Skip to main content
eScholarship
Open Access Publications from the University of California

Geographic Variation in a Spider’s Ability to Solve a Confinement Problem by Trial and Error

  • Author(s): Jackson, Robert R.
  • Cross, Fiona R.
  • Carter, Chris M.
  • et al.
Abstract

Portia is a genus of web-invading araneophagic (spider eating) jumping spiders known from earlier studies to derive aggressive-mimicry signals by using a generate-and-test (trial and error) algorithm. We studied individuals of Portia labiata from two populations (Los Baños and Sagada) in the Philippines that have previously been shown to differ in the level to which they rely on trial-and-error derivation of signals for prey capture (Los Baños relied on trial and error more strongly than Sagada P. labiata). Here we investigated P. labiata’s use of trial and error in a novel situation (a confinement problem: how to escape from an island surrounded by water) that is unlikely to correspond closely to anything the spider would encounter in nature. During Experiment 1, spiders chose between two potential escape tactics (leap or swim), one of which was set at random to fail (brought spider no closer to edge of tray) and the other of which was set for partially succeeding (brought spider closer to edge of tray). By using trial and error, the Los Baños P. labiata solved the confinement problem significantly more often than the Sagada P. labiata in Experiment 1, both when the correct choices were positively reinforced (i.e., when the spider was moved closer to edge of tray) and when incorrect choices were punished (i.e., when the spider got no closer to edge of tray). In Experiment 2, the test individual’s first choice was always set to fail, and P. labiata was given repeated opportunities to respond to feedback, yet the Sagada P. labiata continued to place little reliance on trial and error for solving the confinement problem. That the Los Baños P. labiat a relied more strongly on trial-anderror problem solving than the Sagada P. labiata has now been demonstrated across two different tasks.

Main Content
Current View