June 2001 / Vol. 51 No. 6 • BioScience 459ArticlesPractical problem solving in complex, human-dominated ecosystems requires the integration of threeelements: (1) active and ongoing envisioning of both how theworld works and how we would like the world to be, (2) sys-tematic analysis appropriate to and consistent with the vision,and (3) implementation appropriate to the vision. Scientistsgenerally focus on only the second of these steps, but inte-grating all three is essential to both good science and effectivemanagement. Subjective values enter in the vision element,both in terms of the formation of broad social goals and inthe creation of a preanalytic vision, which necessarily precedesany form of scientific analysis.Because of this need for vision, completely objective sci-entific analysis is impossible. Joseph Schumpeter (1954) putit this way:In practice we all start our own research from the workof our predecessors, that is, we hardly ever start fromscratch. But suppose we did start from scratch, what arethe steps we should have to take? Obviously, in order tobe able to posit to ourselves any problems at all, weshould first have to visualize a distinct set of coherentphenomena as a worthwhile object of our analyticeffort. In other words, analytic effort is of necessity pre-ceded by a preanalytic cognitive act that supplies theraw material for the analytic effort. In this book, thispreanalytic cognitive act will be called Vision. It is inter-esting to note that vision of this kind not only mustprecede historically the emergence of analytic effort inany field, but also may reenter the history of everyestablished science each time somebody teaches us tosee things in a light of which the source is not to befound in the facts, methods, and results of the preexist-ing state of the science. (p. 41)Nevertheless, it is possible to separate the process into themore subjective, or normative, envisioning component andthe more systematic, less subjective analysis component(which is based on the vision). “Good” science is that whichmakes clear its underlying preanalytic vision, and whoseanalysis is consistent with that vision.A changing vision of scienceThe task would be simpler if the vision of science were staticand unchanging. But as the quote from Schumpeter makesclear, this vision is itself evolving as we learn more. This doesnot invalidate science, as some deconstructionists would haveit. Quite the contrary: By being explicit about their underly-ing preanalytic vision, scientists can enhance their honesty andthereby their credibility. Scientific credibility proceeds fromhonest discussion of this underlying vision and its inher-ently subjective elements, as well as from constant, pragmatictesting of conclusions against real-world problems, ratherthan by appealing to a nonexistent objectivity.The preanalytic vision of science is changing from the“logical positivist”view, which holds that science can discoverultimate truth by falsification of hypotheses, to the morepragmatic view that we do not have access to any ultimate, uni-versal truths, but only to useful, abstract representations(models) of small parts of the world. Science, in both the log-ical positivist and this new “pragmatic modeling” vision,works by building models and testing them. But the new Robert Costanza (e-mail: [email protected]) is a professor atthe Center for Environmental Science and the Department of Biol-ogy, and director of the Institute for Ecological Economics, at the Uni-versity of Maryland, Solomons, MD 20688-0038. © 2001 AmericanInstitute of Biological Sciences.Visions, Values, Valuation,and the Need for anEcological EconomicsROBERT COSTANZAALL SCIENTIFIC ANALYSIS IS BASED ON A“PREANALYTIC VISION,” AND THE MAJORSOURCE OF UNCERTAINTY ABOUTCURRENT ENVIRONMENTAL POLICIESRESULTS FROM DIFFERENCES IN VISIONSAND WORLD VIEWS460 BioScience • June 2001 / Vol. 51 No. 6Articlesvision recognizes that the tests are rarely, if ever, conclusive(especially in the life sciences and the social sciences); the mod-els can only apply to a limited part of the real world; and theultimate goal is therefore not truth, but quality and utility. Inthe words of William Deming, “All models are wrong, butsome models are useful” (McCoy 1994).The primary goal of science, then, is the creation of use-ful models whose utility and quality can be tested against real-world applications. The criteria by which one judges the util-ity and quality of models are themselves social constructs thatevolve over time. There is, however, fairly broad and consis-tent consensus in the scientific community about what thesecriteria are. They include (1) testability, (2) repeatability, (3)predictability, and (4) elegance (i.e., Occam’s razor: Themodel should be as simple as possible, but no simpler!). Butbecause of the nature of real-world problems, there are manyapplications for which some of these criteria are difficult orimpossible to apply. These applications may nevertheless stillbe judged as “good” science. For example, some purely the-oretical models are not directly testable, but they may providefertile ground for thought and debate and lead to more ex-plicit models that are testable. Likewise, field studies of wa-tersheds are not repeatable, strictly speaking, because no twowatersheds are identical. But there is much we can learn fromfield studies that can be applied to other watersheds andtested against the other criteria of predictability and ele-gance. How simple a model can be depends on the questionsbeing asked. If we ask a more complex or more detailed ques-tion, the model will probably have to be more complex anddetailed. As science progresses and the range of applicationsexpands, the criteria by which utility and quality are judgedmust also adapt to the changing applications. This inher-ently subjective process goes on constantly within the scien-tific community.Vision and changeResearch concerning the way change proceeds in various or-ganizations and communities suggests that the most effectiveingredient to move change in a particular direction is a clear,shared vision of the desired goal (Senge 1990, Wiesbord1992, Wiesbord and Janoff 1995). Or, as Yogi Berra oncesaid, “If you don’t know where you’re going, you end upsomewhere else.”Yankelovich (1991) has described the crisis in governancefacing modern societies as one of moving from public opin-ion to public judgment.Public opinion is notoriously