Counter-Punishment, Communication, and Cooperation among Partners.

We study how communication affects cooperation in an experimental public goods environment with punishment and counter-punishment opportunities. Participants interacted over 30 rounds in fixed groups with fixed identifiers that allowed them to trace other group members' behavior over time. The two dimensions of communication we study are asking for a specific contribution level and having to express oneself when choosing to counter-punish. We conduct four experimental treatments, all involving a contribution stage, a punishment stage, and a counter-punishment stage in each round. In the first treatment communication is not possible at any of the stages. The second treatment allows participants to ask for a contribution level at the punishment stage and in the third treatment participants are required to send a message if they decide to counter-punish. The fourth combines the two communication channels of the second and third treatments. We find that the three treatments involving communication at any of the two relevant stages lead to significantly higher contributions than the baseline treatment. We find no difference between the three treatments with communication. We also relate our results to previous results from treatments without counter-punishment opportunities and do not find that the presence of counter-punishment leads to lower cooperation level. The overall pattern of results shows that given fixed identifiers the key factor is the presence of communication. Whenever communication is possible contributions and earnings are higher than when it is not, regardless of counter-punishment opportunities.

On agent-based modeling and computational social science.

In the first part of the paper, the field of agent-based modeling (ABM) is discussed focusing on the role of generative theories, aiming at explaining phenomena by growing them. After a brief analysis of the major strengths of the field some crucial weaknesses are analyzed. In particular, the generative power of ABM is found to have been underexploited, as the pressure for simple recipes has prevailed and shadowed the application of rich cognitive models. In the second part of the paper, the renewal of interest for Computational Social Science (CSS) is focused upon, and several of its variants, such as deductive, generative, and complex CSS, are identified and described. In the concluding remarks, an interdisciplinary variant, which takes after ABM, reconciling it with the quantitative one, is proposed as a fundamental requirement for a new program of the CSS.

Punish and voice: punishment enhances cooperation when combined with norm-signalling.

Material punishment has been suggested to play a key role in sustaining human cooperation. Experimental findings, however, show that inflicting mere material costs does not always increase cooperation and may even have detrimental effects. Indeed, ethnographic evidence suggests that the most typical punishing strategies in human ecologies (e.g., gossip, derision, blame and criticism) naturally combine normative information with material punishment. Using laboratory experiments with humans, we show that the interaction of norm communication and material punishment leads to higher and more stable cooperation at a lower cost for the group than when used separately. In this work, we argue and provide experimental evidence that successful human cooperation is the outcome of the interaction between instrumental decision-making and the norm psychology humans are provided with. Norm psychology is a cognitive machinery to detect and reason upon norms that is characterized by a salience mechanism devoted to track how much a norm is prominent within a group. We test our hypothesis both in the laboratory and with an agent-based model. The agent-based model incorporates fundamental aspects of norm psychology absent from previous work. The combination of these methods allows us to provide an explanation for the proximate mechanisms behind the observed cooperative behaviour. The consistency between the two sources of data supports our hypothesis that cooperation is a product of norm psychology solicited by norm-signalling and coercive devices.

Agent-based modeling for understanding social intelligence.

The advantages of agent-based modeling for a general theory of intelligence at the individual and social level are emphasized over other existing approaches mainly relying on rationality theories. As was pointed out during the National Academy of Sciences Sackler Colloquium session "Implications of Agent-Based Modeling for Understanding Human Rationality and Learning," held in October 2001, properties of social intelligent agents include adaptability and learning capacity, as well as the capacity to produce and employ artifacts and manipulate symbols.