From Opinion Polarization to Climate Action: A Social-Climate Model of the Opinion Spectrum

TL;DR

The paper addresses how a population's continuous climate opinions influence, and are influenced by, a coupled climate system. It develops a coupled social-climate model that integrates a continuous opinion dynamics module with an Earth System Model, using a temperature-driven feedback $R(T)$ and emissions $\frac{dC}{dt}=E_0\left(0.5\left(1-\frac{\sum_i o_i}{N}\right)\right)-\delta C$ in discrete time. Key contributions include the first demonstration of opinion polarization arising in a human-environment system, and the identification of policy-relevant levers—lower mitigation costs, enhanced social learning, and heightened sensitivity to temperature—that can promote mitigation even under polarization. The findings show that polarization does not preclude mitigation if the population responds strongly to temperature and costs are manageable, and they highlight how external perturbations can slow climate action, offering guidance for climate-goal achievement through peer influence and lifestyle changes.

Abstract

We developed a coupled social-climate network model to understand the interaction between climate change opinion spread and the climate system and determine the role of this interaction in shaping collective actions and global temperature changes. In contrast to previous social-climate models that discretized opinions, we assumed opinions on climate change form a continuum, and were thereby able to capture more nuanced interactions. The model shows that resistance to behaviour change, elevated mitigation costs, and slow response to climate events can result in a global temperature anomaly in excess of 2°C. However, this outcome could be avoided by lowering mitigation costs and increasing the rate of interactions between individuals with differing opinions (social learning). Our model is the first to demonstrate the emergence of opinion polarization in a human-environment system. We predict that polarization of opinions in a population can be extinguished, and the population will adopt mitigation practices, when the response to temperature change is sensitive, even at higher mitigation costs. It also indicates that even with polarized opinion, an average pro-mitigative opinion in the population can reduce emissions. Finally, our model underscores how frequent and unexpected social or environmental changes, such as policy changes or extreme weather events, can slow climate change mitigation. This analysis helps identify the factors that support achieving international climate goals, such as leveraging peer influence and decreasing stubbornness in individuals, reducing mitigation costs, and encouraging climate-friendly lifestyles. Our model offers a valuable new framework for exploring the integration of social and natural sciences, particularly in the domain of human behavioural change.

Figures (21)

• Figure 1: Schematic representation of the model with input variables for each component
• Figure 2: Time series for the coupled social-climate model Times series showing (a) the opinion dynamics (b) emission scenario (GtCO2yr-1) and (c) temperature anomaly ($^\circ$C) for baseline parameter values. Time series for (d) average opinion (e) emission scenario (GtCO2yr-1), and (f) temperature anomaly ($^\circ$C) for different parameter scenario.
• Figure 3: Opinion evolution over time Opinion distribution of the population at different time steps.
• Figure 4: Effect of variation in temperature response and cost of mitigation on opinion formation and climate change Influence on (a) peak temperature levels, (c) average opinions, and (e) polarization using bimodality coefficient by varying $R_{max}$. Influence on (b) peak temperature levels, (d) average opinions, and (f) polarization using bimodality coefficient by varying $m_{cost}$.
• Figure 5: Temperature response and cost of mitigation have a major influence on emission and peak temperature Heatmap showing the influence of variation in $R_{max}$ and $m_{cost}$ on (a) emission levels and (b) peak temperature anomaly.