Between Uncertainty and Responsibility: A Philosophical Inquiry into Climate Change Projections

Abstract

This paper explores how uncertainty in climate change projections both shapes and is shaped by key epistemic, methodological, ethical, and political concerns. Drawing on a review of key philosophical sources, it examines the relationship between scientific objectivity and the influence of social, ethical, and political values in contexts of deep uncertainty. Authors such as Wendy Parker and Heather Douglas debate the role that social values play in the estimation and communication of scientific uncertainty, particularly when decisions carry significant ethical and political consequences. At the same time, several studies emphasize that, beyond the influence of values, there are structural limitations inherent to complex climate models that prevent uncertainty from being fully reduced. Taken together, these perspectives suggest that both evaluative judgments and technical constraints must be considered when interpreting and managing uncertainty in climate science, especially insofar as it informs collective decision-making processes. The article also examines how certain institutional practices tend to downplay uncertainty, generating biases that affect both scientific communication and public decision-making. It, therefore, explores potential solutions through more integrative approaches, such as robust modeling, risk assessments focused on low-probability but high-impact events (HILL), and collective ethical deliberation. The paper further discusses the concept of normative uncertainty, illustrated through the case of the Tempisque River water conflict, which highlights the difficulty of reconciling competing values. It concludes that, far from being eliminated, uncertainty must be managed through tools that integrate technical rationality, ethical sensitivity, and adaptive governance.

1. Introduction

Contemporary philosophy of science has found in climate change a fertile ground to examine the relationships between knowledge, uncertainty, values, and political decision-making. While climate science has achieved robust consensus regarding the reality of global warming and its anthropogenic causes, the predictive capacity of climate models faces profound levels of uncertainty. This uncertainty is not solely due to technical or statistical limitations, but also to the fact that these models attempt to represent complex, nonlinear, and highly interdependent natural phenomena. In this regard, climate prediction stands as one of the scientific fields with the most extensive experience in managing uncertainty, precisely because it has had to confront the limits of knowledge when modeling dynamic and open systems such as the atmosphere and the ocean. This recognition has led to a stronger adoption of the precautionary principle as a guiding criterion for assessing risks and making decisions under uncertain scenarios.

In this context, this article argues that uncertainty in climate projections should not be understood merely as a technical problem, but rather as a structural challenge that demands integrated responses from epistemology, ethics, and governance. It contends that, given the impossibility of eliminating uncertainty, the crucial task is to conceptualize and manage it through tools that combine rational judgment, ethical deliberation, and institutional prudence. This requires, among other strategies, the strengthening of broad deliberative processes that incorporate diverse perspectives and acknowledge the value-laden nature of many climate-related decisions. The article develops this thesis across four sections: a discussion of the role of values in scientific interpretation; a review of methodological approaches to structural uncertainty; a normative analysis centered on the concept of normative uncertainty; and a case study of the water conflict in the Tempisque River, Costa Rica, which illustrates how such tensions manifest in concrete decision-making contexts. This case study draws on empirical research conducted by the authors in the Tempisque River Basin, including hydrological modeling and participatory engagement with local stakeholders [1]. Finally, the article proposes a philosophical agenda aimed at integrating responsibility, deliberation, and adaptability in decision-making under climate uncertainty.

2. Methodological Approach

This article adopts a philosophical methodology grounded in systematic conceptual analysis and integrative literature review. Rather than attempting to provide an exhaustive mapping of different philosophical positions on climate uncertainty, it aims to clarify key conceptual, epistemological, and normative dimensions involved in the use of climate projections for decision-making.

The literature reviewed was identified through searches conducted in academic databases such as Scopus, Web of Science, PhilPapers, and MDPI’s own repository. The approach outlined is identified with keywords that include combinations of terms such as “climate models,” “uncertainty,” “epistemology,” “robustness,” “values in science,” “normative assumptions,” “post-normal science,” and “philosophy of climate change”. Inclusion criteria prioritized works that directly engage with epistemological and normative questions in climate science, or with methodological issues related to modeling and decision-making under uncertainty. Recent contributions by leading scholars such as Lloyd, Intemann, Parker, Winsberg, Norton, Frigg, and Saltelli were explicitly considered due to their influence and relevance within the field [2,3,4,5,6,7,8,9].

The article combines integrative review methods with conceptual and comparative analysis. It draws on established philosophical approaches, including systematic philosophical analysis (as proposed by Dellsén [10]), conceptual analysis frameworks (e.g., Machery [11]), and integrative literature review strategies tailored to philosophy (cf. Torraco [12]). This framework supports the clarification of key conceptual distinctions—such as those between types of uncertainty or models of validation—and enables comparison between theoretical positions based on their assumptions, implications, and normative commitments. While no formal comparison table is included, the analysis is structured through an implicit comparative logic.

The scope of this review focuses on contemporary philosophical debates concerning climate science, particularly those developed between 2010 and 2024. Although the literature reviewed is primarily situated within Western academic traditions (notably Anglo-American philosophy of science), the analysis acknowledges the importance of broadening philosophical engagement beyond this scope. Related fields—such as science and technology studies (STS), environmental ethics, and climate policy studies—are drawn upon when relevant to the argument. Some philosophical approaches—such as pragmatist or phenomenological perspectives on climate risk, or those rooted in Indigenous or non-Western epistemologies—are not examined in depth. These exclusions result from the article’s specific focus and space limitations but represent important directions for future inquiry.

3. Epistemological and Normative Dimensions of Climate Uncertainty

The models used to predict climate behavior have been carefully examined by philosophers who question how reliable the knowledge they produce truly is, especially regarding the role of values in the estimation, interpretation, and communication of uncertainty. According to Eric Winsberg, uncertainty estimates in climate predictions are inevitably infused with social values, as normative judgments intervene throughout the process—from problem selection and prioritization, to analytical framing, interpretation, and the communication of results—which cannot be considered purely technical [6].

These concerns have been further developed in the recent literature that addresses the epistemological, institutional, and methodological dimensions of climate modeling under conditions of structural uncertainty. These contributions have significantly expanded the debate on the role of values in climate science, focusing on both the modeling practices themselves and the institutional frameworks within which they operate.

One of the most relevant perspectives comes from the theory of post-normal science, which argues that when facts are uncertain, values are in dispute, decisions are urgent, and stakes are high, traditional scientific approaches become insufficient. Recently revisited by Saltelli [9], this framework calls for a deliberate recognition of uncertainty and promotes its governance through inclusive deliberative processes involving an “extended peer community.” Rather than concealing complexity, post-normal science seeks to confront it through epistemic transparency, reflexivity, and interdisciplinary dialog.

From an institutional perspective, studies on the science–policy interface have shown that the architecture of knowledge exchange shapes both the legitimacy and the effectiveness of policy decisions. In the context of European climate policy, Dupont et al. [13] found that varying levels of politicization affect the mechanisms through which expert knowledge is integrated into political processes. In highly polarized contexts, scientific input is often either instrumentalized or excluded, highlighting the importance of designing institutions capable of maintaining critical and non-utilitarian forms of expertise integration.

Moreover, recent philosophical contributions have emphasized the need to evaluate climate models not solely based on their representational accuracy, but on their adequacy for specific purposes. Parker [5], for instance, argues that models should be assessed according to the epistemic and practical goals they are intended to serve—such as explanation, prediction, or policy support—rather than their correspondence to an idealized climate system. This purpose-driven evaluation aligns with the normative dimensions of model use in decision-making under deep uncertainty.

An important methodological development in this context is the application of robustness analysis to climate modeling. According to Harris and Frigg [7,8,14], a model output can be considered robust when it is consistently reproduced by a diverse set of models. However, they caution that such inferences are not deductively valid and require epistemic justification. Their framework distinguishes between “top-down” strategies—based on assumptions about the ensemble of models—and “bottom-up” justifications—derived from the properties of individual models. While robustness analysis can increase confidence in climate projections, it must be applied critically, acknowledging the shared structural assumptions that may lead to false convergence.

Finally, Winsberg [6] has synthesized many of these developments within a broader philosophical framework, emphasizing that climate models function not only as tools for representing physical systems, but also as normative devices that mediate between facts, values, and societal interests. For Winsberg, uncertainty is not merely a limitation to be reduced, but a structural feature of climate science that requires ethical and epistemic deliberation [6].

This position has not gone unchallenged. Wendy Parker [5] offers a nuanced critique of Winsberg’s interpretation. While she agrees that values can influence climate predictions, she believes Winsberg overstates their impact. She warns that taking this idea too far risks suggesting that all science is inevitably shaped by personal or social biases, a view she considers inaccurate. Parker acknowledges that, in certain instances, scientists make choices—such as which data to use or how to present their findings—that may be influenced by values like prudence or concern for collective well-being. However, she clarifies that this is not the case in all instances or across all levels of scientific work. Therefore, she cautions against generalizing or claiming that all scientific knowledge is necessarily value laden.

This tendency to overstate the influence of values is partly shaped by positions such as that of Heather Douglas [15,16], who argues that scientific decisions under conditions of uncertainty should incorporate social values due to the potential consequences of error. While this argument is highly relevant in socially sensitive contexts such as public health or climate change, Parker [5] maintains that, in many cases, scientists possess methodological resources that allow them to mitigate—or sometimes avoid—direct influence from personal, social, or political values unrelated to the scientific evidence itself.

This is the case, for instance, when uncertainties can be statistically represented or when comparative evaluations can be made using multiple independent models. Parker [5] and Douglas [15,16] also caution that exaggerating the role of values in science may foster a misleading sense of precision, particularly in contexts marked by deep, irreducible uncertainty. Such situations are common in long-term climate projections.

Overemphasizing the influence of values also neglects the existence of internal mechanisms within the scientific community that help control and correct errors. Processes such as peer review, replication, and rigorous methodological scrutiny serve as safeguards against the domination of scientific decisions by personal opinions or external interests. As Elliott [17] notes, while values are indeed present in science, their influence must be kept in check through standards such as logical coherence, data quality, and procedural transparency.

In sum, Parker [5] does not argue that science should be completely free of values—as if that were possible or even desirable. Rather, she advocates for a more balanced view, in which the presence of values in scientific work is acknowledged without undermining objectivity. Exaggerating their influence may have the unintended consequence of eroding public trust in science, even when such distrust is unwarranted. This is particularly problematic in high-stakes contexts such as climate change, where confidence in scientific evidence is essential for coordinated and effective action.

In this regard, Sherwood et al. [18] have conceptualized climate uncertainty as a cascade of interacting sources, where each layer contributes to the amplification of total uncertainty in model projections. This cascade begins with internal climate variability—chaotic fluctuations that occur even in the absence of external forcing—and continues with uncertainty about future emissions scenarios, which reflect unpredictable socioeconomic and political dynamics. It culminates in structural uncertainties arising from limitations in model design, such as parameter choices and simplified process representations. These layers are not isolated but interdependent, meaning that uncertainties at one level can propagate and intensify through subsequent stages of the modeling chain. Understanding this cascade is essential for grasping the complex epistemological and communicative challenges involved in climate forecasting.

More recent studies have explored this issue in greater depth. For instance, Kimpton et al. [19] argue that uncertainty is not merely the result of data scarcity, but is inherent to the operation of complex models, both in climate science and in fields like healthcare. This is due to the inevitable gap between what a model represents and the real-world system it seeks to simulate. Moreover, there are intrinsic limits to what can be calculated or predicted with precision. For this reason, the authors suggest that rather than focusing solely on model accuracy, it is equally important to assess their reliability. They recommend rigorous procedures to verify, validate, and quantify the different levels of uncertainty associated with each model.

Rather than striving for fully precise and deterministic predictions, some experts propose combining different approaches—such as simulation, statistical modeling, and multi-scale frameworks—to develop tools that, even if they do not yield exact answers, are nonetheless robust and clear enough to support sound decision-making, especially in high-stakes contexts [20].

On the other hand, Judith Curry and Peter Webster [21] use the metaphor of the “uncertainty monster” to describe how scientific institutions react to the complexity of climate change. According to them, rather than openly acknowledging the limits of scientific knowledge, there is often a tendency to conceal, downplay, or ignore them in order to maintain an appearance of certainty. This can be observed, for instance, in reports by the Intergovernmental Panel on Climate Change (IPCC), which—although they appear neutral and technical—actually involve implicit decisions about which possible futures are prioritized and how such futures should be communicated to the public [5]. Thus, the critique is not only aimed at the way science is conducted but also at how it is used to shape political decision-making.

Building on the concerns raised by Curry and Webster [21] about the institutional minimization or concealment of uncertainty in climate science, Sherwood et al. [18] propose a more fundamental shift. Instead of focusing primarily on “average” or most likely scenarios, they advocate for a climate science that explicitly attends to low-probability risks with potentially severe consequences. While they do not use the term directly, their concerns align with what has been elsewhere described as High Impact, Low Likelihood (HILL) scenarios. The authors argue that such events, though rare, must not be dismissed precisely because of their potentially catastrophic implications [18].

Sherwood et al. [18] further argue that current reports, including those by the IPCC, still rely heavily on models grounded in linear assumptions, predefined scenarios, and disciplinary silos. According to the authors, this limits the ability to anticipate real climate risks, particularly those involving complex interactions and feedbacks that are not easily captured by conventional models. For instance, the IPCC itself warns that exceeding 1.5 °C increases the risk of crossing climate tipping points, such as the abrupt collapse of the Greenland or Antarctic ice sheets [22], and that permafrost thawing may trigger nonlinear state changes [23]—risks that could be underrepresented in models designed for gradual projections.

They therefore recommend adopting more diverse and integrative approaches—for example, developing narrative storylines of possible futures, conducting participatory climate simulation games, and working with interdisciplinary teams. Such approaches are consistent with the present reflection, not only because they improve preparedness and decision-making in the face of extreme scenarios, but also because they enhance communication. In contrast to highly technical reports designed by and for experts, these methods promote more accessible and context-sensitive forms of conveying uncertainty.

These conceptual and normative reflections, while illuminating, remain incomplete without considering the institutional contexts in which climate science is produced, communicated, and used. To fully understand how uncertainty is framed and managed in practice, it is essential to examine how organizational incentives, policy pressures, and communication norms shape scientific representations. The next section explores these institutional dynamics and their role in the systematic downplaying of uncertainty.

4. Making Just Decisions Amid Uncertainty

The challenge of scientific uncertainty is not merely about assessing whether data are reliable; it also entails an ethical dimension. This becomes especially salient when science-based decisions may have serious consequences for society, as is the case with climate change. In such contexts, some authors—such as Mathias Frisch [24] and Behnam Taebi et al. [25]—argue that scientists should not limit themselves to communicating probabilities or couching their findings in conditional statements such as “if this happens, then that might follow.” Instead, they should assume a more active moral responsibility by taking a stand on which options or solutions are ethically acceptable.

This position has gained traction over time. In 1953, Richard Rudner [26] argued that accepting or rejecting a scientific hypothesis cannot be separated from the potential consequences of being wrong. Heather Douglas [16] revived this discussion in 2009 and applied it to cases such as climate science, where values like justice and safety inevitably influence standards of evidence. However, Douglas contends that scientists can, in many cases, limit their role to clearly communicating the degree of certainty in their findings, leaving normative or policy decisions to other actors [16].

Frisch [24], by contrast, maintains that in contexts of high uncertainty but elevated risk, scientists have a duty to issue clear normative judgments that can serve as moral guidance. This view is supported by Hopster [27], who argues that when science does not provide clear answers, moral confusion intensifies, making it even more urgent to have well-defined ethical principles for responsible decision-making.

Contrasting with the view that scientists should directly take on an active ethical role—as proposed by Frisch [24] and Hopster [27]—Taebi et al. [25] offer an alternative perspective. They argue that, in many cases, the true challenge lies not in data scarcity but in the presence of conflicting moral values—such as social justice, environmental sustainability, and economic development—which give rise to what they term normative uncertainty: situations where multiple ethically legitimate courses of action are difficult to reconcile and it remains unclear which is morally preferable.

Instead of asking scientists to make moral decisions on their own, Taebi et al. [25] propose the creation of collective deliberation spaces in which diverse voices and values can be heard and negotiated. This approach, which they term reflective equilibrium, is a flexible and participatory process that allows decisions to be revised over time as both information and ethical priorities evolve. In this way, rather than relying on a single expert judgment, decision-making becomes more democratic and open to dialog, even enabling the ethical values at stake to develop over time. Thus, uncertainty is not seen merely as a limitation but as an opportunity to enrich ethical debate and strengthen the legitimacy of scientific decisions in complex contexts such as climate change.

Some authors have proposed using expected utility maximization as a tool for decision-making under climate uncertainty, given that it is a traditional approach in contexts of incomplete information. This method, formulated by Von Neumann and Morgenstern [15], involves calculating which option yields the greatest average benefit, weighting each possible outcome by its probability. However, its application to climate change has been questioned. Heal and Millner [28] warn that this approach requires well-defined and reliable probability estimates—something that is often unattainable in complex climate scenarios where uncertainty is deep and difficult to quantify. Therefore, although useful in some settings, expected utility maximization faces important limitations when applied to decisions under profound uncertainty, such as those posed by climate change.

Therefore, in the context of climate change—where neither certainty nor well-defined probabilities are available—what applies is the concept known as deep uncertainty or structural ambiguity. These refer to situations where we not only lack knowledge about outcomes but also about the governing rules of the system. In such cases, as Heal and Millner [28] and Xepapadeas [29] have shown, classical mathematical formulas are insufficient. Instead, alternative approaches are needed, such as robust models, which perform reasonably well even under uncertain conditions, or strategies like the maximin principle (which focuses on avoiding the worst-case scenario), as well as the use of multiple probabilistic assumptions to explore diverse possible futures. These methods allow for more prudent decision-making when full confidence in predictions is not feasible.

A particularly illustrative case of normative uncertainty—that is, uncertainty not about facts but about what ought to be carried out—can be found in the ongoing conflict over water use in the Tempisque River Basin in northwestern Costa Rica. The analysis presented here builds on empirical research led by the authors, including hydrological modeling and participatory processes with local stakeholders, conducted between 2019 and 2024 [1]. This region has experienced multiple episodes of severe drought, notably during the extended dry periods of 2014–2016 and 2019–2020, which have been linked to El Niño events and broader shifts in regional climate patterns [1,30]. These droughts intensified competition between agricultural, ecological, and human consumption needs.

The dispute centers around how to allocate increasingly scarce water resources, and reveals a clash between different value frameworks: some prioritize economic productivity and irrigation for large-scale agriculture (primarily sugarcane and rice), while others emphasize ecological conservation, Indigenous water rights, and access to water for rural communities [1]. While the hydrological data and climatic projections inform all parties, they do not resolve the deeper normative disagreements about fairness, sustainability, or long-term responsibility.

This case exemplifies normative uncertainty because even if there were full agreement on the facts—e.g., rainfall deficits, river discharge rates, or crop water demands—there would still be reasonable disagreement about which principles should guide water governance. Should priority go to economic factors that contribute most to GDP? To communities with historical claims to water access? To ecosystems that sustain biodiversity and buffer climate impacts [1]?

By illuminating the entanglement of facts and values, the Tempisque river case demonstrates that climate uncertainty is not only epistemic but also normative and institutional. It underscores the importance of participatory deliberation processes that can surface conflicting values and seek negotiated resolutions, rather than pretending that technical expertise alone can provide optimal answers [8,25,31].

Each of these scenarios is morally defensible—and even technically plausible—but they involve competing values that are difficult to balance. Rather than being strictly incompatible, their tension lies in how different priorities—such as satisfying basic needs, preserving ecological integrity, and promoting distributive fairness—are weighed against each other in decision-making. For instance, prioritizing human consumption aligns with the principle of meeting vital needs, while safeguarding environmental flows supports ecological and intergenerational justice. Promoting equitable water distribution enhances fairness and inclusion, yet may pose risks to long-term sustainability. None of these options is entirely unproblematic once the potential irreversible trade-offs are considered—such as biodiversity loss, reduced agricultural productivity, or increased social conflict. This makes water governance under climate uncertainty a paradigmatic case of normative tension, where decisions must navigate not just factual constraints but also conflicting ethical commitments.

Faced with such deep uncertainty, as emphasized by Taebi et al. [25], broad deliberative processes are needed—ones that integrate diverse actors and enable adaptive governance mechanisms capable of revising decisions as both information and values evolve. This approach emphasizes the need to open up decision-making processes to a plurality of values, avoiding both technocratic scientism and ethical relativism.

Taken together, these approaches not only provide technical responses to structural uncertainty but also underscore the ethical imperative to develop decision-making frameworks that recognize value plurality, promote fairness, and acknowledge the irreducible unpredictability of future climate trajectories. Rather than merely presenting ranges of risk or relying on ensemble averages, the challenge lies in embedding uncertainty into more flexible, inclusive, and reflexive decision-making systems.

Recent critiques have brought to the fore deeper epistemological and normative limitations embedded in current climate modeling practices. Stainforth [32] warns that ensemble modeling approaches, while powerful, may foster a false sense of epistemic closure by blending divergent model outputs into a misleadingly unified projection. Such practices risk masking deep uncertainties—those that arise not from data scarcity but from structural limitations in our scientific understanding of complex, dynamic systems. He argues that climate modeling must distinguish more clearly between what is empirically robust and what remains conjectural, especially in long-term projections where validation is impossible.

Complementing this view, Thompson [33] critiques the methodological detachment that arises when modelers become trapped in what she metaphorically terms “Model Land”: a conceptual space where the internal coherence of mathematical models supplants their empirical correspondence with reality. Within Model Land, models may appear rigorous, but this rigor can obscure their normative content, political implications, and epistemic limits. Thompson insists that decision-relevant modeling must confront the boundary between quantifiable risk and unquantifiable uncertainty. She advocates for “escaping” Model Land by reintegrating qualitative judgments, contextual knowledge, and ethical reflection into model development and use—especially in high-stakes domains such as climate policy, where numerical outputs can carry disproportionate rhetorical and political weight [33].

These concerns intersect powerfully with the work of Rubiano Rivadeneira and Carton [31], who approach climate modeling from a justice-oriented perspective. Their review reveals how Integrated Assessment Models (IAMs), despite their prominence in informing climate mitigation scenarios, often embed normative assumptions that go unacknowledged and unexamined. By relying heavily on technoeconomic optimization frameworks—particularly cost–benefit analyses—IAMs tend to obscure the distributive, recognitional, and participatory dimensions of climate policy. The authors argue that IAMs frequently prioritize efficiency over equity, marginalizing alternative worldviews and reinforcing dominant geopolitical and economic power structures [31]. This results not only in technocratic policy prescriptions but also in the exclusion of epistemologies from the Global South, Indigenous knowledge systems, and broader public deliberation. Their justice-based critique calls for a reframing of climate modeling practices to include cognitive and epistemic justice, making space for diverse ways of knowing and valuing climate futures.

Taken together, these critiques converge on a fundamental insight: that climate models are not neutral instruments, but socio-technical constructs embedded within particular political and epistemological regimes. Addressing the limitations and biases of climate modeling thus requires more than technical refinement—it demands epistemic humility, methodological pluralism, and ethical reflexivity. Models must not only inform policy but do so transparently, acknowledging whose futures they envision, whose voices they exclude, and what normative assumptions they carry.

5. Conclusions

Uncertainty in climate change projections cannot be reduced to a mere lack of data or methodological shortcomings; rather, it constitutes a structural challenge that permeates the epistemology, ethics, and politics of climate science. The appropriate response is not to eliminate uncertainty, but to recognize, conceptualize, and manage it through tools that integrate rational judgment, ethical deliberation, and institutional prudence.

Beyond theoretical analysis, the discussion of uncertainty in climate projections also invites reflection on possible strategies for action. Several approaches can be derived from the literature reviewed. First, the use of robustness analysis, when critically applied, offers a way to strengthen confidence in model-based inferences by identifying patterns of agreement across structurally diverse models. While not a substitute for probabilistic precision, robustness reasoning can support policy under deep uncertainty—provided its epistemic limitations are acknowledged.

Second, the integration of HILL-type reasoning (High-Impact, Low-Likelihood) into climate risk assessment highlights the ethical imperative to consider low-probability but catastrophic outcomes. This calls for precautionary strategies and multi-scenario planning, even when epistemic support for extreme outcomes is limited.

Third, scholars working within the post-normal science framework emphasize the need for collective deliberation and participatory decision-making processes. When uncertainty is structural and stakes are high, broadening the epistemic community—beyond technical experts—to include affected stakeholders, local knowledge holders, and the public enhances both the legitimacy and robustness of policy responses.

Moreover, the implementation of these strategies faces concrete institutional, political, and communicative challenges. Robustness analysis may require computational and methodological resources not always available in policy settings. HILL-based approaches risk being marginalized by cost–benefit frameworks that discount extreme events. Deliberative processes, while normatively appealing, require careful design to avoid tokenism or decision paralysis. Recognizing these tensions is crucial to translating philosophical insights into actionable guidance for climate governance under uncertainty.

Proposals that advocate transparent communication of uncertainty ranges, the incorporation of values into scientific decision-making, and the development of robust models of governance represent the most promising pathways for addressing this challenge. Engaging responsibly with climate uncertainty can, in fact, lay the groundwork for a widely shared risk management culture—one in which not only technical institutions and decision-makers, but also the broader public, understand and embrace their roles in the face of structural uncertainty.

By connecting conceptual, epistemological, and normative analyses with proposals for practical engagement, this article contributes a distinctive philosophical framework for approaching uncertainty in climate science. Unlike approaches that treat uncertainty as a merely technical or communicative deficit, this paper emphasizes its structural and value-laden character, calling for integrated responses that bridge philosophy, ethics, and governance. In doing so, it helps to reframe the debate from a focus on prediction and control toward a richer account of responsibility, deliberation, and adaptive management under conditions of deep uncertainty.