The Undercorrection Effect: Cognitive Busyness and Dispositional Inferences

The Undercorrection Effect

Table 9.1, The Effects of Busyness on Inferences of Dispositional Anxiety, is not reproduced in this excerpt.

Source: Adapted from Gilbert, D. T., Pelham, B. W., & Krull, D. S. (1988).

In a set of follow-up studies (Gilbert, McNulty, Giuliano, & Benson, 1992), we showed that a wide variety of manipulations could produce the same “undercorrection effect” that the rehearsal of word strings produced. For example, in one study, participants viewed what the researchers called the “anxious woman video”. As before, half were assigned to the anxious topics condition and half were assigned to the relaxing topics condition. However, in this study we manipulated cognitive busyness in a rather unusual way. Our model suggested that people effortlessly categorize behavior (“She looks anxious”) before they go on to characterize the actor (“She’s an anxious person”) and then correct that inference (“But she was discussing her sex life, so she’s probably not such an anxious person after all”). We reasoned that if the categorization stage were for some reason to become cognitively demanding, then resources that might normally be devoted to correction would be usurped by categorization, thus producing the undercorrection effect. In other words, the more difficulty observers had categorizing an action, the more likely they should be to draw dispositional inferences about the actor who performed it. To demonstrate this, participants in the non-busy condition were shown a normal version of the anxious woman video, while participants in the busy condition were shown a version of the video that had been visually degraded by misadjusting the tracking mechanisms on two videocassette recorders and then using those machines to make successive copies of the master tape. The final fourth-generation tape had a great deal of visual noise (streaking, blurring, color dropout, and snow) that interrupted the flow of the action and obscured some of its details. Indeed, participants who viewed the sloppy copy drew more dispositional inferences about the actor than did participants who viewed a pristine copy of the tape.

The foregoing research suggests that busy perceivers fail to correct their characterizations because they are too busy to use information about situational constraints, but not because they fail to notice, gather, or recall that information. Memory data support this suggestion. However, if busy perceivers really do have in memory all the information they require to correct their characterizations, then they should be perfectly able to do so at some later “unbusy” time. We examined this possibility in a series of studies (Gilbert & Osborne, 1989) that began by replicating our original demonstration of the undercorrection effect. For example, in one study we asked participants to watch the anxious woman video. As before, half were assigned to the anxious topics condition and half were assigned to the relaxing topics condition. In addition, half were assigned to rehearse an eight-digit number as they watched the video (busy condition) and the remaining half were not (nonbusy condition). After watching the videotape, participants rated the actor’s trait anxiety and then stopped rehearsing the number. These formerly busy participants then spent a few minutes writing an essay about the actor’s personality, and then rerated the actor. As the authors report in Table 9.2, which is not reproduced here, ratings made while participants were busy showed the typical undercorrection effect; but ratings made after the participants had become unbusy did not. Apparently, formerly busy participants were able to use information stored in memory to correct their characterizations when later they had sufficient resources to do so.