People are generally bad at producing random actions, but now it seems that we are all uniquely bad in our own way. This makes it possible to predict how an individual will act randomly, which could have implications for data security and choosing suitably strong passwords.
Psychologists already know that we struggle with randomness ā ask people to name a random colour, and , while the between 1 and 10 is 7.
To understand if this failure at randomness varies across individuals, at Johns Hopkins University and his colleagues asked 143 people to randomly select numbers from 1 to 9, and separately to choose one of nine boxes displayed in a three-by-three grid. For each task, the participants had to make a total of 250 random choices.
Advertisement
The researchers found they could use a computational model to predict the supposedly random choices of boxes based on that personās choice of numbers, and vice versa. Roughly a year later, when participants were asked back to repeat the tasks, the researchers could still predict their choices. Only 53 of the original participants were able to be included in the final analysis, which showed that theĀ model was around 10 per cent more accurate at predicting the supposedly random selections made 12 months on than chance guessing.
āWe found that using someoneās year-old data, we could predict what theyāre doing, both within a task, and across tasks,ā says Boger. āWe were all really surprised just how stable these behaviours were over time. Itās showing this really rich, idiosyncratic behaviour across people.ā
āThis suggests that our minds contain a process of randomness generation which we tap into when producing ārandomā sequences,ā says at the University of Dundee in Scotland. āBecause this process is stable over time and tasks, if we know someoneās random decisions at one point in time, we may be able to predict their ārandomā decisions in the future.ā
This isnāt necessarily surprising, says at University College London, as the choices are likely to reflect individual differences in cognitive executive control, or the ability to control your information processing and behaviour. For instance, to not say something negative when you instinctively want to involves exercising your cognitive executive control.
However, what is surprising is that the purported randomness can be identified to individuals and over time. Benwell points out this could have significant implications, both in terms of trying to predict how people react in areas such as game theory, but also in the small details of our lives.
āThis could potentially render human-generated passwords, data encryption and/or exploration patterns more predictable than we would hope,ā he says. āIt is important for us to be aware of this so that we can avoid being predictable when we donāt want to be, potentially by offloading such decisions to non-human hardware.ā
PsyArXiv
