Conflict and ambiguity: Preliminary models and empirical tests

The proposition that conflict and ambiguity are distinct kinds of uncertainty remains debatable, although there is substantial behavioral and some neurological evidence favoring this claim. Recently formal decisional models that combine ambiguity and conflict have been proposed. This paper presents empirical tests of four hypotheses and five models of uncertainty judgments under ambiguity and conflict, via comparisons between pairs of conflicting and ambiguous interval estimates by a sample of 395 adults. The main findings are as follows. 1. Human judges see conflict even in nested intervals with identical midpoints and symmetrically differing endpoints. 2. Identical envelopes of intervals may not be perceived as equally conflictive. Moreover, sets of intervals whose average widths are identical may not be perceived as equally ambiguous. 3. Perceived degree of conflict does not necessarily covary with the magnitudes of the differences between corresponding pairs of interval endpoints. Indeed, a nested pair of intervals may be regarded as more conflictive than a non-nested overlapping pair whose pairs of endpoints differ identically to the nested pair. 4. Judgments of degrees of conflict and ambiguity both contribute independently to judgments of overall uncertainty. However, judgments of ambiguity and conflict appear to be positively correlated. None of the models pass all empirical tests, but specific suggestions for improving the models are derived from the findings.