In the fourth edition of the Campaign Lab academic series, we welcomed Dr. Lee de Wit, Assistant Professor of Psychology at the University of Cambridge, to explore one of the most urgent challenges in politics today: polarisation. Drawing on Experimental Psychology and Cognitive Neuroscience, and lessons from large-scale UK surveys, Dr. de Wit offered practical and thought-provoking insights for campaigners aiming to better understand - and communicate across - ideological divides.
Kicking off with the famous “white and gold or black and blue” dress illusion, Dr. de Wit illustrated a key psychological insight: we often assume we share the same perceptions as others, but in reality, even basic interpretations of the world can differ widely. This analogy sets the stage for understanding how people with different political beliefs may not just disagree on values - they may process the same information in fundamentally different ways.
Many see the dress as black and blue but others see it as blue and gold but it's significant not only do we perceive the dress differently we are also surprised by the difference in perception we cannot intuitively understand these perceptual differences.
We see the illusion differently but we may not know why we see it differently and our intuition is to suggest that those who see the dress differently must be deficient in their epistemic framework. In fact though these perceptual differences result from a complex neurological and visual phenomenon. If this can happen in colour perception why should we assume that these subtle differences do not reach far beyond.
Dr de Wit also argued that social media also plays a key role in this social phenomena. The dress became a global controversy precisely because of its polarising nature. Controversy attracts attention and this mirrors the power of politically polarised content online.
While group loyalty and tribalism play a role in political polarisation, Dr. de Wit emphasised that people are often being rational when they discount information from opposing sources. His research finds that perceptions of source bias - and whether different sources are seen as independent or merely echoing each other - strongly influence how people update their beliefs.
In other words, when people reject a claim from “the other side,” they may not be irrational - they may simply believe the source is not credible or is part of a coordinated message network.
Dr de Wit Argues that this can be a positive for campaigners. He highlighted the example of a recent academic paper that showed that an AI designed to talk to Conspiracy theorists had unprecedentedly high levels of persuasion.1 The key takeaway from this being that most people are reasonable and can be persuaded to engage with the right messaging.
Dr. de Wit shared findings from a long-running UK polarisation tracker, revealing that:
Issue-based polarisation has remained relatively stable since 2020.
Emotional or “affective” polarisation - dislike of political opponents - is often stronger on key issues like immigration than between party identities.
Perceived bias and interdependence of information sources are major predictors of how polarised someone becomes on an issue, even when controlling for partisanship or ideology.
Notably, Labour and Conservative voters both believe their own side’s sources are more independent and trustworthy than those of their opponents. This “information silo” perception can drive deeper divergence over time. Due to the lack of trust in sources from the other side of a political divide creating reasons to distrust future output in a vicious circle.
Dr. de Wit offered several practical takeaways for campaigners:
1. Make the abstract tangible: people are more open to changing their minds when messages connect to their lived experience (e.g., GP appointment wait times, potholes) rather than abstract economic statistics.
2. Show reasoning, not spin: persuasion is more likely when people believe you’ve reached a view independently. Overly coordinated party messaging may backfire if it signals bias.
3. Broaden your messenger network: campaigns can benefit from endorsements — not just featuring friendly voices in Labour-branded materials, but enabling trusted community figures to communicate in their own voice.
4. Don’t mistake disagreement for ignorance: seek common ground and understand the logic behind people’s beliefs. Polarisation often stems from how people judge sources, not a refusal to think.
5. Take strategic lessons from “unbiased” communicators: figures like Martin Lewis command trust not because they’re neutral on every issue, but because they explain their reasoning transparently.
Dr. de Wit’s team continues to investigate how perceptions of bias and information dependence evolve across time and countries, and how they affect trust in science communication on urgent issues like climate and health. His call to campaigners was clear: if we want to reduce polarisation and improve persuasion, we need to meet people where they are — and understand why they believe what they do.
To watch a recording or access the slides, stay tuned to Campaign Lab’s website or sign up to our newsletter. Our next academic session is on May 28th with Leonard Metson, exploring petition strategies and email engagement
------
Durably reducing conspiracy beliefs through dialogues with AI, Costello et al