Planning routes across economic terrains: Maximizing utility, following heuristics

Hang Zhang, Soumya V. Maddula, Laurence T. Maloney

Research output: Contribution to journalArticlepeer-review


We designed an economic task to investigate human planning of routes in landscapes where travel in different kinds of terrain incurs different costs. Participants moved their finger across a touch screen from a starting point to a destination. The screen was divided into distinct kinds of terrain and travel within each kind of terrain imposed a cost proportional to distance traveled. We varied costs and spatial configurations of terrains and participants received fixed bonuses minus the total cost of the routes they chose. We first compared performance to a model maximizing gain. All but one of 12 participants failed to adopt least-cost routes and their failure to do so reduced their winnings by about 30% (median value). We tested in detail whether participants' choices of routes satisfied three necessary conditions (heuristics) for a route to maximize gain. We report failures of one heuristic for 7 out of 12 participants. Last of all, we modeled human performance with the assumption that participants assign subjective utilities to costs and maximize utility. For 7 out 12 participants, the fitted utility function was an accelerating power function of actual cost and for the remaining 5, a decelerating power function. We discuss connections between utility aggregation in route planning and decision under risk. Our task could be adapted to investigate human strategy and optimality of route planning in full-scale landscapes.

Original languageEnglish (US)
Article numberArticle 214
JournalFrontiers in Psychology
Issue numberDEC
StatePublished - 2010


  • Bayesian decision theory
  • Decision making
  • Heuristics
  • Navigation
  • Optimality
  • Route selection
  • Utility

ASJC Scopus subject areas

  • General Psychology


Dive into the research topics of 'Planning routes across economic terrains: Maximizing utility, following heuristics'. Together they form a unique fingerprint.

Cite this