Uncertainty Visualization Study Group Notes – 2/11/2013
Alternative Visions of Reality
We can go through systematic reasoning but often don’t; rather we are often influenced emotionally by our values and beliefs.
Subjective Expected Utility relies on the omniscience of the decision maker – i.e. it’s not possible
Decision-maker:
has a well-defined utility function
has a well-defined set of alternatives
can assign a probability distribution to future sets of events
will choose the alternative which maximizes expected value in terms of utility function
SEU is an “Olympian model”:
assumes decision-maker thinks comprehensively about all aspects of a decision
requires infinite computational power; is not how humans make decisions
Alternative: Bounded Rationality
“nearly empty world” – there are millions of variables that could influence each other, but they generally don’t
decisions are generally concerned with specific instances, not comprehensive choices
approximation to subjective expected utility; only works if bounded to a very limited context
evidence that this is closer to how people make decisions
emotion factors into perceived importance, focuses attention
consequence is that we don’t optimize our decisions, in fact decisions may not even be consistent across different points in time
Expertise leads to intuition leads to good decision making
Experts will “intuit” answers to questions much more quickly than novices will decipher them
Discussion Points
[Miriah] In visualization, why not show the space of possibilities? If people have difficulty generating the set of alternatives, why not present it to them directly?
[Sarah] Comprehension might require too much computational power on the part of the user; need to provide focus
[Mike] How important is it to have a sense of the space of possibilities? (especially in regard to answering ill-formed questions)
Person assessing risk needs more than certain/uncertain; but maybe in some circumstances binary qualifier of certain/uncertain is enough
Effect of “emotion” on the context of visualization
If “emotion” or some other factor plays huge role in decision making, should we be attempting to integrate considerations of said factor in the development of visualizations?
Should we simply be providing users with the data and not worrying about additional factors?
Is visualization there for a specific task or simply to provide information?
Perhaps visualization systems should be comprised of multiple perspectives, allowing users to sort through those perspectives
Can we make people spend more time and think more deeply? Is that an alternative measure of success?
[Miriah] Current studies: priming individuals for emotional states are better/ worse for specific types of tasks