In 1903, sociologist W. E. B. Du Bois published a book called The Souls of Black Folk, which marked a turning point not just in sociology but also in the lens through which Americans saw themselves. In writing about his experience as an African American, he coined the term “double consciousness,” which is “the sense of always looking at one’s self through the eyes of others, of measuring one’s soul by the tape of a world that looks on in amused contempt and pity” (8). Although earlier authors had also argued for civil rights, Du Bois was one of the first to turn inwards and examine how an unequal world shaped the self-image of the oppressed. It is in this key respect, he argues, that White and Black Americans diverge – in their existence as a singular or divided self.

By Du Bois’s line of thought, the dismantling of discriminatory structures, although important, is just an outward manifestation of true change – change in consciousness. More than 100 years after its publication, Du Bois’s book has become increasingly urgent. Even as the formal structures of oppression (like racialized bans on voting or segregated school systems) have crumbled, systemic inequalities seem to persist through the unspoken biases of our society. To rectify these biases will require altering our fundamental beliefs.

At this point, a skeptic of the consciousness-change view of social change might beg to differ. What does it matter if the world changes its attitudes? Many people recognize, for example, that being eco-friendly is important. But that hasn’t stopped the world from consuming ever more fossil fuels — we still make petroleum-based plastics, drive gasoline cars, and eat ammonia-supported agriculture. Climate action is just one area among many where our beliefs don’t quite match up to our actions. Indeed, if the goal is to change behavior, changing people’s beliefs can only be part of the story. What else is holding us back?

Beyond our outward beliefs, there are two main factors that could make behaviors resistant to change: implicit beliefs and habits. Regarding implicit beliefs, the most widely gathered data comes from the implicit association test (IAT). The IAT works as follows: first, the participant must classify good and bad words. Then, they classify some category of interest (e.g. “Black” or “White” faces). Finally, in the experimental conditions, they are asked to classify Black and good on the left and White and bad on the right, and then vice versa. As the logic goes, classifying faces co-located with “good” (e.g. Black and good) more quickly implies an automatic preference for that kind of face. The IAT was first introduced in 1998, and has since become wildly popular, used everywhere from schools to offices to governments to measure signs of implicit bias.

Because the IAT has been around for so long, we can use it as a rough measure for the way implicit biases have changed over time. In 2019, Charlesworth and Banaji did just this, using 4.4 million data points from 2007 to 2016. Their main goal was to determine how various biases changed over time, in both explicit (self-reported) and implicit (IAT) responses. They found that, while all explicit responses moved towards neutrality over the study period, only the implicit responses for sexual orientation, race, and skin-tone did the same. Age and disability attitudes showed little change, and body weight attitudes moved away from neutrality.

Assuming you think neutrality towards the above attributes is good, these results imply several encouraging shifts in implicit attitudes. Attitudes towards body weight are the notable exception, and one easy explanation is that it arises from a rise in social media use. Meanwhile, attitudes towards sexuality showed a dramatic shift towards neutrality, paralleling a wave of state-level legal decisions in favor of same-sex marriage and a landmark Supreme Court approval in 2015. This evidence supports the idea that legal legitimacy can drive shifts in both explicit and implicit attitudes. Each of these insights – the effect of social media on body weight attitudes and the effect of legal legitimacy on attitudes towards sexuality – yields actionable consequences for people seeking to change these attitudes.

But before we start making all these inferences, we must ask two questions: first, is the IAT a valid basis for judging “implicit” attitudes? Second, do these attitudes drive real-world behaviors, or do they lag them?

On the first point – whether the IAT is an accurate judge of “implicit” attitudes – it first bears asking why we are interested in implicit attitudes at all. The alternative would be to only consider explicit attributes – that is, the attributes that can be measured in direct reports, involving questions like “do you think you prefer gay or straight people?” The idea behind assessing implicit attitudes is that direct answers to these questions yields an answer that reflects what is societally acceptable more than it reflects how a person will behave. For example, someone might say they have no preference for or against women, and still judge equally competent women as inferior to men at doing a white-collar job.

While the divergence of implicit and explicit beliefs is interesting in theory, there is little evidence that implicit beliefs are better predictors of behavior than explicit beliefs are. Verplanken and Orbell note in their 2022 review article that syntheses of research on implicit beliefs have demonstrated “minimally consequential and highly heterogenous effects” on behavior change. In other words, beliefs can only account for part of behavior change. Just as important is the role of habit – that is, repeated patterns of thought invoked by contexts in a person’s life. Even when beliefs lead a person to want some behavior change, habits often resist that change, as in smokers who unthinkingly pick up a cigarette and light it despite believing in the benefits of quitting smoking.

When habitual contexts are uprooted, people have a unique opportunity for behavior change. In a 2016 study by Verplanken and Roy, households targeted to adopt more sustainable behaviors were more likely to adopt (and sustain) these behaviors when they had moved in the last three months. This supports the idea that the lack of existing habits – such as commuting by car to work – facilitated the process of behavior change.

Research on habitual contexts, and the relationship between attitudes and behavior, has become increasingly urgent in the world of the internet. For many users, checking social media has become as compulsive as a cigarette habit. What are the attitudes and values of a society hooked on the social media habit? The answer to this question likely depends on which values we’re referring to.

Consider, for example, people’s attitudes towards sexuality. Why, in the past 20 years, have we seen such a sudden rise in the acceptance of same-sex marriage in the United States? From a cursory examination, it seems like this acceptance was the result of a generation of gay rights activism, culminating in a cascade of state-wide approvals of same-sex marriage in the early 2000s. This change seems to have occurred before social media became dominant, making it unlikely to be the determining factor.

What about the changes in values around body weight? The 2019 IAT study by Charlesworth and Banaji showed a move towards negative attitudes from 2007 to 2016, which was a shift concurrent with the rise of social media. Although a concurrent change does not imply causation, the specific format of social media – the way thin bodies are presented as representative – does seem conducive to forming such biases.

Another candidate for suspicion is political polarization. Intuitively, we might imagine that social media feeds create filter bubbles that only show people the views they agree with. While this is a believable explanation for political polarization, it is likely wrong. Social media recommends content that maximizes engagement – which, for a conservative, might just as likely be an inflammatory Bernie Sanders tweet as a Fox News article, and for a liberal, might just as likely be a Donald Trump tweet as a CNN article. Indeed, a 2018 study paid conservatives to look at a liberal Twitter feed and liberals to look at a conservative Twitter feed, only to find that each side became more entrenched in their existing views.

Even if the underlying content stems from the desires of content creators, its influence would not be possible without social media. In fact, one might argue that the very format of Twitter, as an engagement-driven platform, demands this kind of value change. In the same way that economic conditions determine our values, informational conditions determine our values. Seemingly mundane things like the monetary policy of the Federal Reserve in 1929 came to fundamentally determine our views towards individual rights, government control, and the social contract. Similarly, seemingly mundane things like the format of a Twitter feed can fundamentally determine our views towards us-them dichotomies, foreign relations, and the rigidity of our morals.

One interesting theory this points to is an analogue for historical materialism called historical informationalism, in which we describe a historical narrative for the modern age in terms of the informational structures that we inhabit. Indeed, this could be quite a prescient story, for in many cases media consumption may drive behavior just as much as material conditions do.

Fortunately for W.E.B. Du Bois (and for society), his practical recommendations, like the universal right to vote and the right to receive a good education, have been accepted as by most Americans as an ideal to strive for. Although our world remains systemically unequal, both explicit and implicit beliefs seem to have moved towards neutrality. As beliefs move in the right direction, the task ahead is to design incentives and habit contexts such that we can align our real-world behavior with the ideals we hold to be true.

References

Du Bois, W. E. B. (2008). The souls of black folk (B. H. Edwards, Ed.). Oxford University Press.

Charlesworth, T. E., & Banaji, M. R. (2019). Patterns of implicit and explicit attitudes: I. Long-term change and stability from 2007 to 2016. Psychological science, 30(2), 174-192.

Charlesworth, T. E., & Banaji, M. R. (2021). Patterns of implicit and explicit attitudes II. Long-term change and stability, regardless of group membership. American Psychologist, 76(6), 851.

Corneille, O., & Hütter, M. (2020). Implicit? What do you mean? A comprehensive review of the delusive implicitness construct in attitude research. Personality and Social Psychology Review, 24(3), 212-232.

Verplanken, B., & Orbell, S. (2022). Attitudes, habits, and behavior change. Annual review of psychology, 73, 327-352.

Verplanken, B., & Roy, D. (2016). Empowering interventions to promote sustainable lifestyles: Testing the habit discontinuity hypothesis in a field experiment. Journal of Environmental Psychology, 45, 127–134. https://doi.org/10.1016/j.jenvp.2015.11.008

Bail, C. A., Argyle, L. P., Brown, T. W., Bumpus, J. P., Chen, H., Fallin Hunzaker, M. B., Lee, J., Mann, M., Merhout, F., & Volfovsky, A. (2018). Exposure to opposing views on social media can increase political polarization. Proceedings of the National Academy of Sciences, 115(37), 9216-9221. https://doi.org/10.1073/pnas.1804840115