Skip to main content

Verified by Psychology Today

Social Learning and Political Polarization

How social networks can create a wiser, more informed electorate.

In 2013, NASA published the latest data showing historical trends in Arctic sea ice levels over the last 30 years. Arctic sea ice is one of the strongest indicators of global climate trends, and the data show that sea ice is decreasing rapidly. However, to NASA’s dismay, conservative pundits and laypeople incorrectly interpreted the figure as showing that sea ice levels were increasing.

Social media discussions about these findings only appeared to worsen the problem, as opinions on climate change appeared to become increasingly polarized. Contrary to Allport’s “contact hypothesis,” it appeared that the more interaction there was among Republicans and Democrats on the topic of climate change, the more polarized, and incorrect, opinions about climate change appeared to become.

What causes polarization, and how does it affect beliefs about climate change?

In the last couple of years, many researchers have been blaming increasing polarization in people’s beliefs about climate change on the process of “biased assimilation,” in which cross-party interactions reinforce partisan bias, and exacerbate animosity across party lines.

In our recent study published in the Proceedings of the National Academy of Sciences, we found that the opposite is true. We found that bipartisan social media networks significantly increased consensus about climate change – leading nearly 90 percent of both Republicans and Democrats to agree that Arctic sea ice is in fact decreasing. Not only was polarization eliminated, but both groups came away with a better understanding of climate trends.

We also were able to identify the source of the problem in earlier studies of social media. The problem comes from the “priming effects” of political imagery. When we used the exact same social media setup that worked so beautifully to eliminate polarization, but this time included the political party logos (elephant and donkey) on the bottom of the screen, then we found that political polarization remained strong, and people’s opinions about climate trends remained inaccurate.

Our study included 2,400 subjects and was replicated a dozen times using independent social media communities. This enabled us to identify a causal finding. Our results show that the cause of polarization is not cross-party communication, but rather communication in a politicized social context. The priming effects of political imagery cause people to become entrenched in their partisan views.

The main recommendation from our study is that cross-party communications can be vastly improved by removing (or reducing) political symbols and partisan logos from social media settings. This is also true for broadcast news programs in which people discuss climate change. When symbols of partisan identity are removed from the screen, people become much more willing to learn from each other, and as a result, they come to an informed consensus.

Political priming creates emotional arousal and, therefore, makes social media spaces more entertaining and engaging. It also directly reduces the efficacy of these spaces for facilitating social learning and informational accuracy.

What does this mean for science communicators who want to bring about social change?

One of the main problems with communicating about climate change is that even if you do get people to watch your program or listen to your new scientific information, they can nevertheless interpret it in exactly the opposite way that you intended. Giving people information is not enough to change their minds. Partisan bias can still lead them to hold onto their old beliefs.

Social media networks act as a filter that can shape how people interpret and perceive new information. The good news is that communication networks that are designed to create social influences in the absence of partisan priming can harness social learning to significantly improve cross-party exchange of information. If partisan priming is removed from a social setting, then new information that reaches people as part of their democratic discourse can arrive in a way that can effectively change their minds.

As a result of these findings, social media giants like Facebook and Twitter are now starting to take seriously the idea that they might need to design their interfaces in ways that promote improved social learning across the aisles. The challenge they face is that political imagery and contentious interactions are emotionally arousing and highly engaging. In short, they are good for business. The question for the next generation of social media is what these companies should do when what is good for business is not good for democracy.


Douglas Guilbeault, Joshua Becker and Damon Centola. 2018. Proceedings of the National Academy of Sciences. “Social Learning and Partisan Bias in the Interpretation of Climate Trends.” 115 (39) 9714-9719.