Learning in honeybees as a function of amount of reward: Tests of the equal-asymptote assumption

Abstract

In Experiments 1 and 2, honeybee foragers visiting the laboratory were fed on targets of two different colors, one containing 5 ILl and the other containing 20 ILl of 50% sucrose solution. The targets were presented singly in quasi-random sequences on the training visits, after which preference was measured in an unrewarded choice test. In Experiment 1, 16 differentially rewarded training trials with each color were followed by the same number of trials with the coloramount relation reversed; no preference for either color was found in the subsequent choice test. In Experiment 2, 20 differentially rewarded training trials with each color-enough to produce a clear preference for the 20-ILI color when given directly after pretraining-were given after 10 feedings to repletion on each color that were calculated to generate near-asymptotic associative strength; no preference for either color was found in the subsequent choice test. In Experiment 3, there were 12 feedings to repletion on one color and, on the other, 12 feedings to repletion followed by 15 trials with a small (5 ILl) reward; no preference was found in a subsequent choice test. The results of all three experiments support a nonrepresentational interpretation of the role of amount of reward in the learning of honeybees.In a continuing series of experiments on learning in honeybees as compared with learning in vertebrates (Bitterman, 1988), we began recently to study amount of reward (Buchanan & Bitterman, 1988), an important parameter oflearning in vertebrates (Mackintosh, 1974). Some previous work hadsuggested that learning in honeybees might be independent of amount of reward (Menzel & Erber, 1972), but the method employed in that work seemed inadequate for a variety of reasons which need not be considered again here. Our own method was to train honeybees with two targets of different colors that were presented successively in quasi-random sequences, one target always containing 5 ILl of a 50% sucrose solution, the other containing 20 ILl of the same solution, and then to measure persistence of response to the targets in an extinction test, that is, when they no longer contained sucrose. If the number of training trials with the two targets was the same, the animals responded more in extinction to the ZO-/Ll target than to the 5-ILI target, but the difference in amount of reward could be offset by a difference in frequency. If the number of 5-1L1 training trials was at least three times the number of 20-1L1 trials, the animals responded more in extinction to the 5-1L1 target; if the number of 5-1L1 training trials was twice the number of ZD-ILI trials, the animals responded equally often to the two targets in extinction.

Learning in honeybees as a function of amount of reward: Further experiments with color

“…Of the various tests of the equal-asymptote assumption employed by Buchanan and Bitterman (1989), the reversal test is perhaps the most powerful. The procedure is to compare responses to two stimuli after one of them has been paired with large reward followed by small reward and the other with small reward followed by large reward.…”

“…In the odor experiments, the possibility of differential delay was ruled out by marking the location of each drop with a readily discriminable white dot, but dots could not be used in the color experiments because they seemed to overshadow the colors (Lee & Bitterman, 1990b). On the assumption that the associative strength acquired by a stimulus paired with sucrose depends on the duration of concurrent exposure, which is longer for larger rewards because the ingestion time is longer (Buchanan & Bitterman, 1989), the amount of reward per se actually may have played no role at all in the color experiments, because stationary honeybees, which readily Copyright 1993 Psychonornic Society, Inc. process odors, do not seem to process visual stimuli very well under the same conditions (Walker, Baird, & Bitterman, 1989).…”

“…The results for both color and odor can be understood on the simple nonrepresentational assumption that the attractiveness of a stimulus depends on the strength of its association with reward and that the strength of association depends on the amount as well as the frequency of the reward. In another respect, however, the results are different, those for color suggesting that the effect of the amount of the reward is on the rate of acquisition (the equal-asymptote assumption), and those for odor that the effect is on asymptotic strength.The information about color comes from a series of three experiments by Buchanan and Bitterman (1989). In the first-a reversal experiment-one of two targets contained 5 iLl of sucrose solution on Trials 1-16 and 20 iLl on Trials 17-32, while the opposite was true of the other.…”

“…The information about color comes from a series of three experiments by Buchanan and Bitterman (1989). In the first-a reversal experiment-one of two targets contained 5 iLl of sucrose solution on Trials 1-16 and 20 iLl on Trials 17-32, while the opposite was true of the other.…”

Comparative analysis of learning in honeybees

“…The results of initial experiments (Buchanan & Bitterman, 1989) suggested that the effect of amount of reward is only on the rate of growth in associative strength (the equal-asymptote assumption), which if true would argue Figure 2. Cumulative number of responses in an unreinforced choice test with two targets different in odor, one (5-20) having previously contained first 5-and then 20-pl drops of50% sucrose solution, and the other (20-5) the two amounts in the opposite sequence.…”