Dataset Information

Impact of network structure on collective learning: An experimental study in a data science competition.

ABSTRACT: Do efficient communication networks accelerate solution discovery? The most prominent theory of organizational design for collective learning maintains that informationally efficient collaboration networks increase a group's ability to find innovative solutions to complex problems. We test this idea against a competing theory that argues that communication networks that are less efficient for information transfer will increase the discovery of novel solutions to complex problems. We conducted a series of experimentally designed Data Science Competitions, in which we manipulated the efficiency of the communication networks among distributed groups of data scientists attempting to find better solutions for complex statistical modeling problems. We present findings from 16 independent competitions, where individuals conduct greedy search and only adopt better solutions. We show that groups with inefficient communication networks consistently discovered better solutions. In every experimental trial, groups with inefficient networks outperformed groups with efficient networks, as measured by both the group's average solution quality and the best solution found by a group member.

SUBMITTER: Brackbill D

PROVIDER: S-EPMC7473554 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Impact of network structure on collective learning: An experimental study in a data science competition.

Brackbill Devon D Centola Damon D

PloS one 20200904 9

Do efficient communication networks accelerate solution discovery? The most prominent theory of organizational design for collective learning maintains that informationally efficient collaboration networks increase a group's ability to find innovative solutions to complex problems. We test this idea against a competing theory that argues that communication networks that are less efficient for information transfer will increase the discovery of novel solutions to complex problems. We conducted a ...[more]

PMID: 32886685

Similar Datasets

Project description:BackgroundCommunity-based interventions have shown promise in reducing childhood overweight and obesity. However, they have been critiqued for using linear logic models. Participatory community-based systems approaches are posited as addressing the complexity of non-linear relationships in a local context. Community members are empowered to understand and describe obesity causation, identify and prioritise possible solutions. The application of such approaches to childhood obesity is in its infancy.AimTo describe the first 12 months of a participatory whole-of-community systems approach to creating collective action to tackle childhood obesity, called GenR8 Change, in a local government area of Victoria, Australia.MethodsThree group model building (GMB) sessions focused on the development of a causal loop diagram (CLD), prioritised evidence-informed actions, and developed implementation strategies. The collective impact framework underpinned the approach, with a local backbone group supporting community members to implement prioritised actions.ResultsThe first two GMB sessions included 20 key community leaders where a CLD examining the factors contributing to childhood obesity in the community was constructed and refined (22 variables GMB1, 53 variables GMB2). In the third session, 171 members of the wider community further refined the CLD, identified priorities for childhood obesity prevention (72 variables in final CLD). One-hundred and thirteen individuals signed up across 13 working groups to plan and implement 53 prioritised actions. Agreed community actions included creating sugar free zones; developing healthy policies; increasing breastfeeding rates; improving drinking water access; and increasing physical activity options. Twelve months post-GMB3, 115 actions had been implemented.ConclusionGenR8 Change is one of the first communities to apply systems thinking to childhood obesity prevention. Knowledge on how to collectively identify relevant leverage points to tackle childhood obesity can now be shared with other communities.

Project description:This study investigates if and under which conditions humans are able to identify and follow the most advantageous leader who will them provide with the most resources. In an iterated economic game with the aim of earning monetary reward, 150 participants were asked to repeatedly choose one out of four leaders. Unbeknownst to participants, the leaders were computer-controlled and programmed to yield different expected payout values that participants had to infer from repeated interaction over 30 rounds. Additionally, participants were randomly assigned to one of three conditions: single, independent, or cohesion. The conditions were designed to investigate the ideal circumstances that lead to identifying the most advantageous leader: when participants are alone (single condition), in a group that lets individuals sample information about leaders independently (independent condition), or in a group that is rewarded for cohesive behavior (cohesion condition). Our results show that participants are generally able to identify the most advantageous leader. However, participants who were incentivized to act cohesively in a group were more likely to settle on a less advantageous leader. This suggests that cohesion might have a detrimental effect on group decision making. To test the validity of this finding, we explore possible explanations for this pattern, such as the length of exploration and exploitation phases, and present techniques to check for confounding factors in group experiments in order to identify or exclude them as alternative explanations. Finally, we show that the chosen reward structure of the game strongly affects the observed following behavior in the group and possibly occludes other effects. We conclude with a recommendation to carefully choose reward structures and evaluate possible alternative explanations in experimental group research that should further pursue the study of exploration/exploitation phases and the influence of group cohesion on group decision making as promising topics for further research.

Dataset Information

Impact of network structure on collective learning: An experimental study in a data science competition.

Publications

Impact of network structure on collective learning: An experimental study in a data science competition.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets