tally from a description of the world as viewed by large-bodied mammals such as ourselves. If our descriptions of multiscalar natural systems inevitably involve a choice regarding the perspective from which, and scale on which, to measure and monitor natural systems, how can we avoid the apparent implication that our human values and perspectives may influence the scales we observe and describe in nature? If this implication is accepted, it seems also to follow that the "descriptive" models we use to characterize natural processes actually express, in a less explicit but nevertheless profound sense, the values we pursue and the actions we take in pursuing them.
While a full exploration of this less objectivist approach to scale would be beyond the scope of this paper, these general ideas set the background for the more particular explorations undertaken here. I believe that conservation biology, restoration ecology, and ecological engineering are all "normative sciences" and that choices of models to understand interactions between humans and nature are either explicitly or implicitly based on value considerations. One of the important ways—perhaps the most important way—in which values affect science is in our choice of scales on which to characterize and address ecological risks and problems. For this reason I undertake, with some trepidation, an examination of scalar aspects of human values as my contribution to the discussions of engineering and ecological constraints central to this volume.
This paper addresses three important aspects of scalar problems in environmental values and policy. First, I explore the idea that the nonnative disciplines of conservation biology, restoration ecology, and ecological engineering use a "scientific" language that must have normative as well as descriptive content. Further, I believe that this valuational element is often embodied in decisions regarding the scale we choose to employ and the scale of the models we construct in our observation and manipulation of our environment. Second, building on empirical and theoretical work by Holling (1978, 1992, 1994, and in this volume), I propose a multiscalar analysis of social values and argue for a pluralistic approach to environmental policy. This approach recognizes multiple, irreducible values derived from nature by humans, seeks to associate particular values and classes of values with specific natural dynamics that are dominant on various scales of the environment, and organizes human values according to scale, providing multiple criteria of good management guided by multiple values. Finally, I will offer a series of devices called risk decision squares, which help to sort decisions affecting the environment according to an ecologically—that is, spatially and temporally—sensitive typology of risks involved in a given decision.
One important consequence of the rise of modernism and the Newtonian, objectivist model of the physical universe is that choices of scale and perspective become an essential element in every description of nature. Observation is nec-