In or out – how brain activity can predict your vote on Brexit

Surveys and opinion polls are notoriously bad at predicting election results, as a chain of rather unexpected events last year demonstrated. These instruments usually ask people about their explicit attitudes and opinions. Often, however, these "external" proxies are not entirely representative of what a person is really thinking. For example, several studies have shown that implicit attitudes – that is, subtle preferences or biases outside the realm of our consciousness – can be more useful in predicting our future choices.

As scary as this may sound, there is also mounting evidence that our physiological responses can be even more accurate in revealing how we're likely to vote. In a new paper in Social Cognitive and Affective Neuroscience, researchers from Kingston University and the University of Essex have taken a closer look at a voting outcome in the UK that, last year, came as a surprise to a lot of people. Their findings suggest that people's brain responses to statements about the EU were a more accurate predictor of they way they went on to vote in the Brexit referendum than their stated intentions.

The researchers used electroencephalography (EEG), which measures the brain's electrical activity at the scalp, to record a particular brain signal called the N400. These are brainwaves that peak roughly 400ms after a relevant stimulus, such as unexpected words in a sentence (e.g., "I drink my coffee with milk and socks"), or even violations of our values. For example, a particularly illustrative study found that devout Christians showed a strong N400 signal when reading statements that contradicted their core beliefs (e.g., "I think that euthanasia is an acceptable course of action").

The researchers recorded the N400 of 62 participants in the five weeks preceding the UK's EU referendum last year. All participants were eligible to vote, with 18 intending to vote leave and remain, respectively, and 26 saying they were undecided. The three groups had similar years of education and gender composition, but the leave group (mean age 33.1 years) was significantly older than the remain group (mean age 22.8 years).

To elicit N400 brain waves, participants were exposed to both "remain" (e.g., "Free access to healthcare for all EU migrants should be allowed) and "leave" (e.g., "If Britain leaves Europe our quality of life will be enhanced") positions. All statements had been retrieved from official material used by both sides of the campaign, and their number was split evenly between the two positions. To make sure that the statements were politically unambiguous, they had been rated by an independent group of participants and only the 112 strongest were retained for the actual experiment.

Each statement was presented one word at a time, with each word appearing for 200ms followed by a white screen for 300ms, which is a common set-up for EEG experiments. Participants had to indicate consent or dissent with each statement by pressing a key on a keyboard in front of them. There was a clear-cut difference between agreement and disagreement to statements by "leavers" and "remainers", showing that the experiment captured political differences reliably.

The researchers hypothesised that "leavers" and "remainers" would show higher N400 signal strength in response to statements from the political position opposite to their own. And this is exactly what happened, at least among decided voters.  Participants with the intention to vote leave showed a higher N400 signal to remain- compared to leave-associated statements, while remainers showed the opposite pattern.

Next, the authors looked at whether N400 strength was able to predict the actual vote participants casted several weeks later. For this, they contacted the original participants again after the referendum and fed their vote into a statistical model (two participants were excluded because they had not voted in the referendum). And, indeed, the magnitude of the N400 response predicted (i.e. correlated with) the final vote in the referendum. What is even more interesting, this did not depend on whether the participants had already made up their mind when they came in for the EEG measurement. In fact, the N400 brain signal was a significantly better predictor of voting behaviour than participants' explicit agreement or disagreement with pro- and contra-EU statements. This means that even though participants might not have been explicitly aware of their intentions at the time of the experiment, their brain responses were indicative of them having, in fact, already made up their mind.

For the majority of undecided participants, their ultimate vote was to remain (22 out of 25) – which was representative of the London area where they were recruited. This is the study's main shortcoming because we do not know if N400 would have also predicted voting behaviour among undecided people who went on to vote leave. Still, the study offers some very fascinating insights. It shows that under some circumstances physiological parameters provide a more accurate reflection of personal values and in this case, voting behaviour, than people's overt opinions. This has crucial implications for opinions polls and surveys, which in the majority of cases rely on explicit information such as agree/disagree responses. On a side note, but no less fascinating than the study's main results, political information seems to be processed in the brain within as little as 400ms, which is outside our scope of consciousness. This should give pollsters some food for thought.

About the author

Post written for BPS Research Digest by Helge Hasselmann. Helge studied psychology and clinical neurosciences. Since 2014, he is a PhD student in medical neurosciences at Charité University Hospital in Berlin, Germany, with a focus on understanding the role of the immune system in major depression.