By Zac Baynham-Herd, Analyst at Ogilvy Consulting’s Behavioural Science Practice


Groups of animals can make crazily smart or critically stupid decisions. In beehives, thousands of individuals collectively switch towards better foraging locations. This behaviour is neither triggered by a control centre, nor enforced by hierarchy. Rather, it results from effective communication and copying – otherwise known as ‘social learning’. These very behaviours which enable wisdom in crowds can, however, also lead to madness in herds. Just ask the shepherds mourning their flocks or investors their bubble stocks. Determined to crack this ancient conundrum, Wataru Toyokawa and colleagues of the University of St Andrews in Scotland, placed human ‘swarms’ under the microscope in work published last month in Nature Human Behaviour.[1]

As social animals, we seek information from others when making decisions. Good marketers perhaps know this best. Our wardrobes, career choices and even political inclinations are all shaped by social observations. As it happens, bees behave similarly. But rather than write online reviews, bees perform a waggle dance when they return to the hive. The better the food, the longer the waggle, and the more followers it attracts as a result.[2] Inspired by this, biologist turned psychologist Dr Toyokawa and colleagues built a model to capture how similar social learning strategies might also shape collective intelligence in humans.

To test their model, the team then ran an experiment. They repurposed an established learning-and-decision problem, called a ‘multi-armed bandit’, as an online slot-machine game. For 70 consecutive rounds, participants chose one of three slot machines to play, receiving pay-outs each time. Given that one slot usually paid out more, and as the decisions of other group members were always visible, players could learn to choose the best one, or just copy others. The catch was that half way through the game, one bad slot suddenly became the best slot, and stayed that way for the remainder of the experiment.

The team found that the more uncertain (harder) the slots, the poorer the players’ decisions, and the later they made the intelligent switch between slots. Using their model, the team then made the discovery that conformity increased in larger groups. The more people exhibited a given choice, the more people followed in tow – in the same way that people might choose apps based on download figures or trust Twitter accounts with large followings. Just as players in bigger groups were slower to respond to the pay-out change, groupthink can prevent market-leaders, politicians or even sports coaches from challenging dogma, dooming them to ‘poor slot’ decision-making.

For difficult tasks and within big groups, Dr Toyokawa concludes, people may rely more heavily on following the herd than learning for themselves. Crucially, such behaviour risks not only personal costs – backing the wrong horse – but societal ills like the spread of fad diets or fake news. Here, it pays to be a black sheep. Conversely, in smaller groups where individuals question prevailing logic, and for easy tasks where others are usually right, the wise decision may be to follow the flock.  

We all want to make better decisions. Previous studies have identified the beneficial roles of thought-diversity, sub-grouping and flat-leadership structures in optimising ideation and problem-solving.[3] According to Dr Toyokawa, stimulating independent thought may reduce the risk of collective madness. But, beyond informing management practices, future insights and similar behavioural experiments could help illuminate consumer behaviour, understand viral trends and help us discover ‘unseen opportunities’.

As online activity grows however, so does the potential for both ‘black sheep’ and ‘herd mentality’ to proliferate. So next time you are buying stocks, or indeed socks, consider whether you are blindly following the herd. Perhaps, as Robert Frost famously suggests, taking the road less travelled by will make all the difference.


[1] Toyokawa, W., Whalen, A. and Laland, K.N., 2019. Social learning strategies regulate the wisdom and madness of interactive crowds. Nature Human Behaviour

[2] Toyokawa, W., 2019. What smart bees can teach humans about collective intelligence. The Conversation

[3] Colvin, G., 2016. Humans are underrated: What high achievers know that brilliant machines never will. Penguin.