Quantitative tests of an associative theory of risk-sensitivity in honeybees.

Risk-sensitivity was studied in free-flying honeybees trained individually to choose between two scented targets (A and B) with varying amounts and concentrations of sucrose solution as reward. In the first phase of experiment 1, the animals showed "risk-aversion," preferring A, which provided 5 microl of a 40 % sucrose solution on every trial, to B, which provided 30 microl of the same solution once in every six trials (mean amount per trial 5 microl for each alternative). In the second phase, the preference reversed with reversal of the reward assignments. In experiment 2, the consistently rewarded A (5 microl of 40 % sucrose solution per trial) was again preferred, although the inconsistently rewarded B now provided twice the amount of sucrose solution on average (30 microl on two of every six trials, mean amount per trial 10 microl). In experiment 3, with A providing 10 microl of a 15 % sucrose solution on every trial and B providing 10 microl of a 60 % sucrose solution on two of every four trials (mean concentration per trial 30 %), the animals preferred B. In Experiment 4, patterned after experiment 1, similar results were obtained under more natural conditions in which the animals were no longer constrained (as they were in the first three experiments) to go equally often to each alternative. The results of all four experiments were predicted quantitatively and with considerable accuracy by a simple associative theory of discriminative learning in honeybees.