Cognition, Motivation, and Strategy of Individuals in Networked Environments
The massive amounts of available digital trace data enable studies of population-level human interaction on an unprecedented scale. This project offers groundbreaking insights into how multidimensional network configurations shape the success of value-creation processes within crowdsourcing systems and online communities. Furthermore, this project offers new computational social science approaches to theorizing and researching the roles of social structure and influence within technology-mediated value creation processes.
Projects in this stream center on the question of how networked interactions affect crowd decision-making and problem-solving. This research stream explores how to best support individual decision-making to most effectively harness the “wisdom of crowds.” For example, we study the design of IT-based decision-making support using a rating scale (representing a judgment task) and a prediction market (representing a choice task) in a randomized experiment. We also explore I explore individual-level cognitive biases in the evaluation of new ideas and frontier research proposals. We also study how individuals participating in repeated innovation contests improve their performance through social learning.
Ongoing work in this research stream is focused on understanding problem solving behavior as a search process in which humans actively seek out information that maximizes the evidence for their model of the problem at hand. That is, by running human experiments as well as agent-based simulations, we are trying to understand how people represent complex problems and use that representation in order to find the best solution to a problem. Throughout this work, we model optimal decision-making as a process of minimizing the surprises that would be generated by our actions, and in doing so, we are able to understand search behavior in a way resolves traditional dilemmas in behavioral economics about whether people tend to exploit known solutions to problems or explore for new ones.
Funding: Nation Science Foundation
- "Learning from Mixed Signals in Online Innovation Communities" (Organization Science)
- "Looking Across and Looking Beyond the Knowledge Frontier" (Management Science)
- "Rate or Trade" (Information Systems Research)
Social Network Processes in Collaborative Decision-Making
Projects in this research stream investigate the emergence of collaboration in teams. Team decision-making is among the most commonly studied topics in management and organizational behavior, with implications for businesses, governments, security, and sustainability. However, research frequently focuses on summative performance outcomes, ignoring the emergent processes and endogenous communication effects that produce those outcomes. We address this challenge using the digital trace data left by computer-mediated communication, and rich temporal data collected from unobtrusive sensors to examine emergent processes and endogenous communication effects. This project studies how team can be organized efficiently and how emergent properties and processes within a team affect team performance. This project makes important contributions to our fundamental understanding of collaborative decision-making and the organization of work in general.
Funding: DARPA, Northeastern Seed Grant (Tier 1)
- "Teams vs. Crowds" (Academy of Management Discoveries)
Social Influence and Information Diffusion in Networks
In most areas of human activity, from science to business, we expect performance to uniquely determine access to resources and reward. Yet, in areas where performance is difficult to quantify or depends on tastes and preferences, reputation and invisible networks of influence play an important role. This can lead to severe inequalities such that certain individuals or products are more successful than others. We are conducting a series of studies to examine how social influence and information diffusion shape outcomes on the global scale. In one study in this stream, we explore how social networks play a crucial role in dissemination and adoption of innovations using data from two country-level field experiments. We discover a novel mechanism for the diffusion of innovations which results in linear adoption curves. This is contrary to the typical S-shaped curves promoted in the diffusion of innovation literature. As a result, we propose a new model for the diffusion of innovation through on-demand information-seeking behavior. In another study in this research program we investigates these dynamic processes on networks more formally through agent-based modeling. Using an evolutionary game theoretical model with dynamic network updating, we examine how cooperation emerges and persists. We show, for example, that outcomes associated with host-guest norms (play aggressive away, not at home), rather than ownership norms (play aggressive at home, not away), tend to evolve on dynamic networks. This result challenges the commonly held view in the literature on which conventions emerge to resolve conflicts. Furthermore, it adds generality to research showing that learning in social networks has profound effects on the resulting dynamics which often favor cooperation, and I propose a new model of network formation.
Funding: National Science Foundation, Army Research Office, Office of Naval Research
- "Conflict and Convention in Dynamic Networks" (Journal of the Royal Society Interface)
- "Product Diffusion Through On-Demand Information-Seeking Behavior” (Journal of the Royal Society Interface)
In this work, we ask a number of questions about how to optimally design experiments—in particular, rapidly-deployed experiments run on online experimentation platforms like VolunteerScience and nodeGame. For one project, we’ve designed an optimal experimental design protocol that can rapidly search a vast space of possible experiments to find the design that will maximize the information gained from running the experiment. We use a similar approach in another project studying social networks and decision-making, where we optimize for the particular network structure that would produce the greatest effect size in our experiment.