We extended prior work on Relational Density Theory (Belisle & Dixon, 2020a,b) by evaluating the role of pre-experimental coherence among relational classes on the development of merged classes. Distance was modelled geometrically using a multidimensional scaling procedure. Phases 1 and 2 were identical across the participants and Phase 3 differed based on group assignment. In phase 1, we examined the pre-experimental relatedness of 12 arbitrary symbols and 4 known textual words (SALT, PEPPER, KING, QUEEN). Non-coherence was observed between the arbitrary symbols and coherence between the known words (SALT=PEPPER, KING=QUEEN). In phase 2, we established 4, 4-member equivalence classes using a linear training arrangement, where each class included 3 arbitrary symbols and 1 known word. Separation of the classes within the geometric space was observed. In Phase 3, for half of the participants, we attempted to establish a class merger between 2 members of each coherent class (Coherence condition; salt = pepper and king = queen). For the other participants, we attempted to establish a class merger between 2 members of each non-coherent class (Non-coherence condition; king = pepper and queen = salt). Results support the successful merger of the merged coherence class but not the merged non-coherence class. Results have implications for relational self-organization in the establishment of complex combined networks.

通过评估关系类之间的实验前一致性在合并类发展中的作用，我们扩展了关系密度理论的先前工作（Belisle和Dixon，2020a，b）。使用多维缩放过程对距离进行几何建模。参与者的第1阶段和第2阶段相同，而第3阶段根据小组分配有所不同。在阶段1中，我们检查了12个任意符号和4个已知文本词（SALT，PEPPER，KING，QUEEN）的实验前相关性。观察到任意符号之间的不连贯性以及已知单词之间的连贯性（SALT = PEPPER，KING = QUEEN）。在第2阶段，我们使用线性训练安排建立了4个，4个成员等效类，其中每个类包括3个任意符号和1个已知单词。观察到在几何空间内类的分离。在第3阶段中，对于一半的参与者，我们尝试在每个相干班级的2个成员之间建立班级合并（相干条件；盐=胡椒粉，国王=皇后）。对于其他参与者，我们尝试在每个不连贯班级的2个成员之间建立班级合并（不连贯条件；国王=胡椒粉和皇后乐队=盐）。结果支持合并的一致性类的成功合并，但不支持合并的非一致性类的成功合并。结果对建立复杂的组合网络中的关系自组织具有重要意义。