Computational psychiatrists have broken new ground in understanding how humans learn, by modelling a software program to behave just behave like us. The study, published in the journal eLife could help explain why people with depression can feel overwhelmed by negative thoughts.
Using a simple shape-choosing task, the study found that humans learn more when events have volatile outcomes. The research was led by Michael Browning from the University of Oxford and his colleague Erdem Pulcu. They investigated whether a tendency to alter how we prioritise the processing of negative and positive events could bias our beliefs. The study is the first to prove that people separately judge how informative good and bad outcomes are. This could mean that people with depression tend to prioritise negative thoughts because they seem more important.
During the study, participants chose between two shapes that flashed up on a screen. Choosing one shape earned them money, the other would take it away. “All you have to do in this task is learn where you think the winner is, so you can win as much money as possible,” Browning explains.
Browning designed the task so either the positive (winning money) or negative (losing money) outcomes were variable. Variable means the outcomes change unexpectedly – for instance getting heads eight times in a row during a coin toss – rather than stably switching between the two.
To better understand what the variability of events means, and why we feel that teaches us more, think about football. Imagine you’re trying to figure out how good your team is. “If your football team is constant and always wins 70 per cent of the matches, then the results of each match doesn’t tell you much. Whatever happens, they've still got a 70 per cent chance of winning the next match,” Browning explains. Yet, if your football team’s success changes often, each match is giving you more information. “If they win, it’s more likely the football teams form is good. So, you pay more attention to the outcome of each match.”
So while a team that always wins being victorious again feels insignificant, a team with more erratic form will have a bigger impact on how a supporter thinks.
In this task, “changeable” means the best thing to choose is changing over time. The variable outcome (which could be the win or the loss) is changing from one shape to another. "You’ve got to learn more when the variable outcome switches sides – it tells you the probability of it has changed, so you have to update your belief,” Browning says.
During this recently published study, Browning programmed a mathematical computer model to complete the same task as the humans. These computer programs had parameters that changed how quickly the computer learned, what it learned and how much it was interested in winning or losing.
The researchers altered these parameters so the computer behaved as much as possible like an individual participant. “We can use these estimates to see how much people learn specifically from wins and losses and how quickly they learn how important wins and losses are,” Browning says. “The computer describes the whole process that is going on in the person's brain.” In short, we pay more attention to events that teach us more. Yet no one has previously shown that people separately judge how informative good and bad outcomes are, and then change their learning to reflect this.
Knowledge of these processes could be used to help understand depression and anxiety. “This study shows that maybe people with depression focus on the negative things in the world because they believe that they hold more information, which would make focusing on them the logical thing to do,” Browning says.
“Now we've shown we can alter how healthy people learn, we are running a study in depressed patients to see whether, if we increase volatility of positive outcomes, it it will make them feel better. It’s just a way of making the positive outcomes more useful, so even though people aren't aware of it they then start to focus on them.”
Browning and Pulcu also monitored the levels of the neurotransmitter noradrenalin, by measuring the size of participants' pupils. Noradrenaline controls pupil dilation, and there is a theory that it acts in the brain to make people learn more from the events they experience.
“We tested whether people learned more from outcomes when they were changeable, and we thought this might be controlled by the noradrenaline system. We found evidence of this for losses, but not really for wins,” says Browning. This result suggests that drugs which change the function of the noradrenaline system might be helpful as a novel treatment for depression.
This article was originally published by WIRED UK
