Emergence and collapse of reciprocity in semiautomatic driving coordination experiments with humans

Forms of both simple and complex machine intelligence are increasingly acting within human groups in order to affect collective outcomes. Considering the nature of collective action problems, however, such involvement could paradoxically and unintentionally suppress existing beneficial social norms in humans, such as those involving cooperation. Here, we test theoretical predictions about such an effect using a unique cyber-physical lab experiment where online participants (N = 300 in 150 dyads) drive robotic vehicles remotely in a coordination game. We show that autobraking assistance increases human altruism, such as giving way to others, and that communication helps people to make mutual concessions. On the other hand, autosteering assistance completely inhibits the emergence of reciprocity between people in favor of self-interest maximization. The negative social repercussions persist even after the assistance system is deactivated. Furthermore, adding communication capabilities does not relieve this inhibition of reciprocity because people rarely communicate in the presence of autosteering assistance. Our findings suggest that active safety assistance (a form of simple AI support) can alter the dynamics of social coordination between people, including by affecting the trade-off between individual safety and social reciprocity. The difference between autobraking and autosteering assistance appears to relate to whether the assistive technology supports or replaces human agency in social coordination dilemmas. Humans have developed norms of reciprocity to address collective challenges, but such tacit understandings could break down in situations where machine intelligence is involved in human decision-making without having any normative commitments.

Stingy bots can improve human welfare in experimental sharing networks

Machines powered by artificial intelligence increasingly permeate social networks with control over resources. However, machine allocation behavior might offer little benefit to human welfare over networks when it ignores the specific network mechanism of social exchange. Here, we perform an online experiment involving simple networks of humans (496 participants in 120 networks) playing a resource-sharing game to which we sometimes add artificial agents (bots). The experiment examines two opposite policies of machine allocation behavior: reciprocal bots, which share all resources reciprocally; and stingy bots, which share no resources at all. We also manipulate the bot's network position. We show that reciprocal bots make little changes in unequal resource distribution among people. On the other hand, stingy bots balance structural power and improve collective welfare in human groups when placed in a specific network position, although they bestow no wealth on people. Our findings highlight the need to incorporate the human nature of reciprocity and relational interdependence in designing machine behavior in sharing networks. Conscientious machines do not always work for human welfare, depending on the network structure where they interact.

Individual and collective learning in groups facing danger

While social networks jeopardize people’s well-being by working as diffusion pathways of falsehood, they may also help people overcome the challenge of misinformation with time and experience. Here I examine how social networks provide learning facilitation using an experiment involving an iterated decision-making game simulating an unpredictable situation faced by a group (2786 subjects in 120 groups). This study shows that, while social networks initially spread false information and suppress necessary actions, with tie rewiring, on the other hand, they facilitate improvement in people's decision-making across time. It also shows that the network's learning facilitation results from the integration of individual experiences into structural changes. In sum, social networks can support collective learning when they are built through people's experiences and accumulated relationships.

Network Engineering Using Autonomous Agents Increases Cooperation in Human Groups

Cooperation in human groups is challenging, and various mechanisms are required to sustain it, although it nevertheless usually decays over time. Here, we perform theoretically informed experiments involving networks of humans (1,024 subjects in 64 networks) playing a public-goods game to which we sometimes added autonomous agents (bots) programmed to use only local knowledge. We show that cooperation can not only be stabilized, but even promoted, when the bots intervene in the partner selections made by the humans, re-shaping social connections locally within a larger group. Cooperation rates increased from 60.4% at baseline to 79.4% at the end. This network-intervention strategy outperformed other strategies, such as adding bots playing tit-for-tat. We also confirm that even a single bot can foster cooperation in human groups by using a mixed strategy designed to support the development of cooperative clusters. Simple artificial intelligence can increase the cooperation of groups.

Collective communication and behaviour in response to uncertain 'Danger' in network experiments

In emergencies, social coordination is especially challenging. People connected with each other may respond better or worse to an uncertain danger than isolated individuals. We performed experiments involving a novel scenario simulating an unpredictable situation faced by a group in which 2480 subjects in 108 groups had to both communicate information and decide whether to ‘evacuate’. We manipulated the permissible sorts of interpersonal communication and varied group topology and size. Compared to groups of isolated individuals, we find that communication networks suppress necessary evacuations because of the spontaneous and diffuse emergence of false reassurance; yet, communication networks also restrain unnecessary evacuations in situations without disasters. At the individual level, subjects have thresholds for responding to social information that are sensitive to the negativity, but not the actual accuracy, of the signals being transmitted. Social networks can function poorly as pathways for inconvenient truths that people would rather ignore.

Resource sharing in technologically defined social networks

Technologically enabled sharing-economy networks are changing the way humans trade and collaborate. Here, using a novel 'Wi-Fi sharing' game, we explored determinants of human sharing strategy. Subjects (N = 1,950) participated in a networked game in which they could choose how to allocate a limited, but personally not usable, resource (representing unused Wi-Fi bandwidth) to immediate network neighbors. We first embedded N = 600 subjects into 30 networks, experimentally manipulating the range over which subjects could connect. We find that denser networks decrease any wealth inequality, but that this effect saturates. Individuals' benefit is shaped by their network position, with having many partners who in turn have few partners being especially beneficial. We propose a new, simplified "sharing centrality" metric for quantifying this. Further experiments (N = 1,200) confirm the robustness of the effect of network structure on sharing behavior. Our findings suggest the possibility of interventions to help more evenly distribute shared resources over networks.

Locally noisy autonomous agents improve global human coordination in network experiments

Coordination in groups faces a sub-optimization problem and theory suggests that some randomness may help to achieve global optima. Here we performed experiments involving a networked colour coordination game in which groups of humans interacted with autonomous software agents (known as bots). Subjects (n = 4,000) were embedded in networks (n = 230) of 20 nodes, to which we sometimes added 3 bots. The bots were programmed with varying levels of behavioural randomness and different geodesic locations. We show that bots acting with small levels of random noise and placed in central locations meaningfully improve the collective performance of human groups, accelerating the median solution time by 55.6%. This is especially the case when the coordination problem is hard. Behavioural randomness worked not only by making the task of humans to whom the bots were connected easier, but also by affecting the gameplay of the humans among themselves and hence creating further cascades of benefit in global coordination in these heterogeneous systems.

How to identify water from thickener aqueous solutions by touch

Water detection is one of the most crucial psychological processes for many animals. However, nobody knows the perception mechanism of water through our tactile sense. In the present study, we found that a characteristic frictional stimulus with large acceleration is one of the cues to differentiate water from water contaminated with thickener. When subjects applied small amounts of water to a glass plate, strong stick-slip phenomena with a friction force of 0.46 ± 0.30 N and a vertical force of 0.57 ± 0.36 N were observed at the skin surface, as shown in previous studies. Surprisingly, periodic shears with acceleration seven times greater than gravitational acceleration occurred during the application process. Finite-element analyses predicted that these strong stimuli could activate tactile receptors: Meissner's corpuscle and Pacinians. When such stimuli were applied to the fingertips by an ultrasonic vibrator, a water-like tactile texture was perceived by some subjects, even though no liquid was present between the fingertip and the vibrator surface. These findings could potentially be applied in the following areas: materials science, information technology, medical treatment and entertainment.