key: cord-1006727-30aknq84
authors: Makhanova, Anastasia; Shepherd, Melissa A.
title: Behavioral immune system linked to responses to the threat of COVID-19
date: 2020-07-01
journal: Pers Individ Dif
DOI: 10.1016/j.paid.2020.110221
sha: b888411d8356481be4ea470829703bed271a8e66
doc_id: 1006727
cord_uid: 30aknq84

People possess psychological processes that help them avoid pathogens, which is particularly important when novel infectious diseases (e.g., COVID-19) spread through the population. Across two studies we examined whether trait pathogen avoidance (operationalized as perceived vulnerability to disease; PVD) was linked with responses to COVID-19 and preventative behaviors. In Study 1, PVD was positively associated with stronger reactions to the threat of COVID-19, including increased anxiety, perceptions that people should alter their typical behavior, as well as reported importance of engaging in proactive and social distancing behaviors. In Study 2, PVD was again associated with increased anxiety, as well as more vigilant behavior when grocery shopping, fewer trips to the store, and fewer face-to-face interactions. These associations remained significant when controlling for the Big-5 personality traits. Although the two subscales of PVD (germ aversion and perceived infectability) were often parallel predictors, several differences between the subscales emerged. Germ aversion may be more associated with behaviors whereas perceived infectability with vigilance.

J o u r n a l P r e -p r o o f Behavioral Immune System Linked to Responses to the Threat of COVID-19

The motivation to avoid pathogens underpins myriad aspects of people's psychology (Ackerman et al., 2018) . Although people encounter pathogens daily, pathogens threat is accentuated by situations such as the flu season, or as was happening at the time of data collection: the COVID-19 pandemic. To slow down the propagation of novel infectious diseases like COVID-19, it is crucial to engage in preventative behaviors such as social distancing and strict hygiene practices, and earlier engagement is better. Yet, many disparage these recommendations at the critical, early stages because signs of illness are not evident in their immediate environment. Therefore, understanding factors that predict perceptions of pandemic threat and importance of preventative behaviors is imperative. One such factor may be people's trait pathogen avoidance.

People possess various affective, cognitive, and behavioral processes-often collectively termed the -behavioral immune system‖-that help them navigate their environments in ways that reduce the risk of pathogen contagion (Schaller & Park, 2011) . To the extent that some people are more chronically attuned to the potential threat of pathogens that others, research has demonstrated that there are individual differences in people's trait levels of pathogen avoidance (Duncan et al., 2009 , Tybur et al., 2009 . These individual differences have been proposed to stem from genetic, developmental, and sociocultural factors (Olatunji et al., 2019; Rosin et al., J o u r n a l P r e -p r o o f stronger early responses to the COVID-19 pandemic.

Prior research, however, has primarily examined hypothetical pathogen threats. Only one study to our knowledge focused on pathogen avoidance in the context of an actual threat. Beall and colleagues (2016) linked the Ebola pandemic to downstream processes of pathogen avoidance-more conservative voting patterns (c.f. Tiokhin & Kruschka, 2017; Schaller et al., 2017) . Other researchers, however, may have also noticed increased pathogen avoidance in their data (e.g., Prokosch et al., 2019) . Nevertheless, no prior research has linked trait pathogen avoidance to people's psychological reactions to and behaviors during a pandemic.

Capitalizing on current events occurring at the time of data collection, the present research provides a critical test of the theory that trait pathogen avoidance attunes people to increasing pathogen threat. In Study 1, we surveyed a large, primarily American sample at the beginning of the COVID-19 pandemic and examined whether trait pathogen avoidance was associated with perceptions of the severity of the evolving threat, the importance of engaging in preventative behaviors, and selection of media information. We focused on proactive rather than reactive pathogen avoidance processes (see Ackerman et al., 2018) , because at this time most people were not exposed to others who were sick with COVID-19. We predicted that trait pathogen avoidance-assessed using the Perceived Vulnerability to Disease questionnaire (PVD; J o u r n a l P r e -p r o o f 

On the first page of the online survey, participants were given study information and provided informed consent. 

In addition to age, gender, and race/ethnicity, participants reported their religious beliefs J o u r n a l P r e -p r o o f (10-point scale from 1 Not religious at all to 10 Very religious) and political orientation (10point scale from 1 Very liberal to 10 Very conservative). We only examined political orientation for the U.S. subsample.

We assessed individual differences in pathogen avoidance using the PVD questionnaire (Duncan et al., 2009) , which includes two subscales: germ aversion (GA) and perceived infectability (PI). GA reflects people's affective and behavioral responses to potential pathogens (8 items; α=.74; e.g., -I don't like to write with a pencil someone else has obviously chewed on‖). PI reflects people's general perception about their susceptibility to disease (7 items; α=.92; e.g., -If an illness is going around, I will get it‖). Participants rated their agreement with each statement using a 7-point scale (1=Strongly Disagree, 7=Strongly Agree). The scale was originally developed using a large sample of undergraduate students from Canada and the Netherlands (Duncan et al., 2009) , and has since then been used in studies of young adults in other countries such as Spain (Diaz et al., 2016) , Czech Republic (Prokop & Kubiatko, 2014) , and Iran (Ahmadzadeh et al., 2013) . Thus, it was suitable for the present research on an international (albeit primarily American) sample.

To ensure that we were assessing trait levels of pathogen avoidance, participants were told to think about how they act typically, not necessarily how they would act right now and to think about how they would have answered these items 6 months ago. Although the reliabilities of the two subscales were similar to those reported in past research, we nevertheless conducted a confirmatory factor analysis (CFA) in MPlus to examine how well the data fit the two-factor structure hypothesized. The two-factor solution, χ 2 (89)=973.40, fit the data better than the onefactor solution, χ 2 (90)=2731.13, Δχ 2 (1)=1757.73, p<.001. The two-factor solution fit the data J o u r n a l P r e -p r o o f reasonably well; see supplemental materials for fit indices for both models as well as the factor loadings. Thus, our findings are consistent with prior factor analyses suggesting that GA and PI are independent factors (Diaz et al., 2016) . Indeed, GA (M=4.16, SD=1.10) and PI (M=3.48, SD=1.32) were only modestly correlated (r=.19, p<.001) . Consistent with the theoretical perspective suggesting the subscales assess different aspects of trait pathogen avoidance (Duncan et al., 2009) , research has demonstrated that GA and PI tend to predict different behaviors (e.g., Makhanova et al., 2018; Young et al., 2011) although this distinction has not yet been fully clarified. Consequently, we predicted that PVD would be positively associated with responses to COVID-19, but we did not have strong predictions about whether GA or PI would be associated with the dependent measures (except for choices of media articles, the hypotheses for which are clarified below).

Participants responded to a question asking how the COVID-19 outbreak compares to the seasonal flu using a 7-point scale (1=It is much less of a problem; 7=It is much more of a problem). Most participants reported that COVID-19 was more serious than the seasonal flu, M=5.40, SD=1.34, t(1588)=41.82, p<.001 .

Participants responded to a question asking how necessary it is for them to alter their typical daily behavior to combat the spread of COVID-19 using a 7-point scale (1=Not at all J o u r n a l P r e -p r o o f number between 0 and 100. On average, people were a little less anxious than the midpoint (M=45.33, SD=25.43).

Participants reported whether certain preventative behaviors were important for lowering their risk of becoming sick with the virus. Twelve behaviors were presented randomly.

Participants responded how important did they think it was to (1) wash their hands for 20+ seconds, (2) clean their phone and other objects they touched often, (3) avoid touching their face, (4) have enough food in their house for a 14-day self-quarantine, (5) have enough disinfecting cleaning products, (6) avoid shaking hands with people, (7) wear a face mask when they were in public, (8) avoid international airplane travel, (9) avoid domestic airplane travel, (10) avoid going on cruises, (11) avoid large gatherings, and (12) getting enough sleep at night. Participants responded using a 7-point scale (1=Not important at all; 7=Extremely important). On average, people rated the importance of engaging in these behaviors somewhat higher than the midpoint (M=5.05, SD=1.02).

An exploratory factor analysis suggested that there may be two factors (Factor One: items 1, 2, 3, 5, 12; Factor Two: items 4, 6, 7, 8, 9, 10, 11 ) that were somewhat correlated (r=.42). We performed a CFA using participants (n=1555) with no missing data for the behaviors, GA, PI, and SPA using MPlus. The two-factor solution, χ 2 (53)=609.77, fit the data better than the onefactor solution, χ 2 (54)=1295.85, Δχ 2 (1)=686.08, p<.001. We will refer to Factor One as -proactive behaviors‖ and Factor Two as -social distancing behaviors.‖

Participants were also asked an open-ended question about their actual behavior. Most participants (n=1339) provided a valid answer. Responses were coded for whether the following J o u r n a l P r e -p r o o f behaviors were mentioned: hand washing, concern with touching (face, objects, people), cleaning objects and surfaces, social distancing (broadly construed), stocking up on food, wearing gloves or masks, and other (e.g., eating well, exercising, taking supplements, cancelling travel). For analyses, we considered the total number of categories mentioned (M=2.31, SD=1.22, range: 0 to 6). Additionally, because some participants wrote longer, more detailed responses than others, we analyzed the number of characters in each response as a proxy for level of engagement with the question (M=200.00, SD=199.45, range: 4 to 2000). We used this dependent measure for exploratory analyses.

Participants responded how important it was to avoid going out in public if they were sick (even if they thought it was just a cold) using a 7-point scale (1=Not important at all;

7=Extremely important). On average, people rated the importance of this behavior somewhat higher than the midpoint (M=5.79, SD=1.43).

We asked participants to make two choices regarding online media content they would prefer to read. One question asked participants to choose between seeing a list focused on how to avoid coming into contact with COVID-19 pathogens versus a list focused on how to protect your health during the COVID-19 outbreak. The majority of participants (75.4%, n=1201) said they would prefer the second list. We hypothesized that GA may predict selecting the first list, whereas PI may predict selecting the second list.

Participants were also asked to choose which full article they would prefer to read:

-COVID-19 isn't as deadly as we think‖ or -COVID-19 spread is reaching a new level.‖ Most participants (58.4%, n=928) selected the first article. We hypothesized that PI may predict J o u r n a l P r e -p r o o f selecting the second, more threatening article. We did not have any predictions for GA.

For exploratory analyses, participants completed a scale that assesses situational activation of pathogen avoidance processes. Scale details and results are reported in supplemental materials.

We first report the descriptive associations between demographic variables and GA and PI, as well as their responses to the threat of COVID-19. For brevity, we will use the phrase -responses to the threat of COVID-19‖ for situations in which a variable predicts all 6 dependent measures: perceptions that COVID-19 is a bigger problem than the seasonal flu, perceptions that people need to change their behavior to combat the spread of COVID-19, overall anxiety about COVID-19, importance of both factors of protective behaviors, and the importance of staying home when sick. Correlations are reported in Table 2 .

Although age was not associated with GA, and was unexpectedly negatively associated with PI, older participants reported stronger responses to the threat of COVID-19 than younger participants. Similar to past research (e.g., Duncan et al., 2009; Haidt et al., 1994; Tybur et al., 2011) , women reported higher levels of GA and PI than men. Furthermore, women generally reported stronger responses to the threat of COVID-19. This is particularly interesting given that the current data suggest that men are more likely to experience complications and die compared to women (Sun et al., 2020).

Being more religious was associated with higher GA, but there was no link between religiosity and PI. Religiosity was positively associated with engaging in proactive behaviors Journal Pre-proof J o u r n a l P r e -p r o o f Table 2 Correlations between demographics and key variables.

Variable 2 3 4 5 6 7 8 9 10 11 12 1. Age .09*** .10** .18*** .03 -.07** .07* .14*** .14*** .19*** .18*** .15*** 2. Gender (male = 0, female = 1)

-.13*** .08** .10*** .10*** -.06* 0.05 † .12*** .26*** .12*** .09*** 3. Political orientation .54*** .14*** -.12*** -.23*** -.15*** -.16*** -.05 -.11*** -.10** J o u r n a l P r e -p r o o f (e.g., washing hands and avoiding touching one's face) but not social distancing behaviors (e.g., avoiding travel and large gatherings). People who were less religious were more likely to report that COVID-19 was more problematic than the seasonal flu compared to people who were more religious. In the U.S. subsample, being more politically conservative was associated with higher GA and lower PI, as well as weaker responses to the threat of COVID-19 (except proactive behaviors such as washing hands).

We also examined how the announcement by the World Health Organization labeling 

Results for our primary analyses using multiple regression are reported in Table 3 . 1 GA was positively associated with people thinking that they needed to change their daily behavior to combat the spread of COVID-19, people's general anxiety about COVID-19, the extent to which people rated engaging in protective behaviors as important to them (both factors), and the extent J o u r n a l P r e -p r o o f to which people reported that it is important to avoid going out in public if they are sick (even if they thought it was just a cold). However, GA was not associated with perceiving that COVID-19 is a bigger problem than the seasonal flu.

PI, on the other hand, was positively associated with perceptions that COVID-19 is a bigger problem than the seasonal flu. Likewise, PI predicted people thinking that they needed to change their daily behavior to combat the spread of COVID-19 and people's general anxiety about COVID-19. That is, for these two analyses, both GA and PI were independent, significant predictors. Unlike the association with GA, PI was only associated with the second factor of preventative behaviors (social distancing) but not the first factor (proactive behaviors). PI was also not linked to the extent to which people reported that it is important to avoid going out in public if they were sick. Thus, our predictions regarding article choice were only partially supported.

We followed up the primary analyses to correct for multiple hypothesis testing using the Benjamini-Hochberg procedure with a false discovery rate set to 5% (Benjamini & Hochberg, 1995) ; all of the associations reported as significant above remained significant when using the correction procedure.

We explored whether people's free response data were associated with GA and PI. touching public objects with a paper towel‖).

At the start of the pandemic, trait pathogen avoidance was associated with stronger reactions to the threat of COVID-19, including higher overall anxiety about COVID-19, greater perceptions that people needed to alter their typical behavior, and greater rated importance of engaging in proactive and social distancing behaviors. Although both GA and PI were associated with responses to COVID-19 some differences emerged: GA demonstrated stronger links to behaviors whereas PI demonstrated stronger links to increased vigilance. One limitation, however, was our focus on hypothetical behaviors. Thus, in Study 2 we used more concrete dependent measures.

In Study 2 we examined associations between people's trait pathogen avoidance and behavior during social distancing efforts. We assessed whether GA and PI are linked to people's J o u r n a l P r e -p r o o f behaviors when grocery shopping, whether these behaviors were more vigilant than their typical behaviors, and the number of times in the last 7 days they went out to the store or talked to people face-to-face. Furthermore, we isolated effects of trait pathogen avoidance by measuring and controlling for the Big 5 personality traits.

Participants were recruited online through social media platforms to participate in a 15minute study with the opportunity to win one of five Amazon gift cards valued at $50; 1308 people completed the study. Data were excluded from 12 participants who were under 18. We additionally excluded 59 participants from analyses. Fifty-eight incorrectly answered the same catch question as in Study 1 and one participant told us not to use their data on the last page of the survey. Demographic information for the final sample (n=1237) is reported in Table 4 . Most participants were located in the United States (n=882, 71%).

We conducted a sensitivity analysis using G*Power for our main models: multiple regressions with two predictors. These analyses indicated that we had 80% power to detect effect sizes larger than r=.08.

Data collection occurred from April 7, 2020 until April 16, 2020. On the first page of the online survey, participants were given study information and provided informed consent. Then, participants completed 6 blocks of questions presented in random order. 2 Table 4 2 One additional block assessed sleep disturbances and was for a separate research hypothesis. 

As in Study 1, participants reported their religious beliefs and U.S. participants reported their political orientation on 10-point scales.

As in Study 1, participants completed the PVD questionnaire (Duncan et al., 2009) J o u r n a l P r e -p r o o f of this change, conducted a CFA to verify that the two-factor solution was still appropriate. The two-factor solution, χ 2 (89)=963.79, fit the data better than the one-factor solution, χ 2 (90)=2653.05, Δχ 2 (1)=1689.26, p<.001. The two-factor solution fit the data reasonably well; see supplemental materials for fit indices for both models as well as the factor loadings.

Participants completed a 50-item scale to measure their trait levels of agreeableness, conscientiousness, extraversion, neuroticism, and openness to experiences (Goldberg, 1999) .

Each personality factor was assessed by 10 items (α's ranged from .78 to .92).

As in Study 1, we asked participants to report their anxiety. This time, however, we asked about their anxiety about COVID-19 as well as their anxiety about social distancing. Participants answered using a slider from 0 to 100. Participants reported anxiety levels around the midpoint for both questions (disease: M=55.91, SD=25.39; social distancing: M=45.93, SD=29.72; r=.43, p<.001) .

Participants were asked to imagine themselves at the grocery store at this moment and imagine how likely they would be to engage in seven behaviors: (1) avoid touching more items than necessary (select the first item you touch), (2) bag your own groceries/go through selfcheckout rather than going to the cashier, (3) wait to go down an aisle until there were no/few customers, (4) wear a face mask, (5) wear gloves, (6), avoid buying foods that aren't prepackaged (e.g., fresh produce), and (7) bring your own hand sanitizer or wipes to use while at the store. Participants responded using a 7-point scale (1=Extremely unlikely; 7=Extremely likely). We created a composite of these behaviors (α=.64). Although the reliability of this J o u r n a l P r e -p r o o f composite was not ideal, removing any one item did not increase the reliability so we decided to retain all items.

Participants were also asked to consider how their behavior would be different from their typical behavior: -Taking into account your usual behavior at the grocery store, how much more careful would you be right now when it comes to avoiding germs?‖ Participants responded using a 7-point scale (1=Much less careful than usual; 7=Much more careful than usual). On average, all participants reported that they would be much more careful than usual (M=6.41, SD=0.90), one sample t(1233)= 94.25, p<.001. We predicted that PVD scores would be positively associated with being more vigilant than usual; we hypothesized that PI may be significantly associated with behaving differently whereas GA may not be.

Participants were asked to respond how often they engaged in four different behaviors in the last 7 days: (1) -how many times have you left the house to go to a store (grocery store, pharmacy, home improvement store, etc.),‖ (2) -how many times have you picked up food from a restaurant or had food delivered to you,‖ (3) -how many times have you left the house to go for a walk outdoors,‖ and (4) "how many times have you had a face-to-face conversation (even if 6ft apart) with someone who does not live with you.‖ For the last item, we elaborated that a friend or family member would be a relevant example but a cashier or a coworker (if still going to work)

would not be a relevant example. We also assessed whether participants had a dog and used the binary variable as a covariate for analyses focusing on the number of days participants went for a walk. Participants answered on a 9-point scale (1=zero, 9=eight or more) and responses for each question ranged from 1 to 9. On average, participants reported the following frequency of 

Participants reported how likely they were to engage in various behaviors tomorrow and how many toilet paper rolls they had at their house. Details about these measures and results are reported in supplemental materials.

Our primary analyses are multiple regression models in which the dependent measure is regressed onto both GA and PI simultaneously. We also conducted ancillary analyses where we included a number of covariates: agreeableness, conscientiousness, extraversion, neuroticism, openness to experience, whether participants were living in the United States (0=not in United

States; 1=in United States), and whether participants were under stay at home order (0=no stay at home order or unsure, n=225; 1=under stay at home order, n=1010). In text, we mention if the associations with PI and GA were different between the primary and ancillary analyses; see supplemental materials for full results from the ancillary analyses.

The results of the primary analyses are reported in Table 5 . GA and PI were associated with anxiety about COVID-19. In terms of anxiety about social distancing, only GA was associated with anxiety about social distancing. Both GA and PI were positively associated with more vigilant behaviors at the grocery store; however, only PI was associated with people saying that they would be more vigilant at the grocery store than they are typically. These associations remained significant in the ancillary analyses.

In terms of behaviors in the past 7 days, PI was negatively associated with the number of J o u r n a l P r e -p r o o f times participants went to the store and went for a walk. GA was not associated with either behaviors. However, GA was negatively associated with the number of face-to-face interactions in the past 7 days whereas there was no association between face-to-face interactions and PI.

Neither GA nor PI were associated with ordering food at restaurants. 

Across two studies, trait pathogen avoidance was linked to stronger reactions to a real pathogen threat and greater proclivity to engage in preventative behaviors that both protect the individual from contagion and help slow the spread of the virus. Study 1 documented people's responses during the initial stages of the pandemic. Trait pathogen avoidance predicted people taking COVID-19 more seriously: PVD was positively associated with increased anxiety, perceptions that COVID-19 was more serious than the seasonal flu, and how important people thought engaging in preventative behaviors was for limiting virus propagation. Study 2 documented people's behavior several weeks later when most people were under stay-at-home orders. Trait pathogen avoidance predicted greater compliance with social distancing efforts:

PVD was positively associated with vigilant behavior at the grocery store, number of times people left the house to go to the store or for walks, and the number of face-to-face interactions people reported having in the last 7 days. The associations between PVD and social distancing behaviors emerged when controlling for conscientiousness, extraversion, and neuroticism, highlighting the importance of individual differences specific to pathogen avoidance.

These data are consistent with prior research suggesting that people's pathogen avoidance psychology reacts to actual disease threats (Beall et al., 2016) . Critically, findings extend the pathogen avoidance literature by demonstrating a link between trait pathogen avoidance and individuals' perceptions of a real-world threat, as well as the importance of engaging in hygiene and social distancing practices necessary to combat the contraction and spread of the disease.

Furthermore, our findings are consistent with recent evidence suggesting that GA and PI are associated with differential behavioral outcomes (e.g., Makhanova et al., 2018; Young et al., 2011) . Indeed, although GA and PI often demonstrated parallel associations, at times differences between subscales emerged. In Study 1, GA was associated with perceptions of the importance J o u r n a l P r e -p r o o f of proactive protective behaviors and staying home if one is sick, as well as greater preference for information about how to avoid germs (rather than protecting their health). In Study 2, GA was associated with fewer face-to-face interactions during social distancing and greater anxiety about social distancing. Taken together, GA seems to be more strongly associated with behaviors than PI. On the other hand, in Study 1, PI was associated with perceptions that COVID-19 was a greater threat than the seasonal flu and the desire to read an article that talked about the increasing risk of COVID-19 (rather than an article downplaying the risk). In Study 2, PI was associated with a greater change from typical in vigilant behavior at the grocery store and fewer trips outside of the house (to either the store or to walk outside). Taken together, PI seems to be more strongly associated with increased monitoring and vigilance than GA. Notably, both PI and GA were associated with responses to the threat of COVID-19 even though people were not reacting to an actual, visibly sick person in their environment. Findings are thus consistent with the notion that pathogen avoidance promotes proactive strategies aimed at mitigating pathogen transmission (Ackerman et al., 2018) .

It is important to acknowledge several limitations of the present research. First, although we recruited online using platforms that have wide dissemination capabilities, our participants tended to be relatively young, primarily American, primarily female, mostly White, and largely liberal in their political orientations. 3 Thus, we did not have the range to fully examine how different cultural and demographic variables may have affected people's responses to the threat of COVID-19, and how those factors may influence trait pathogen avoidance. Future research would benefit from increasing sociodemographic diversity of samples in order to examine such 3 We did measure other demographic information that were not analyzed for or reported in the present manuscript. The study codebooks and data are publicly available on OSF (blinded for review: https://osf.io/ja86s/?view_only=43c2a30e67ea4672b459db687fa3dc63).

J o u r n a l P r e -p r o o f associations more systematically.

We found that people's trait levels of pathogen avoidance, both germ aversion and perceived infectability, predicted responses to the threat of COVID-19 and social distancing behaviors. However, it remains largely unknown what shapes people's trait pathogen avoidance levels. Research suggests that various factors may be important: past history of illness (Makhanova et al., 2020) , cultural and developmental factors (Rosin et al., 2008) , as well as genetic factors (Olatunji et al., 2019; Sherlock et al., 2016) . Importantly these factors may be different for men and women. For example, Sherlock and colleagues (2016) studied female twin pairs and found that approximately 50% of the variation in disgust sensitivity is due to genetic components. However, a recent study that examined both female and male twin pairs, suggests that disgust sensitivity may not have genetic components for men (Olatunji et al., 2019) .

Considering women tend to have higher overall levels of disgust sensitivity than men (Duncan et al., 2009; Haidt et al., 1994; Tybur et al., 2011) , it is important to examine whether other factors that shape people's trait pathogen avoidance may be moderated by sex or gender. Overall, because trait pathogen avoidance predicts important outcomes for social behavior during a pandemic, it is important to understand the determinants of variance in this trait.

Social behavior during a pandemic may also be perceived and enacted differently by men and women. In our samples, women tended to take COVID-19 and social distancing more seriously than men. Meanwhile, medical research demonstrates that men are more likely to experience worse consequences of getting sick with COVID-19 (Sun et al., 2020) . It is thus important to understand what factors may influence men's and women's diverging perceptions of important social distancing behaviors. Some factors may be biological, stemming from greater evolutionary pressures to avoid pathogens faced by women compared to men. Over the J o u r n a l P r e -p r o o f evolutionary past, women spent a large proportion of their lives taking care of children, and maternal illness or mortality was of tremendous cost for offspring survival (Campbell, 2013) .

Alternatively, some of the factors may be due to sociocultural norms reflecting femininity and masculinity (Courtenay, 2000) . Men may see observable social distancing behaviors as less masculine, because often masculinity is associated with taking risks rather than being cautious.

Masculinity norms tend to promote unhealthy behaviors across many domains (Courtenay, 2000) . Although in exploratory analyses of Study 2 we did not find that the associations between dependent measures and perceived vulnerability to disease were moderated by sex, 4 it is nonetheless important for future research to disentangle how evolutionary and sociocultural pressures may affect men's and women's health risk behaviors.

Another factor that may affect both trait pathogen avoidance and health behaviors is life history strategy. Research using Life History Theory has demonstrated that early childhood environments affect the ways that people allocate their time and energy between goals relevant to reproduction, survival, and somatic development (Belsky et al., 1991; Ellis et al., 2009; Del Guidice et al., 2016) . Specifically, harsh and unpredictable early environments tend to promote a -fast‖ life history strategy associated with the allocation of greater resources toward earlier reproduction, and thereby the allocation of fewer resources toward long-term somatic development. More plentiful and predictable early environments on the other hand tend to promote a -slow‖ life history strategy associated with delayed reproduction and the allocation of greater resources toward somatic development. Recently several theoretical arguments have outlined ways in which early environments may affect pathogen avoidance behaviors (Hill et al., 2016; Ackerman et al., 2018) . Moreover, research informed by life history theory has J o u r n a l P r e -p r o o f demonstrated that situations connoting increased pathogen threat promote less risk-taking (Prokosch et al., 2019) . To the extent that life history strategy may sensitize people to pathogen threats and affect their responses to those threats, people from harsh and unpredictable early environments may respond differently to pathogen threats that people from plentiful and predictable environments. We did not assess any markers of life history strategy so we are unable to answer these questions with the present data. Thus, it would be imperative for future research to examine how early environments may sensitize people to the threat of potential pathogens and how life history strategies may affect people's social distancing behaviors during situations of acute pathogen threat like the COVID-19 pandemic.

Finally, the link between trait pathogen avoidance and stronger reactions to the threat of COVID-19 has broad implications, which were not directly assessed in the present studies.

Pathogen avoidance is linked to many social behaviors including decision-making, morality, and prejudice (Ackerman et al., 2018) . Therefore, the upregulation of pathogen avoidance processes described in these data may correspond to problematic downstream behaviors that emerged during the beginning of the COVID-19 pandemic, such as toilet paper hoarding 5 and negative social behavior toward people of Asian descent. Thus, the downstream effects of the reported associations are likely widespread beyond individuals' self-isolation.

Across two studies, trait pathogen avoidance was associated with taking the threat of COVID-19 more seriously and with greater engagement in social distancing. Thus, the findings buttress theories of pathogen avoidance by illustrating that trait pathogen avoidance predicts reactions to the real, looming threat of a pandemic. Given that trait pathogen avoidance was associated with greater vigilance toward a rather invisible threat (because most participants did

The behavioral immune system: Current concerns and future directions

The investigation of validity and reliability of a scale of perceived vulnerability to disease in Iran

Infections and elections: Did an Ebola outbreak influence the 2014 US federal elections (and if so, how)?

Childhood experience, interpersonal development, and reproductive strategy: An evolutionary theory of socialization

Controlling the false discovery rate: a practical and powerful approach to multiple testing

Italy's coronavirus crisis could be America's

The evolutionary psychology of women's aggression

Constructions of masculinity and their influence on men's well-being: Journal Pre-proof of gender and health

Life history theory and evolutionary psychology

Perceived vulnerability to disease: Development and validation of a 15-item self-report instrument

Fundamental dimensions of environmental risk

Sex, germs, and health: pathogen-avoidance motives and health-protective behaviour

Individual differences in sensitivity to disgust: A scale sampling seven domains of disgust elicitors

Vulnerability to disease as a predictor of faster life history strategies

Binding together to avoid illness: Pathogen avoidance and moral worldviews

Childhood Illness as an Antecedent of Perceived Vulnerability to Disease

Implications of life-history strategies for obesity

Sex differences in the etiology of disgust sensitivity: A preliminary behavioral genetic analysis

Perceived vulnerability to disease predicts environmental attitudes

Caution in the time of cholera: Pathogen threats decrease risk tolerance

Handbook of emotions

Evidence that an Ebola outbreak influenced voting preferences, even after controlling (mindfully) for autocorrelation: Reply to Tiokhin and Hruschka

Implications of the behavioural immune system for social behaviour and human health in the modern world

The behavioral immune system (and why it matters)