Diversity discovers better problems, not just better solutions

November 19, 2021

Solving problems should benefit from a diversity of perspectives, knowledge, values and skills. Disappointingly, however, empirical studies do not always match this expectation. Even though they try and abstract away from systemic and psychological obstacles to diversity which exists in real life, experts in collective intelligence have so far struggled to show that more diverse groups reliably produce better solutions.

In a paper published in Perspectives on Psychological Science on Oct. 4, Justin Sulik, Bahador Baharami and NOMIS researcher Ophelia Deroy explain how to get over this gap.

The benefits of diversity are less likely to show in well-defined problems, such as logical puzzles, pub quizzes, or yes-no political forecasts, which constitute the bulk of tasks tested in the collective intelligence literature. There, knowing the right answer is all that matters. Starting with diverse information can make it more likely that someone in the group has the right answer, but also comes with costs: interactions can be more difficult, slower, and lead to group-think, over-confidence or red herrings.

Instead, the authors’ review of the vast literature on diversity in collective intelligence and new models shows that diversity is most beneficial for complex and open-ended problems. Here, current popular strategies might just reflect premature convergence on a suboptimal solution. Diversity can provide the creative impetus to break out of such stagnation, prompting further exploration.

As suggested by Kay and King, in political and economic domains, we are rarely interested in yes-no questions such as, “Will Saad Hariri win the next elections in Lebanon?” or “Will there be a parity between the pound and the euro in January?” Rather, we deal with broad, weakly defined questions such as, “What will happen in the Middle East?” or “How will Brexit affect the economy?” There, getting more diverse perspectives matters for identifying which specific problems to tackle, for deciding how to conceptualize those problems, what factors matter, and what outcomes count as relevant.

Crucially, science is also such a complex and open-ended problem. It is not enough to just make science diverse. The core challenge is to ensure that diversity is cultivated at the right junctures: when raising problems and setting the agenda, and not just when solving them.

Read the Perspectives on Psychological Science publication: The Diversity Gap: When Diversity Matters for Knowledge