Transactive Systems Model of Collective Intelligence: The Emergence and Regulation of Collective Attention, Memory, and Reasoning

Modern organizations face a highly complex and dynamic environment. In this dissertation, I argue that collectives (teams, organizations, or communities) facing highly complex and dynamic situations need to be designed for collective intelligence, or the ability to achieve goals in a wide range of environments. Integrating and building on extant work, I theorize a Transactive Systems Model of Collective Intelligence, guided by agent-based modeling, and test it with data from open-source software teams. I also explore and propose a set of diagnostic indicators as an extension for further development, to enable monitoring and intervention by algorithmic tools and/or leaders.

I take a complex adaptive systems view of collectives and describe how transactive attention, memory, and reasoning systems (TAS, TMS, and TRS) emerge from individual-level cognitive processes and member interactions to shape the emergence of collective intelligence. I further theorize how these systems interact with each other and respond to dynamic environmental complexity. I complement this narrative theory with agent-based modeling (ABM) to validate the sufficiency of the proposed transactive process. Once validated, I use the ABM to conduct two virtual experiments to demonstrate the co-regulation of TMS and TAS, which is complemented by TRS. Based on these virtual experiments, I derive hypotheses about the critical environmental threats that each transactive system is specifically equipped to address, highlighting the co-regulation necessary in response to changes in the environment which we theorize underlie the development and maintenance of collective intelligence. Finally, I empirically test corresponding hypotheses for aggregated system behavior by analyzing 18 months of archival data from 476 open-source software teams. Consistent with predictions, I find evidence confirming the hypotheses and providing initial support for the transactive systems theory.

In the next chapter, I build on the socio-cognitive architecture of collective intelligence articulated in the Transactive Systems Model and theorize three observable collaborative processes that are related to the transactive system processes. I propose that these observable processes can serve as diagnostic indicators to provide real-time information about the functioning of the underlying, largely unobservable complex adaptive system. I explore these collaborative process indicators in another virtual experiment that supports their general utility in signaling the level of functioning of different transactive systems. I also propose combining them into a single metric that acts as a leading indicator of collective intelligence. It can be monitored in real-time and guide the diagnosis of underlying problems by looking at its component parts. Articulation and refinement of such metrics provide useful guides for intervention by humans or algorithmic tools and together with transactive systems lay a foundation for a Machine Theory of Collective Intelligence. While much remains to be learned about the nature of collective intelligence, this dissertation presents a multi-method systems approach for investigating its emergence, adaptation, and diagnosis, laying the groundwork for future research.