This article reviews the research relevant to negativity and extremity biases in impression formation and discusses that research as it relates to the major theories that explain these biases. We also describe a model for these biases that draws on principles of natural object categorization. This model explains negativity and extremity biases in terms of the perceived diagnostitity of different kinds of cues for alternative categorizations of the stimulus. The model not only accounts for existing evidence regarding negativity and extremity biases but also suggests circumstances (a) in which positivity biases should occur and (b) in which single cues might be sufficient to prompt categorizations resistant to counterevidence.Two of the more interesting findings arising from research into impression formation are the negativity bias and the extremity bias. Studies have consistently indicated that impression judgments are inordinately influenced by an actor's more negative and more extreme attributes, traits, and behaviors

In this article, we report two experiments that examine a cue-diagnosticity explanation (Skowronski & Carlston, 1986) for the extremity and negativity biases that are commonly found in impressionformation research. The first experiment demonstrates that negative behaviors are generally perceived as more diagnostic than positive behaviors when the former are morality related (honestydishonesty) but that these perceptions are reversed when negative behaviors are ability related (intelligence-stupidity). Furthermore, as predicted, extreme behaviors relating to either morality or ability were perceived as more diagnostic than were moderate behaviors. As a consequence of these perceptions, it was predicted that the impression-formation task used in Experiment 2 would detect negativity and extremity biases for the morality judgment, but positivity and extremity biases for the ability judgment. The data confirmed these predictions, providing support for the cue-diagnosticity approach and posing problems for alternative approaches to judgment biases. Experiment 2 also found recall to be characterized by negativity biases for stimulus behaviors relating to both morality and ability.Data for these studies were collected as part of the dissertation project conducted by John J. Skowronski under the direction of Dona] E. Carlston. Thanks are extended to the members of the dissertation committee-Elizabeth Altmaier, Robert S. Baron, Irwin S. Levin, and Milton E. Rosenbaum-for their helpful comments throughout the project. Thanks are also extended to Augustana College, Rock Island, Illinois, for facilities and subjects used in the early phases of this research.

People's recollections of the past are often positively biased. This bias has 2 causes. The 1st cause lies in people's perceptions of events. The authors review the results of several studies and present several new comparative analyses of these studies, all of which indicate that people perceive events in their lives to more often be pleasant than unpleasant. A 2nd cause is the fading affect bias: The affect associated with unpleasant events fades faster than the affect associated with pleasant events. The authors review the results of several studies documenting this bias and present evidence indicating that dysphoria (mild depression) disrupts such bias. Taken together, this evidence suggests that autobiographical memory represents an important exception to the theoretical claim that bad is stronger than good.

Five experiments used a relearning paradigm to determine whether people spontaneously make trait inferences from behavior descriptions. In each experiment, Ss learned actors' traits more readily after prior exposure to congruent descriptive stimuli (a "savings effect"), suggesting that implicit trait knowledge had been distilled from those descriptions. Moreover, this savings effect (a) was unaffected by Ss' processing objectives, (b) persisted for as long as a week after stimulus presentation, (c) occurred even when the original stimuli could not be recognized, and (d) could not be accounted for by priming mechanisms or differential familiarity with experimental materials. This evidence thus suggests that people do spontaneously derive trait knowledge from behavioral stimuli.

Spontaneous trait transference occurs when communicators are perceived as possessing the very traits they describe in others. Study 1 confirmed that communicators become associated with the trait implications of their descriptions of others and that such associations persist over time. Study 2 demonstrated that these associations influence specific trait impressions of communicators. Study 3 suggested that spontaneous trait transference reflects simple associative processes that occur even when there are no logical bases for making inferences. Finally, Study 4 used more naturalistic stimuli and provided additional evidence that the phenomenon reflects mindless associations rather than logical attributions. Together these studies demonstrate that spontaneous trait transference is a reliable phenomenon that plays a previously unrecognized role in social perception and interaction.