[The importance of balance, generalization and positivity as cognitive structural rules]

Three formal schemata (the cognitive rules of balance, generalization and positivity) were examined to determine their impact on the retention of information about social configurations. First, under standard conditions (12 relations) better retention for relations stemming from balanced configurations was demonstrated, in accordance with the Heiderian memory hypothesis (Experiment 1). Neither generalization nor positivity were observed. The second experiment, which involved 24 relations, showed both generalization and positivity to be effective. A balance effect therefore appears to be limited to a maximum of about 18-21 to-be-learned relations. Beyond this limit, however, the simpler rules (positivity and generalization) can still have an impact on memory. Thus, learning remains schema-driven and does not become "rote learning". The failure of balance to be effective when a larger amount of material is involved is discussed with reference to two alternatives: interference by the number of relations and interference by the number of triads. The significance of the regarding models of representation for social relations in memory is also discussed.