Are two heads better than one? Or do too many cooks spoil the broth?

Can we predict whether a team of interacting individuals will outperform a group of individuals working alone?

In their new paper “Task complexity moderates group synergy,” Abdullah Almaatouq and Duncan Watts, along with co-authors Mohammed Alsobay and Ming Yin, tackle this question at the task level. Key to their approach is a measurable interpretation of task complexity, defined in terms of the number of task components and the interdependencies between them. Using this framework, they find that collaborative teams are more efficient than even the most efficient individuals when the task is complex—but not when the task is simple.

Selecting a task class for complexity manipulation

Almaatouq et al.’s first step towards measuring the moderating effect of task complexity was to identify a type of task which could be made more or less complex while keeping all other elements constant. In this way, the authors could easily measure performance as a function of complexity without worrying about the effects of other aspects of the task.

The authors settled upon constraint satisfaction and optimization problems (CSOPs), a class of tasks commonly studied in artificial intelligence and operations research. In particular, they chose a room assignment problem: participants assigned students to rooms depending on their level of utility in each room and a series of constraints (e.g., “Students A and J may not share a room or be in adjacent rooms”). With this task setup, the authors could systematically alter the complexity by adjusting the number of students, rooms, and constraints while keeping all other task components fixed.

The authors observed that groups working on the highest-complexity tasks were as fast as their fastest and most efficient group members, or “strongest links.” Moreover, they found that groups’ increasing speed outweighed their decreased solution quality as tasks grew more complex, resulting in higher overall efficiency.

Further exploration of group dynamics

So, do too many cooks actually spoil the broth? Almaatouq et al. show that the answer is: it depends. Task complexity should be taken into account, as it plays a crucial role in team performance. The authors find that, at least for tasks with many moving parts, the benefits of distributing work to a group might outweigh the process losses that such teamwork incurs.

This key finding also qualifies previous research on group performance. Since laboratory studies of group performance generally rely upon very simple tasks, they may overemphasize group process losses while understating synergistic gains. Almaatouq et al. highlight the need for a more thorough understanding of the role of complexity, paving the way for future research that expands upon the one-size-fits-all task model.

Read the full paper published in PNAS here.