level of precaution may increase or decrease as uncertainty is resolved in a broad class of models
with unknown thresholds, whether these thresholds are reversible, hysteretic, or irreversible. Fi-
nally, Peterson et al. [25] suggest that mistaken overconfidence in system parameters leads naive
managers to overload pollutant inputs to ecosystems because the optimal strategy is insensitive to
the existence of a threshold, causing cycles of collapse and recovery. While recurrent overloading
of an ecosystem may well lead to its collapse, in Peterson et al.’s model mismanagement results
because the decisionmaker is not allowed to recognize that a threshold exists, rather than as an
inherent feature of utility maximization. An important result of our study is that if the existence of
a threshold is suspected, some precaution is warranted even under uncertainty about its location.
Many versions of the precautionary principle have been proposed to justify reductions in the
level of polluting activities in the face of large uncertainty about the consequences of such ac-
tivities [13]. Our analysis gives both an economic justification for precautionary reductions in
pollutant loading - in terms of the expected economic gains - and a more nuanced view of how
beliefs about system thresholds should affect precautionary behavior. In particular, a decrease in
uncertainty about the location of a threshold may either increase or decrease the desirable level
of precaution, depending on the expected resilience of the ecosystem as well as the initial level
of uncertainty. Thus, stakeholder conflict in some kinds of environmental dispute may be a result
of different beliefs about threshold proximity and uncertainty, even when there is broad consensus
about underlying processes and system dynamics. For example, current views on climate change
policy can be broadly divided into those that are pessimistic and those that are optimistic. Climate
change pessimists advocate for an immediate reduction in the production of greenhouse gases until
uncertainty about the processes of climate change is reduced. Conversely, climate change optimists
suggest that no costly reductions in greenhouse gas production should be undertaken until the same
uncertainty about climate change is reduced. Our analysis suggests that there may be much more
common ground between these two views than might otherwise be thought: both optimists’ and
pessimists’ views can be consistent with the same underlying economic or ecologic objectives and
expected system resilience, and their differences can be attributed to different beliefs about the un-
certainty with which important thresholds are known (as an illustrative example, if both pessimists
and optimists assume a system resilience Xc - Xtarget = 0.2 in Figure 2, but pessimists believe
σXc = 0.4 while optimists believe σXc = 0.1, then a reduction in uncertainty σXc would warrant
increased precaution for the former and decreased precaution for the latter).
Our analysis contains several key insights for the choice and implementation of ecosystem
18