Negative first impressions resist correction simply because they end the interaction

Evidence

Person Perception More recently, Smith and Collins (2009) used ABM to simulate the processes by which impressions of people are constructed, transmitted, and filtered through social networks. Among other variables, their model assigned values to agents representing how probable they were to “behave” in a positive or negative manner. The model assumed that agents will repeatedly interact only with agents who act positively toward them. One outcome was that agents formed more negative impressions of other agents than was warranted by the agents’ probabilities of positive and negative behaviors. This result is due to the fact that positive interactions with an “unlikeable” agent could be corrected through repeated interaction, whereas negative interactions with an objectively “likeable” agent are not corrected because they result in decreased future interaction (see Denrell, 2005, for a mathematical model of this process). — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 828

Categorization

• XREF: Mechanism echoes real outgroup-psychology findings about one bad stereotype-confirming contact overriding many good contacts; connects to the reader’s own prior Allport thread and to general negativity-bias findings.
• corroborated: 2 independent passages support this claim — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 828