Walter Kerri, Bex Peter
Northeastern University, Boston, MA, United States of America.
PLoS One. 2024 Dec 23;19(12):e0315998. doi: 10.1371/journal.pone.0315998. eCollection 2024.
Face discrimination ability has been widely studied in psychology, however a self-administered, adaptive method has not yet been developed. In this series of studies, we utilize Foraging Interactive D-prime (FInD) in conjunction with the Basel Face Model to quantify thresholds of face discrimination ability both in-lab and remotely. In Experiment 1, we measured sensitivity to changes for all 199 structural Principal Components of the Basel Face Model and found observers were most sensitive to the first 10 components, so we focused on these for the remaining studies. In Experiment 2, we remotely investigated how thresholds varied when one component changed, compared to when two components changed in combination. Thresholds measured remotely were not significantly different from those measured in-lab (t(14) = 0.23, p = .821), and thresholds were significantly lower for components in combination than alone (t(7) = 2.90, p = .023), consistent with probability summation and Euclidean distance between faces, but not superadditivity. In Experiment 3, we replicated Experiment 2 with slight rotation to the faces to prevent pointwise comparisons. Thresholds were higher with rotation (t(30) = 4.32, p < .001) and for single than combined components, but did not reach significance (t(7) = 2.24, p = .061). Charts were measured in approximately 25.90 ± 8.10 seconds.
面部辨别能力在心理学中已得到广泛研究，然而，一种自我管理的自适应方法尚未被开发出来。在这一系列研究中，我们将觅食交互D-prime（FInD）与巴塞尔面部模型结合使用，以量化实验室环境和远程环境下的面部辨别能力阈值。在实验1中，我们测量了巴塞尔面部模型所有199个结构主成分变化的敏感度，发现观察者对前10个成分最为敏感，因此在后续研究中我们专注于这些成分。在实验2中，我们远程研究了与两个成分组合变化时相比，单个成分变化时阈值如何变化。远程测量的阈值与实验室测量的阈值没有显著差异（t(14) = 0.23，p = 0.821），并且组合成分的阈值显著低于单个成分的阈值（t(7) = 2.90，p = 0.023），这与概率求和以及面部之间的欧几里得距离一致，但不符合超可加性。在实验3中，我们对人脸进行了轻微旋转以防止逐点比较，重复了实验2。旋转后人脸的阈值更高（t(30) = 4.32，p < 0.001），单个成分的阈值高于组合成分，但未达到显著水平（t(7) = 2.24，p = 0.061）。图表测量大约用时25.90 ± 8.10秒。
