space. If environmental problems are problems of scale, it makes sense to think explicitly about how we construct mental models of the space in which we value nature.
In this spirit, I emphasize the horizon of concern expressed in environmental problems as setting a rough context for understanding and interpreting data in the search for a more harmonious relationship with ecological and physical systems. Horizons of concern are temporal scaling devices. Our operating assumption here is that, if Holling is correct, we should be able to construct an objective model of crucial ecological processes as viewed from the perspective of an animal of human lifespan and body size. Could we (a) define a set of dynamic, physical models that describe nature from a human perspective and then (b) choose that subset of possible spaces that reflect human values. This process of analysis, if successful, might result in an "association" of important social values with particular temporal horizons, and in turn an association of temporal horizons with physical dynamics of a particular scale. For example, Norton and Ulanowicz (1992) have shown, using a hierarchical analysis, that the protection of biodiversity is best modeled and pursued at the landscape ecosystem level. This follows from the temporal horizon of the goal of biodiversity protection—to protect biological resources for many human generations into the future.
But we have now deserted traditional "pure" and "value-free" science; we have recognized that the choice of boundaries for our physical models can express values and concerns that are shaped by our value-laden experience of space. The goal of this examination is to think more explicitly about this interaction between values and modeling and the ways in which our representations of natural processes and environmental problems embody spatial aspects of an action-oriented model for articulating environmental policy problems (a process that is represented abstractly in Figure 1).
Since it is a goal of model building in environmental management that the models inform environmental decision making to improve communication between scientists and the public, we conclude that any model for this purpose must be fairly simple in structure. It must, that is, be a simple enough representation of multiscaled natural processes to serve as an aid in public discussion of the goals of a forest management plan or a plan for ecological restoration of a river system. Our prescriptive, multiscalar models must provide a publicly useful vocabulary for discussing environmental goals. We can in this way shape our models of management by associating them with the temporal and spatial scales of the natural dynamics that generate the values guiding our choice of goals. In this sense we are searching for a spatiotemporally organized, and ecologically informed, phenomenology of the space in which individuals formulate and pursue personal and environmental values.
To initiate discussion, I suggest three basic scales, each of which corresponds to a temporally distinct policy horizon: (1) locally developed values that express the preferences of individuals, given the established limits and rules—