Behavioral and Rater Prediction Data Associated with "Predicting Cooperation with Minimal Information: Convergence Between Evolved and Artificial Mindreading Heuristics"

This dataset contains behavioral and rater data collected across two studies, along with performance benchmarks from artificial prediction heuristics applied to the same behavioral data. The first component comprises behavioral data from a repeated Split-or-Take-All Prisoner's Dilemma experiment conducted at a university experimental economics laboratory. Anonymously paired participants chose to either cooperate ("Split") or defect ("Take All") across two rounds, with payoffs determined by the combination of both players' choices: mutual Split yielded 5 USD to each player, mutual Take All yielded 0 USD to each, and mismatched choices awarded 10 USD to the defector and 0 USD to the cooperator. The dataset includes each participant's Round 1 and Round 2 choices, their partner's choices, and round-by-round payoff outcomes. The second component comprises data from an online rater prediction study conducted via Prolific, in which raters predicted participants' Round 2 choices across conditions varying access to behavioral information. This component includes rater predictions, prediction accuracy measures, and demographic variables. The third component comprises prediction accuracy benchmarks generated by a set of artificial heuristics — including base rate, strategic reciprocator, and optimizing heuristic strategies — applied to the same participant behavioral data. These benchmarks allow direct comparison of human rater prediction performance against rule-based artificial prediction strategies. These three components above are all packaged into a single tab of the .csv file in this record. A second tab serves as the data dictionary, defining variable names in the first tab. The fourth component is the survey instrument, including participant instructions, consent form, attention checks, and all survey questions administered to raters, which is included as a separate .pdf file in this record. The instrument retains Qualtrics display logic and branching notation to fully document the conditional structure of the survey flow across experimental conditions. These data support research on cooperative behavior, reciprocity, social prediction, and strategic decision-making in repeated social dilemmas. Researchers interested in prisoner's dilemma dynamics, conditional cooperation, impression formation, behavioral prediction, or human-AI comparison may find this dataset useful. Associated publications are listed under Related Identifiers. Thin-slice video clips of participants recorded immediately prior to their Round 2 decision were shown to raters in the prediction experiment and are archived separately (see Related Identifiers). A restricted-access record linking visual stimuli to participant behavioral data is available upon request. For questions about the data, contact the corresponding author via the associated publication.本数据集包含两项研究中收集的行为数据与评分者数据，以及应用于同一套行为数据的人工预测启发式算法性能基准。 第一部分为某大学实验经济学实验室开展的重复“分裂-全拿”囚徒困境（Split-or-Take-All Prisoner's Dilemma）实验的行为数据。匿名配对的参与者在两轮博弈中均需选择合作（“分裂”）或背叛（“全拿”），收益由两名参与者的选择组合决定：双方均选择分裂时，每名参与者可获得5美元；双方均选择全拿时，每名参与者收益为0；当双方选择不匹配时，背叛者可获得10美元，合作者收益为0。本数据集包含每名参与者第一轮与第二轮的选择、其搭档的选择，以及每一轮的收益结果。 第二部分为通过Prolific平台开展的在线评分者预测研究数据。在该研究中，评分者需根据不同行为信息获取条件，预测参与者的第二轮选择。该部分数据包含评分者的预测结果、预测准确率指标，以及人口统计学变量。 第三部分为应用于同一套参与者行为数据的多个人工启发式算法生成的预测准确率基准，其中包括基础率策略、战略互惠者策略与优化启发式策略。上述基准可用于直接对比人类评分者的预测表现与基于规则的人工预测策略的表现。 上述三部分数据均打包于本记录中.csv文件的第一个工作表内；第二个工作表为数据字典，用于定义第一个工作表中的变量名称。 第四部分为调查问卷工具，包含面向实验参与者的说明、知情同意书、注意力测试题，以及面向评分者的全部调查问卷问题。该工具以单独的.pdf文件形式包含在本记录中，保留了Qualtrics的显示逻辑与分支标注，以完整记录不同实验条件下调查流程的条件结构。 本数据集可为重复社会困境中的合作行为、互惠行为、社会预测与战略决策相关研究提供支持。对囚徒困境动态、条件性合作、印象形成、行为预测或人机对比研究感兴趣的研究者可使用本数据集。相关已发表文献已列于“相关标识符”项下。 预测实验中，评分者会观看在参与者做出第二轮决策前即刻录制的参与者薄片视频剪辑（thin-slice video clips），此类视频已单独归档（详见相关标识符）。如需获取将视觉刺激与参与者行为数据相关联的受限访问数据集，可提交申请。 如有关于本数据集的疑问，请通过关联发表文献联系通讯作者。