Biased perceptions explain collective action deadlocks and suggest new mechanisms to prompt cooperation (1)

All the perception biases we consider here can be situated in a two-dimensional space (χ,δ), defined by a bias in the level of cooperators by cooperators (χ, where χ 0 an overestimation of the number of other cooperators in the population) and a bias in the level of cooperators by defectors (δ, where δ 0 an overestimation in the number of cooperators by defectors).

We control δ and χ such that we navigate from a scenario of under-trust into a scenario of over-trust. We can observe that increasing cooperation bias (i.e., increasing both δ and χ) eases the coordination toward full cooperation entailed by the non-linear public goods with f > 1. The effect of over- and under-trust on coordination toward cooperation can be grasped by the position of the coordination point for different values of M, increasing δ and χ reduces the position of the coordination point (represented with dashed lines), implying that a smaller fraction of cooperators is needed to evolve toward full cooperation.

Social perception bias can, however, affect cooperators or defectors in different directions. In the case of false consensus, individuals overestimate the adoption of their own strategy in a population, meaning that cooperators will overestimate the fraction of cooperators and defectors will overestimate the fraction of defectors. The effects of heterodirectional bias on cooperation dynamics, considering false uniqueness (δ = −χ, χ 0, right half). We can observe that false uniqueness induces a stable coexistence of cooperators and defectors, which may not be sufficient to support high levels of collective success.

We show that increasing the magnitude of punishment has a different effect depending on the nature of bias prevailing in a population. For instance, a lower punishment is necessary to sustain collective success under false consensus, compared with false uniqueness (for the combination of parameters analyzed, in particular, high value M = 8). Additionally, we can observe that an effort to reduce individuals’ perception biases can render high levels of collective success, even in situations where low incentives (low ι) are not effective—as a baseline, we show the group success characterizing a situation where neither biases nor incentives lead to the coordination in virtually all groups.

The fact that biases generate new, stable equilibria can have strong implications for the functioning of society. The workings and efficiency of markets and market regulation rely on the bottom-up ability of selfish agents to achieve socially desired outcomes and not get stuck in deadlocks as the ones we identify. Considering the social network of interacting individuals not only may prove desirable to re-create realistic settings but also can be instrumental in explaining the origin of social perception biases and reconciling the apparently contradictory ones.

1. F. P. Santos, S. A. Levin, V. V. Vasconcelos, Biased perceptions explain collective action deadlocks and suggest new mechanisms to prompt cooperation. iScience. 24 (2021), doi:10.1016/j.isci.2021.102375.