A Motivational Account of the Impact Bias
Abstract (summary)
The impact bias is the tendency for people to overestimate the intensity and duration of their feelings in response to some future outcome (Gilbert, Driver-Linn, & Wilson, 2002; Wilson & Gilbert, 2003; Wilson & Gilbert, 2005). Most explanations for the bias conceptualize it as an undesirable side effect of how the cognitive system goes about making predictions. Yet, perhaps such mispredictions can serve an important self-regulatory function, helping people achieve their goals.
This research argued for a motivated impact bias. That is, there are times when people may be driven to predict feelings more extreme than they actually will experience. By overestimating one's feelings to some future outcome, the stakes of succeeding or failing at gaining that outcome are raised. As such, this may energize people to pursue their goals to achieve the desired outcomes. The more one wants to gain (or avoid) some outcome, the more likely s/he will be to exhibit such an impact bias.
To investigate this question, participants were asked to predict their feelings in response to the outcome of a performance task (visual responsiveness task in studies 1-3, creativity task in study 4, and verbal test in study 5) they would complete. It was hypothesized that a motivated impact bias would most likely manifest itself when the motivation to achieve some outcome was high. Using an expectancy-value framework, motivation was conceptualized as a product of expectancy and value (e.g. Tolman, 1932; Atkinson, 1957).
It was predicted that by measuring (value in study 1; value and expectancy in study 3; value in study 5) and manipulating (value in study 2; value and expectancy in study 4; expectancy in study 5) motivation to achieve some outcome (good visual responsiveness in studies 1-3; high creativity in study 4; successful performance on verbal ability test in study 5) that the impact bias would be observed.
For studies 1-3, there was no overall impact bias, and motivation had no effect on forecasting accuracy. One potential problem was lack of evidence for any motivated behavior on the task. For studies 4 and 5, there was evidence of an overall impact bias, but it was not moderated by expectancy and value. Again, there was no evidence of expectancy-value effects on the task.
However, by using performance on the tasks in studies 4 and 5 as a behavioral measure of motivation, there was evidence to suggest that people who are motivated to achieve (or avoid) some outcome overestimate their future feelings. This is not sufficiently strong evidence to support a motivated impact bias, but it does suggest that the relationship between motivational variables and the impact bias warrant further investigation. These results should encourage researchers to consider that the impact bias is not always an undesired error, but can instead be a desired strategy.