In a ‘provocative’ new study, researchers set out to test a strange but intriguing hypothesis.
Could three monkeys predict whether Kamala Harris or Donald Trump will win the US presidential election simply by looking at photos of the candidates?
The idea may seem ludicrous, but the findings suggest otherwise.
Researchers tracked monkeys’ eye movements while showing them hundreds of senate and governor candidate races across the US.
They found the macaques stared longer at the faces of losing US political candidates a majority of the time.
A BANANAS STUDY: Researchers tested whether monkeys could predict the outcome of US political elections, including the 2024 presidential race
They’re detecting something purely based on the picture,’ said co-author Yaoguang Jiang, a neuroscientist University of Pennsylvania.
Jiang and his colleagues have been studying facial preference in monkeys for years.
The pre-print for this most recent study was published this month on bioRxiv and has not yet been reviewed by other scientists.
Previously, they performed ‘monkey Tinder’ experiments in which macaques were shown pictures of monkeys they have seen before.
They found that the monkeys would only glance briefly at a high-status male, but would stare at an image of a low-status male or a female.
This is likely because monkeys tend to interpret staring as a sign of aggression, the researchers explained.
Jiang and his team wondered whether the monkeys would exhibit the same behavior while looking at human faces.
To test this theory, they showed three adult male macaques photos of candidate pairs from US elections for senate, governor, and president while tracking how their eyes moved across the image.
The monkeys tended to fix their gaze on one candidate in each pair.
When shown photos of candidate pairs from nearly 300 senate and governor races from 1995 to 2008, the monkeys stared longer at the losing candidate 54 percent of the time.
Three adult male macaques were shown side-by-side pictures of pairs of candidates from past elections for senate, governor and president while researchers tracked their eye movements
When shown photos of candidates from swing state elections, the monkeys chose the loser 58 percent of the time.
But for presidential elections from 2000 to 2020, the monkeys indicated the loser only 50 percent of the time – which is no more accurate than flipping a coin.
And when shown photos of this year’s presidential candidates, ‘it was a toss-up,’ co-author Michael Platt, a neuroscientist University of Pennsylvania, told Science.
But the monkeys did predict that Trump’s vice presidential pick, Senator J. D. Vance (R–OH), would lose if he ran face-to-face against Minnesota Governor Tim Walz (D), Harris’s running mate.
Altogether, the researchers’ findings suggest that candidate’s faces convey information that is somehow related to how voters cast their ballot, Platt said.
In general, the monkeys tended to stare longer at losing candidates. But this correlation did not manifest for presidential candidates
The researchers think some of that information could come from the candidates’ facial structure.
For example, the size and shape of a person’s jaw can be an indicator of social dominance, leading the monkeys to stare at the candidate with a weaker jawline.
The team found that on average, winning candidates had 2 percent more ‘prominent’ jaws, based on the proportion of the jaw to the cheek.
But prominent election forecasters are skeptical of the monkey prediction method.
‘How many elections in advance, where the outcome is unknown – as opposed to past elections where the outcome is known – has the method predicted?’ Allen Lichtman, a historian who has accurately predicted the outcome of almost every presidential election over the past 40 years, asked Science.
‘If the answer is none, I’m done. I have no further interest,’ he added.
Gary King, a political scientist who analyzed 28 years of congressional election outcomes to create his own forecasting model, told Science that the relationships that the study found are ‘sort of interesting.’
But he doubts that this monkey method would be any more effective than existing election forecasting methods, which analyze voter income, ideology, prior voting patterns and other key factors.
Platt, however, says that the study illustrates a clear link – whether you believe in the monkeys’ forecasting abilities or not.
‘When it comes to voting – and pretty much all of our other behaviors – there’s a little monkey in all of us,’ he said.