How Chess Expertise Reshapes Event Segmentation

How do chess players perceive events in a chess game, as these events unfold in real time? The study builds upon the hierarchical bias hypothesis, stating that observers instinctively segment activities in alignment with a partonomic hierarchy. Eighteen participants (nine experts, nine novices) watched chess game videos and identified event boundaries at both fine and coarse grains. Data were analysed using discrete and continuous methods as well as an agreement index. Experts identified roughly twice as many boundaries as novices at both grains, showed descriptively but not significantly higher within-group agreement on boundary placement than novices (mean Agreement Index 0.607 vs. 0.506), and produced fine and coarse boundaries that aligned into a partonomic hierarchy more strongly than chance and more strongly than novices did. Verbal descriptions further showed that experts produced markedly more action-coded vocabulary than novices (9.8 vs. 2.1 terms per description). However, when descriptions containing such vocabulary were compared directly, experts and novices distributed it between basic-movement and higher-order strategic content in similar proportions. Expertise effects were uniform across grains, inconsistent with the specific prediction that differentiation would dominate at the fine grain and unitization at the coarse grain, though the present data cannot rule out joint contributions of both mechanisms to expert event representations. The results extend event segmentation research to a strategically cumulative domain and replicate the Zacks et al. (2001b) alignment effect with an additional expertise interaction.在棋局实时展开的过程中，国际象棋选手如何感知其中的各类事件？本研究基于层级偏向假说（hierarchical bias hypothesis）展开，该假说指出观察者会本能地依据部分论层级结构（partonomic hierarchy）对活动进行分段。18名被试（9名专业棋手、9名新手）观看棋局视频，并分别在细粒度与粗粒度层面识别事件边界。研究采用离散法、连续法以及一致性指数（agreement index）对数据展开分析。结果显示，在两个粒度层面上，专业棋手识别出的事件边界数量均约为新手的两倍；在边界定位的组内一致性上，专业棋手的表现虽数值更高，但未达到显著水平（平均一致性指数分别为0.607与0.506）；且专业棋手生成的细粒度与粗粒度边界，其匹配形成部分论层级结构的程度显著高于随机水平，同时也优于新手。进一步的言语描述分析表明，专业棋手生成的动作编码词汇（action-coded vocabulary）量显著高于新手（每份描述平均分别为9.8个与2.1个词汇）。不过，若直接对比包含此类词汇的描述内容，专业棋手与新手在基础走法与高阶战略内容间分配该类词汇的比例却较为相似。专业熟练度效应在两个粒度层面上均保持一致，这与“细粒度层面以分化为主、粗粒度层面以整合为主”的特定预测不符；不过本研究数据无法排除两种机制共同作用于专业棋手事件表征的可能性。本研究结果将事件分段（event segmentation）研究拓展至具有战略累积性的领域，并在纳入熟练度交互效应的前提下，复现了Zacks等人（2001b）提出的匹配效应。