Lisa Feldman Barrett, Rhea Eskew
Date of Award
Doctor of Philosophy
Department or Academic Unit
College of Science. Department of Psychology.
Base Rates, Emotion, Expectancies, Threat Detection
Cognition and Perception | Personality and Social Contexts | Psychology
Emotions are known to influence judgments and decision making, even on tasks or in situations unrelated to the original source of the emotional experience (Schwarz & Clore, 2007; Clore, Gasper, & Garvin, 2001). However, despite the wealth of research on incidental emotions (those elicited by a source other than the task at hand), it remains unclear whether such emotion effects extend to purely objective judgments for which there is a clear correct and incorrect response. To test this possibility, we had participants complete an emotion induction procedure and then a threat detection task in which they were shown images of White males holding either neutral everyday objects (e.g., wallets, cameras, cellphones, soda cans) or guns. Participants were asked, under time pressure, to identify whether each individual was holding a gun or a neutral object. In a series of background experiments, we showed that participants induced to experience anger demonstrated a bias on the threat detection task whereby they made more errors claiming that neutral objects were guns than vice versa. Neutral participants did not exhibit any such bias. Importantly, the effect appeared to be emotion-specific, as several other positive and negative emotional states (disgust, happiness, sadness) failed to produce any effect on threat detection performance. We believe that, of these emotions, anger alone produced a bias because it was the only emotion that was relevant to the gun/no-gun decision. That is, anger is an emotion that might typically be elicited in situations involving the potential for violence or aggression, and so the experience of anger could more readily be misattributed to the decision at hand.
Building off of these background studies, this dissertation is an extensive investigation into the process by which anger influences performance on the threat detection task. Using signal detection theory, we were able to show that angry participants are not more or less sensitive to the distinction between guns and neutral objects; they did not make more or less errors on the task overall. Instead, angry participants appear to make more false alarms, mistakenly misidentifying neutral objects as guns, in order to ensure that guns are accurately identified when present. In the background studies, we assumed that the signal distribution (the distribution for the gun trials) and the noise distribution (the distribution for the neutral object trials) were Gaussian distributions of equal variance for both neutral and angry participants. If this assumption is accurate, then angry participants set a lower criterion for deciding that a gun is present. That is, they need less evidence that a threat is present before they are willing to claim that a given trial contains a gun. However, it is also possible that anger influences the variance of the underlying distributions, in which case shifts in the distributions could explain the biased error rates without any movement in the placement of angry participants' criterions. For instance, past research has suggested an advantage among individuals high in trait anger or trait anxiety on threat detection tasks whereby they attend to threat-relevant stimuli more readily (Mogg et al., 1997; Rinck et al., 2005). If angry participants in our study were able to more effectively attend to the gun stimuli than the neutral object stimuli throughout the task, this could potentially shrink the variance of the signal distribution among angry individuals. We explored this possibility in two experiments by calculating receiver operating characteristics (ROCs) for both angry and neutral participants for the threat detection task. Results failed to reveal any evidence of unequal variance for both neutral and angry participants. Thus, it appears that experiencing anger is indeed causing participants to lower their criterion so that less information is needed in order to declare the presence of a threat.
Three additional studies explored what might cause angry participants to lower their criterion relative to neutral participants. Given past work on the influence of emotions on estimates of the likelihood of encountering emotion-relevant events (DeSteno et al., 2000), we believed that anger may be causing participants to feel that they were more likely to encounter guns (an anger-relevant stimuli) compared to neutral participants. Two experiments demonstrated that anger does indeed seem to increase participants' estimates of the number of trials that contain guns in the current threat detection task as well as the perceived likelihood that one will encounter aggressive and violent stimuli in the real world. A final experiment demonstrated that such emotion-based expectancies mediate anger's impact on threat detection performance. By experimentally manipulating the base rates of guns and neutral objects in the task and explicitly providing that information to participants, we were able to block participants from utilizing their own subjective expectancies for the number of guns in the task. Using this manipulation we were able to block the bias from occurring among angry participants as well as produce a matching bias among neutral participants. In other words, by manipulating the proposed mediator we were able to successfully eliminate the main effect of emotion on threat detection performance.
Taken together, these experiments represent a concerted effort to understand how emotions may influence simple, objective decisions like object recognition that are thought to rely primarily on automatic processing. Results suggest that angry participants believe they are more likely to encounter anger-relevant stimuli, like guns, and so they exhibit a bias whereby they need less evidence that a threat is present before they are willing to claim an object is a gun. While this process is meant to serve an adaptive purpose—signaling the presence of actual threats in order to help an individual cope more efficiently—when an emotional state gets carried over from its eliciting situation and is utilized as a source of information in a novel, unrelated situation emotions can introduce sources of error into people's judgments.
Baumann, Jolie Elizabeth, "Anger and threat detection: increased expectancy for emotion-relevant stimuli influences object recognition" (2012). Psychology Dissertations. Paper 25. http://hdl.handle.net/2047/d20002727
Click button above to open, or right-click to save.COinS