Research Article: Adolescents show collective intelligence which can be driven by a geometric mean rule of thumb

Date Published: September 24, 2018

Publisher: Public Library of Science

Author(s): Christos C. Ioannou, Gabriel Madirolas, Faith S. Brammer, Hannah A. Rapley, Gonzalo G. de Polavieja, Dan Braha.


How effective groups are in making decisions is a long-standing question in studying human and animal behaviour. Despite the limited social and cognitive abilities of younger people, skills which are often required for collective intelligence, studies of group performance have been limited to adults. Using a simple task of estimating the number of sweets in jars, we show in two experiments that adolescents at least as young as 11 years old improve their estimation accuracy after a period of group discussion, demonstrating collective intelligence. Although this effect was robust to the overall distribution of initial estimates, when the task generated positively skewed estimates, the geometric mean of initial estimates gave the best fit to the data compared to other tested aggregation rules. A geometric mean heuristic in consensus decision making is also likely to apply to adults, as it provides a robust and well-performing rule for aggregating different opinions. The geometric mean rule is likely to be based on an intuitive logarithmic-like number representation, and our study suggests that this mental number scaling may be beneficial in collective decisions.

Partial Text

Making decisions in groups can greatly improve cognitive performance [1]. This effect is of widespread interest in psychology, management and political science, partly due to the importance of social interactions in society from small everyday decisions to governmental panels deciding issues of policy [2]. Studies in this area of research range from exploring optimal methods to statistically aggregate large samples of estimates [3], to how individuals use information from others [4], to decision making in groups of freely interacting individuals [5]. Despite this extensive research, our understanding is however incomplete as this previous work has been limited to studies on adults; little is known about how collective intelligence develops during the approach to adulthood.

In the type of cognitive task used here, collective intelligence appears already well developed by the pre-teens and stable during adolescence, from at least 11 years of age. The changes in social and cognitive (specifically numerical) skills that occur during adolescence [7–9,12] do not appear to affect collective intelligence during this period of development. Especially when individuals disagreed regarding their (independently formed) initial estimates, the aggregation rule used by the groups to reach their group consensus estimate approximated the geometric mean. To our knowledge, this use of a geometric mean approximation in group consensus has not been demonstrated before, even in adults. Although this rule seems reasonable given the right-skewed overall distribution of initial estimates, aggregation rules such as the arithmetic mean, or even choosing a single one of the initial estimates, would be easier to implement. Studies of adults have in fact shown that individuals tend to choose a single estimate rather than taking the average when given multiple pieces of social information, despite averaging frequently being the more accurate strategy [25]. Our results show that groups of freely-interacting adolescents can implement a ‘wisdom of crowds’, where multiple, independent opinions are processed using a suitable averaging rule. However, further research is needed to determine which factors are needed for adolescents to approximate a geometric mean rule, such as face-to-face interaction [10], and how the pre-existing social friendship network of participants affect consensus decision making. Such an effect may explain differences between results in experiments 1 and 2, as groups were randomly formed across the school year in experiment 1, but participants were allowed to form their own groups in experiment 2. Additionally, although the average improvement in estimates after group discussion was not statistically different between females and males, when performance was measured as a binary ‘improved’ versus ‘not improved’, female participants were more likely to improve their estimate. This is consistent with previous work [26] where it was shown that the proportion of females in a group was positively related to performance in collective tasks, and that this could be explained by higher social sensitivity scores in females.




0 0 vote
Article Rating
Notify of
Inline Feedbacks
View all comments