Social Movements as Parsimonious Explanations for Implicit and Explicit Attitude Change

The Dynamic Unity of Psychology and Society

The connection between individuals and their social world is often split into psychological study of minds and sociological study of societal structures, each discipline addressing one side of this dynamic interplay. Given this dilemma, Solomon Asch (1952/1987) identified attitudes as a vital nexus: “The study of attitudes may open a way to clarification. . . Here is a critical point at which social and personal processes join each other. . . at which social events become personally significant and personal events become of social moment” (p. 593). From an even broader perspective, attitudes can be viewed as one of many intertwining threads between psychological and sociological processes.

In cultural-historical activity theory (CHAT), for example, the unity of consciousness and activity is a fundamental principle, with all psychological activity regarded as inherently social. CHAT originated in the Soviet Union in the late 1920s, with Lev Vygotsky, Alexei Leontiev, and Alexander Luria among its founders, and sought to analyze the mind as it develops in reciprocal connection with societal processes (Chaiklin, 2019). Activity theory centers “the activity of man [sic] who, as a species and as individual, acts to change reality and in doing so is himself changed” (McLeish, 1975, pp. 264–265). For our purposes, this notion provides broad philosophical grounding for investigating the psychological ramifications of collectively acting to change society. This type of reciprocal feedback from activity to cognition is almost never studied in intergroup psychology (Louis, 2009), indicating a need for theoretical development and empirical investigation. The link between political movements and attitude change may serve as a vivid illustration of the interweaving of psychological transformation and collective activity.

More typical than investigations of change in society and attitudes have been models emphasizing social and attitudinal stability. System justification theory, for example, foregrounds the idea that people tend to defend and justify existing social, economic, and political systems, thereby satisfying psychological needs to reduce uncertainty and conflict (Jost, 2019). Other work suggests that institutional structures and sociopolitical configurations are largely stable and thus so are aggregate attitudes (Payne et al., 2017). Although there is an acknowledgment that widespread changes in bias “will require culture-wide changes in social organization and practice” (Hardin & Banaji, 2013, p. 21), little has been explored in terms of concrete social or political forces that could catalyze such a transformation, or mechanisms through which this process might occur.

From Individual Differences to Aggregate Analysis

Until recently, the individual-differences framework adopted in most implicit attitude research has been a barrier to studying societal attitude change. This perspective construes bias as a personal characteristic—like a personality trait—that is largely stable in an individual’s mind. Of late, however, the view that biases are primarily a feature of social contexts has gained ascendance. The “bias of crowds” model (BoC; Payne et al., 2017), for instance, highlights that while individual-level bias is transitory and unstable, aggregate levels of implicit bias tend to be highly consistent over time. Highlighting the short-term volatility of individual bias, a recent meta-analysis found a test–retest reliability of only r = .42 over a few weeks (Gawronski et al., 2017). BoC argues that the seeming contradiction of individual fluctuation with aggregate stability is resolved by conceiving of implicit bias as the result of an environment with a relatively stable level of systemic inequalities that maintains ready accessibility of biased associations and stereotypes. This concept accessibility fluctuates as individuals experience diverse interactions and situations in their daily lives, making implicit attitudes fleeting and idiosyncratic at the individual level, but reflective of a stable overall social environment in aggregate form. In a nutshell, implicit racial bias is taken to be “a cognitive reflection of racialized associations in the culture” (Payne et al., 2019, p. 11694).

These systemic inequalities offer an explanation for robust correlations between aggregate implicit and explicit bias and sociological disparities. For example, aggregate county-level racial bias correlates with racial disparities in birth weight (Orchard & Price, 2017), and poorer health care and health outcomes for Blacks (Leitner et al., 2016). Regional racial bias is associated with lethal force by police against Blacks (Hehman et al., 2018), and the prevalence of national-level gender stereotypes predict gender disparities in Science, Technology, Engineering, and Math (STEM) fields (Nosek et al., 2009). These aggregate-level correlations with structural inequalities are typically larger than individual-level correlations with discriminatory behavior (Greenwald et al., 2009; Oswald et al., 2013; Payne et al., 2017). This suggests that conceptualizing aggregate-level bias as a reflection of structural inequality better illuminates relations between sociological and psychological phenomenon than an individual-differences approach.

Notwithstanding recent criticisms of BoC that argue implicit bias is more likely a noisily measured individual-level construct (Connor & Evers, 2020), aggregate analysis provides a reliable base from which to investigate attitude change in relation to large-scale sociopolitical factors. For example, aggregated implicit and explicit measures show much stronger predictive validity for meaningful real-world outcomes related to racial disparities than individual measures (Hehman et al., 2019). Thus, aggregation, whether by minimizing individual-level measurement error or more effectively capturing the social environment, appears especially promising for investigating the impact of political struggle on structural inequalities and widespread biases.

Continuity and Change in Aggregate Attitudes

If aggregate implicit biases are indeed relatively stable over time, the persistence, transmission, and reproduction of historical inequalities are likely causes. Payne et al. (2019) found that White residents of counties more dependent on slavery in 1860 showed higher levels of pro-White implicit bias 156 years later, arguing that contemporary racial inequalities in these counties resulted from the transmission of pre-Civil War inequalities. Against this backdrop of historical continuity of racial bias and inequality, however, there is also evidence for substantial shifts in racial attitudes within the past few generations (Newport, 2013). Massive shifts in political norms and explicit attitudes are not uncommon in U.S. history, as Charlesworth and Banaji (2019, p. 1) showed,

From Puritan America through the 19th century, same-sex relations were punishable by death; today, same-sex marriage is legal across the United States. In 1958, only 4% of White Americans approved of Black–White marriages; today, 87% of White Americans approve. (Newport, 2013)

We suggest there is much to be gained by investigating such implicit and explicit attitude changes in concert with the political forces that can generate them.

A handful of recent studies have investigated aggregate biases over multiyear periods, and their mixed findings are worth reviewing and reconsidering in the context of social movements. Two of these studies investigated a hypothesized “Obama effect” (Schmidt & Axt, 2016; Schmidt & Nosek, 2010), proposing that former U.S. President Barack Obama, as a high-status Black exemplar with unprecedented visibility, may have caused a societal shift in implicit racial attitudes. The hypothesis cited experimental evidence that exposure to Obama as a counter-stereotypical exemplar reduced anti-Black implicit bias (Columb & Plant, 2011, 2016; Plant et al., 2009). However, after examining national IAT-Race data and accounting for demographic shifts, the studies concluded that implicit and explicit attitudes did not change meaningfully during Obama’s campaign or presidency. In conclusion, the authors suggested that racial attitudes may be less malleable than previously thought. As we will argue later, emphasis on the political figure of Obama, rather than on an anti-racist social movement that arose late in his presidency—Black Lives Matter—appears to have obscured a genuine reduction in racial bias.

In another relevant study, Westgate et al. (2015) found an aggregate reduction in U.S. implicit and explicit anti-gay bias between 2006 and 2013. Speculating about potential causes, the authors suggested legal changes (e.g., legalization of same-sex marriage) and more favorable media representations of gay and lesbian people as likely contributors. Examining a slightly longer span (2005–2016), Ofosu et al. (2019) found associations between state-level legalization of same-sex marriage and aggregate reductions in U.S. anti-gay implicit bias. Although bias had been decreasing prior to this legislation, it declined even more sharply afterward. The proposed mechanisms were new social norms signaled by the legislation, to which people were motivated to conform by changing their attitudes and behaviors. The study suggested that attitudes and legislation may be mutually reinforcing, such that changing attitudes may have spurred the legislation, which in turn further reduced bias. However, the notion that public attitudes are sufficient or even a key spur for new laws is complicated by the myriad of political issues around which attitudes have shifted even more dramatically (e.g., support for Medicare for All or raising the federal minimum wage), but which have decidedly not resulted in new legislation. The study attempted little explanation for why anti-gay bias was declining prior to the passage of same-sex marriage, but we suggest that the large and vibrant marriage equality movement during this period is a strong possibility.

Finally, in the most comprehensive study to date of aggregate attitude change, Charlesworth and Banaji (2019) examined IAT data for several different social categories from 2007 to 2016 in the United States. Although explicit attitudes became uniformly more egalitarian, implicit attitudes showed a mixed pattern. Implicit attitudes toward sexual orientation, race, and skin-tone became less biased, attitudes toward age and disability did not change, and attitudes toward body weight became more biased. The study examined hypotheses derived from theories of attitude strength and change (Petty & Krosnick, 1995), finding attitudes that changed had lower levels of initial bias, stronger implicit–explicit (IE) correlations, and higher perceived societal priority than more stable attitudes.

To summarize, the causal explanations offered for attitude changes in these studies are: (a) the proposed mechanism failed (Obama effect); (b) changing media portrayals and norms signaled by local legislation (anti-gay bias reduction); or (c) the attitude’s strength and social priority (sexual orientation, race, skin-tone, age, disability, body weight). Despite reference to historical and contemporary examples of large, progressive attitude shifts, none of these studies mention or empirically incorporate social movements—a conspicuous potential driver of such changes. Contesting structural inequalities and transforming attitudes has long been a central goal of social movements. As a legal scholar put it,

The structural diagnosis of inequality, exclusion, and subordination—and the concurrent need to radically reshape those structures—is not a new realization among either scholars or social change activists. Indeed, this structural lens is at the heart of some of the most ambitious, transformative, and egalitarian movements in American politics. (Rahman, 2018, pp. 123–124)

Because concrete political movements remain largely in the background within psychology, we believe that theoretically elaborating their central characteristics will facilitate psychological investigation.

Defining Social Movements and Their Key Characteristics

The interdisciplinary study of social movements offers valuable tools for framing psychological enquiry into attitude change. Distilling this work, the foundation of social movements can be conceptualized as contentious collective action (Tarrow, 2011). Such action is contentious in that it tends to be wielded by people who lack institutional power, makes new or unaccepted claims on authority, and fundamentally challenges others. This activity rises to the level of a movement when ordinary people coordinate their actions around shared demands in sustained confrontation with political opponents. Thus, social movements can be defined more precisely as “collective challenges, based on common purposes and social solidarities, in sustained interaction with elites, opponents, and authorities” (Tarrow, 2011, p. 9; cf. Tilly & Wood, 2009). Although not all movements succeed in winning their intended changes, those that do characteristically create social, political, and economic disruptions that are powerful enough to compel elites to grant reforms (Piven & Cloward, 1979). Contentious, disruptive activity can engulf an entire society, creating a cycle of contention (Tarrow, 2011, p. 16) that when organized and channeled toward overturning power can lead to revolution.

This theory of social movements features four empirical pillars: collective political challenge, common purpose, social solidarity, and sustained political action. We offer this as a working definition that is open to modification through developments in social movement theory or empirical evidence accumulated within this framework. For psychologists adopting this working definition, social movements should meet all four of these criteria. Figure 1 depicts the overall process through which these four pillars of social movements influence attitudes among direct participants, movement supporters, neutral bystanders, and movement opponents. The model provisionally specifies the empirical pillars of movements thought to relate to each mechanism of psychological change; these mechanisms and their supporting evidence will be elaborated later in the paper. In general, a movement should have a greater effect the stronger its contentious challenge, the more cohesive its purpose, the greater its social solidarity, and the longer it sustains itself.

OPEN IN VIEWER

Figure 1.

Process by which social movements influence attitudes of participants, supporters, bystanders, and opponents.

Regarding the first pillar, movements propel contentious challenges, usually through disruptive direct actions against authorities, elites, other social groups, or sociocultural norms. This may involve protests, sit-ins, boycotts, demonstrations, speak-outs, interrupting public events, blocking traffic, street theater, strikes and other job actions, or even running movement-connected candidates in elections. These actions may be designed to win new supporters, shift public opinion on key issues, increase awareness and attention to movement claims and messaging, demonstrate the movement’s strength and commitment, or pressure decision-makers to act on behalf of the movement’s interests. As Tarrow (2011) wrote,

Despite their growing expertise in lobbying, legal challenges, and public relations, the most characteristic actions of social movements continue to be contentious challenges. . . because [movement leaders] lack the stable resources—money, organization, access to the state—that interest groups and parties control. (p. 10)

Second, regarding common purpose, movements arise out of shared or overlapping interests and values among participants. This shared purpose emboldens people to act collectively, even when it entails personal risk or other costs. Third, movements involve social actors consciously recognizing their common interests, and acting in solidarity to protect, defend, and further each other’s empowerment and well-being. Finally, movements must sustain their engagement in political confrontation over time. To summarize, a social movement should entail collective political challenges against authorities, elites, or norms, from people with common goals, expressing social solidarity in ongoing action against political opponents.

Historically, riots, rebellions, and general social turmoil preceded organized movements, and these forms of contentious politics continue today. Riots are not typically considered movements because they often involve only fleeting solidarity and are very short-lived. However, the urban riots of the 1960s or Los Angeles in 1992 arose from a shared sense of injustice and the need to act against racism and police violence, causing an enduring liberal shift in policy support by mobilizing Black and White voters (Enos et al., 2019). Spontaneous actions such as riots or other uprisings may be discrete moments within a larger movement, or indications that a movement is forming (Sherif, 1970; Tarrow, 2011). This seems to have been the case with the early riots in response to the 2020 police murder of George Floyd, which later broadened into a new phase of the Black Lives Matter (BLM) movement.

In U.S. history, the movements that created the deepest sociopolitical transformations succeeded in sustaining struggles against opponents with more resources and connections to power (Tarrow, 2011). Movements have been a potent and recurring force, from abolitionism to the Civil War and Reconstruction, to the labor movement, Civil Rights and Black Power movements, Latinx and Chicano movements, American Indian movements, anti-war movements, women’s suffrage and liberation movements, LGBTQ (lesbian, gay, bisexual, transgender, and questioning) liberation and marriage equality movements, Occupy Wall Street, and BLM. Not only have movements created a more just society, but they have had well-documented impacts on Americans’ explicit attitudes (Zinn, 2015). For example, White Americans’ attitudes toward racial desegregation, interracial marriage, and Blacks in general changed in a progressive direction during and after the Civil Rights and Black Power movements of the 1960s to 1970s (Greeley & Sheatsley, 1971; Taylor et al., 1978). Importantly, the Civil Rights movement created this widespread attitude change despite violent backlash from the state and racist counter-movements (Condran, 1979; Taylor et al., 1978). A recent political science study found progressive effects of nonviolent civil rights activism on public opinion and vote share from 1960 to 1972, especially when activism was met with repression by government or vigilantes, which drove a positive media framing of the movement (Wasow, 2020).

Social Movements Parsimoniously Explain Changes in Aggregate Attitudes

Political movements are one of the few social forces with a historically demonstrated capacity to change public attitudes on a mass scale. Thus, as psychologists increasingly analyze attitudes in aggregate national and regional forms, the potential influence of social movements looms large. Thus far, studies on large-scale implicit and explicit attitude change have posited media representations, social norms, legal and policy changes, and social priority of attitudes as causal factors. We believe that social movements offer a more parsimonious explanation for such attitude changes, and that they can be conceptualized as underlying and propelling many of the intermediate causal factors proposed above.

When discussing evidence for policy changes driving attitude change, Ofosu et al. (2019) cited the finding that after Donald Trump was elected U.S. President in 2016, participants supporting both Clinton and Trump reported increased acceptability of prejudice against stigmatized groups (Crandall et al., 2018). We suggest this finding speaks more to the potency of movements than policy changes. Applying a social movement lens, it is apparent that Trump’s campaign, election, and inflammatory rhetoric—while not a social movement in itself—energized and breathed life into a variety of white supremacist, “alt-right,” and neo-fascist social movements (Blout & Burkart, 2020; Eddington, 2018). Moreover, this effect occurred well before Trump took office and had a chance to enact policy changes (Hawley, 2019). Thus, while Obama’s election had the effect of demobilizing movements of the political left, Trump’s election encouraged movements of the political right. As the bitter fruit of these movements, the Southern Poverty Law Center reported more than 400 bias-related incidents in the week after Trump’s election and more than 1,000 incidents in the following month, a significant spike (Hatewatch Staff, 2016a, 2016b). Further work on right-wing movements could help to substantiate how not just progressive social movements but reactionary movements may fuel attitude change, though in the direction of increased bias and bigotry.

Ofosu et al. (2019) also highlighted legislation and legal rulings as key causal factors in shifting sexual orientation attitudes. They stated that evidence for the impact of legislation on attitudes is somewhat scant and mixed, but they emphasize that the Supreme Court’s 1978 upholding of legalized interracial marriage was followed by increasing support for interracial marriages, suggesting a causal link. However, inspection of the relevant Gallup Poll (Newport, 2013) shows this long linear increase in support actually began years earlier, during the Civil Rights movement of the early 1960s, suggesting a more seminal role than the court decision in propelling attitudes. This suggests that legal and legislative changes may often be better conceptualized as intermediate achievements of successful movements—or as one causal pathway through which movements can operate—than as key causal drivers in and of themselves.

Taken together, Ofosu et al. (2019) and Westgate et al. (2015) discussed cultural/media and legal/legislative changes to account for societal reductions in anti-gay implicit and explicit bias. However, this historical timeframe also included the mass movement for lesbian, gay, bisexual, and transgender (LGBT) rights. Beginning in 2008, days after Proposition 8 in California outlawed previously legal same-sex marriages, tens of thousands of demonstrators across the U.S. voiced support for same-sex marriage (McKinley, 2008). A year later, hundreds of thousands participated in the 2009 National Equality March on Washington, D.C., transforming the national conversation and political tide. Although indeed many legal rulings and legislative victories followed the rise of the LGBT movement, and media portrayals continued to improve, the marriage equality movement has been argued to be the driving force behind these changes (Kowal, 2015). We suggest that psychology can gain by directly exploring the tectonic sociopolitical shifts associated with social movements that often undergird changes in laws, legislation, media representations, social norms, and patterns of societal activities.

Charlesworth and Banaji (2019) provided the strongest evidence of ongoing progressive (and sometimes regressive) shifts in aggregate attitudes. From 2007 to 2016, attitudes regarding sexual orientation, race, and skin-tone became less biased. In contrast, attitudes concerning age and disability remained stable, and body-weight attitudes became even more anti-fat. To account for autocorrelations and seasonality in implicit attitude data over time, the study effectively utilized auto-regressive integrated moving average (ARIMA) models. These forecasting models still leave the problem of explaining changes in trends, however (Jebb et al., 2015). The study found that, consistent with theories of attitude strength, attitudes that became less biased had higher social priority. However, the finding that changing attitudes had stronger I-E correlations contradicted this theory, because strong I-E correlations indicate robust intra-attitudinal structures thought to be more stable. To account for this discrepancy, the study offered an alternative explanation that strong I-E associations may represent attitudes that are frequently discussed and thus better elaborated (Nosek, 2007).

We suggest that a social movement framework also provides a parsimonious explanation for these seemingly contradictory findings. First, social movements tend to raise the social priority of related issues and attitudes. They spur increased discussion—in official political and media outlets and everyday life—that encourages attitude elaboration on a mass scale. For instance, when BLM arose it exploded the notion of “post-racial” America and refocused national political and media discourse on police brutality and racism (Petersen-Smith, 2015). This generated conversations, arguments, and political statements regarding the value of Black lives, the reality of racial oppression, and the virtues of fighting for racial justice. Charlesworth and Banaji (2019) briefly mentioned historical examples of social issues becoming “priorities” in U.S. society, including slavery during the Civil War, women’s right to vote during the suffrage movement, and racial justice during BLM. This reinforces the idea that all of these aspects of attitudes related to change—social priority, elaboration, and IE correspondence—are catalyzed by the mutual influence of social movements on all of them.

A social movement hypothesis also accounts for the finding that attitudes that became less biased over time had lower levels of initial bias (Charlesworth & Banaji, 2019). This is because attitudes related to social issues targeted by recent movements would have a less biased “starting point.” To illustrate, U.S. history has witnessed multiple movements against racism and homophobia, with BLM and the struggle for marriage equality the latest incarnations. From this perspective, the finding that attitudes around race, skin-tone, and sexuality have a less biased starting point is not surprising. A social movement explanation is further supported—in the negative—by evidence that biases around age, disability, and body weight remained stable or became more biased. None of these issues had been addressed by mass social movements in the decades prior to the study. The absence of relevant social movements may also relate to “hollow” attitude changes—identified as reductions in explicit bias but stagnation or increases in implicit bias observed among socially neglected issues (Charlesworth & Banaji, 2019). In sum, social movements are a highly plausible generator of historical and contemporary attitude change, and they also offer an efficient theoretical explanation for the intermediate cultural and attitudinal features identified in these studies.

As a final point of comparison to a social movement explanation, we consider some additional speculations on attitude change and stability offered in Charlesworth and Banaji (2019):

Rapid change in sexuality attitudes may arise from the unique concealability of sexual orientation (Pachankis et al., 2018), enabling positive contact before the stigma is revealed. . . Furthermore, we note that age, disability, and body-weight attitudes involve a perceived but measurable decline of the body and may therefore be seen to have an objective basis. In contrast, race, skin-tone, and sexuality attitudes are not rooted in objective evidence but have emerged for arbitrary and historic reasons (e.g., Sidanius & Pratto, 1999). Such differences in perceived objectivity may contribute to the relative stability of age, disability, and bodyweight attitudes. (p. 17)

We find these suggested sources of attitude change unconvincing. First, the ostensible concealability of sexual orientation cannot plausibly explain rapid change in sexuality attitudes. Not only is it questionable how concealable sexual orientation truly is (Sylva et al., 2010), but its level of concealability has not changed in recent years. What has changed are the social and political circumstances around “coming out,” along with the rights and public respect won by the marriage equality movement. Second, it is true that age, disability, and body-weight attitudes may be perceived as having an objective bodily basis, though conceptualizations and attitudes around these categories are widely culturally and historically variable. In fact, a critical insight spawned by the U.S. disability rights movement of the 1970s is that disability is not fundamentally determined by a person’s body (and many disabilities are not physical), but by whether society is arranged to include or marginalize that person. Another problem is that perceived objectivity of the attitude object also applies to race, skin-tone, and sexuality. Most people mistakenly view race as a biologically real category, visibly evident in skin-tone and other physical characteristics. Sexual orientation is also commonly assumed to be biologically determined. In other words, each of these categories, while socially constructed, are popularly perceived as objectively instantiated in the body. This renders perceived objectivity an inadequate explanation for differences in the attitude trajectories. A social movement hypothesis appears to offer firmer ground for explaining attitude changes, and even for explaining common notions of what is biologically objective versus socially constructed in the first place.

How Might Social Movements Change Attitudes?

A key reason for the potential of social movements to create far-reaching and lasting change appears to be their capacity to create multiple pathways for changing attitudes. As displayed in Figure 1, our survey of the psychological literature identifies six categories of mechanisms through which movements may affect attitudes: remaking societal structures and activities; influencing media representations; creating favorable intergroup contact conditions and affiliation dynamics; encouraging empathy, perspective-taking, and reduced intergroup anxiety; spurring social recategorization; and fostering social identification and self-efficacy processes. These mechanisms are postulated to operate on both aggregate and/or individual levels.

In Table 1, we elaborate the evidence for each mechanism in relation to implicit and/or explicit attitude change. In cases where empirical evidence for a theoretical mechanism is lacking, we write “hypothesized to apply to. . .” the specified attitude (implicit, explicit, or both). We describe how the mechanism is expected to function within movements and whether it applies to participants, supporters, bystanders, and/or opponents. Finally, we identify the relevant literature, along with key concepts, variables, and measures used in the studies. Table 1 and Figure 1 propose levels of analysis (aggregate and/or individual) for each mechanism, based on the supporting literature. These hierarchical levels are both conceptual in terms of plane of casual operation, and empirical with regard to the type of data best suited for analysis of each mechanism. We explicate the mechanism categories in descending order from aggregate to individual levels.

OPEN IN VIEWER

Table 1.

Potential Mechanisms Through Which Social Movements Can Change Attitudes.

Mechanism	Who is influenced?	Attitude changed	Description of hypothesized mechanism in relation to social movements	Relevant literature	Key variables and measures
Changing Society (aggregate level)
Removing environmental cues of social inequality.	Participants; supporters; bystanders; opponents.	Implicit.	Altering social environments in durable ways (e.g., removing Confederate statues) changes available mental associations that temporarily pass through the mind.	Bias of crowds model (BoC; Payne et al., 2017); Vuletich & Payne, 2019.	Variables: Inequality—group disparities in housing, education, income, wealth, social status, criminal justice, civil rights, health, or other outcomes. Measures: Regional sociological measures of disparities, including: lethal force databases (Hehman et al., 2018); public health records including access to health care, disease diagnosis rates, and death rates (Leitner et al., 2016) expenditure per disabled Medicaid enrollee (Leitner et al., 2016); natality records including infant mortality, incidence of preterm births and low birth weight (Orchard & Price, 2017).
Changing social structures.	Participants; supporters; bystanders; opponents.	Implicit.	Because implicit bias is a cognitive reflection of associations stemming from structural inequalities, a movement winning structural changes to a society should result in aggregate bias change.	Implication of BoC model (Payne et al., 2017).
Reducing regional sociological disparities.	Participants; supporters; bystanders; opponents.	Implicit, explicit, and possible latent factor.	Movements reducing sociological disparities (e.g., residential segregation, health, wealth, policing) should reduce aggregate regional implicit and explicit bias (or single latent phenomenon of regional bias).	Implied by correspondence between regional sociological disparities and aggregate implicit and explicit attitudes (summarized in Hehman et al., 2019)
Internalization of new sociopolitical relations and societal activity.	Participants; supporters; bystanders; opponents.	Hypothesized to apply to implicit and explicit attitudes.	Because psychology is the totality of social relations internalized, humans can alter their own psychology as they collectively act to transform sociopolitical relations.	Implied by Vygotsky, 1997 and cultural-historical activity theory.
Media Representation (aggregate level)
Media can amplify a movement’s message and generate political discussion.	Participants; supporters; bystanders; opponents.	Explicit.	Media often highlights a social movement’s slogans, actors, and arguments, raising awareness of its purpose. This catalyzes political discussion and propositions that can persuade people to change explicit attitudes.	Gamson & Wolfsfeld, 1993; Implication of APE model (Gawronski & Bodenhausen, 2006, 2014); Sawyer & Gampa, 2018.	Variables: Traditional print and broadcast media data; social media data. Measures: Traditional media data are measured by references in Factiva and LexisNexis (Sawyer & Gampa, 2018) and NY Times Index (Wasow, 2020). Social media are measured using Twitter data corpus (Giorgi et al., 2020).
Agenda seeding of the media.	Participants; supporters; bystanders; opponents.	Explicit.	Social movement tactics can shift public opinion, discourse, attitudes, and politics by generating (favorable or unfavorable) media coverage.	Wasow, 2020 on agenda seeding.
Media pairs a social group with positive language and imagery.	Participants; supporters; bystanders; opponents.	Implicit.	A movement can drive positive language (e.g., “Black Lives Matter”) and imagery to be associated with a social group (Black people), changing associations with that group.	Implication of APE model (Gawronski & Bodenhausen, 2006, 2014); Sawyer & Gampa, 2018.
Contact and Affiliation (aggregate and individual levels)
Intergroup contact under specific beneficial conditions encouraged by social movements.	Participants; supporters.	Implicit; explicit.	Social movements can encourage contact in conditions of (a) cooperation, (b) equal status during interaction, (c) common goals, and (d) support of authorities, law, or custom, thus changing attitudes.	Allport, 1954.	Variables: Intergroup contact; conditions of contact; affiliation or liking; evaluations of out-group members. Measures: Mere contact and optimal conditions of contact (Allport, 1954; Pettigrew & Tropp, 2008). Perceptions of conditions of contact: equal status, self-revealing interaction, cooperation, equalitarian norms (Gaertner et al., 2016). Positive or negative explicit evaluations of out-group members and ratings of liking for them (used in several studies). Willingness to cross-group boundaries in making friends—strong influences in producing more positive attitudes toward out-groups (Pettigrew, 1998). Self-disclosure to and helping of out-group members among college students; degree of preference for out-group playmates in children (Dovidio et al., 2017).
Additional benefits of contact under beneficial conditions.	Participants; supporters.	Implicit; explicit.	Movements also afford opportunities to (e) get to know other group members personally and to form intergroup friendships, which change attitudes.	Dovidio et al., 2017.
Extended contact encouraged by movements.	Participants; supporters; bystanders; opponents.	Implicit; explicit.	Knowing an ingroup member formed a positive relationship with an out-group member reduces bias.	Wright et al., 1997.
Vicarious contact encouraged by movements.	Supporters; bystanders; opponents.	Implicit; explicit.	Observing ingroup members’ positive impressions of other groups changes attitudes.	Lemmer & Wagner, 2015.
Virtual contact encouraged by movements.	Supporters; bystanders; opponents.	Implicit; explicit.	Intergroup interactions across great distances via social media can shift attitudes.	Dixon et al., 2005.
Synchronous walking during movement actions.	Participants.	Implicit; explicit.	Synchronous walking beside an out-group member reduces implicit and explicit bias against that group.	Atherton et al., 2019
Coordinated movement increases affiliation.	Participants.	Explicit.	Coordinated movements like walking, chanting, and music-making—common at protests—increases affiliation with others, which can shift attitudes.	Hove & Risen, 2009; Wiltermuth & Heath, 2009.
Coordinated movement encourages social recategorization.	Participants.	Explicit.	Coordinated movements—common at protests—can trigger recategorization of others from out-group to in-group members.	Cirelli, 2018; Cross et al., 2019; Good et al., 2017.
Perspective-Taking, Empathy, and Anxiety Reduction (individual level)
Increasing empathy with oppressed groups.	Participants; supporters; bystanders; opponents.	Implicit; explicit.	Movements can increase empathy for a group’s social plight and their lived experiences, which improves attitudes toward that group.	Lepe, 2019; Pettigrew & Tropp, 2008; Sternadori, 2017.	Variables: Empathy; intergroup anxiety; perspective-taking; humanization. Measures: Threat/anxiety response during interracial interactions via cardiac and hemodynamic physiological reactions (Blascovitch et al., 2001). Behavioral measure of anxiety through a cooperative interracial problem-solving activity (Blascovitch et al., 2001). Humanization scale, with human uniqueness and human nature subscales (Bastian & Haslam, 2010).
Reducing anxiety about intergroup interaction.	Participants; supporters.	Implicit; explicit.	Movements create opportunities for positive intergroup interactions, which can alleviate anxiety about future interactions and reduce bias.	Blascovitch et al., 2001; Pettigrew & Tropp, 2008.
Increasing intergroup perspective-taking.	Participants; supporters; bystanders; opponents.	Implicit; explicit.	Movements can increase understanding of out-group perspectives, which lessens anxiety and improves implicit and explicit attitudes.	Aberson & Haag, 2007; Pettigrew & Tropp, 2008.
Hearing movement actors voice their opinions, arguments, and proposals.	Participants; supporters; bystanders; opponents.	Explicit.	Hearing someone voice their opinion humanizes that person in the minds of those hearing it, even if they currently disagree with that opinion.	Schroeder et al., 2017.
Social Recategorization (individual level)
Social recategorization (common ingroup identity model).	Participants; supporters.	Implicit; explicit.	Social movements catalyze the reclassification of formerly out-group members into a new superordinate ingroup category. “Us” and “Them” become “We.”	For a review of research, see Gaertner & Dovidio, 2005.	Variables: Social recategorization; superordinate ingroup; subgroup identities; intergroup attitudes. Measures: Social recategorization is measured via self-reported cognitive representations of intergroup relations and subgroup identities (i.e., one group, two subgroups within a larger group, two separate groups, or separate individuals; Gaertner & Dovidio, 2005). Measures of intergroup harmony or bias (Gaertner et al., 1990); compliance with a request from out-group member (Nier et al., 2001, Study 2).
Forming new ingroup identities reduces bias against former out-groups.	Participants; supporters.	Implicit; explicit.	Recategorization may be especially effective when meaningful subgroup memberships are retained within the superordinate identity (e.g., Black, White, native-born, and immigrant workers within the labor movement).	Arneson, 1994; Gaertner & Dovidio, 2005.
Intergroup solidarity during a larger social movement can shift attitudes.	Participants; supporters.	Explicit.	Multiracial labor unions decrease Whites’ racial resentment and increase support for affirmative action to benefit Black workers.	Frymer & Grumbach, 2021
Social Identity and Self-Efficacy (individual level)
Change in self-definition of movement participants.	Participants	Hypothesized to apply to implicit and explicit attitudes.	Enlargement of personal identity; attitudes shift due to fulfillment and realization of self through movement participation.	Gamson, 1992.	Variables: Social identification; politicized collective identification; self-evaluation; perceived self-efficacy; explicit racial attitudes. Measures: Social identification with movement and collective action (Simon et al., 1998). Politicized collective identity (Simon & Klandermans, 2001). Self-evaluation ratings (liking and positive/negative feelings; Walther & Traselli, 2003). Attraction measures (liking and rating of how enjoyable to work with another person) reviewed in Greenwald & Banaji (1995). Perceived self-efficacy rating scales (Brunstig & Postmes, 2002; Hornsey et al., 2006). 5-item White Guilt Scale (Branscombe et al., 2004). 20-item Attitudes Toward Blacks Scale (Brigham, 1993).
Self-identification with a movement.	Participants; supporters.	Implicit; explicit.	Positive self-evaluation is transferred to groups within the movement with whom one socially identifies. Politicization of collective social identification should intensify this process.	Greenwald et al., 2002; Simon & Klandermans, 2001; Walther & Traselli, 2003.
Holding opinions in common with others in a movement.	Participants; supporters.	Explicit.	Finding you share opinions with others increases mutual attraction and positive intergroup evaluation.	Reviewed in Greenwald & Banaji, 1995.
Raising activist self-efficacy.	Participants.	Explicit	Movements can increase people’s self-efficacy to solve social problems, promoting further activism and associated attitude changes.	Edwards & Oskamp, 1992; Klandermans, 1989; Van Zomeren et al., 2004.
Raising self-efficacy to ameliorate or solve social problems.	Participants.	Explicit.	Whites who engage in anti-racist work enhance their self-efficacy to reduce racial inequality and develop more positive attitudes toward oppressed groups.	Stewart et al., 2010

Note. BoC = bias of crowds; APE = associative–propositional evaluation; NY = New York.

A Differential Impact on Implicit and Explicit Attitudes?

Nearly all research linking attitudes with social movements has measured explicit attitudes, often through public opinion polls. Although self-reported attitudes are subject to social desirability pressures, implicit attitudes—typically considered less controllable—may offer unique insight into the psychological impact of movements. This distinction between implicit and explicit attitudes appears to be reflected in the striking fact that every explicit attitude studied by Charlesworth and Banaji (2019) has trended toward neutral over past years in the United States, while this is decidedly untrue of implicit attitudes. This divergence seems to conflict with the high correlation between implicit and explicit racial attitudes when aggregated over long time periods (13 years) at the state level, and an interpretation of implicit and explicit attitudes as two measures of a broader latent construct: regional racial bias (Hehman et al., 2019). This seeming contradiction should be further explored, as such “snapshots” of regional IE correlations when aggregated across several years may not capture diverging and dynamic trends over time. Alternately, regional trends in implicit and explicit attitudes may differ from national trends. We suggest that the effects of social movements on implicit and explicit attitudes should initially be examined separately to explore differential effects, and potentially as a composite or latent outcome if this appears warranted.

There is reason to think that social movements may affect implicit and explicit attitudes through different underlying processes. The associative–propositional evaluation (APE) model, for example, theorizes that implicit attitude change can result from overall changes to associative evaluations or temporary shifts in activation of preexisting associations (Gawronski & Bodenhausen, 2006, 2014). The APE suggests that exposure to familiar exemplars like Obama may activate preexisting associations (e.g., powerful, accomplished) that temporarily influence associative evaluations of Blacks in general. However, because Obama is a specific individual, activation of these associations may not generalize in a durable way to one’s overall associative structure for Black people. In contrast, a social movement that repeatedly pairs Black people as a social group with positive language (“Black Lives Matter”), positive imagery, and positive traits (courageous, agentic, fighting for justice) appears more likely to systematically change one’s overall associations with Black people, potentially generating more fundamental and lasting change. Regarding explicit attitudes, an anti-racist movement—with its message amplified by media—may catalyze widespread and extensive political discussion, thus generating persuasive arguments about the value of Black lives or the virtue of struggling against racism. APE specifies that such arguments and propositions can change explicit attitudes, and that explicit propositional processes can alter implicit attitudes via top-down construction of new associative evaluations. Equally, changes in implicit associations can have bottom-up influence on explicit attitudes. In an ongoing social movement, it is probable that both processes operate reciprocally, amplifying the potential effect on both types of attitudes. Therefore, implicit and explicit attitudes could be examined in dynamic relation to each other as they change during a movement, to see if implicit attitudes may mediate later changes in explicit attitudes or vice versa.

In terms of available evidence, implicit and explicit attitudes changed during BLM (2013–2016) in the same direction (less anti-Black bias) and to a similar degree. However, there were differences in these attitudes’ trajectory among White and Black subgroups. Although Whites’ implicit attitudes became less pro-White, Blacks’ implicit attitudes did not change. Because Blacks demonstrated little implicit bias prior to BLM (being close to “no preference”) and Whites exhibited greater bias, the larger change among Whites may reflect their greater room for bias reduction. In terms of explicit attitudes, Whites and Blacks showed opposite directions of change, with Whites becoming less explicitly pro-White, and Blacks becoming less pro-Black. Rather than opposing trends, this can alternately be conceptualized as a mutual shift toward a no preference position.

A possible explanation for these patterns is that anti-racist social movements may move all racial groups toward egalitarian attitudes, if they are not already egalitarian. Oppressed groups often express strong explicit ingroup preferences to counteract hostile out-group attitudes, and these ingroup preferences are likely to weaken if societal attitudes become less hostile (Gawronski & Bodenhausen, 2006; Westgate et al., 2015). Thus, decreasing bias among Whites during BLM may have alleviated the need for Blacks to maintain explicit pro-Black attitudes. Future research can explore the possibility that anti-racist social movements, rather than polarizing Whites and Blacks, may be capable of moving both groups toward more egalitarian implicit and explicit racial attitudes. Changes could be examined nationally, and/or using data aggregated regional data, which demonstrates high reliability and validity in relating to regional measures of inequality (Hehman et al., 2019).

Hypotheses, Questions, and Methods for Future Research

The current era of politically volatility appears favorable for an “activist turn” in the study of implicit and explicit attitudes. Social movements offer hope for transforming the sociopolitical terrain, along with the attitudes we form to navigate this terrain, for better or worse. We suggest psychology can further this process by theorizing social movements as parsimonious causal agents and empirically examining their impact on social relations and attitudes. In this section, we lay out some key questions, testable hypotheses, and methodological approaches that may be useful in this work.

Conclusion

Amid public health crises, economic shocks, impending climate catastrophe, and ongoing oppression, social movements raise new possibilities and have a demonstrated record of changing history (Solnit, 2016). Although changes in attitudes should be expected to occur during and after social movements—especially those winning tangible structural changes—Zerhouni et al. (2016) found that cities with lower preexisting aggregate anti-Arab bias had higher rates of participation in rallies to honor Charlie Hebdo victims and support free speech. This suggests that social movements and changes in aggregate attitudes can be reciprocally reinforcing. In one hypothetical causal chain, social movements reduce aggregate bias, which encourages further mass mobilization, which in turn further decreases bias. We recommend future research explore different starting points in various reciprocal causal paths to assess prospects for dynamic interplay between attitudes and social movements. These links may vividly illustrate how psychological transformation is deeply interwoven with collective political activity.