In the summer of 2007, 200 participants were brought to a computer laboratory in Lincoln, Nebraska, USA, to complete a survey soliciting their political, personality and demographic information subsequent to their having been contacted by phone at random by a professional survey organization. Though in no way a representative sample, this group has the advantage of not being restricted to college undergraduates and, relatedly, having reasonably representative demographic characteristics given the target population: mean age = 42; 52 per cent female; mean income in the $40 000–60 000 range; and mean educational level = some college. These 200 were intended to serve as a pool from which smaller groups could be culled for physiological testing. The particular group employed in the analysis here consisted of 48 individuals who were called back later that summer. They were selected because of availability and because they were the individuals most clearly falling on either the political left or the political right according to the survey responses provided during their first visit. Participants were paid $50 for each of their two separate trips to the laboratory. The data on two participants had to be removed, one owing to a health issue, the other owing to a mechanical problem with a sensor.
To measure political orientation, several variables were combined. Since a US sample was used in our analyses, we used party labels, ideological labels and individual political issues that would be familiar to such a group. Thus, participants were asked to (i) report their ideological position on a scale running from strong liberal (left) to strong conservative (right), (ii) report their partisan affiliation, from strong Democrat (left) to strong Republican (right), (iii) answer 28 items on their specific policy preferences presented in the well-known Wilson–Patterson format [43
], and (iv) complete a social principles index. The latter presented subjects with 15 forced choices between basic principles of social organization. As an example of items in this last category, participants indicated whether ‘society works best when … those who break the rules are punished … or … when those who break the rules are forgiven’ ([44
]; a full listing of these and the Wilson–Patterson items can be found in electronic supplementary material, appendices A and B). For both the Wilson–Patterson issue items and the ‘society works best’ items, an additive index was constructed (with the position on the political right always given the higher coding). These four diverse measures of political orientation are fairly strongly related, with bivariate correlations ranging from 0.57 to 0.75 (p
< 0.05 in all cases) and with a factor analysis confirming that these four measures tap into a single dimension (a principal components analysis yielded a single factor accounting for approx. 75% of the variance, and factor loadings for the individual variables were 0.79 or higher). The four indicators were weighted equally and added together to create a broad measure of left–right political orientation.
In the physiological session, participants were shown a series of 33 still images. Each image was shown once and was preceded by a fixation point that was displayed during an inter-stimulus interval. The order of slides was initially randomized and then presented in the same order to all participants. During the slide show, electrodermal activity (in the form of skin conductance readings) was collected using a pair of Ag|AgCl electrodes and standard psychophysiological equipment. Since eccrine glands release moisture as part of sympathetic nervous system activation, and since the rate of movement of electricity across the surface of the skin is a good indicator of the presence of moisture, electrodermal activity has long been accepted as a fairly direct and pure representation of sympathetic activity, making it a good measure of the psychological concepts of emotion, arousal and attention. [45
]. There are a number of approaches to measure skin conductance level (SCL) response to a stimulus; a common approach is to measure SCL at two different time points, which can be reported either as a raw or adjusted difference, or as a percentage or proportion [45
]. This approach has the advantage of providing a means to control for wide variation in baseline electrodermal activity and is the approach followed here. SCLs for each image were measured as a proportion of the SCL recorded while the participant was viewing the fixation point prior to image exposure. This creates a standardized measure where 1 denotes no change in SCL between viewing a fixation point and an appetitive/aversive image, and numbers greater than 1 indicate an SCL increase. Several other measurement approaches were constructed to capture the difference between SCL during fixation point and SCL during image exposure, including calculating raw first differences and differences in logged means. The resulting variables were all correlated at levels greater than 0.90, and we report the proportion measure because it lends itself to easy and intuitive interpretation.
All of the images used in the present study were rated by 126 independent judges (none of whom was a participant in the studies reported here) who were asked to rate image valence on a nine-point scale whether each image gave them ‘happy/positive’ [1
] or ‘unhappy/negative’ [9
] feelings and to rate how strongly they felt an emotional reaction when looking at the image. Based on these ratings, the three most negatively valenced and the three most positively valenced images were selected for use during the physiological session. The negative (aversive) images were a spider on a man's face (mean valence rating 7.65, s.d. = 1.68), an open wound with maggots in it (mean valence rating 7.94, s.d. = 1.15), and a crowd fighting with a man (mean valence rating 7.83, s.d. = 1.16). The three images judged to be the most positive (appetitive) were of a happy child (mean valence rating = 4.94, s.d. = 2.33), a bowl of fruit (4.36, s.d. = 2.21) and a cute rabbit (4.62, s.d. = 2.33). Raters were also asked to report the specific emotion they felt when looking at each image. The most frequently reported emotion for the spider image was fear (78% of raters reported the image evoked this emotion), for the maggot image, the most frequently reported emotion was disgust (96%), and for the crowd fighting with a man, the most frequently evoked emotion was anger (76%). Thus, these images would seem to capture an array of different negative emotional responses. Positive emotions have fewer discrete categories and, according to the raters, the most frequently evoked emotion for all three appetitive images was happiness.
The relation between political temperament and electordermal increases in response to aversive/appetitive images was initially examined by dividing participants at the mean on the composite measure of political orientations and then plotting separately the physiological response for the left-of-centre and right-of-centre groups. The consistent empirical finding in psychophysiology is that, while participants exhibit an enhanced physiological response to both appetitive and aversive stimuli, the response is typically greater for aversive stimuli [39
]. Consistent with the theory outlined above and with previous research on narrower emotions [18
], the hypothesis is that individuals on the right side of the political spectrum will exhibit increased electrodermal activity when viewing aversive images while those on the left side will exhibit increased electrodermal activity, in relative terms, when viewing the appetitive images. This prediction is confirmed by a 2 (image type: appetitive versus aversive) × 2 (ideology: left versus right) mixed analysis of variance (ANOVA) as there is a significant interaction between image type and ideology (F
= 5.60; p
< 0.05). As can be seen in , electrodermal increases for those on the political right are greater for aversive relative to appetitive images, whereas for those on the political left the opposite pattern of results is exhibited.
Mean skin conductance change (in microsiemens) as a function of political temperament (left versus right) and image type (appetitive versus aversive). Triangles with solid line, right-of-centre; squares with solid line, left-of-centre.
Though these initial findings are suggestive, political orientations are better characterized as continuous rather than dichotomous since many individuals are political moderates rather than ideologues. Moreover, other variables besides physiological patterns are likely to be relevant to political orientations. Therefore, we regressed the continuous measure of political orientation on the mean difference in physiological response depending on stimulus type (skin conductance increase in response to appetitive subtracted from skin conductance increase in response to aversive), as well as on four standard demographic controls: age, gender, income and education. Higher values on the composite measure of political ideology indicate right-of-centre orientations and higher values on the physiological measure indicate relatively greater electrodermal increases to aversive stimuli, meaning that a positive relationship is expected. As can be seen in , only one of the control variables is significantly related to political orientations: increasing levels of education correlate with left-of-centre political orientations. Importantly, however, relatively greater electrodermal increases when viewing aversive stimuli are indeed a strong predictor of right-of-centre political beliefs (b = 12.17; p < 0.01).
Table 1. Predicting political orientations with differential skin conductance reactivity to appetitive and aversive images. Dependent variable is an aggregate of standardized scores on the Wilson–Patterson index, society works best items, a seven-point (more ...)
A parallel but more politically focused test of this hypothesis is afforded by the fact that people who care about politics (such as the group of participants being analysed here) are likely to find visible political figures to be either appetitive or aversive. Though there could be numerous reasons for a politician to be viewed favourably or unfavourably, an important factor for most politically attuned individuals is the degree of ideological similarity between themselves and the politician in question. A politician with an ideology that is consistent with that of the respondent is more likely to be viewed as appetitive, whereas a politician with an ideology that is inconsistent with that of the respondent is more likely to be viewed as aversive. Given the results in , we hypothesized that the electrodermal responses of individuals on the right would be greater, in relative terms, to ideologically dissimilar politicians, whereas the electrodermal responses of individuals on the left would be greater, in relative terms, to ideologically similar politicians. Kaplan et al
] report that neural activity in the dorsolateral pre-frontal cortex, the anterior cingulate cortex and the insula increased when political partisans viewed images of candidates from the opposing party (compared with images of the favoured party), but these researchers did not analyse partisan groups separately, meaning it is unknown whether activation to the opposing party was more noticeable among those on the right than among those on the left.
Images of well-known American political figures were included in the 33 stimuli presented; specifically, pictures of Ronald Reagan, Bill Clinton, Hillary Clinton and George W. Bush. These four seem appropriate since at the time of the study (summer of 2007), Barack Obama, John McCain and Sarah Palin had not yet arrived on the national political scene and pre-tests indicated that many participants could not identify pictures of other important national politicians, including (then Vice President) Richard Cheney, (then Speaker of the House) Nancy Pelosi and the previous two Democratic presidential nominees: John Kerry and Al Gore.
To test the hypothesis that left-of-centre participants respond more to ideologically similar (appetitive) politicians while right-of-centre participants respond more to ideologically dissimilar (aversive) politicians, a definition of ideological similarity is necessary. The ideology of the aforementioned four politicians is relatively easy to categorize. As of mid-2007, Bill and Hillary Clinton were nationally visible politicians associated with the left, just as George W. Bush and Ronald Reagan were highly salient touchstones of the right. Indeed, polls at the time suggested that George W. Bush and Hillary Clinton were the most polarizing political figures in American politics—and Ronald Reagan and Bill Clinton, as former two-term presidents on quite different sides of the ideological ledger, were still able to incite passions. The ideology of those participating in the physiological exercise was assessed with the same composite measure as before and the central measure of physiological change was again mean increase in the participants' SCLs from the preceding inter-stimulus interval to the images (pictures of politicians) in question.
presents the electrodermal response of participants on the left and participants on the right to images of politicians either ideologically similar to or different from the participant. As expected, the pattern of responses is similar to what was observed in . The increase in electrodermal activity of right-of-centre participants is greater for politicians with whom they are in ideological disagreement than for politicians with whom they are in ideological agreement, whereas the electrodermal activity of left-of-centre participants is greater for politicians with whom they are in ideological agreement than for politicians with whom they are in ideological disagreement. This was confirmed by a 2 (image type: appetitive versus aversive) × 2 (ideology: left versus right) ANOVA as there is a significant interaction between image type and ideology (F = 10.86; p < 0.01) but no other significant effects or interactions. Thus, whether the focus is on generically aversive/appetitive stimuli or on specifically political stimuli, the results suggest that individuals on the left are more responsive to appetitive relative to aversive stimuli, while individuals on the right are more responsive to aversive relative to appetitive stimuli.
Mean skin conductance change (in microsiemens) as a function of political temperament (left versus right) and political image type (appetitive versus aversive). Triangles with solid line, right-of-centre; squares with solid line, left-of-centre.
Converting political orientation from a dichotomous to a more appropriate continuous form and adding the same controls as included in underscores these conclusions. As can be seen in , political orientation is strongly predicted by electrodermal response to ideologically similar and dissimilar political figures. The further respondents are to the political right, the more their electrodermal response to negative images tends to outstrip their response to positive images (b = 8.99; p < 0.01). The control variables are, again, not significantly related to political orientation with the exception that increasing levels of education are associated with left-of-centre political orientations. Further tests should be run in case there is something particular about the political images employed here but these initial indications are consistent with the results in and .
Table 2. Predicting political orientations with differential skin conductance reactivity to ideologically similar and ideologically dissimilar political images. Dependent variable is an aggregate of standardized scores on the Wilson–Patterson index, society (more ...)