| Summary: | When choosing
between immediate and future rewards, how do people deal with uncertainty about
the value of the future outcome or the delay until its occurrence? Skylark et
al. (2020) suggested that people employ a delay-reward heuristic: the inferred
value of an ambiguous future reward is a function of the stated delay, and
vice-versa. The present paper investigates the role of this heuristic in choice
behaviour. In Studies 1a--2b, participants inferred the value of an ambiguous
future reward or delay before the true value was revealed and a choice made.
Preference for the future option was predicted by the discrepancy between the
estimated and true values: the more pleasantly surprising the delayed option,
the greater the willingness to choose it. Studies 3a--3c examined the
association between inference and preference when the ambiguous values remained
unknown. As predicted by the use of a delay-reward heuristic, inferred delays
and rewards were positively related to stated rewards and delays, respectively.
More importantly, choices were associated with inferred rewards and, in some
circumstances, delays. Critically, estimates and choices were both
order-dependent: when estimates preceded choices, estimates were more
optimistic (people inferred smaller delays and larger rewards) and were
subsequently more likely to choose the delayed option than when choices were
made before estimates. These order effects argue against a simple model in
which people deal with ambiguity by first estimating the unknown value and then
using their estimate as the basis for decision. Rather, it seems that
inferences are partly constructed from choices, and the role of inference in
choice depends on whether an explicit estimate is made prior to choosing.
Finally, we also find that inferences about ambiguous delays depend on whether
the estimate has to be made in "days" or in a self-selected temporal unit, and
replicate previous findings that older participants make more pessimistic
inferences than younger ones. We discuss the implications and possible
mechanisms for these findings.
|
|---|