Daniel Thomas A, Wright Anthony A, Katz Jeffrey S
Auburn University, Auburn, AL, USA.
College of William & Mary, P.O. Box 8795, Williamsburg, VA, 23187, USA.
Learn Behav. 2016 Dec;44(4):320-328. doi: 10.3758/s13420-016-0219-0.
Previous work in discrimination learning has shown that nonmatching (oddity) tasks are learned faster and more accurately than comparable matching tasks. This learning advantage has been coined the oddity preference effect (Wright & Delius in Journal of Experimental Psychology: Animal Behavior Processes, 31, 425-432. doi: 10.1037/0097-7403.31.4.425 , 2005). Pigeons trained in a nonmatching task, following training in a same/different (S/D) task, learned the abstract concept of difference (Daniel et al., in Animal Cognition, 18(4), 831-837, 2015), but they did not show the expected faster acquisition or high levels of transfer from the oddity preference effect. In the present study, experimentally naïve pigeons were trained in an identical nonmatching task to examine whether they would show the oddity preference effect on abstract-concept learning. These experimentally naïve pigeons did show an oddity preference effect; their transfer to novel configurations was above chance with the initial (smallest) set size (3-item set) and was substantially more accurate than novel transfer in similar match-to-sample (MTS) or S/D tasks (Bodily et al., in Journal of Experimental Psychology: Animal Behavior Processes, 34, 178-184. doi: 10.1037/0097-7403.34.1.178 , 2008; Katz & Wright in Journal of Experimental Psychology: Animal Behavior Processes, 32, 80-86. doi: 10.1037/0097-7403.32.1.80 , 2006). As the number exemplars in the training set increased, transfer to novel configurations increased and reached equivalence to trained-stimulus performance with a 24-item set. Despite this transfer being equal to baseline performance with a 24-item set, subsequent transfers following training with larger set sizes declined before eventually rising again to baseline performance. This unusual set-size function (with inflection points at the 24- and 96-set sizes) suggests that these pigeons may have combined item-specific and relational learning strategies with differing emphasis as they acquired the abstract concept.
先前在辨别学习方面的研究表明，非匹配（奇特性）任务比类似的匹配任务学得更快、更准确。这种学习优势被称为奇特性偏好效应（赖特和德利乌斯，《实验心理学杂志：动物行为过程》，第31卷，第425 - 432页。doi: 10.1037/0097 - 7403.31.4.425，2005年）。在相同/不同（S/D）任务训练后接受非匹配任务训练的鸽子学会了差异的抽象概念（丹尼尔等人，《动物认知》，第18卷第4期，第831 - 837页，2015年），但它们并未表现出奇特性偏好效应所预期的更快习得速度或高水平的迁移。在本研究中，对未经实验训练的鸽子进行了相同的非匹配任务训练，以检验它们在抽象概念学习上是否会表现出奇特性偏好效应。这些未经实验训练的鸽子确实表现出了奇特性偏好效应；它们向新构型的迁移在初始（最小）集合大小（3项集合）时高于随机水平，并且比类似的样本匹配（MTS）或S/D任务中的新迁移准确得多（博迪利等人，《实验心理学杂志：动物行为过程》，第34卷，第178 - 184页。doi: 10.1037/0097 - 7403.34.1.178，2008年；卡茨和赖特，《实验心理学杂志：动物行为过程》，第32卷，第80 - 86页。doi: 10.1037/0097 - 7403.32.1.80，2006年）。随着训练集中示例数量的增加，向新构型的迁移增加，并且在24项集合时达到与训练刺激表现相当的水平。尽管这种迁移在24项集合时与基线表现相当，但在使用更大集合大小进行训练后的后续迁移在最终再次上升到基线表现之前有所下降。这种不寻常的集合大小函数（在24项和96项集合大小时有拐点）表明，这些鸽子在获取抽象概念时可能以不同的侧重点结合了特定项目和关系学习策略。
