I’d like to thank audiences at the Princeton Environmental Institute, the International Association for Environmental Philosophy and International Society for Environmental Ethics joint conference in Allenspark, Colorado in 2004, and also at the University of Utah. I am especially grateful to Chrisoula Andreaou, Paul Baer, Roger Crisp, Nir Eyal, Geoffrey Frasz, Dale Jamieson, Josh MacLeod, John Meyer, Elijah Millgram, Klaus Nehring, Michael Oppenheimer, Peter Singer, Cass Sunstein, Elizabeth Willott and an anonymous referee for this journal for comments on various versions of the paper. I thank the Center for Human Values at Princeton University and the University of Washington for research support.
Jordan and O’Riordan 1999, p. 22.
“Vorsorge” means “foresight” or “taking care;” the “Vorsorgeprinzip” is the “foresight principle.” At the core of early conceptions of this principle in Germany was the belief that society should seek to avoid environmental damage by careful “forward-looking” planning, blocking the flow of potentially harmful activities. The Vorsorgeprinzip has been invoked to justify the implementation of vigorous policies to tackle river contamination, acid rain, global warming and North Sea pollution. See Raffensberger and Tickner 1999b, p. 4.
For example, the Wingspread Statement (1998) declares that “existing environmental regulations and other decisions, particularly those based on risk assessment, have failed to adequately protect human health and the environment, as well as the larger system of which humans are but a part.”
Soule points out that despite the fact that Robert Shapiro, the CEO of Monsanto, is “perhaps the most passionate proponent” of genetically-modified food crops, he warns: “when you start talking about large-scale introduction of dramatic traits in combination with each other, you are dealing with systems that are so complicated that no one can effectively model them. You can start with running field trials, just as when you introduce a new drug you run clinical trials to see if people really keel over. But, just as the human body is a subtle and complicated thing, it may be that only one time in a million some side effect happens” (Soule 2000, p. 311).
The Wingspread Statement (1998) asserts that “there is compelling evidence that damage to humans and the worldwide environment is of such a magnitude and seriousness that new principles for conducting human activities are necessary.” On my view, the conflict between cost-benefit analysis and the precautionary principle need not be deep, and may be more apparent than real. See Gardiner forthcoming.
Prestowitz 2003. Prestowitz describes himself as “a former Reagan administration trade hawk.” However, conservative resistance to the principle is not universal. Raffensberger and Tickner report that in 1998 the Indiana Republican Committee adopted the precautionary principle as the basis for its environmental platform (Raffensberger and Tickner 1999b, p. 9).
The label “ultraconservative” is appropriate, since the triggers are conceived of in the most minimal way possible, whilst the response is conceived of in the most robust way possible.
An anonymous referee points out to me that some proponents of precaution have used rhetoric that may suggest the UCPP. He cites the following claim by a Greenpeace spokeperson on BBC2 Nature, cited in North 1995, p. 256: “The precautionary principle is a principle that puts the burden of proof onto the polluter rather than the environment. If a polluter cannot prove that what he is discharging will not damage the environment and will not harm the environment then he simply isn’t allowed to discard that sort of waste.”
Jordan and O’Riordan 1999, p. 32.
ibid., p. 16.
ibid., p. 15.
ibid., p. 16.
Some proponents of the precautionary principle have ambitions of this type. For example, Raffensberger and Tickner say: “Silent Spring was a call to forestall [the] rapid development and deployment of pesticides and to return to an ethic of working with nature, not against it. That is the ultimate goal of precaution” (Raffensberger and Tickner 1999b, p. 11). And another author in the same volume claims: “Implementing precaution is not a discreet task but is intimately enmeshed in the development of an ecological society” (M’Gonigle 1999, p. 142). Both of these are grand ideals, suggesting that the precautionary principle is derived from a wider perspective on environmental ethics and policy.
“[P]recaution is a culturally framed concept that has evolved along different pathways and at different rates in different countries. Searching for a single, all-encompassing definition is, therefore, likely to be a fruitless endeavor because individuals will never agree upon what is or is not precautionary in a given situation. Cultural theory tells us that there is no one single context of risk perception. We all “‘see’ the world in a different way, although four broad archetypes can be distinguished. . . . So those who regard the environment as inherently robust and capable of withstanding sustained human impact will tend to be less precautionary than those who regard human impact on nature as unpredictable and potentially calamitous. These value positions are deeply entrenched, and scientists and policy makers need to be more sensitive to this when they communicate risk to the body politic” (Jordan and O’Riordan 1999, p. 18).
Jordan and O’Riordan 1999, p. 19.
This result seems welcome. For one thing, many enthusiasts of precaution, including Jordan and O’Riordan, believe that democratic consultation is a core component of the approach. For another, it suggests that, though the precautionary principle itself is empty, it is empty in a useful kind of way, and one that is amenable to practical resolution.
The pessimism rests on some questionable assumptions and a dubious argument. Jordan and O’Riordan appear to reason as follows:
(1) Precaution has evolved differently in different countries
(2) Therefore, individuals will never agree on what is precaution in a given situation.
(3) Therefore, search for a single, all-encompassing definition is likely to be fruitless.
But this argument is disputable. First, the meaning of premise (1) is unclear. This is partly because Jordan and O’Riordan stress commonalities as well as differences between cultures in their understandings of precaution. In particular, they assert not only that there are different pathways which countries can share, but that there is something like progress along a pathway, so that different countries can progress at different rates along the same pathway. Second, this undermines the crucial claim in premise (2). For one thing, the existence of common pathways and rates of progress along the same pathway would suggest the possibility of convergence in at least some cases. For another, many environmental problems, and especially global ones, may constitute new contexts for thinking about precaution, and ones that different countries can share. Third, the inference to premise (2) is in any case invalid. It commits a genetic fallacy: one cannot infer from the fact that people's beliefs have different origins that there is no prospect of their coming to an agreement. (For example, there might be independent standards to which they might refer in generating new beliefs, such as other beliefs that they have (e.g., scientific beliefs).) Fourth, even if the conclusion in (3) could be established on other grounds, this would not necessarily defeat the precautionary principle, since it does not requires “a single, all-encompassing definition” of precaution. For example, it might be enough for many purposes that the precautionary principle is shown to be reasonable in a certain set of core cases; and there is a much better chance of agreement here.
Consider some relevant considerations. First, Jordan and O’Riordan make strong claims about attitudes to risk: that risk perception is a “deeply cultural phenomenon;” that this is because of “entrenched values;” and that this means that precaution is “essentially contested,” in the sense that issues about it are not resolvable by appeal to a set of commonly recognizable standards or cases. But this is all disputable. Second, their claims about environmental value need to be questioned. For one thing, it is not clear why issues of risk should be interpreted as issues of value (as opposed, say, to issues of rationality); for another, even if it were, it is not clear why this would make practical issues which invoke such attitudes irresolvable. (Of course, it is likely that Jordan and O’Riordan's claims simply emerge from a more through-going skepticism about value. But many environmentalists would want to resist such a metaethics. More importantly, we ought not simply to build such assumptions into a general analysis of the precautionary principle.) Third, it is not clear why we should accept that all that is at stake in such matters is “cultural framing.” For example, is it really true that any possible view about the fragility of the earth is acceptable? Science surely plays some role here. For example, many environmental philosophers seem to think that ecology in particular is transformative of our ethical views. Finally, we could accept some level of cultural framing and not give up the precautionary principle. After all, what counts as a harm and scientific uncertainty is a matter external to the precautionary principle itself. Hence, there might be ways to fit such factors into a precautionary account other than by giving up on the core notion of precaution.
For example, the destruction of the earth ought not be the necessary result of applying the precautionary principle.
Soule 2000 , p. 313.
On Soule's view it is not clear how this is to be done. The most likely candidate, CBA, is ruled out, since Soule says that the WPP can conflict with CBA. Perhaps the idea is that there is a process of intuitionistic weighing. But then, given that he says nothing about either what should be weighed or how the weighing should be understood, it is unclear why Soule is as sympathetic as he is to the WPP. For example, why does he say that it is “innocuous”? Surely there is some chance that the process may be hijacked by the personal agendas of the regulators.
Soule (2000, p. 313) offers as an example of the WPP a communication from the Commission of the European Communities.
ibid., p. 315.
Soule does think the WPP can be problematic in international contexts.
For example, it provides no way in which to guard against the ultraminimal or ultraconservative interpretations of the precautionary principle, and so no defense against the charge of extremism. Furthermore, it does not seem distinct from CBA in the way proponents of the precautionary principle envisage. For, although it allows for the possibility of acting against CBA, this is simply a result of its general permissiveness about how decisions are justified: it is equally true that it does not exclude the possibility that the CBA will be one's only basis for decision-making (since the CBA allows for an accounting of risks, in terms of probability, and then a weighing, as the WPP envisages).
Soule 2000 , p. 317.
One common objection is that it relies on an incoherent attitude to science: on the one hand, proponents of the SPP want to say that there is enough scientific evidence to show that a given policy poses genuine risks; on the other hand, they claim that the science is too impoverished to establish that the policy is safe. Hence, opponents accuse advocates of SPP of wanting to have it both ways with science, and claim that this is incoherent. This criticism should be rejected. As Soule argues, there are often different evidentiary demands for demonstrating risk versus demonstrating safety (Soule 2000, p. 319).
There are actually two concerns with agrochemicals. One—which the current objection relies on—is about their known environmental costs. But another is that they may pose environmental risks of their own which are comparable in importance and uncertainty to the risks of GMPPPs. Soule emphasizes the second concern in the latter part of his article.
Soule 2000, p. 324.
I learned this specific example from T.H. Irwin. For similar illustrations, and a more general discussion, see Freeman (2003, pp. 14–18).
By this I mean not only that it fits our behavior but that we continue to endorse such behavior even on learning of the problems of applying the MP more generally.
Harsanyi 1975, p. 39.
ibid., p. 40.
Situations where there is a very low probability of loss, and a very high probability of substantial gains.
Rawls (1999b, p. 133) says: “Clearly the maximin rule is not, in general, a suitable guide for choices under uncertainty. But it holds only in situations marked by certain special features.”
In the technical jargon of economics, this makes the situation one of genuine uncertainty, not mere risk.
Rawls 1999b, p. 134.
In particular, it seems plausible that Rawls’ restricted maximin covers a much more limited domain than the idea of precaution.
One important issue is to what kinds of circumstances the principle can be applied. Rawls (1974, p. 226) himself asserted that the maximin principle ought not to be used for “small-scale situations.” My own view is that in environmental policy the precautionary principle should apply in the first instance at the global level. I hope to explain and develop this view in future work.
Other approaches to decision under uncertainty include Savage's minimax risk criterion and the Hurwicz pessimism-optimism index. See Luce and Raiffa 1957, pp. 278–86.
This is partly because this is all I am in a position to make. But it is also because I do not want to get bogged down in, or distracted by, the details. Since my main aim is to defend the plausibility of some form of core precautionary principle through defending the criterial approach as such, I want my argument here to be as general as possible, and from this point of view it is good to leave many details to be filled in later, because this could be done in a variety of different ways and still be compatible with the main aim.
One principle which would yield this would be that of maximax (trying to maximize the maximum possible outcome). Maximax might be rational under other conditions, such as where we don’t care much about losses, but do care a lot about gains. But it does not seem rational under the Rawlsian conditions. A more likely principle would be that of trying to maximize the value of the average outcome. But without probabilities to guide one's assessment this seems like an odd strategy. If the possibility set were well-determined, it would be like trying to maximize the median outcome. But it is not clear that this is reasonable under the disaster and care little for gains conditions. Furthermore, it seems likely that many actual cases where the Rawlsian conditions are met would not allow such a fine-grained individuation of the alternatives.
Harsanyi 1975, p. 40; emphasis in original.
Strict Bayesians will not like this response, since they think that all probability assignments are ultimately subjective, and such assignments can always be made. I defer discussion on that issue until later.
Manson 2002, p. 273.
European Environment Agency 2001, p. 170, box 16.1.
This kind of case may be important, and may be an important part of a general precautionary movement. But the RCPP does not address it. Since the RCPP aims to capture only part of precautionary thinking, it leaves it open what the proponent of precaution should say under such circumstances. So, the point here is part of my core case strategy.
One might think it should be simply a realistic threat criterion. But one should not be asymmetric. In particular, it does not seem reasonable to have a demanding standard for the “unacceptable outcomes” criterion and an undemanding one for the “care little for gains” criterion.
Actually, it is not clear if “supplement” is the right word. The criteria may be intended by Rawls himself to apply only in situations of uncertainty rather than ignorance. Hence, they may simply presuppose a background account of the outcome set to be considered, which presupposes a notion of realistic threat. Alternatively, perhaps the problem is implicitly dealt with by the conjunction of the “care little for gains,” and “disaster” conditions. For it may be that the real objection to taking the bare possibilities seriously is not their epistemic status, but the costs involved in doing so. The idea might be that precautionary measures in those cases would be prohibitively expensive or restrictive on normal human life—so much so as to constitute a disaster in their own right. (This fits well with an expected utility account—but I argue later that, at least at the present level of analysis, the RCPP need not oppose, and indeed should be compatible with, such a picture.)
Manson seems to suggest that mere logical possibility is enough for the Catastrophe Principle. But proponents of precaution usually demand some kind of scientific reason for thinking that there may be a problem. So they envisage natural possibilities. (The standard cases involve scientific evidence for the threat. For example, in climate change, there is scientific knowledge of the basic greenhouse mechanism, together with a body of empirical evidence; with genetically-modified foods, there is some previous experience with newly introduced species.)
This, I would claim, is the case for climate change. There we have a basic theoretical understanding of the greenhouse mechanism, strong empirical evidence that the concentrations of greenhouse gases has increased through industrialization and some evidence which suggests that this is already resulting in climate change. See Gardiner 2004.
Indeed, this probably accounts for the usefulness of distinguishing a range of opinion about levels of confidence or proof: from “scientifically based concern” to “reasonable grounds for concern” to “balance of evidence” to “beyond reasonable doubt.” Different levels of proof can be appropriate for different purposes (European Environment Agency 2001, p. 184).
There are some theoretical problems for maximin principles in determining how different courses of action are individuated. But this does not count decisively against such principles. For one thing, similar problems beset other decision procedures that require action-individuation. For another, in practice one has to be very careful to imagine the real alternatives including “safe exits” precautions one might take.
Harsanyi 1975, pp. 595–6; emphasis in original.
It is not clear whether this undermines the uncertainty condition. In any case, the example is relevant since the other conditions appear to be met.
The RCPP also does well with many cases where the conditions are not met and where precautionary action seems unwarranted: e.g., insurance against tennis shoe malfunction. In the standard cases, insurance seems rational because we care comparatively little about the cost of insurance relative to the costs of losing one's home or belongings, whereas insurance in the tennis shoe case seems irrational because the relative cost of a new pair of tennis shoes does not constitute a disaster.
In a book that came out just as this article was going to press, Cass Sunstein worries that the “care little for gains” condition threatens to confine the RCPP to trivial cases, and this undermines the application to global warming (see Sunstein 2005, p. 112). I cannot offer a full response here. But we should note that Rawls is speaking of gains that can be made above some minimum we can guarantee through eliminating the worst case scenario. Hence, much depends on how one understands the alternative options. Suppose, for example, that we could avoid the possibility of catastrophic climate change and guarantee a decent quality of life for all, all at the cost of slowing down our rate of accumulation of purely luxury goods by two years. This might satisfy the Rawlsian condition even if the cost of those luxury goods in dollar terms were very large.
Some projections suggest costs of around 2% of global product per year. This is a large amount of money in absolute terms, but many writers point out that it constitutes only a delay of a year or two in projected growth. Other models claim that such projections underestimate the indirect benefits of reduced fossil fuel consumption, and that, once this is factored in, mitigation might actually be beneficial even if there were no climate change.
At least, this is true for those inclined to be against precautionary action. The less skeptical might dispute it. For example, IPCC 2001 does assign probabilities to the outcomes it considers, and these are of considerable magnitude.
On the first issue, Lomborg is almost amazingly optimistic. (For example, he envisions a worldwide shift to solar power by 2060. See Lomborg 2001, p. 286.) On the second issue, his claim that the money would be better spent on the world's poor seems to falsely assume that these are mutually exclusive options.
Nir Eyal points out to me that perhaps Lomborg might accept the RCPP but believe that global poverty is the “unacceptable outcome” in relation to which we should “care little about the gains” to be made from climate change mitigation.
Similarly, the criteria allow us to avoid the obvious counterexamples. We do not refuse to fly to Chicago in order to take an excellent job, because we do care a lot about the potential gains in that case, and because we do think we have a decent grasp of the probabilities.
Hence, it accommodates some of the procedural concerns of Jordan and O’Riordan whilst avoiding some of the drawbacks of their proposal.
This is important. Even though it captures salient features, a principle of this kind may not exhaust the considerations relevant to the judgment. Hence, one may not be able to simply lift the considerations over from one case to another and say that the same approach is warranted. There may be background conditions that are unexpressed, or perhaps that have not even been directly considered, which differ between the situations and so can disrupt the attempted generalization. (Jonathan Dancy (1993) calls such considerations “defeaters.”)