We often want to know what’s driving other people’s actions. Does the politician who visited a refugee camp on the eve of elections truly care for the poverty-stricken? In reality of course, our mind reading skills are pretty limited and something as complex as an apparent act of altruism can disguise a huge diversity of motives. Most of the time, these motives remain entirely private to the individual – a driving force in a black box.
For a new paper published in Science, however, researchers have prised open the box by decoding patterns of brain activity to reveal the hidden motives underlying people’s altruistic decisions. Participants engaged in a financial game that required them to make choices about how to allocate money between themselves and their experimental partner. A decision could either be selfish, in that it primarily benefitted the participant (eg. £10 for me, £2 for my partner) or altruistic, in that it maximised the partner’s payoff at a cost to the participant (eg. £4 for me, £10 for my partner). Occasionally, participants made self-sacrificial decisions – but why?
The researchers pinned down the participants’ motives for altruism by creating these motives themselves. To do this, they put the participants through two different conditions. Half of the participants were made to feel a sense of compassion towards one of their partners, by having them repeatedly observe this partner receive aversive electric shocks – the researchers reasoned that this would encourage the participants to be generous to this partner out of empathy. The other half of the participants were provoked into feeling a desire to reciprocate their partner’s kindness – they observed one of their experimental partners sacrificing their own profit on several trials in order to prevent the participant him or herself from receiving painful electric shocks. The researchers anticipated that this would encourage these participants to make altruistic decisions towards this partner as a way of repaying the kindness.
Indeed, both motive inductions pulled at the heartstrings – participants playing with a partner towards whom they felt empathy or in debt made more altruistic decisions than when they had to allocate money to a neutral partner. As the participants made these decisions, the research team examined their brain activity using functional magnetic resonance imaging (fMRI). The scans revealed that there was no brain region in particular that became more or less active under the influence of a particular motive. Thus, a quick glance at brain activity couldn’t tell the researchers whether a person’s altruistic decision was rooted in empathy or a desire to reciprocate.
Instead, what did appear to be critically different between the two motives was how various brain regions communicated with each other. However, the ultimate test of the consistency and usefulness of this finding lay with this question: could a computer be trained to use information about these connections to judge whether a person made a selfless choice due to a state of empathy or a wish to reciprocate?
The researchers investigated this by providing a computer programme (a kind of “learning algorithm”) with multiple "learning experiences" – this involved showing it examples of the kind of brain connection patterns that were associated with each type of motive. Crucially, the researchers then measured how often the algorithm was able to accurately identify a person’s motive when it was given a brain scan it had never been exposed to. Using this approach, the computer could predict individuals’ hidden motives with an accuracy of 68 per cent.
Interestingly, the researchers found a remarkable similarity between connection patterns that characterised altruism driven by empathy, and altruism that participants occasionally displayed towards a completely neutral partner. This raises the intriguing possibility that what the authors call "home-grown altruism" – our intrinsic impetus for kind behaviour – is primarily rooted in a sense of compassion.
Now, these findings appear to show that machines can have insight into the richness of human motivation, even when behaviour itself lends us no clues. But let’s temper that excitement! While we might find it intuitively impressive that a machine accurately judged a person’s hidden motive in 68 per cent of cases, we should keep in mind that if it were 50 per cent accurate, it would be no better than random guessing (after all, because of the way the researchers designed things, the computer programme only needed to choose between two possible motives).
As we delve into the realm of real social interactions, hidden motives gain a far greater complexity. People may lend a helping hand due to a sense of compassion, because they expect the return of good karma, as a means of boosting their public image, or perhaps for some other more obscure reason. Faced with such intricate thought processes, no doubt rooted in incredibly complex sets of connections in the brain, the guessing of the machine might be no better than our own! But that remains to be seen.
Hein G, Morishima Y, Leiberg S, Sul S, & Fehr E (2016). The brain's functional network architecture reveals human motives. Science (New York, N.Y.), 351 (6277), 1074-8 PMID: 26941317
How to cheat a brain-scan-based lie detector
First brain scan study to feature THAT dress
Post written by Sofia Deleniv for the BPS Research Digest. Sofia holds a degree in Experimental Psychology and is currently working towards a PhD in Neuroscience at the University of Oxford, where she investigates multisensory processing in the mouse brain. In 2015, she decided to try her hand at science writing by starting her blog 'The Neurosphere'. You can visit her Facebook page or Twitter feed for updates on her written work and other exciting bits of science.
Our free fortnightly email will keep you up-to-date with all the psychology research we digest: Sign up!