Bot interventions in human cooperation and coordination

Hirokazu’s current research aims to understand the role that individuals play in facilitating cooperation and coordination in networks. To study this phenomena, he has introduced anonymous artificial intelligence “bots” into social network experiments, allowing human subjects to interact with them. These bots have various methodological benefits.

For one, bots can facilitate causal inference: bots with specific, known, programmed responses can be used to test human responses to targeted behavioral “treatments.” In another respect, the bot-injection study helps to demonstrate the positive potential for “heterogeneous systems” involving humans and machines. In such systems, artificial intelligence can work as catalyst of social reaction among humans (rather than as a human replacement). Through this study, Hirokazu is developing a new design concept for machines to help group of humans help themselves.

Using this approach, Hirokazu examined the functional role of erratic behaviors in human coordination. He performed experiments involving a classic networked color coordination game in which groups of humans interacted with bots. The bots were programmed with varying levels of behavioral randomness and different geodesic locations. Theses experiments showed that bots, placed at central locations within the network and acting with small levels of random noise, significantly improved the collective performance of human groups. This behavioral randomness in the bots affected the gameplay of the humans, suggesting that this spillover effect can benefit global coordination.

Color coordination game with bots: This game’s goal is for every player in a network to select a color different than all of their neighbors as quickly as possible. Left group is composed entirely of human players. Right group includes three “noisy” nots (depicted in squares). Red connections indicate when the colors conflict. Quicker solutions to the color conflict reflects greater coordination capability of the group.

Online experiment with social networks

In pursuit of his research, Hirokazu has worked to develop online experimental framework in social science that make a theoretical contribution, while also offering practical insights. The advantages to us Internet are many, including the ability to recruit large samples quickly, and the ease of replicating experimental treatments. By employing novel software and recruiting subjects online from around the world via online labor markets, Hirokazu is able to conduct a series of experiments that spanned the whole range of possible parameters.

Using the experimental tool “breadboard”, for instance, Hirokazu revealed the relationships between network dynamics and cooperative behaviors.   When the rate of change in social ties is too low, subjects opt to retain many ties, at the risk of attaching to defectors.  When the turnover rate is too high, cooperators cannot detach from defectors as quickly as defectors can re-attach, prompting a switch in individual behavior to defection.  These results suggest that optimal levels of cooperation are achieved at intermediate levels of change in social ties. 

Cooperation game with network rewiring: Each human subject is initially assigned a location in a network and then give a choice: either cooperate (blue) or defect (red). Subjects who cooperate pay 50 units for each network neighbors and each of their neighbors receive 100 units. Subjects who defect pay 0 units and their neighbors receive 0 units. After subjects choose whether to cooperate, a fixed percentage of subject pairs (“rewiring rate”) are chosen at random in which one individual in the pair is allowed to decide whether to form a new tie if one do not exist or to cut a tie if one do exist. These interactions are repeated for 15 rounds.

In another experiment, Hirokazu planned and executed a series of experiments to better understand the dynamics of emergency evacuation. In these experiments, he introduced a novel decision-making game with interpersonal communication, simulating an unpredictable situation caused by a disaster. This game has reproduced numerous network phenomena observable under real circumstances, including: the spreading of false rumors, network disintegration, and appearance of panic. In the human subjects gameplay, the diffusion of false information was generated almost spontaneously, which, when applied to real-world disaster scenarios,  could translate to life-threatening danger.

Hirokazu Shirado hopes to use these findings to manipulate various qualities inherent to social networks, working to minimize the negative effects of inequality, while stimulating cooperation and innovation. His ultimate objective is to integrate these strategies and solutions in to the filed of system and computational engineering to advance the human condition.

Evacuation game: during the game, subjects need to make a decision on whether to evacuate from a “disaster” which may or may not strike them. To find out the truth, subjects who have not yet evacuated can communicate with neighbors by sending “Safe” signal (turning their node blue) and “Danger” signal (turning their node red). Since a randomly-selected subject (the “informant” shown as “i” in the video) is informed in advance whether a disaster will indeed strike or not, subjects can choose a correct action by disseminating the accurate information originating from the informant. In the experimental session shown in the video, a disaster will strike in 95 seconds. Subjects who have evacuated are shown with bold nodes.