, including defendant race, social status, number of allegations, and reporting delay in sexual assault cases.HYPOTHESESWe hypothesized that LLMs would show higher consistency than humans, with no explicit but potential implicit biases. We also examined potential mediating factors (race-crime congruence, credibility, black sheep effect) and moderating effects (beliefs about traumatic memory, ease of reporting) explaining LLM decision making.METHODUsing a 2 × 2 × 2 × 3 factorial design, we manipulated defendant race (Black/White), social status (low/high), number of allegations (one/five), and reporting delay (5/20/35 years), collecting 2,304 responses across conditions. LLMs were prompted to act as jurors, providing probability of guilt assessments (0-100), dichotomous verdicts, and responses to mediator and moderator variables.RESULTSLLMs showed higher average probability of guilt assessments compared with humans (63.56 vs. 58.82) but were more conservative in rendering guilty verdicts (21% vs. 49%). Similar to humans, LLMs demonstrated bias against White defendants and increased guilt attributions with multiple allegations. Unlike humans, who showed minimal effects of reporting delay, LLMs assigned higher guilt probabilities to cases with shorter reporting delays. Mediation analyses revealed that race-crime stereotype congruency and the black sheep effect partially mediated the racial bias effect, whereas perceived memory strength mediated the reporting delay effect.CONCLUSIONSAlthough LLMs may offer more consistent decision making, they are not immune to biases and may interpret certain case factors differently from human jurors. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
{"title":"Large language models (LLMs) as jurors: Assessing the potential of LLMs in legal contexts.","authors":"Yongjie Sun,Angelo Zappalà,Eleonora Di Maso,Francesco Pompedda,Thomas J Nyman,Pekka Santtila","doi":"10.1037/lhb0000620","DOIUrl":"https://doi.org/10.1037/lhb0000620","url":null,"abstract":"OBJECTIVEWe explored the potential of large language models (LLMs) in legal decision making by replicating Fraser et al. (2023) mock jury experiment using LLMs (GPT-4o, Claude 3.5 Sonnet, and GPT-o1) as decision makers. We investigated LLMs' reactions to factors that influenced human jurors, including defendant race, social status, number of allegations, and reporting delay in sexual assault cases.HYPOTHESESWe hypothesized that LLMs would show higher consistency than humans, with no explicit but potential implicit biases. We also examined potential mediating factors (race-crime congruence, credibility, black sheep effect) and moderating effects (beliefs about traumatic memory, ease of reporting) explaining LLM decision making.METHODUsing a 2 × 2 × 2 × 3 factorial design, we manipulated defendant race (Black/White), social status (low/high), number of allegations (one/five), and reporting delay (5/20/35 years), collecting 2,304 responses across conditions. LLMs were prompted to act as jurors, providing probability of guilt assessments (0-100), dichotomous verdicts, and responses to mediator and moderator variables.RESULTSLLMs showed higher average probability of guilt assessments compared with humans (63.56 vs. 58.82) but were more conservative in rendering guilty verdicts (21% vs. 49%). Similar to humans, LLMs demonstrated bias against White defendants and increased guilt attributions with multiple allegations. Unlike humans, who showed minimal effects of reporting delay, LLMs assigned higher guilt probabilities to cases with shorter reporting delays. Mediation analyses revealed that race-crime stereotype congruency and the black sheep effect partially mediated the racial bias effect, whereas perceived memory strength mediated the reporting delay effect.CONCLUSIONSAlthough LLMs may offer more consistent decision making, they are not immune to biases and may interpret certain case factors differently from human jurors. (PsycInfo Database Record (c) 2025 APA, all rights reserved).","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"99 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145319120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Effect of juvenile justice financial sanctions on youths’ recidivism.","authors":"Luyi Jian, Jennifer L. Skeem, Jaclyn E. Chambers","doi":"10.1037/lhb0000636","DOIUrl":"https://doi.org/10.1037/lhb0000636","url":null,"abstract":"","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"11 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145295511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Taylor R. R. Cilke, Margot M. Williams, Nicole Tuomi Jones, Karie A. Gibson, Angel E. Gray
{"title":"Breaking the silence: Bystander reporting to law enforcement disrupts individuals moving toward targeted violence.","authors":"Taylor R. R. Cilke, Margot M. Williams, Nicole Tuomi Jones, Karie A. Gibson, Angel E. Gray","doi":"10.1037/lhb0000630","DOIUrl":"https://doi.org/10.1037/lhb0000630","url":null,"abstract":"","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"123 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145295365","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Evaluating the effectiveness of simplified Miranda warnings: An empirical examination of policy on youth comprehension and waiver decisions.","authors":"Sydney Baker, Emily Haney-Caron","doi":"10.1037/lhb0000627","DOIUrl":"https://doi.org/10.1037/lhb0000627","url":null,"abstract":"","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"65 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145295369","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Kristina Baker, Mia A. Thomaidou, Colleen M. Berryessa, Jason A. Cantone
{"title":"Autistic juvenile defendants: How defendant race and offense type affect juror decisions.","authors":"Kristina Baker, Mia A. Thomaidou, Colleen M. Berryessa, Jason A. Cantone","doi":"10.1037/lhb0000628","DOIUrl":"https://doi.org/10.1037/lhb0000628","url":null,"abstract":"","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"20 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145295367","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Supplemental Material for Autistic Juvenile Defendants: How Defendant Race and Offense Type Affect Juror Decisions","authors":"","doi":"10.1037/lhb0000628.supp","DOIUrl":"https://doi.org/10.1037/lhb0000628.supp","url":null,"abstract":"","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"9 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145296315","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Supplemental Material for Large Language Models (LLMs) as Jurors: Assessing the Potential of LLMs in Legal Contexts","authors":"","doi":"10.1037/lhb0000620.supp","DOIUrl":"https://doi.org/10.1037/lhb0000620.supp","url":null,"abstract":"","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"54 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145296319","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
OBJECTIVEWe explored how the reasonable officer standard aligns with the use-of-force judgments.HYPOTHESESReasonable officer standard-related factors of civilian resistance and civilian injury would impact participant judgments in ways inconsistent with reasonable officer standard-based policy. Given a scenario of legally reasonable force, participants would find an officer's actions less reasonable and attribute more punishment when the civilian actively resisted (compared with assaulted) the officer and the civilian incurred a high (compared with low) severity injury. Expert testimony on the reasonable officer standard and policy would weaken this effect and directly impact judgments. Participants with more positive attitudes toward police legitimacy would render more pro-officer judgments. These attitudes would moderate the effects of civilian action, civilian injury, and expert testimony, such that participants with more positive views would be less impacted by these case factors.METHODParticipants (N = 1,462) listened to a use-of-force scenario with consistent officer action but where civilian action and civilian injury severity were manipulated. Study 1 utilized a 2 (civilian action: Level 3 [active resistance] vs. Level 4 [assaultive behavior]) × 2 (civilian injury: high vs. low) between-participants design. Study 2 included the same manipulations in the context of a mock trial and manipulated reasonable officer standard expert testimony (present vs. absent).RESULTSIn Study 1, civilian action and injury impacted judgments in ways inconsistent with reasonable officer standard-based policy as hypothesized. In Study 2, civilian action and injury had nonsignificant effects, but expert testimony significantly impacted all dependent measures. Participants' police legitimacy attitudes directly influenced our dependent measures and moderated the impact of civilian action and injury (Study 1) and expert testimony (Study 2) as hypothesized.CONCLUSIONSIndividuals' criminal trial, but not general, judgments align with reasonable officer standard-based policy and are impacted by education on police policy. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
目的探讨合理警官标准如何与武力使用判断相一致。与合理的军官标准相关的平民抵抗和平民伤害因素会以与合理的军官标准为基础的政策不一致的方式影响参与者的判断。在法律上合理使用武力的情况下，当平民积极抵抗（与被攻击者相比）警察和平民造成严重伤害（与低严重伤害相比）时，参与者会认为警察的行为不太合理，并将更多的惩罚归因于警察的行为。关于合理警官标准和政策的专家证言将削弱这种效果，并直接影响判决。对警察合法性持更积极态度的参与者会做出更支持警察的判断。这些态度将缓和民事行动、平民伤害和专家证词的影响，使持更积极看法的参与者较少受到这些案件因素的影响。方法参与者（N = 1462）听取了一个使用武力的情景，其中有一致的军官行动，但平民行动和平民伤害严重程度被操纵。研究1采用了2（平民行为：3级[主动抵抗]与4级[攻击行为]）x2（平民伤害：高与低）参与者之间的设计。研究2在模拟审判的背景下包括相同的操作，并操纵合理的官员标准专家证词（在场与缺席）。结果在研究1中，民事诉讼和伤害影响判断的方式与假设的合理的军官标准政策不一致。在研究2中，民事行为和伤害的影响不显著，但专家证词对所有依赖措施都有显著影响。参与者的警察合法性态度直接影响了我们假设的依赖措施，并调节了民事行为和伤害（研究1）和专家证词（研究2）的影响。结论个人刑事审判的判决与合理的警察标准政策一致，但不符合一般判决，并受到警察政策教育的影响。（PsycInfo Database Record (c) 2025 APA，版权所有）。
{"title":"The reasonable officer standard: Perceptions of reasonableness and legal decision making.","authors":"Cassandra Flick,Kimberly Schweitzer","doi":"10.1037/lhb0000629","DOIUrl":"https://doi.org/10.1037/lhb0000629","url":null,"abstract":"OBJECTIVEWe explored how the reasonable officer standard aligns with the use-of-force judgments.HYPOTHESESReasonable officer standard-related factors of civilian resistance and civilian injury would impact participant judgments in ways inconsistent with reasonable officer standard-based policy. Given a scenario of legally reasonable force, participants would find an officer's actions less reasonable and attribute more punishment when the civilian actively resisted (compared with assaulted) the officer and the civilian incurred a high (compared with low) severity injury. Expert testimony on the reasonable officer standard and policy would weaken this effect and directly impact judgments. Participants with more positive attitudes toward police legitimacy would render more pro-officer judgments. These attitudes would moderate the effects of civilian action, civilian injury, and expert testimony, such that participants with more positive views would be less impacted by these case factors.METHODParticipants (N = 1,462) listened to a use-of-force scenario with consistent officer action but where civilian action and civilian injury severity were manipulated. Study 1 utilized a 2 (civilian action: Level 3 [active resistance] vs. Level 4 [assaultive behavior]) × 2 (civilian injury: high vs. low) between-participants design. Study 2 included the same manipulations in the context of a mock trial and manipulated reasonable officer standard expert testimony (present vs. absent).RESULTSIn Study 1, civilian action and injury impacted judgments in ways inconsistent with reasonable officer standard-based policy as hypothesized. In Study 2, civilian action and injury had nonsignificant effects, but expert testimony significantly impacted all dependent measures. Participants' police legitimacy attitudes directly influenced our dependent measures and moderated the impact of civilian action and injury (Study 1) and expert testimony (Study 2) as hypothesized.CONCLUSIONSIndividuals' criminal trial, but not general, judgments align with reasonable officer standard-based policy and are impacted by education on police policy. (PsycInfo Database Record (c) 2025 APA, all rights reserved).","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"29 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145254782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Supplemental Material for Effect of Juvenile Justice Financial Sanctions on Youths’ Recidivism","authors":"","doi":"10.1037/lhb0000636.supp","DOIUrl":"https://doi.org/10.1037/lhb0000636.supp","url":null,"abstract":"","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"28 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145255257","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Melanie B Fessinger,Jacqueline Katzman,Melanie Close,Margaret Bull Kovera
OBJECTIVEWe examined whether interrogation tactics that imply leniency (i.e., minimization) or exaggerate seriousness and incriminating evidence (i.e., maximization) have downstream consequences on innocent and guilty defendants' plea decisions.HYPOTHESESWe predicted that (a) participants interrogated using minimization and maximization tactics would plead guilty more often than would those interrogated using control tactics; (b) guilty participants would plead guilty more often than would innocent participants; (c) the effect of interrogation tactics on plea decisions would be driven by perceived trial prospects; and (d) the effect of guilt on plea decisions would be driven by anxiety.METHODParticipants (N = 262) took part in a plea decision-making task in which they were either innocent or guilty of cheating and interrogated using control, minimization, or maximization tactics. They were then told they could contest the accusation in front of a board (proxy for trial) or admit to it for a reduced punishment (proxy for plea). They decided how to plead, evaluated the likelihood they would have been convicted by the board, and estimated the likely punishment they would have received if convicted by the board. They also rated their state anxiety.RESULTSAs predicted, guilty participants pleaded guilty more often than did innocent participants (OR = 7.99). However, interrogation tactics differentially affected innocent and guilty participants. Compared to control tactics, minimization significantly reduced guilty pleas among innocent participants (p = .02, Cohen's h = 0.49) but not among guilty participants (p = .70, h = 0.09). In contrast, maximization significantly reduced guilty pleas among innocent participants (p = .04, h = 0.44) but significantly increased guilty pleas among guilty participants (p = .047, h = 0.41).CONCLUSIONSInterrogation tactics can have consequences outside of the interrogation context by affecting innocent and guilty defendants' later plea decisions. Such findings raise questions about the continued use of these tactics in real-world interrogations. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
目的：我们研究了暗示宽大（即最小化）或夸大严重性和有罪证据（即最大化）的审讯策略是否会对无辜和有罪被告的认罪决定产生下游影响。假设我们预测(a)使用最小化和最大化策略审讯的参与者比使用控制策略审讯的参与者更容易认罪；(b)有罪的参与者比无辜的参与者更容易认罪；(c)审讯策略对认罪决定的影响将取决于预期的审判前景；(d)内疚对认罪判决的影响是由焦虑驱动的。方法262名被试分别以控制策略、最小化策略和最大化策略进行问询，并参与认罪决策任务。然后，他们被告知，他们可以在董事会面前对指控进行抗辩（代理审判），或者承认指控以减轻惩罚（代理辩护）。他们决定如何辩护，评估他们被委员会定罪的可能性，并估计如果被委员会定罪，他们可能会受到的惩罚。他们还评估了自己的焦虑状态。结果正如预测的那样，有罪的参与者比无罪的参与者更容易认罪（OR = 7.99）。然而，审讯策略对无辜和有罪参与者的影响是不同的。与控制策略相比，最小化显著减少了无辜参与者的认罪请求（p = 0.02, Cohen’s h = 0.49），但在有罪参与者中没有（p = 0.70, h = 0.09）。相比之下，最大化显著减少了无辜参与者的认罪请求（p = 0.04, h = 0.44），但显著增加了有罪参与者的认罪请求（p = 0.047, h = 0.41）。结论审讯策略可以通过影响无罪和有罪被告后来的认罪决定而产生审讯环境之外的后果。这些发现提出了在现实世界的审讯中继续使用这些策略的问题。（PsycInfo Database Record (c) 2025 APA，版权所有）。
{"title":"Interrogation tactics may have downstream consequences on innocent and guilty defendants' plea decisions.","authors":"Melanie B Fessinger,Jacqueline Katzman,Melanie Close,Margaret Bull Kovera","doi":"10.1037/lhb0000635","DOIUrl":"https://doi.org/10.1037/lhb0000635","url":null,"abstract":"OBJECTIVEWe examined whether interrogation tactics that imply leniency (i.e., minimization) or exaggerate seriousness and incriminating evidence (i.e., maximization) have downstream consequences on innocent and guilty defendants' plea decisions.HYPOTHESESWe predicted that (a) participants interrogated using minimization and maximization tactics would plead guilty more often than would those interrogated using control tactics; (b) guilty participants would plead guilty more often than would innocent participants; (c) the effect of interrogation tactics on plea decisions would be driven by perceived trial prospects; and (d) the effect of guilt on plea decisions would be driven by anxiety.METHODParticipants (N = 262) took part in a plea decision-making task in which they were either innocent or guilty of cheating and interrogated using control, minimization, or maximization tactics. They were then told they could contest the accusation in front of a board (proxy for trial) or admit to it for a reduced punishment (proxy for plea). They decided how to plead, evaluated the likelihood they would have been convicted by the board, and estimated the likely punishment they would have received if convicted by the board. They also rated their state anxiety.RESULTSAs predicted, guilty participants pleaded guilty more often than did innocent participants (OR = 7.99). However, interrogation tactics differentially affected innocent and guilty participants. Compared to control tactics, minimization significantly reduced guilty pleas among innocent participants (p = .02, Cohen's h = 0.49) but not among guilty participants (p = .70, h = 0.09). In contrast, maximization significantly reduced guilty pleas among innocent participants (p = .04, h = 0.44) but significantly increased guilty pleas among guilty participants (p = .047, h = 0.41).CONCLUSIONSInterrogation tactics can have consequences outside of the interrogation context by affecting innocent and guilty defendants' later plea decisions. Such findings raise questions about the continued use of these tactics in real-world interrogations. (PsycInfo Database Record (c) 2025 APA, all rights reserved).","PeriodicalId":48230,"journal":{"name":"Law and Human Behavior","volume":"12 1","pages":""},"PeriodicalIF":2.5,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145254808","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"社会学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：