Updating Spontaneous Trait Inferences: An Analysis of Memory Reconsolidation as a Mechanism of Changing First Impressions

Abstract Previous research showed that people infer traits from others' behaviors without any intention or awareness being necessary (Uleman et al., 2012). Nevertheless, this phenomenon has only been measured through limited behavioral information about others received at a single point in time. The persistence of or change in spontaneously formed trait inferences (STIs) over time and in the presence of new trait-inconsistent information has not been investigated previously. The present work aimed to examine this question by presenting a theoretical model of STI formation and updating and testing this model through five experiments.Drawing on past research on implicit evaluations and explicit trait attributions, we hypothesized that STIs should be robust and resistant to change over time and in the presence of new trait-inconsistent information. This would be due to perceivers forming and storing new STIs (including the ones that are semantically inconsistent with the initial STIs) independently from the initial STIs in memory. Consistent with this hypothesis, we showed that STIs were not affected by trait-inconsistent information subsequently learned about the same target actor (Experiment 1a and 1b), and were persistent over 48 hours after initial formation. STIs were also formed from the subsequently learned behavioral information (even trait-inconsistent information), yet, stored independently from the initial STIs as hypothesized. Participants held multiple implicit inferences about the same actor, even when the traits were inconsistent.Next, the present study investigated memory reconsolidation as a possible mechanism for updating spontaneous impressions. Existing work on memory reconsolidation established that memories become flexible when they are reactivated, even after they are consolidated over time. Similarly, we hypothesized that having initial associations about persons reactivated when learning about new information about those persons should facilitate memory restructuring, and thus impression updating. The updating of implicit impressions was expected to manifest itself in two ways; first, participants would decrease the strength of STIs they formed from the initial behaviors, and second, they would increase the strength of SGI (context-dependent, spontaneous goal inferences) formed from these initial behaviors. The results for STI updating were mixed. When participants learned about the initial behaviors via three cycles of exposure (and thus strongly encoded in participants' memory), they did not update their STIs formed from these initial behaviors in light of new trait-inconsistent information presented after memory reactivation (Experiment 2). The expected pattern of updating (the disappearance of the STI effect in a false recognition task within this particular learning condition) was observed in Experiment 4, when participants learned about the positive behaviors of actors initially (by being exposed to each behavior once), and were asked to make quick judgments about the actors who performed those behaviors. Also, consistent with our predictions about updating of SGIs, in Experiment 3, participants' pattern of SGIs (especially of those with a better memory of actors' behaviors) were stronger after having learned about trait-inconsistent new information upon memory reactivation. In addition, those with stronger SGIs tended to form weaker STIs about the same actors in Experiment 3, suggesting that SGI might have been construed as an alternative to STI. All experiments that measured behavior memory showed a primacy effect; old (initial) behaviors were recalled better than new behaviors of the actors. Among the new behaviors, those that were consistent with the old behaviors in terms of implied trait were remembered better than those that implied inconsistent traits or did not imply any trait information. The implications of the findings on the stickiness of first impressions and primacy effect in behavior memory are discussed.