Relational coherence in ambiguous and unambiguous relational networks.

Clinical theories often appeal to general cognitive styles in explaining psychopathology, but without describing in detail how the patterns are formed. In the present investigation, two experiments were conducted to examine how individuals respond to ambiguous relational networks. In both experiments, the participants learned two 3-stimulus networks (A1 LESS THAN B1, A1 GREATER THAN C1 and A2 GREATER THAN B2, C2 LESS THAN A2). Participants were presented with test trials to examine if they classified the combinatorial relations (B1 ↔ C1 and B2 ↔ C2) as SAME or DIFFERENT and as GREATER THAN or LESS THAN. Although the B-C combinatorial relation in Network 1 is derivable in a readily coherent way (B1 GREATER THAN C1 and thus also B1 DIFFERENT C1), in Network 2 the combinatorial relation is ambiguous. When participants were required to specify the Network 2 B-C relation as either SAME or DIFFERENT, those who chose DIFFERENT, also consistently chose B2 as either GREATER THAN or LESS THAN C2. Conversely, those who classified the B-C relation as SAME were inconsistent within themselves in choosing B2 as GREATER THAN or LESS THAN C2. In Experiment 2, nonarbitrary multiple exemplar pretraining was used to bias SAME versus DIFFERENT as a response for ambiguous combinatorial relations. In accord with the pattern seen in Experiment 1, those biased toward DIFFERENT consistently chose a comparative relation between B2 and C2 while those biased toward SAME were inconsistent in their comparative choices. The findings provide support for the importance of history and coherence in establishing patterns of responding to ambiguous relational networks, providing a beginning behavioral model of cognitive styles and errors.