Hibbing et al. propose that variations in political ideology, along the left–right (liberal–conservative) dimension reflect deep-seated, stable, inter-individual differences in physiological and psychological affective responses to negative stimuli. Specifically, they argue that conservatives (those with a right-leaning political orientation) have a negativity bias (they react more strongly to negative stimuli). Hibbing et al. convincingly show that one can find evidence to support this proposal. At the same time, they overlook much of the counterevidence contained within the psychological research literature. Building on recent findings, we challenge their claim that patterns of behavior moderated by (or correlated with) political orientation are attributable to a stable, affect-based negativity bias among conservatives.
First, a number of behavioral tendencies associated with right-leaning (conservative) political views are uncorrelated with the negativity of the stimuli being considered. For example, right-leaning Americans (i.e., Republicans) are more likely to vote for political candidates who have stereotypically Republican-looking faces, whereas left-leaning U.S. voters (i.e., Democrats) are not influenced by these political facial stereotypes (Olivola et al. Reference Olivola, Sussman, Tsetsos, Kang and Todorov2012). Yet the voting preferences of both Democrats and Republicans are strongly (and positively) predicted by another facial stereotype: how competent the candidates look (Olivola et al. Reference Olivola, Sussman, Tsetsos, Kang and Todorov2012). This pattern of results – Republicans being more influenced by one set of facial features, but equally influenced by another – is difficult to explain within Hibbing et al.'s theoretical framework. It is unlikely that Republican voters react more strongly to “facial-conservatism” than “facial-competence” because the latter is actually a stronger predictor of voters' preferences (Olivola & Todorov Reference Olivola and Todorov2010); even among Republican voters (Olivola et al. Reference Olivola, Sussman, Tsetsos, Kang and Todorov2012). As another example, consider political consumerism, whereby people deliberately abstain from purchasing products that conflict with their political views (boycotting) and/or deliberately purchase products that accord with their political views (“buycotting”). Hibbing et al.'s assertion that “[c]onservatives have a negativity bias, whereas liberals do not have a positivity bias” (sect. 6.3, para. 3) implies distinct predictions concerning the tendency for liberals and conservatives to engage in political boycotting (negative product avoidance) versus buycotting (positive product approach). If conservatives react more negatively (than liberals) to products that are incongruent with their political views then they should be more likely (than liberals) to engage in political boycotting. In contrast, if conservatives are distinguished solely by a negativity bias and liberals do not show a positivity bias then both groups should react just as positively to politically congruent products, and therefore be equally likely to engage in political buycotting. However, the data are inconsistent with these predictions: Conservatives are less or equally likely (than liberals) to engage in political boycotting, whereas liberals are more likely (than conservatives) to engage in political buycotting (Katz Reference Katz2011; Newman & Bartels Reference Newman and Bartels2011). Taken together, these examples almost suggest a double-dissociation between politically differentiated behavioral patterns and stimulus negativity.
Second, behavioral tendencies associated with right-leaning (conservative) political views that are correlated with stimulus negativity do not necessarily reflect differences in affective appraisal. For example, we found that people who identify with right-leaning political parties are more likely (than their left-leaning counterparts) to exhibit an irrational aversion to taxes, in the specific sense that they will go to greater lengths to avoid tax-related costs than to avoid equivalent (or larger) financial costs that are unrelated to taxes (Sussman & Olivola Reference Sussman and Olivola2011). However, this “tax aversion” is driven by differing beliefs about tax usage and not by differing emotional appraisals: left- and right-leaning respondents reported feeling equally angry when their tax dollars were used in ways that they disapproved of, but left-leaning respondents were more likely to believe that their tax dollars were being used in ways that they approved of (Sussman & Olivola Reference Sussman and Olivola2011). Similarly, although Republicans are less willing to pay a surcharge for emitted carbon dioxide that is framed as a “carbon tax” (vs. “carbon offset”), this tendency is driven by memory retrieval processes and not by affective reactions (Hardisty et al. Reference Hardisty, Johnson and Weber2010).
Third, many behavioral tendencies associated with right-leaning (conservative) political views are too malleable to be the product of deep-seated physiological reflexes. For example, we found that we can decrease (increase) tax aversion among right-leaning participants, simply by asking them to list a few positive (negative) uses of their tax payments (these manipulations did not influence left-leaning participants; Sussman & Olivola Reference Sussman and Olivola2011). In fact, asking right-leaning participants to consider positive uses of their tax payments eliminated the left-right difference in tax aversion altogether (Sussman & Olivola Reference Sussman and Olivola2011). Similarly, merely priming political identity (by asking participants which political party and presidential candidate they supported) increased the likelihood that Republicans (but not Democrats) preferred lower-risk (vs. higher-risk) monetary gambles when these were labeled as “conservative” (vs. “risk-tolerant”) choice options (Morris et al. Reference Morris, Carranza and Fox2008).
The above studies demonstrate failures of negativity bias to explain several important behavioral and attitudinal correlates of political ideology, and thus represent empirical challenges to the theory advanced in the target article. More generally, we are skeptical that any single variable or dimension can provide a unifying account of something so complex, malleable, and nebulous as political ideology. Hibbing et al. acknowledge that their theory faces many difficult questions (e.g., “If conservatives are universally more averse to negativity, it would seem that heightened response and attention to the negative should lead to equal amounts of concern over a leveled rainforest and a hostile out-group.” [sect. 6.2, para. 6]). Unfortunately, their attempts to address these issues (e.g., “…it may be the case that conservatives are particularly attuned to threats by an identifiable, malevolent, volitional force such as a bad guy with a gun. Or, perhaps attitudes toward longer term and arguably more amorphous threats such as climate change, pollution, and income inequality are not as connected to negativity biases” [sect. 6.2, para. 6]) lead them to contradict their earlier claims (e.g., “Environmental stimuli that are unexpected, ambiguous, uncertain, or disorderly also appear to generate more response and attention from conservatives than liberals at a variety of levels” [sect. 6, para. 6]). It seems the only way to accommodate existing data is to adopt a definition of “negativity bias” that is so flexible it risks becoming unfalsifiable.
Hibbing et al. propose that variations in political ideology, along the left–right (liberal–conservative) dimension reflect deep-seated, stable, inter-individual differences in physiological and psychological affective responses to negative stimuli. Specifically, they argue that conservatives (those with a right-leaning political orientation) have a negativity bias (they react more strongly to negative stimuli). Hibbing et al. convincingly show that one can find evidence to support this proposal. At the same time, they overlook much of the counterevidence contained within the psychological research literature. Building on recent findings, we challenge their claim that patterns of behavior moderated by (or correlated with) political orientation are attributable to a stable, affect-based negativity bias among conservatives.
First, a number of behavioral tendencies associated with right-leaning (conservative) political views are uncorrelated with the negativity of the stimuli being considered. For example, right-leaning Americans (i.e., Republicans) are more likely to vote for political candidates who have stereotypically Republican-looking faces, whereas left-leaning U.S. voters (i.e., Democrats) are not influenced by these political facial stereotypes (Olivola et al. Reference Olivola, Sussman, Tsetsos, Kang and Todorov2012). Yet the voting preferences of both Democrats and Republicans are strongly (and positively) predicted by another facial stereotype: how competent the candidates look (Olivola et al. Reference Olivola, Sussman, Tsetsos, Kang and Todorov2012). This pattern of results – Republicans being more influenced by one set of facial features, but equally influenced by another – is difficult to explain within Hibbing et al.'s theoretical framework. It is unlikely that Republican voters react more strongly to “facial-conservatism” than “facial-competence” because the latter is actually a stronger predictor of voters' preferences (Olivola & Todorov Reference Olivola and Todorov2010); even among Republican voters (Olivola et al. Reference Olivola, Sussman, Tsetsos, Kang and Todorov2012). As another example, consider political consumerism, whereby people deliberately abstain from purchasing products that conflict with their political views (boycotting) and/or deliberately purchase products that accord with their political views (“buycotting”). Hibbing et al.'s assertion that “[c]onservatives have a negativity bias, whereas liberals do not have a positivity bias” (sect. 6.3, para. 3) implies distinct predictions concerning the tendency for liberals and conservatives to engage in political boycotting (negative product avoidance) versus buycotting (positive product approach). If conservatives react more negatively (than liberals) to products that are incongruent with their political views then they should be more likely (than liberals) to engage in political boycotting. In contrast, if conservatives are distinguished solely by a negativity bias and liberals do not show a positivity bias then both groups should react just as positively to politically congruent products, and therefore be equally likely to engage in political buycotting. However, the data are inconsistent with these predictions: Conservatives are less or equally likely (than liberals) to engage in political boycotting, whereas liberals are more likely (than conservatives) to engage in political buycotting (Katz Reference Katz2011; Newman & Bartels Reference Newman and Bartels2011). Taken together, these examples almost suggest a double-dissociation between politically differentiated behavioral patterns and stimulus negativity.
Second, behavioral tendencies associated with right-leaning (conservative) political views that are correlated with stimulus negativity do not necessarily reflect differences in affective appraisal. For example, we found that people who identify with right-leaning political parties are more likely (than their left-leaning counterparts) to exhibit an irrational aversion to taxes, in the specific sense that they will go to greater lengths to avoid tax-related costs than to avoid equivalent (or larger) financial costs that are unrelated to taxes (Sussman & Olivola Reference Sussman and Olivola2011). However, this “tax aversion” is driven by differing beliefs about tax usage and not by differing emotional appraisals: left- and right-leaning respondents reported feeling equally angry when their tax dollars were used in ways that they disapproved of, but left-leaning respondents were more likely to believe that their tax dollars were being used in ways that they approved of (Sussman & Olivola Reference Sussman and Olivola2011). Similarly, although Republicans are less willing to pay a surcharge for emitted carbon dioxide that is framed as a “carbon tax” (vs. “carbon offset”), this tendency is driven by memory retrieval processes and not by affective reactions (Hardisty et al. Reference Hardisty, Johnson and Weber2010).
Third, many behavioral tendencies associated with right-leaning (conservative) political views are too malleable to be the product of deep-seated physiological reflexes. For example, we found that we can decrease (increase) tax aversion among right-leaning participants, simply by asking them to list a few positive (negative) uses of their tax payments (these manipulations did not influence left-leaning participants; Sussman & Olivola Reference Sussman and Olivola2011). In fact, asking right-leaning participants to consider positive uses of their tax payments eliminated the left-right difference in tax aversion altogether (Sussman & Olivola Reference Sussman and Olivola2011). Similarly, merely priming political identity (by asking participants which political party and presidential candidate they supported) increased the likelihood that Republicans (but not Democrats) preferred lower-risk (vs. higher-risk) monetary gambles when these were labeled as “conservative” (vs. “risk-tolerant”) choice options (Morris et al. Reference Morris, Carranza and Fox2008).
The above studies demonstrate failures of negativity bias to explain several important behavioral and attitudinal correlates of political ideology, and thus represent empirical challenges to the theory advanced in the target article. More generally, we are skeptical that any single variable or dimension can provide a unifying account of something so complex, malleable, and nebulous as political ideology. Hibbing et al. acknowledge that their theory faces many difficult questions (e.g., “If conservatives are universally more averse to negativity, it would seem that heightened response and attention to the negative should lead to equal amounts of concern over a leveled rainforest and a hostile out-group.” [sect. 6.2, para. 6]). Unfortunately, their attempts to address these issues (e.g., “…it may be the case that conservatives are particularly attuned to threats by an identifiable, malevolent, volitional force such as a bad guy with a gun. Or, perhaps attitudes toward longer term and arguably more amorphous threats such as climate change, pollution, and income inequality are not as connected to negativity biases” [sect. 6.2, para. 6]) lead them to contradict their earlier claims (e.g., “Environmental stimuli that are unexpected, ambiguous, uncertain, or disorderly also appear to generate more response and attention from conservatives than liberals at a variety of levels” [sect. 6, para. 6]). It seems the only way to accommodate existing data is to adopt a definition of “negativity bias” that is so flexible it risks becoming unfalsifiable.
ACKNOWLEDGMENTS
Christopher Y. Olivola acknowledges the support of The British Academy and the Royal Society.