Neural Bases of Motivated Reasoning: an Fmri Study of Emotional Constraints on Partisan Political Judgment in the 2004 U.S
Total Page:16
File Type:pdf, Size:1020Kb
Neural Bases of Motivated Reasoning: An fMRI Study of Emotional Constraints on Partisan Political Judgment in the 2004 U.S. Presidential Election Drew Westen, Pavel S. Blagov, Keith Harenski, Clint Kilts, and Stephan Hamann Abstract & Research on political judgment and decision-making has formation threatening to their own candidate, the opposing converged with decades of research in clinical and social psy- candidate, or neutral control targets. Motivated reasoning was chology suggesting the ubiquity of emotion-biased motivated associated with activations of the ventromedial prefrontal cor- reasoning. Motivated reasoning is a form of implicit emotion tex, anterior cingulate cortex, posterior cingulate cortex, in- regulation in which the brain converges on judgments that sular cortex, and lateral orbital cortex. As predicted, motivated minimize negative and maximize positive affect states associ- reasoning was not associated with neural activity in regions ated with threat to or attainment of motives. To what extent previously linked to cold reasoning tasks and conscious motivated reasoning engages neural circuits involved in ‘‘cold’’ (explicit) emotion regulation. The findings provide the first reasoning and conscious emotion regulation (e.g., suppres- neuroimaging evidence for phenomena variously described sion) is, however, unknown. We used functional neuroimag- as motivated reasoning, implicit emotion regulation, and psy- ing to study the neural responses of 30 committed partisans chological defense. They suggest that motivated reasoning is during the U.S. Presidential election of 2004. We presented qualitatively distinct from reasoning when people do not have subjects with reasoning tasks involving judgments about in- a strong emotional stake in the conclusions reached. & INTRODUCTION avoid aversive feelings such as anxiety and guilt. We use In political science, cognitive science, economics, law, the term motivated reasoning here because of its wide- and business, the predominant models of judgment and spread use (although, strictly speaking, all reasoning is decision-making today are ‘‘bounded rationality’’ models typically motivated by emotions such as interest, excite- (Simon, 1990). These models suggest that people are ment, anxiety, etc.; see Marcus, 2002; Westen, 1985). rational within limits imposed by cognitive shortcuts and Neural network models of motivated reasoning sug- heuristics (Westen, Weinberger, & Bradley, in press; gest that in affectively relevant situations, the brain Gigerenzer & Selten, 2001; Kahneman & Tversky, 2000). equilibrates to solutions that simultaneously satisfy two In political science, a long-standing body of research sets of constraints: cognitive constraints, which maxi- on ‘‘partisan’’ biases in political judgment (e.g., Taber, mize goodness of fit to the data, and emotional con- Lodge, & Glathar, 2001; Campbell & Converse, 1960) straints, which maximize positive affect and minimize points to another set of limits to rational judgment im- negative affect (Westen, Feit, Arkowitz, & Blagov, 2005; posed by emotion-biased or motivated reasoning (i.e., Thagard, 2003; Westen, 1998). Decision theorists have reasoning biased to produce emotionally preferable con- long argued that people gravitate toward decisions that clusions; Kunda, 1990; Lord, Ross, & Lepper, 1979). maximize expected utility (or in emotional terms, that Motivated reasoning can be viewed as a form of implicit optimize current or anticipated affect; Simon, Krawczyk, affect regulation in which the brain converges on solu- & Holyoak, 2004; Mellers, 2000; Westen, 1985). Contem- tions that minimize negative and maximize positive porary views of motivation similarly emphasize approach affect states (Westen & Blagov, in press; Westen, 1985, and avoidance systems motivated by positive and nega- 1994, 1998). Freud (1933) described such processes tive affect (Carver, 2001; Davidson, Jackson, & Kalin, decades ago, using the term ‘‘defense’’ to denote the 2000). The same processes of approach and avoidance, processes by which people can adjust their cognitions to motivated by affect or anticipated affect, may apply to motivated reasoning, such that people will implicitly approach and avoid judgments based on their emotional Emory University associations. D 2006 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 18:11, pp. 1947–1958 A series of studies involving political crises in the Kerry, Bill Clinton, Dick Cheney, the Democratic Party, United States spanning the past 8 years (the impeach- and the Republican party using a 0–100 ‘‘feeling ther- ment of Bill Clinton, the disputed presidential election mometer’’ (from cold to warm); obtained 4-point rat- of 2000, and the discovery of torture by the United ings of how often Bush and Kerry made them feel angry, States at Abu Ghraib prison in Iraq) supports this model hopeful, afraid, proud, and disgusted; and obtained (Westen et al., 2005). These studies, along with simu- 4-point ratings of the extent to which they saw the lations using a connectionist network designed to ad- two candidates as moral, intelligent, dishonest, and out dress ‘‘hot cognition’’ (Thagard, 2003), suggest that of touch with ordinary people. To be included sub- political reasoning can be strongly influenced by the jects had to rate themselves as a strong Democrat or emotional consequences of drawing one conclusion or Republican and to endorse a difference between the the other. Although research has begun to examine ex- two parties or the two candidates 30 points on the plicit (conscious) processes used to regulate emotion, feeling thermometer. notably suppression and distraction (Anderson et al., 2004; Hariri, Mattay, Tessitore, Fera, & Weinberger, 2003; Ochsner, Bunge, Gross, & Gabrieli, 2002), no studies Measures and Procedures have yet examined the neural processes involved in motivated reasoning or implicit affect regulation. The To simulate the constraint satisfaction processes that present study is also among the first to explore the occur as citizens confront political information, we neural basis of any form of political decision-making. devised six sets of statements regarding each of the In this study, conducted during the U.S. Presidential following targets: George Bush, John Kerry, and polit- election of 2004, we observed the reasoning processes ically neutral male targets (e.g., Tom Hanks, Hank of committed partisans as they were presented with Aaron, William Styron). (We tried to generate well- threatening information about their own candidate, the known targets who were emotionally neutral but ulti- opposing candidate, and neutral control individuals. We mately selected politically neutral but largely mildly hypothesized that reasoning about threatening informa- positive targets because of the difficulty identifying tion about one’s own candidate would activate regions well-known figures of any kind about whom people likely to be involved in implicit emotion regulation, nota- have no feelings.) Although many of the statements bly the ventromedial prefrontal cortex (VMPFC) and and quotations were edited or fictionalized, we maxi- the anterior cingulate cortex (ACC), as well as regions mized their believability by embedding them in actual reflecting elicitation of negative emotion (the insula, quotes or descriptions of actual events. Subjects were lateral orbital frontal cortex, and amygdala) (see, e.g., given detailed instructions prior to scanning and a Ochsner & Gross, 2005). practice run to familiarize them with procedure prior to imaging trials. Each statement set consisted of seven slides present- METHODS ing verbal material, designed to present a clear contra- diction between the target person’s words and actions Participants and then to resolve that contradiction (Figure 1). Pre- We recruited subjects (n = 30, 15 Democrats and sentation duration of the slides ranged from 5 to 15 sec, 15 Republicans) by placing flyers at local political party depending on the length and demands of the material offices, public places, and cars and houses with po- or task, and reflected pretesting on pilot subjects. Slide litical endorsements (e.g., bumper stickers); posting 1 (15 sec) presented an initial statement, typically a information on Internet political discussion groups quote from the target individual. Slide 2 (12 sec) pre- and local political and party listservs; and placing sented a contradictory statement suggesting that the newspaper and radio advertisements. Recruitment ma- target’s words and actions were inconsistent. Slide 3 terials requested right-handed men, ages 22–55 years, (7 sec) asked subjects to consider whether the target’s who were ‘‘committed Republicans or Democrats.’’ ‘‘statements and actions are inconsistent with each We conducted all screening and scanning from late other,’’ and Slide 4 (5 sec) asked them to rate the extent August through early October 2004. Subjects received to which they agreed that the target’s words and deeds $50 compensation. were contradictory, from 1 (strongly disagree)to4 Potential subjects were screened by phone using a (strongly agree) by using a four-button pad. Slide 5 magnetic resonance imaging (MRI) screener (to rule out (12 sec) presented an exculpatory statement that logi- safety risks, neurological conditions, etc.) and a political cally explained away the inconsistency. Slide 6 (7 sec) attitudes questionnaire, using items from the National then asked subjects to consider whether the target’s Election Studies (NES,