Geoscience Reference
In-Depth Information
given context, for which the term 'equifi nality' has been used (e.g., Beven, 2002).
Here, the distinction between possible
outcomes
and
probabilities
of particular
outcomes is useful, as it points to different types and levels of uncertainty, which
should be addressed with different methods (Brown, 2004). For example, uncertain-
ties associated with climate scenarios are of a different type and magnitude to those
associated with numerical parameters in a climate model, and are addressed with
correspondingly different methods (Houghten et al., 2001). Indeed, for some types
of outcomes, commonly assessed in the physical sciences, the distinction between
possibilities and probabilities is largely a question of methodology. Here, formal
methods are available at varying levels of detail to describe possible outcomes (sce-
nario analysis, possibility theory) and probabilities of outcomes based on precise
defi nitions of 'events' (subjective and frequency-based probability methods).
Common to these formalisms is an assumption that
only one
outcome exists in
principle. In contrast, other types of outcomes, commonly addressed in the social
sciences, are non-unique in principle, such that discussions of probability are irrel-
evant (e.g., perceptions on the fair distribution of wealth). This can lead to tensions
between social and physical scientists on issues of uncertainty.
Causes of Uncertainty and Risk
Uncertainties and risks may be psychological, social or situational in origin. Psycho-
logical factors include the propensity for risk aversion and fear of the unknown.
Social factors include language, and the development of scientifi c networks, which
are built on trust and consensus. Situational factors concern the types of problem
addressed, including their transparency, scale, variability, and complexity. Clearly,
these factors are closely related in practice; for example, trust (a social factor) is built
on personality (an individual factor) and depends on the complexity of the problem
in question (a situational factor). Establishing the relationships among these sources
of uncertainty is a key research challenge, as the accumulated uncertainties in deci-
sion making will be sensitive to these relationships. For example, the overall uncer-
tainty in a fl ood inundation model will depend on the type (e.g., linearity versus
non-linearity) and degree of association between physical parameters in adjacent
locations. Similarly, the levels of uncertainty associated with a fl ood warning will
depend on the modes of message construction and dissemination (e.g., expert-driven
versus community-driven) between planning authorities, the emergency services, and
the general public, who develop, issue and respond to those warnings (Handmer,
2001). It follows that the sources of uncertainty are manifest both in the outcomes
of research and decision making, i.e., 'what we know', and in the processes through
which those outcomes are produced, i.e., 'how we came to know'.
Research in the cognitive sciences has shown that an individual's perceptions of
uncertainty and risk are determined partly by the structure of the human brain and
partly by their experiences and personality. Knowledge is embodied in cognitive
structures that are commonly referred to as 'mental models' (Morgan et al., 2001).
These models are implicit, intuitive, and frequently wrong. In particular, they are
sensitive to 'framing effects', which originate from the presentation of a single
problem in different ways (e.g., glass half full versus half empty). Other biases
include the positive weighting of events that are easily remembered ('availability
heuristic'); the tendency to rate two events more probable than a single event
('conjunction fallacy'); the selective processing of information that confi rms an
