Socio-political dynamics in clean energy transition

A rapid and effective transition to low-carbon energy production is essential to limit climate change impacts. While the scientific community has mostly focused on research and development and techno-economic aspects, quantifying the role of public acceptability and policy in shaping emission trajectories has been much more elusive. This study investigates the coupled dynamics of nonlinear socio-political acceptance and anthropogenic CO2 emissions, with implications for climate policies and clean energy investments. Our findings show that a top-down policy approach alone may not be sufficient for effective emission cuts, highlighting the need for a multi-level strategy that combines top-down and bottom-up approaches. Additionally, opinion polarization can trigger detrimental CO2 emission oscillations when governments decide to take heavy-handed policy interventions in highly polarized socio-political systems. Delayed perception of climate change damage or abrupt reactions to extreme weather events may also significantly affect emission reduction efforts, although in the opposite direction. Integrating these socio-political dynamics into climate models can enhance our understanding of the complex interplay between human and natural systems, enabling the development of more effective and resilient mitigation strategies.


Introduction
Achieving ambitious climate change mitigation goals requires curtailing emissions driven by energy production sufficiently to enable natural sinks, naturebased climate solutions, and CO 2 removal technologies (e.g.Direct Air Capture) to reduce the concentration of CO 2 in the atmosphere [1][2][3].Although emissions pathways emerge from complex bi-directional interactions between natural and human systems, such interacting elements are often analyzed separately and then coupled via static interactions based on fixed behavioral rules, resulting in a wide range of plausible greenhouse gas emissions scenarios [4][5][6][7][8][9][10].Instead, since anthropogenic CO 2 emissions comprise the most critical contribution to global warming, their dynamics should be coupled to social processes determining whether socio-political systems are willing to step away from fossil fuels [11][12][13].
Understanding the co-evolution of human and environmental dynamics in the context of decarbonization is essential to identifying the impact of different governance and behavioral scenarios on the transition to a given sustainability target [12,14,15].The conventional approach, as seen in Shared Socioeconomic Pathways (SSPs)-based Representative Concentration Pathways (RCPs) scenarios, relies on exogenously defined RCPs.These scenarios impose different emission trajectories based on factors such as changes in population, economic growth, education, urbanization, and technological development.Nevertheless, these scenarios lack the capacity to dynamically generate emissions over time, failing to incorporate the dynamic, endogenous reactions of socio-political systems to evolving environmental circumstances [6,16,17].
Emission cuts have remained well below the goals set by international agreements [18,19], primarily because of a lack of national economic and sociopolitical commitment [20,21].At the global level, strong inequalities between the global north and the global south induce polarization regarding the need to take immediate actions and hinder cooperation on climate change [22].At the national (or local) level, the decision of whether and when to shift from fossil fuel to low-emission energy sourcescommonly called clean energy-is inherently influenced by the perceived socioeconomic cost of atmospheric CO 2 concentration [23,24].Poor air quality associated with fossil-fuel combustion and extreme weather events (EWEs)-partly attributed to rising CO 2 levels-increase the perceived damage of climate change [25,26].As such a cost rises, so does the willingness to invest in clean energy due to the increasing environmental and economic damage caused by climate change [27].However, despite a general understanding that immediate mitigation actions are needed, local hindrances to change in fossil fuelbased economies, myopic economic priorities, and opinion polarization are causing delays in the transition to clean energy [24,[28][29][30].Furthermore, the efficacy of achieving emission cuts through either a top-down approach, characterized by authoritative and extensive policy interventions, or a bottomup strategy, involving shifts in public acceptance and social norms, remains an open question.Improving climate-model realism requires accounting for such socio-political feedbacks and endogenizing anthropogenic emissions [6].
Endogenizing greenhouse gas emissions is further complicated by strongly non-linear social behaviors and technology adoption [5,[31][32][33][34].Social norms, for example, can result in self-reinforcing or self-undermining actions and social tipping points [35,36].Opinion polarization, impacting national commitment to international climate change mitigation agreements [37], further complicates matters.Here, we propose a coupled human-natural system model that captures opinion formation and dynamics and its crucial role in guiding the consensus for climate policy [38,39].Suppose we focus on emergent behavior rather than individual interactions among agents.In that case, opinion dynamics concerning climate action may be represented as a bistable system, where public opinion tends to polarize to either an exploitation or a mitigation state (figure 1).This is a reasonable assumption considering how opposite political trends can shape and polarize public opinion [40], as also observed in other co-evolving humanenvironmental systems such as the public response to the COVID-19 pandemic [41].
When considered in isolation from the influence of other variables, such dynamics are driven by a double-well potential function, as shown in figure 1: in the unbiased case, the potential has two minima corresponding to a stable state of negative (exploitation) or positive (mitigation) feedback.Public opinion can shift abruptly between these two stable states, pushing the anthropogenic forcing toward investing in clean energy or maintaining a fossil fuel-based energy system.Coupling this representation of opinion dynamics with socio-political drivers such as perceived economic damage and hindrance to change, we obtain a more general formulation of the dynamics of socio-political acceptance of clean energy investments.

Methods
We model two-way interactions within a coupled human-natural system describing clean-energy transitions and the possible impacts of non-linear social dynamics.The proposed model is a 3D dynamical system, where the three governing differential equations describe (1) the ecosystem state (CO 2 concentration in the atmosphere), (2) the anthropogenic forcing (mitigation action through clean energy investments), (3) the social-political feedback on (1) and (2).Our model can take into account different policy scenarios, the impact of a hindrance to change and perceived economic damage, opinion polarization, and social interaction (figure 2(A)).

The drivers of socio-political acceptance
The climate crisis is forcing society to face the negative consequences of global warming [43,44].As atmospheric CO 2 concentration increases, so does the associated perceived economic damage PED [45].Several integrated assessment models have estimated the economic impact of anthropogenic CO 2 emission, or social cost of carbon, as the discounted value of economic damages caused by the emission of one additional ton of CO 2 equivalent [46][47][48].Yet, the actual societal perception of climate change risk cannot be confined to its technical definition [49][50][51].Dynamic social norms, attitudes, values, cultural identities, and political influence, such as perceptions and polarization, determine when it is acceptable to promote climate mitigation actions [52,53].As shown in figure 2(B), PED tends to push public opinion toward consensus for mitigation actions (green curve).However, the hindrance due to diverse sensitivities to environmental issues, economic priorities, and vulnerability to climate change can favor maintaining the status quo and accepting exploitation (orange curve).
We identify and investigate two main drivers of opinion dynamics: opinion polarization and social interaction.Polarization results in diverging opinions about whether to take mitigation actions or not.The stronger the opinion polarization, the further apart the two stable states of opinion tend to be (figure 2(C)).Social interaction can amplify or decrease the influence of opinion dynamics on social acceptance.It shapes and reinforces societal norms, facilitates the exchange of information and knowledge, and promotes social learning and behavioral contagion [54,55].In this modeling framework, it affects the depth of the two wells in the potential function.Strong social interaction means a potentially rapid transition from exploitation to mitigation, but also strong perturbations are required to initiate and support the transition.It also reflects the general attitude towards a more or less fast transition to clean energy.

Modeling the impact of acceptance on CO 2 emission cuts
We propose a dynamical systems approach to study the interaction between CO 2 emission trajectories and non-linear socio-political dynamics that control clean energy investments.We aim to reproduce the two-way interactions focusing on the interplay between perceived damage caused by climate change, the mitigation actions taken to reduce such damage, and the opinion dynamics on to what extent and when we should take mitigation actions.We assume that energy demand is the major driver for global anthropogenic emissions [56], and, therefore, model the transition from fossil fuel to clean energy as the main mitigation action.
The natural system component is represented by the dynamics of atmospheric dioxide concentration [CO 2 ].This depends on the difference between anthropogenic emissions (η[E 0 − E CL ]), anthropogenic sinks (or offset OF), and net natural sinks (NS) as [24] d where ξ [CO2] is a term that accounts for uncertainty and short-term variability in the atmospheric carbon budget [56].The parameter η represents the increase in [CO 2 ] per unit of energy generated by fossil fuels, E 0 the global energy demand, and E CL the clean energy production.Therefore, E 0 − E CL represents the global energy demand met by CO 2 -emitting fossil fuels.In this context, the mitigation action is represented by reducing emissions by meeting the global energy demand with higher shares for low-emitting energy production systems.Natural sinks are expected to increase with [CO 2 ] as a result of the carbon fertilization effect [57][58][59].However, the fertilization effect can be limited by various factors like water and nutrient availability, soil quality, and the capacity of plants to photosynthesize [60][61][62][63].Additionally, changes in land use and habitat loss may limit these sinks in the future [64][65][66].The relation between natural sinks and [CO 2 ] is represented here by a simple power-law function [24].This relation can describe different degrees of non-linearity, with a and b capturing the sensitivity of natural sinks to CO 2 levels and allowing for the possibility of temporal variation.
The dynamics of clean energy production are governed by socio-political acceptance (SA) -or general mitigation opinion-the depreciation and maintenance costs (θE CL ), and economic loss directly related to rising where α is the energy produced per unit of investment in clean-energy infrastructure, γ is the coefficient of capital reduction due to climate change impacts, and τ is a fundamental time-scale parameter that represents the reduction in [CO 2 ] per unit of effort, and ξ ECL accounts for uncertainty.The parameter τ models the policy scenario that can push for a clean-energy transition for a given level of socio-political acceptance.The latter increases with the perceived socioeconomical damage PED caused by rising [CO 2 ] and depends on social interactions or opinion dynamics.Such interactions correspond to the level of public engagement that can reinforce or undermine the mitigation effort as a result of the common perception of the ecosystem and economic status [32].For a more detailed representation of how these different variables and processes interact in our model, we refer to Perri et al [24].
As discussed previously, we model socio-political acceptance as a self-reinforcing or self-undermining process in which SA tends to increase with the perceived economic damage caused by climate change (PED), and it could be limited by the hindrance to change of fossil fuel-based economies (µE CL ) The term ξ SA accounts for uncertainty in SA [67] and we assume that the PED can be defined as the sum of a compounding effect of global warming PED [CO2] (t − ω∆t)-deterministically linked to rising [CO 2 ]-and stochastic jumps in the perceived damage PED EWE due to the occurrence of EWEs.The deterministic component of PED is defined similarly to the social cost of carbon of the DICE model [68,69] as , where ϕ 1 and ϕ 2 are coefficients that quantify the relative economic impact of changes in the radiative forcing due to changes in [CO 2 ] with respect to the reference pre-industrial value [CO 2 ] Ref .The delay term ω∆t accounts for the possible time lag between emissions and their impact; in fact, while some consequences of fossil fuel combustion are immediate, such as air pollution, others are observed only after several years.For instance, the maximum impact of a CO 2 emission pulse may take several decades to be observed [70][71][72].Such a time lag could affect the emission trajectory, as governments may delay their actions until the effects of past emissions are apparent.Here, the delay in perceived damage can be turned off by setting the parameter ω = 0.
The stochastic component PED EWE accounts for abrupt changes in perceived damage due to EWEs and the 'collective memory' of such events.EWEs are known to influence the decision-making process concerning climate change adaptations [73][74][75].EWEs, such as floods, hurricanes, widespread wildfires, and protracted droughts, raise awareness about the impact of climate change on people's lives and countries' economies.It is clear that governments tend to take adaptive actions to reduce future impacts once faced with this increasing frequency and damage caused by EWEs.What is not clear is (i) to what extent EWEs also trigger mitigation actions by forcing governments and stakeholders to cope with the adverse effects of climate change and (ii) what is the impact of the collective memory of EWEs.The effect of an EWE on public perception is likely to fade in a few years.This effect is here modeled as a linear decay of the spike in concern over time PED EWE = ξ EWE − K EWE PED EWE , with ξ EWE being stochastic jumps in the PED, and K EWE the slope of the decay.The random jumps are modeled as a marked Poisson process of frequency λ EWE (equal to the frequency of the EWEs considered) and exponentially distributed size of mean α EWE [76,77].The functional relation in equation ( 3) is chosen to describe the non-linear behavior typical of the adoption of new technologies or new social norms, which is represented by a logistic function [78,79].
In summary, our framework assumes that the acceptance of a mitigation action dynamically responds to the environmental status and its perceived economic implications and it can be reinforced or weakened by public opinion and social interactions [80].The parameters of equation ( 3), namely β, ϵ, and µ describe the degree of social interaction, opinion polarization, and hindrance to change, respectively (see figure 2).This feedback is characterized by strong non-linearity: as social norms reinforce majority behavior, mitigation actions encounter great resistance at the early stage [74,81,82].This is particularly true in case of uncertainty in the information about the ecosystem status and its implication [83].
Reducing the uncertainty about the consequences of climate change is fundamental to promoting coordinated mitigation actions [84].Moreover, according to social comparison theory [85], individuals tend to compare themselves to others to form an opinion in case of uncertain outcomes.Also, a change in opinion does not necessarily imply a shift in behavior (Cognitive Dissonance theory [85,86]).It is plausible that shifts in opinion take place much earlier than changes in behavior, which become more likely when the majority starts to adopt new social norms [74,87].For these reasons, social scientists have observed that behavior changes, social norms, and technologies tend to follow tipping point dynamics [15,88].
Similar non-linear dynamics have also been observed in the socio-technical dynamics of clean energy transitions [89,90].It is widely recognized that adopting new technologies, including a shift to clean energy, tends to follow a logistic development [91][92][93][94].While here we focus on social dynamics, equation (3) could also describe the 'technical acceptability' of clean energy investments.In this context, public incentives trigger a self-reinforcing process of clean energy transition that results in the typical logistic growth [95,96].

Results
The dynamical system model presented in equations ( 1)-( 3) was numerically solved to simulate CO 2 emission trajectories under different policy, behavioral, and perception scenarios (figure 3).We ran N = 1000 simulations with white noise in each equation and then estimated average trajectories (solid and dashed curves in figure 3) and their standard error (here represented by the shaded areas).The model's parameters were estimated using recent emission and clean energy investment trends [56] (see table S1).Global Energy Outlook's projections were used to account for the rising global energy demand [97].
The impact of policy and behavior changes on emission trajectories is explored in figures 3(A) and (B).As expected, a business-as-usual scenario with no additional mitigation policy (blue curve in figure 3(A)) results in a rising concentration.Strong climate policies can boost clean energy production (red curve in figure 3(A)) to an extent to which immediate emission reduction is theoretically possible (black curve in figure 3(A)).However, immediate reduction coupled with behavioral changedriven, for example, by social interaction-can result in periodic oscillations (figure 3

(B)). In these scenarios, intermediate values of social interaction (orange and red curves) result in the highest risk of rebound effect of [CO 2 ].
The effect of non-linear opinion dynamics on emission trajectories highly depends on the policy scenario.Two interesting cases are a moderate political push for mitigation policy (mild policy scenario) and a strong political commitment (immediate emission cuts scenario).In a scenario of mild climate policy to reduce emissions, strongly non-linear opinion dynamics and social interaction may slow down the transition to clean energy (figure SI(A)).High social interaction (β → 1) may cause a runaway effect with increasing [CO 2 ] despite a generally elevated Panel (A) shows [CO2] over time in case of no climate mitigation policy (blue line), mild (red line), strong (orange line), and immediate emission cuts (black line), corresponding to an ideal strict policy able to achieve a reduction of global of [CO2] in a few years.These scenarios were obtained without accounting for non-linear opinion dynamics, namely setting β = 0. Panel (B) was obtained considering the most extreme policy scenario (immediate reduction) and varying the social interaction in the opinion dynamics.The case of no social interaction (black line) corresponds to the one obtained in panel (A) with β = 0.In contrast, mild (red), strong (orange), and very strong (blue) social interaction scenarios were obtained with β = 0.1, β = 0.25, and β = 0.75, respectively.(C) Shows the impact of opinion polarization under a scenario of immediate emission cuts and mild social interaction (β = 0.25), and varying polarization parameter ϵ to simulate no polarization (ϵ = 1/3; orange line), mild (ϵ = 1; red line), and strong polarization (ϵ = 1.5; blue line).(D) Displays global atmospheric carbon dioxide concentration ([CO2]) accounting for a delay ∆t between emissions and perceived impacts up to 20 years.The dashed blue curve represents the reference trajectory without delay, while the red and orange curves display the [CO2] trends with 10 and 20 years delay, respectively.(E) shows the impact of sensitivity to EWEs.The trajectories of increasing sensitivities are obtained by increasing the amplitude of the jumps in perceived damage (αEWE).Finally, (F) explores the memory effect of EWEs.Different trajectories were obtained by varying KEWE.

level of acceptance of clean energy investments due to elevated PED (figure SI(B)).
Our results indicate that pushing for immediate [CO 2 ] reduction through a vigorous top-down policy approach presents some risks related to opinion polarization.Strong opinion polarization can favor the periodic oscillations discussed above (blue and red curves in figure 3(C)).However, if there is no opinion polarization in the perception of climate change and consequent mitigation actions, emission rebounds are successfully avoided (orange curve).The time-lag ∆t between CO 2 emissions and the consequent global warming could also increase the risk of further warming.As expected, the larger the delay, the higher the peak in concentration (figure 3(D)).Similarly, EWEs could increase public awareness of the impact of climate change and trigger more immediate actions (figure 3(E)).The impact of EWEs on SA depends on public sensitivity to the impact of extreme events and on how long a change in SA can last ('memory effect'; figure 3(F)).Ultimately, socio-political dynamics exert a multilevel feedback on emission trajectories.To better disentangle the effects of specific drivers, we ran a large number of simulations to reconstruct the impacts of scenarios of policy change, opinion dynamics, and various aspects of the PED. Figure 4 shows the main statistics of [CO 2 ] obtained from varying six parameters that characterize the sociopolitical feedback.
It is worth mentioning that in spite of the model's simplicity, the simulations in figures 3(D)-(F) resemble the IPCC's global emission trajectories resulting from various SSPs scenarios.

Impact of different policy and behavioral scenarios
We have shown that the dynamics of socio-political acceptance can shape CO 2 emission trajectories to the extent that can reinforce or undermine climate policies such as carbon pricing or incentives for investments in clean energy.If a decisive top-down policy approach could favor a rapid and nearly immediate reduction of CO 2 concentration, such a reduction would be effective only without nonlinear sociopolitical feedback (black curves in figures 3(A) and (B)).Even in this case, small oscillations in [CO 2 ] are possible due to differences in the intrinsic timescales of social and environmental dynamics.Such timescale differences can shrink in the presence of reinforcing opinion dynamics.
When the natural and the human systems coevolve at very similar time scales, dueling dynamics can emerge and result in periodic oscillations.Under specific circumstances (strict climate policy aimed at immediate emission reduction and mild to intense social interaction), such fluctuations can lead to a second peak of [CO 2 ] much higher than the peak at net zero emissions (figure 3(B)).These oscillations emerge because the acceptance of clean energy investments decreases as [CO 2 ] level declines.If strict top-down climate policies were applied and the level of atmospheric CO 2 rapidly decreased, the sociopolitical acceptance could become negative, meaning that there would be a decrease in clean energy investments.This investment overshooting can be favored and reinforced by non-linear opinion dynamics, undermining further climate mitigation.

Analogy with the socio-political response to the COVID-19 pandemic
Recognizing the significance of co-evolving social and ecological non-linear dynamics, influenced by policy and governance, becomes apparent when the time scales of action and reaction align.An example of this co-evolution is vividly demonstrated in the varied governmental and societal responses to the COVID-19 pandemic.During the first outbreak, governments and health authorities imposed strict and immediate mitigation measures to limit the spread of the SARS-CoV-2 virus.While the measures were often extreme and constrained individual economic and personal freedom, they were still accepted by public opinion as the damage caused by the virus diffusion was perceived to be devastating.However, as soon as the positive effects of the lockdowns, mask mandates, and social distancing started to pay off regarding infection reduction, the perceived damage rapidly decreased, and public opinion began to push for relaxing the restriction measures.The latter resulted in a second (and at times third and fourth) wave of infections with a peak in damage probably even higher than the first one.Notably, these oscillations were mitigated by increasing natural immunity post-exposure and through vaccination efforts, averting further cycles of resurgence and decline [98].
Similar dynamics could, in principle, also emerge from the interaction between CO 2 emission reductions and the socio-political acceptability of such reductions.The combination of strict measures and shift in opinion due to changes in the perceived damage could result in a rebound effect.Pointing out effective transition pathways (avoiding overshooting) and leading to a sustainable, stable state rather than chaotic oscillations (limit cycle) remains a significant challenge.In the context of decarbonization, it is crucial to consider the temporal dimension of emission cuts and their socio-political acceptance, as these may lead to overshooting mitigation efforts or critical thresholds in the Earth's system [24,99].

A multi-level approach to mitigate the effects of opinion polarization
Our results support the general understanding that a top-down climate policy may not be sufficient for effective and long-lasting emission cuts [100,101].While the 2015 Paris Agreement [102] operates more as a framework for global cooperation rather than a strict regulatory mechanism, it already suggests a shift in global climate mitigation policy from top-down to bottom-up [103,104].A multi-level approach with a top-down policy that reinforces bottom-up movements seems the most effective strategy to reach net zero emissions in reasonable times, avoiding rebound effects and opinion polarization.
Opinion polarization in climate mitigation represents a limiting factor for effective emission cuts.Variability in climate change sensitivity and inequalities within socio-political groups could induce political polarization [105][106][107].Political polarization often inhibits support for health and environmental issues [108].In the United States, for example, political polarization emerged during the COVID-19 pandemic due to legislation that challenged the status quo and, to some extent, limited economic and personal freedom [41].This political polarization resulted in diverse approaches to the pandemic, with conservative states taking mild measures while liberal ones using a more cautious and strict approach [109].Similar dynamics and diversity of action can be observed in the context of decarbonization.Besides partisan sorting, polarization could be favored by the divergent opinions pushed by mass media.If overoptimistic scenarios induce inertia, over-pessimistic scenarios [33,110] could also delay actions as they push the public opinion in the direction of unrealistic 'climate doomism' , for which mitigation actions may be seen as long past due [111,112].
The model results show that opinion polarization does not necessarily delay action but amplifies the general tendency to either take action or not.But because of that, opinion polarization can trigger periodic and harmful oscillation in emissions (figure 3(C)).These oscillations occur when strong top-down climate policy is enforced on a polarized socio-political system.However, such oscillations tend to disappear if opinion polarization is low (orange curve in figure 3(C)).Again, we can observe that this dynamical system's response is analogous to the global and national response to the COVID-19 pandemic, with public opinion polarizing either in favor or against strong measures and being one of the factors inducing multiple infection waves.

Impact of delayed actions and EWEs
Society's response to emission increments may not be immediate simply because some of the effects of rising [CO 2 ] take many years to manifest.It is well-known, for example, that the maximum warming and disruption of ecological assemblages can occur years to decades after the emission event and can be abrupt [70,113,114].We show that a delay in the perception of the negative effects of climate change causes a setback in emission cuts (figure 3(D)).
Public opinion can also respond abruptly to climate change impacts due to EWEs [75,115].Sudden increments in public awareness can quickly vanish or last for years.The impact of public sensitivity on EWEs and its memory effect is shown in figures 3(E) and (F).Higher sensitivity to extreme events (or, analogously, more frequent extreme events) can trigger more timely emission cuts than low sensitivity (respectively, green and magenta curves in figure 3(E)).The longer the effects of EWEs are perceived, the faster the transition to clean energy alternatives (figure 3(F)).

Key drivers of emission trajectories in sociopolitical dynamics
While all aspects of the socio-political feedback considered here significantly impact the emission path, opinion polarization seems to be one of the main controlling factors (figure 4).On the one hand, public opinion polarized in favor of mitigation actions can speed up the transition to clean energy.On the other hand, changes in environmental conditions (reduction of CO 2 , and, with it, PED) or political will (policy) can rapidly shift the consensus to a more conservative approach that relies heavily on fossil fuels.
Diverse policy interventions also have a remarkably strong impact on emission trajectories.Under favorable socio-political conditions (e.g.low polarization and low hindrance to change), strong policies may be very effective in reducing atmospheric CO 2 concentration.However, they may be ineffective when facing strong opposition and lobbying.The impact of social interaction, while less marked than policy and opinion polarization, can still substantially reinforce or undermine mitigation actions.Finally, delays in the perception of economic damage, as well as EWEs, could play a role in how and when mitigation actions are taken, thus influencing emission scenarios, as shown in figure 4.

Final remarks
Our stylized model implicitly assumes that the energy market responds elastically to fluctuations in demand for clean energy, although with certain delay and resistance.Under this condition, periodic oscillations in clean energy investments and fossil fuel-related emissions are plausible in different scenarios.However, dynamic changes in the supply-demand balance and technology can have long-lasting and irreversible effects [116,117].While dangerous periodic oscillations in coupled human-natural systems are possible, they only occur under specific circumstances (strong opinion polarization and heavy top-down climate policy), and several intrinsic macroeconomic mechanisms can help prevent or hamper them.
While we focus on aggregated behavior and global emission trajectories, our modeling approach could be expanded to account for multiple agents representing nations, states, or sub-regional entities that decide whether to invest in clean energy based on the local perception of climate change impacts and hindrances to change.Besides clean energy investments, this framework could be expanded to account for other possible actions, such as nature-based solutions, shifts to plant-based diets, and large country effects, highlighting the disproportionate influence of major nations in steering the course of global climate mitigation strategies [118,119].Accounting for different technologies, besides policy interventions, would allow us to model the diverse perceptions of different strategies and solutions to reduce emissions.
Emission cuts are a global common good on which most countries agreed, but, in practice, each country decides if and when to invest in clean energy based on its priorities.In this type of strategic game, perfect collaboration is complex, as many actors could feel enticed to take no action while benefiting from the others' effort (free-riding).However, global emissions can be significantly reduced even in a cooperation scenario where actors with less agency or different priorities are incentivized to take actions [24,120].Recent policy interventions, such as the 2022 USA Inflation Reduction Act, promise to drive historic climate action, not only by investing in green energy infrastructure and combining economic growth with decarbonization, but also by reinforcing reciprocal actions of other emitters, aligning with the principles the Paris Agreement.Unfortunately, we observe that such policy interventions often receive strong opposition from local authorities and people directly affected by the changes.This is what is happening in the European Union, where the 2020 Green Deal is facing backlash from protests in the agricultural and automotive industries.
We describe the coupled dynamics of sociopolitical acceptance of climate change mitigation, risk perception, opinion dynamics and polarization, and top-down international policies.Non-linear sociopolitical processes could influence the emission trajectories for the next few decades.To capture the full spectrum of future scenarios, climate, and Earth system models need to start incorporating such dynamical bi-directional interactions between human and natural systems [12,24,121].

Figure 1 .
Figure 1.Schematic representation of an unbiased double-well potential driving the emergent opinion dynamics behavior.The opinion can shift between two stable states corresponding to general public support for exploitation or mitigation action.This potential for opinion dynamics refers to the dynamics of SA for given values of the environmental and technological variables; when these variables are made fully dynamic, the shape of the potential becomes time-dependent accordingly.

Figure 2 .
Figure 2. Schematic representation of opinion dynamics and different social and behavioral processes that affect the socio-political acceptance of clean energy investments.Panel (A) shows a simplified taxonomy of the subsystems included in our modeling framework [42].Panels (B)-(D) show the potential function affected by different degrees of hindrance to change and perceived economic damage (B), opinion polarization (C), and social interaction (D).

Figure 3 .
Figure 3. Examples of simulated average trajectories and variability of global atmospheric carbon dioxide concentration ([CO2]) under different (A) policy and (B) behavioral, (C) opinion polarization, (D) delay in PED, and (E)-(F) EWEs scenarios.The solid and dashed lines represent the average value of N = 1000 simulation with noise, and the shaded areas indicate the standard error.Panel (A) shows [CO2] over time in case of no climate mitigation policy (blue line), mild (red line), strong (orange line), and immediate emission cuts (black line), corresponding to an ideal strict policy able to achieve a reduction of global of [CO2] in a few years.These scenarios were obtained without accounting for non-linear opinion dynamics, namely setting β = 0. Panel (B) was obtained considering the most extreme policy scenario (immediate reduction) and varying the social interaction in the opinion dynamics.The case of no social interaction (black line) corresponds to the one obtained in panel (A) with β = 0.In contrast, mild (red), strong (orange), and very strong (blue) social interaction scenarios were obtained with β = 0.1, β = 0.25, and β = 0.75, respectively.(C) Shows the impact of opinion polarization under a scenario of immediate emission cuts and mild social interaction (β = 0.25), and varying polarization parameter ϵ to simulate no polarization (ϵ = 1/3; orange line), mild (ϵ = 1; red line), and strong polarization (ϵ = 1.5; blue line).(D) Displays global atmospheric carbon dioxide concentration ([CO2]) accounting for a delay ∆t between emissions and perceived impacts up to 20 years.The dashed blue curve represents the reference trajectory without delay, while the red and orange curves display the [CO2] trends with 10 and 20 years delay, respectively.(E) shows the impact of sensitivity to EWEs.The trajectories of increasing sensitivities are obtained by increasing the amplitude of the jumps in perceived damage (αEWE).Finally, (F) explores the memory effect of EWEs.Different trajectories were obtained by varying KEWE.

Figure 4 .
Figure 4. Impact of different policy scenarios, social interaction, opinion polarization, delay in perceived economic damage, sensitivity to and memory of extreme weather events on atmospheric [CO2].In the box plot, the red central marks indicate the median, and the bottom and top edges display the 25th and 75th percentiles, respectively.The whiskers extend to twice the interquartile range of the simulated values.