Bumblebees Learn a Relational Rule but Switch to a Win-Stay/Lose-Switch Heuristic After Extensive Training

Abstract

Mapping animal performance in a behavioral task to underlying cognitive mechanisms and strategies is rarely straightforward, since a task may be solvable in more than one manner. Here, we show that bumblebees perform well on a concept-based visual discrimination task but spontaneously switch from a concept-based solution to a simpler heuristic with extended training, all while continually increasing performance. Bumblebees were trained in an arena to find rewards on displays with shapes of different sizes where they could not use low-level visual cues. One group of bees was rewarded at displays with larger shapes and another group at displays with smaller shapes. Analysis of total choices shows bees increased their performance over 30 bouts to above chance. However, analyses of first and sequential choices suggest that after approximately 20 bouts, bumblebees changed to a win-stay/lose-switch strategy. Comparing bees' behavior to a probabilistic model based on a win-stay/lose-switch strategy further supports the idea that bees changed strategies with extensive training. Analyses of unrewarded tests indicate that bumblebees learned and retained the concept of relative size even after they had already switched to a win-stay, lost-shift strategy. We propose that the reason for this strategy switching may be due to cognitive flexibility and efficiency.

Keywords: abstract concepts; adaptive decision-making; animal cognition; behavioral analyses; cognitive flexibility; cognitive offloading; the law of least effort.

Figure 1

Training and testing protocol. (A) Stimuli options used during training. (B) Stimuli options used for each of the three different unrewarded tests. (C) Training and test protocol. Bees were trained for 30 bouts (visits to the arena before returning to the hive). All stimuli in panel (A) were used randomly across bouts during training. Only two of the possible three sizes of shapes were presented during a single bout. Only one of the possible three colors and one of the possible three shapes were presented each bout. One group of bees (n = 10) was trained to find a 50% sucrose solution at the center of the stimulus containing the larger of the three shapes and bitter quinine solution at the smaller of the three shapes. Another group (n = 8) were trained on the opposite contingency. Once training was complete, bees were subjected to three unrewarded tests (with one or two reminder/training bouts between each test to keep bees motivated). All tests used small and large-sized shapes. The learning test used one randomly chosen type and color used during training. The novel shape test used one randomly chosen color used during training but always a star shape that had not been used during training. The novel color test used one randomly chosen shape used during training but always colored yellow, which had not been used during training.

Figure 2

Bees use a win-stay/lose-switch strategy after extensive training. (A) There was a significant increase in the number of correct choices over the 180 conditioned choices (p = 0.018). (B) Bees’ performance over three blocks of 10 training bouts during the relative size discrimination task. Performance increased gradually over bouts when considering the total number of choices in each bout (black dashed line; p = 1.96e-4). Bees’ first choice performance increased significantly from the first to the second block of training bouts to 72.22% (p = 3.71e-4) but then dropped to chance level from the second to the third block of training bouts (blue dash-dotted line; p = 0.79). The second choice performance was near chance for the first two blocks of training bouts (p > 0.49) but then increased significantly during the third block of training bouts (green dotted line; p = 4.28e-4). These results indicate that bees changed to a win-stay/lose-switch strategy after extensive training. Vertical lines = standard error of the mean. Red dashed line = chance level performance (50%). (C,D) The average conditional probabilities of a bee’s second choice within each bout being correct given the outcome of the bee’s first choice of the bout (either correct or incorrect). Both conditional probabilities increased to above chance during the second and third blocks of bouts (p = 2.27e-4 for win-stay and p = 8.40e-4 for lose-switch). (E) Our win-stay/lose-switch model’s performance matches our bees’ performance on the task during the last block of 10 bouts during training (p = 0.15), again suggesting that after extensive training bees changed to a win-stay/lose-switch strategy (Vertical lines = standard error of the mean). Red dashed line = chance level. *p < 0.05 and n.s., p > 0.05.

Figure 3

Bees learn and retain a relative size rule. (A) The performance of bees during each of the three unrewarded tests shows that they learned and retained the concept of relative size (p < 2.41e-3). (B) The scatter plot displays the correlation between the performance of bees in the learning test and their first choice performance before changing strategies, during the second block of 10 bouts (rho = 0.58, p = 0.01). The red solid line = line of best fit. (C,D) The significant drop in performance from the last bout of training to the learning test (p = 9.30e-4; D) and the difference in performance between the second and first half of choices during each of the tests (p = 0.03 for learning test; p = 0.28 for novel shape transfer test; p = 0.14 for novel color transfer test) suggest that bees had begun the tests with the win-stay/lose-switch strategy. Bars = mean. Vertical lines = standard error of the mean. Red dashed line = chance level (50%). *p < 0.05 and n.s., p > 0.05.

Figure 4

The average of the number of choices on stimuli (correct and incorrect) over three blocks of 10 training bouts. Over training-bouts, bees made fewer choices to visit all three available rewarding stimuli (p = 1.17e-5), indicating that bees continually increased efficiency on solving the task during training. Vertical lines = standard error of the mean. *p < 0.05.