Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2025 Jul 8;15:24363. doi: 10.1038/s41598-025-08616-9

Assessing the limits of delay of gratification in bumble bees

Luigi Baciadonna 1,2,✉, Eleonora Rovegno 3, Giulia Bigazzi 1, David Baracchi 1,✉
PMCID: PMC12238607  PMID: 40628794

Abstract

The ability to resist immediate temptation for a better, delayed outcome is thought to underpin advanced cognition. Traditionally, this ability has been studied in large-brained animals. However, to understand its evolutionary origins, it is necessary to expand the study to species with different life histories and ecological pressures, such as insects. Here, we investigated the ability to endure delayed gratification in the bumble bee Bombus terrestris using a delay maintenance task that assessed the capacity to sustain a delay for a large reward. Bees were first individually trained in a foraging arena to associate two differently coloured flowers either with an immediately accessible small reward (a 5 µl drop of 30% sucrose) or a delayed large reward (four 5 µl drops with a 15-s delay). Bumble bees were then tested over 10 delay choices between the two flowers/rewards. Finally, bees were tested in a delay maintenance test in which the 15-s delay was extended indefinitely. Overall, our findings suggest that bumble bees did not consistently choose the delayed reward over the immediate one across the ten trials, and their ability to sustain delay was highly variable across individuals. Although a small subset of bees (n = 5) exhibited prolonged waiting times comparable to those observed in mammals, birds, and fish (with one individual resisting for up to 364-s), the overall lack of consistency suggests that self-control in bumble bees may be limited and related to either ecological constraints or their foraging strategy. These results provide new insights into the evolutionary origins and drivers of self-control.

Keywords: Comparative cognition, Cognitive evolution, Foraging strategies, Hymenoptera, Invertebrate cognition, Self-control

Subject terms: Animal behaviour, Entomology

Introduction

Self-control, the ability to discard an immediate gratification in favour of a better (in terms of quality and/or quantity) but delayed reward, is one of the most challenging forms of inhibition that allows adaptive decision making, goal-directed behaviour, and future planning1–5. In humans, self-control has been associated with general intelligence and has been found to predict later success in life, particularly in young children of high socioeconomic status3,6,7. In recent years, there has been an increasing focus on comparative studies of self-control in non-human animals with the overall aim of reconstructing the evolutionary origins, main drivers and cognitive mechanisms underlying this ability8–10.

As in humans, self-control plays a pivotal role in animals’ lives in a wide range of social (e.g., mate choice) and non-social (e.g., foraging activities) contexts, and this ability appears to vary across taxa9,11. Comparative studies have highlighted that several factors may account for such variation. For instance, a large comparative study across 36 vertebrate species showed a positive correlation between inhibitory control and absolute brain size12. A phylogenetic analysis indicated that absolute brain volume was the most significant predictor of the ability to inhibit a motor response across these species, explaining a substantial portion of the variance compared to relative brain size. In addition, self-control abilities have been associated with ecological constraints9,13–18. For example, in a self-control paradigm where individuals of two related primate species chose between an immediate small reward and waiting for a variable amount of time for a larger reward, common marmosets waited longer for food compared to cotton-top tamarins. The observed differences in waiting behaviour can be explained by feeding ecology. Common marmosets rely primarily on sap, which requires them to scratch the tree bark and then wait for the fluid to flow, while cotton-top tamarins, being insectivorous, benefit from impulsive actions to quickly locate food19,20. Social complexity can also account for variation within and between species. In non-human primates, species differences in self-control have been linked to differences in fission–fusion dynamics21, and, in corvids, the performance on reversal learning tasks (a form of motor self-regulation task) appears to correlate with levels of sociality22. Finally, physiological conditions such as starvation can affect inhibitory control abilities in small animals with high metabolic rates, including both invertebrates and vertebrates23,24.

Given the challenges in identifying the primary drivers of self-control, there is a pressing need for expanded research efforts encompassing species with different life histories and subject to different socio-ecological pressures than those traditionally studied9. Building upon this need for broader investigations, a study provided valuable insights into the evolution of self-control in cuttlefish25. Cuttlefish are not habitual tool users and do not engage in caching strategies, and their social behaviour does not depend on individual recognition or cooperation. Therefore, cuttlefish may have evolved their ability to delay gratification (e.g., patiently learning to wait in proximity before an attack) as an adaptation to optimize their foraging strategies25. Recent evidence of self-control in cleaner fish also supports the hypothesis that self-control may not involve or require advanced cognitive skills17,26.

Eusocial pollinating insects, such as honey bees and bumble bees, can provide a unique perspective to disentangle the main driver(s) of self-control among social, metabolic, and anatomical (i.e., brain size) factors. Foraging is the most notable and challenging activity for pollinators27,28. Flowers are ephemeral, and their reward values are inconsistent. Pollinators must make economic decisions to ensure a net gain, considering factors like reward quantity and quality, and the costs of handling complex flowers29,30. Their foraging decisions must take into account their hierarchical organisation, where nutritional needs must be considered both at the level of the individual bee and at the level of the colony31–33, and where workers forage for resources in response to social needs rather than their own. The challenge that foraging poses has been the driver for the development of cognitive abilities34, from basic discrimination between flowers to more intricate tasks, such as solving non-natural problems through observation of more knowledgeable individuals35,36. Previous research has shown that honey bee foragers, when tested in a delay choice task, prefer a sweeter or larger reward after a short delay (5-s) to a less sweet or smaller immediate reward11. However, when bees were deprived of food for 18–24 h, honey bees chose the immediate reward24. This suggests that honey bees can demonstrate self-control at least over short intervals. However, it is important to note that the delay choice task does not assess an individual’s ability to maintain the chosen delay, as subjects cannot reverse their decision once made. Therefore, it remains unknown whether insects such as bees can tolerate and sustain waiting for a better but delayed reward. In this study, we addressed this crucial question in bumble bees using a delay maintenance task. Initially, individual bees were trained in a foraging arena in which they experienced two differently coloured flowers in separate trials; one colour was consistently associated with an immediately accessible small reward and the other with a larger reward available after a 15-s delay. During this phase, the bees learned the value and timing of each reward independently. In the preference test phase, however, both flowers were presented simultaneously, allowing us to assess whether bumble bees, over 10 trials, preferentially chose the larger delayed reward over the immediate smaller one. Crucially, we assessed their ability to sustain inhibition in a delay maintenance test, where the 15-s delay was extended indefinitely. This approach allowed us, for the first time in an insect, to examine how often the decision to wait is actually linked to the ability to do so.

Material and methods

Animals and experimental setup

Five queen-right colonies of Bombus terrestris audax (Koppert, Italy) were individually housed in plywood boxes (40 × 30 × 10 cm) consisting of two connected chambers, one containing the nest and the other one with bedding on the floor37. The plywood box was connected to a flight arena (82 × 60 × 40 cm) by a transparent Plexiglas tunnel (25 cm long, 3.5 × 3.5 cm cross-section). The flight arena was covered with a UV-transparent PMMA panel (80 × 50 cm). Several shutters along the tunnel controlled the flow of foraging bees. On the opposite side of the tunnel entrance, the arena had a transparent window (80 × 50 cm) covered with white filter paper so that the bees could not see through it. At the base of the window, the panel had two holes through which the plastic flowers were presented to the bumble bees (see Fig. 1a-e). The whole apparatus was kept in the laboratory under controlled conditions (23 ± 2 °C, 12∶12 light–dark cycle, 50 ± 10% humidity) and was regularly supplied with pollen and a 30% (w/w) sucrose solution. Each day, bumble bees were allowed to freely enter the arena and forage on artificial flowers. The flowers were constructed using a 60 mm petri dish lid glued to the centre of a plastic straw. During this phase, all flowers were uncoloured and rewarded ad libitum with a 30% sucrose solution. Only about a third of the flower surface was accessible from the arena, while the rest remained on the experimenter’s side. On each experimental day, motivated foragers were marked on the thorax with a non-toxic ink (UniPOSCA) for individual identification. A total of 49 bees from the five colonies were tested, with 12, 11, 10, 10 and 6 bees tested per colony, respectively.

Fig. 1.

Fig. 1

Open in a new tab

Schematic representation of the training and test phases. From a to c represent the training phases. a) Bumble bees undertake a total of 50 trials (side counterbalanced) where they encounter a coloured flower paired with a small immediate reward. b) Before transitioning to the delayed large reward training, bumble bees undertake five trials to familiarize themselves with the landing process and the rotation mechanism granting access to a four-times larger delayed reward. c) Bumble bees undertake 50 trials (side counterbalanced) where a coloured flower is associated with a larger reward, requiring the bees to wait for 15-s before accessing the sucrose solution. d) A preference test is represented, in which bumble bees are confronted with two coloured flowers. They can choose to land on a flower offering an immediate reward or select to land and wait 15-s for a larger reward. Once a choice is made, the unselected flower becomes inaccessible. Bumble bees received a total of 10 consecutive trials. e) Bumble bees face a single trial where they must choose between a flower offering an immediately available reward and another flower containing a large reward but inaccessible for an indefinite period. The duration for which individuals sustain their preference for the delayed large-rewarded option before switching was recorded. Trials = number of trials; R = reward type; t = latency between bee landing on flower and access to reward.

Training phase

Coloured flowers were used in the training, preference, and test phases. Two separate groups of bees were tested, with each group exposed to a different colour pair: either blue and yellow or blue and purple. To colour the flowers, a paper dish was placed under the Petri dish containing the flower. The training phase consisted of 100 trials in which the subject had to visit two flowers differing in colour and reward. Each flower type was presented in 50 trials, with 25 trials per side. Only one flower was presented in each trial (Fig. 1a, b). The order of colours and sides was pseudorandomised to avoid consecutive repetitions. One colour was associated with an immediately available small reward (a 5 µl drop of 30% sucrose) (IR), while the other colour was associated with a delayed large reward (four 5 µl drops with a 15-s delay) (DR). The colour-reward associations were counterbalanced across bees. For IR flowers, the reward was readily and always accessible. For DR flowers, the reward was placed on the visible but inaccessible side. To obtain the DR, the bee had to first land on the flower; the reward then became available 15-s after this initial landing, regardless of whether the bee remained on the flower or flew away and returned during that time.

At the beginning of the training, to facilitate the learning process, an additional 5 µl drop was placed on the accessible side of the DR flowers for the first five trials (Fig. 1b). This drop attracted the bumble bee, motivating it to land on the DR flower and notice the four 5 µl drops of reward placed on the inaccessible side of it. Furthermore, this initial training was necessary to reduce the likelihood of the bee flying away when the flower was gently rotated to allow access to the sucrose solution. In these initial trials, the delay was 10-s. From the sixth DR trial, the extra drop was removed, and the delay increased to 15-s (Fig. 1c). Moreover, from the sixth DR trial, the reward was always made available 15-s after the bee’s first landing on the flower, regardless of whether the bee chose to remain on the flower (scored as total wait – bees that indeed waited on the flower for 15-s) or flew away and returned later. Once the bee consumed the reward and flew off, the flower was physically removed from the arena, and the bee was then allowed to exit. At that point, the bee was free to either remain in the Perspex tunnel or return to the nest box to deliver the sucrose. All bumble bees underwent 100 acquisition trials, with the inter-trial interval never shorter than 15-s.

Preference test

After completing the training phase, the individuals were presented with both coloured flowers for the first time and they were required to make a choice between the two (Fig. 1d). Once the bee landed on one flower, the other flower was promptly removed. In total, each bumble bee was tested over 10 choice trials, with at least a 60-s inter-trial interval (ITI) during which it remained in the tunnel or returned to the nest box. After each ITI, the tunnel was opened on both sides, and the individual had the option to proceed to the next trial or return to the nest before continuing. The IR and DR sides were pseudo-randomised. Individuals who did not choose the large reward at least once during the 10 trials (five bees), or who showed a strong side bias (one bee), were considered ineligible for the delay maintenance test.

Delay maintenance test

The bumble bees that waited at least once across ten trials were eligible for the delay-maintenance test. This test consisted of a single alternative choice, in which individuals were presented with both coloured flowers and had to choose between them (Fig. 1e), identically to the preference test. However, the 15-s time limit was removed so that the individuals would never have the possibility to get the reward from the DR flower. The amount of time each individual sustained the delay for the larger reward, whether by waiting on the flower or by flying away and returning, before switching to the smaller reward was recorded. Importantly, during this single-alternative choice the IR was always available and in full view to the bumble bee.

Statistical analyses

Generalised linear mixed models (GLMMs) were used to analyse the data (glmer and emmeans function from the R package lmer4 and emmeans repsectively38,39). R 4.3.1 was used for all analyses. In the training phase, the statistical model included “bee wait” (coded as a binomial response, 0 = bee did not wait 15-s on the DR flower; 1 = bee did wait 15-s on the DR) as a dependent variable, “trial” (the first five trials were excluded as the interval was shorter than 15-s) as a numerical factor, and “delayed colour” as a fixed factor. In the preference test the statistical model included “bee choice” as the dependent variable, “trial” was a numerical factor and “colour pair” (i.e. blue and yellow or blue and purple), and “total wait” (i.e., the number of times the bees waited 15-s on the DR flower during the initial training), were fixed factors. In the delay-maintenance the statistical model included “delay-maintenance” as the dependent variable, “delay choice”, “colour pair” and “total wait” were fixed factors. In both statistical models colony identity was entered as a random factor whereas bee identity was only included in the training and preference test to account for repated measures. In all cases, different model specifications were compared for estimating the dependent variables controlling for other regressors (see Tables 1 and 4) and the significant model with the highest explanatory power (AIC value), if detected, was selected. Otherwise, the full model was selected and p-value of each factor was reported.

Table 1.

GLMM Analysis on bumble bees’ choice during the training. P-values represent the comparison with the model with one more level of complexity.

Models AIC Log-Lik χ2 P (> χ2)
Model 1: Choice ~ (1|Col/ID) 2176.6 -1085.28
Model 2: Choice ~ Trial + (1|Col/ID) 2030.7 -1011.35 147.84 2.2e-16
Model 3: Choice ~ Trial + ColourDR + (1|Col/ID) 1997.1 -992.53 37.65 6.67e-09
Open in a new tab

Table 4.

GLMM Analysis on bumble bees’ choice in the delay maintenance test. P-values represent the comparison with the model with one more level of complexity. All models were not significant, meaning that removing the corresponding factor from the model did not significantly reduce the model fit.

Models AIC Log-Lik χ2 P (> χ2)
Model 1: DelayMaintainance ~ (1|Col) 425.78 -209.89
Model 2: DelayMaintainance ~ DelayChoice + (1|Col) 427.75 -209.88 0.02 0.87
Model 3: DelayMaintainance ~ DelayChoice + totwait + (1|Col) 428.15 -209.07 1.60 0.20
Model 4: DelayMaintainance ~ DelayChoice + totwait + colourpair + (1|Col) 429.76 -208.88 0.39 0.53
Open in a new tab

Ethical statements

There are currently no regulations governing the welfare of insects in research. Nevertheless, our experimental approach adhered to the principles of the 3Rs (Refinement, Reduction, Replacement) and the Animal Sentience Precautionary Principle (ASPP)40–42. While some manipulation is necessary for behavioural studies, we minimised any direct handling to reduce any lasting effects on their well-being. Each bee was tested in a single day, and, after testing, the bees were removed from the colony to allow for the recruitment of new foragers. The bees were kept in a cage, provided ad libitum access to a sucrose solution, and returned to their natal colony only at the conclusion of the experiments.

Results

Training

Bumble bees began to wait on the DR flower for 15-s (without flying away) after an average of 8.58 ± 4.56 (mean ± S.E.) visits during the yellow–blue flower pair training, and after 6 ± 1.64 (mean ± S.E.) visits during the blue–purple flower pair training. Moreover, across 50 DR presentations, bumble bees waited 15-s on the flower (without flying away) in 39.7 ± 4.26 (mean ± S.E.; Fig. 2a) trials during the yellow–blue training, and in 35.4 ± 8.62 (mean ± S.E.) trials during the blue–purple training. The GLMM analysis showed that, over this training period, bumble bees increasingly began to wait for the entire 15-s (GLMM, trial, X2 = 134.8, df = 1, p < 0.0001; Table 1). Flower colour had a significant effect, with yellow being the most attractive, followed by the blue and then the purple (GLMM, X2 = 33.5, df = 2, p < 0.0001; post-hoc test, purple-blue p = 0.0003, yellow-blue p < 0.001, yellow-purple p < 0.001).

Fig. 2.

Fig. 2

Open in a new tab

Results of training, preference test and delay maintenance test. 2a Training. Percentage of bee waiting on the DR flower across the 45 trials for each colour delay flower. The first five trials were excluded because a drop of sucrose was immediately accessible. 2b Number of choices for the DR across 10 trials in the preference test. 2c Latency to switch to the IR during the delay maintenance test. The latency to switch to the small reward ranged from 3-s to 364-s. Box plot: horizontal lines represent the median; boxes extend from lower to upper quartile and whiskers indicate interquartile range above upper quartile (maximum) or below lower quartile (minimum). Coloured filled circles represent individual bumble bees.

Preference test

In the preference test, in which the bees faced both flowers simultaneously for the first time, 44 out of 49 bumble bees (89.8%) chose to wait at least once during the 10 trials. Over the 10 trials, these bees chose to wait an average of 4.6 ± 0.32 (mean ± S.E.) times (median = 4.5; Fig. 2b). A subset of 5 bumble bees (10.2%) consistently preferred the smaller immediate reward over waiting for the larger reward throughout the 10 trials. Overall, the trials did not affect bee choice behaviour (GLMM, X2 = 0.70, df = 1, p = 0.40; Table 2). Furthermore, neither total wait (GLMM; X2 = 0.02, df = 1, p = 0.88) nor flower pair (yellow-blue vs. purple-blue; GLMM; X2 = 1.96, df = 1, p = 0.16) had an effect on preference choice.

Table 2.

GLMM Analysis on bumble bees’ choice in the preference test. P-values represent the comparison with the model with one more level of complexity.

Models AIC Log-Lik χ2 P (> χ2)
Model 1: Choice ~ (1|Col/ID) 641.17 -317.58
Model 2: Choice ~ Trial + (1|Col/ID) 642.47 -317.24 0.69 0.40
Model 3: Choice ~ Trial + TotWait + (1|Col/ID) 644.01 -317.01 0.46 0.49
Model 4: Choice ~ Trial + TotWait + ColourPair + (1|Col/ID) 644.05 -316.03 1.95 0.16
Open in a new tab

In summary, while most bees chose the DR at least once, their overall preference was for IR, and selection of the DR remained highly variable across individuals (Table 3). A small subset of bees consistently preferred the IR, and their choices remained stable across trials. Furthermore, the results suggest that experience had no influence on their decision-making.

Table 3.

Summary of bumble bee performance during the preference test and the delay maintenance test. The squares marked with "DR" indicate instances in which bees chose the DR over the IR.

Bee
ID		 Preference test Delay maintenance test
Trial1 Trial2 Trial3 Trial4 Trial5 Trial6 Trial7 Trial8 Trial9 Trial10 TOT
1 DR DR DR DR DR DR DR DR 8/10 150-s
2 DR DR DR DR DR DR DR 7/10 79-s
3 0/10 IR—Excluded
4 0/10 IR—Excluded
5 DR DR DR DR DR DR 6/10 IR
6 DR DR DR 3/10 IR
7 DR 1/10 187-s
8 DR DR DR DR DR DR DR DR 8/10 IR
9 DR DR DR DR DR DR DR 7/10 364-s
10 DR DR DR 3/10 IR
11 DR DR DR 3/10 IR
12 0/10 IR—Excluded
13 DR DR DR DR DR 5/10 IR
14 DR DR DR DR DR DR 6/10 98-s
15 0/10 IR—Excluded
16 DR DR DR DR DR DR 6/10 IR
17 DR DR DR 3/10 IR
18 DR DR DR 3/10 IR
19 DR DR 2/10 IR
20 DR DR DR DR DR DR DR 7/10 IR
21 DR DR DR DR 4/10 185-s
22 0/10 IR—Excluded
23 DR DR 2/10 IR
24 DR DR DR 3/10 36-s
25 DR DR DR 3/10 IR
26 DR DR DR DR DR DR DR 7/10 6-s
27 DR DR DR DR DR 5/10 4-s
28 DR DR 2/10 60-s
29 DR DR DR DR DR 5/10 49-s
30 DR DR DR DR DR DR 6/10 4-s
31 DR DR DR DR DR DR 6/10 3-s
32 DR DR DR DR DR DR 6/10 IR
33 DR DR 2/10 IR
34 DR DR DR DR DR DR 6/10 IR
35 DR DR DR DR 4/10 IR
36 DR DR 2/10 IR
37 DR DR DR DR DR DR DR DR 8/10 3-s
38 DR DR DR DR DR DR DR 7/10 18-s
39 DR DR DR 3/10 15-s
40 DR DR DR DR 4/10 4-s
41 DR DR DR DR DR 5/10 4-s
42 DR DR DR DR DR DR DR DR DR 9/10 4-s
43 DR DR DR DR 4/10 60-s
44 DR DR DR 3/10 IR
45 DR DR DR DR 4/10 44-s
46 DR 1/10 IR
47 DR DR DR DR DR 5/10 29-s
48 DR DR DR 3/10 260-s
49 DR DR DR DR DR DR 6/10 7-s
Open in a new tab

Delay maintenance test

In the delay maintenance test, 24 out of these 44 bumble bees (54.5%; Table 3) waited for the larger reward, maintaining their delayed choice for an average duration of 70 ± 19.29-s (mean ± S.E; Fig. 2c). Waiting times ranged from a minimum of 3-s to a maximum of 364-s. Nine bumble bees (20%) switched to the smaller reward before 15-s (mean duration 4 ± 0.44-s), while the remaining 15 out of 44 (34%) stayed longer than 15-s before switching to the smaller reward (mean duration 109 ± 26.17-s). The delay maintenance ability was not affected by the preference test (GLMM; X2 = 0.0001, df = 1, p = 0.99; Table 4). Furthermore, neither total wait (GLMM; X2 = 1.055, df = 1, p = 0.30) nor colour pair (yellow-blue vs. purple-blue; GLMM; X2 = 0.35, df = 1, p = 0.55) had an effect on the delay maintenance performance.

In summary, only a few individuals sustained their decision to wait for the larger reward, and the considerable variability suggests a limited ability to endure the delay. In addition, their decision was not influenced by prior experience (preference test) and the colour combination.

Discussion

To date, there has been little research into self-control and how often the decision to wait is linked to the actual ability to do so in small-brained animals. In this study, we investigated self-control in bumble bees through a paradigm that tested, after training, the extent to which bees were able to sustain waiting despite the availability of a smaller, immediately accessible reward. In the preference test, bumble bees showed some ability to choose the larger reward with a delay, but this behaviour was highly variable and overall not statistically significant. In the delay maintenance task, 24 out of 44 bees (54.5%) waited for the delayed reward. Indeed, waiting times in this task, varied widely across individuals. While a few bees demonstrated an ability to wait for longer period of time similar to that seen in some mammals, birds, and fish (with one individual resisting for up to 364-s), the overall ability to sustain waiting was limited and highly variable among other individuals8,9,16,17.

Our results may provide insight into self-control mechanisms. It has been suggested that metabolic rate is indicative of self-control, with smaller species needing to replenish energy more frequently, potentially leading to reduced self-control compared to larger species with lower metabolic rates24. However, our results challenge, to some extent, this hypothesis. Despite bumble bees being larger and having lower metabolic rates compared to honey bees43–47, they were able to wait for the larger reward for an average of 70-s in the delay maintenance task. This delay duration, though variable among individuals, is remarkable for a species with such metabolic rate. In addition, bumble bee colonies are smaller than honey bee colonies, ranging from 50 to 400 individuals45,48. Unlike honey bees, which store large quantities of honey, bumble bees store only small amount of food, with each individual forager playing a crucial role in collecting nectar and pollen daily for the colony48,49. Thus, individual energy levels are closely linked to colony needs, suggesting that the ability to delay immediate rewards may involve a trade-off between individual foraging decisions and long-term colony benefits. However, this likely depends on factors such as the duration of the wait for larger rewards and the availability of alternative resources in the environment.

Delayed gratification is thought to be a sophisticated cognitive ability primarily observed in large-brained vertebrates, which are often characterized by socio-ecological pressures, such as the need for group cooperation to coordinate and maintain social bonds9. However, accumulating evidence in other vertebrates and invertebrates, such as fish and cuttlefish respectively, challenges this assumption9,17, as these species show a comparable ability to inhibit behaviour and do not have complex social lives. Although bumble bees are eusocial insects that live in small colonies with a somewhat organized division of labour and exhibit complex cooperative behaviour, the underlying mechanisms and depth of these interactions are fundamentally different from those of vertebrate social species50,51. Mammals and birds engage in cooperation driven by cognitive and affiliative bonds, reciprocal relationships, and alliances. In contrast, the cooperation in eusocial insects is more rigidly structured, primarily guided by genetically programmed behaviours and chemical signals that coordinate colony activities. Their different social systems, communication mechanisms, and evolutionary pressures reflect alternative strategies used by different species to cope with social interactions and ecological challenges. Therefore, our preliminary findings of the ability of a few bees to show self-control in a delay maintenance task add to ongoing discussions questioning whether delayed gratification arises exclusively from socio-ecological pressures.

In bumble bees, the ability to delay gratification may have evolved, similarly to cuttlefish, to optimise their foraging behaviour52,53. These pollinators encounter various plant species while flying in their natural habitats and must compare the visual cues from specific flowers with stored memories of flower signals and associated rewards54. To accomplish this complex task, many insects, including bumble bees, become temporally specialized on a few, or sometimes even a single, flower species46,55. As a result, when searching for their preferred flowers, bees and other pollinators can disregard other species, even if they might offer equal or greater rewards. This flower constancy is widely accepted as an efficient strategy, even from a cognitive point of view, where sticking to familiar, rewarding flowers proves more efficient despite potential alternatives27,56. Foraging on new flower species requires experience in handling them effectively, which may involve an initial period of inefficiency27,57. Therefore, self-control may improve foraging behaviour in bumble bees, helping them ignore and bypass flowers that are, for example, closer to the colony but less rewarding, in favour of more profitable flowers further away. However, this ability to inhibit must also be balanced with the need for flexibility, allowing bees to remain open to new flowers when familiar ones are no longer rewarding or available. The extent to which individual bees are able to maintain this balance may vary depending on their cognitive traits, prior experiences, and environmental and colony conditions, leading to individual differences in foraging strategies.

Associative learning has also been suggested as a mechanism by which self-control can develop58. Through experience, animals can learn to resist immediate rewards and exert some control. For example, self-control can emerge when waiting behaviour is reinforced by the prospect of receiving a higher or better reward later59,60. In this study, bumble bees were trained through over 100 trials to associate different flowers with colour and reward, with one flower offering a larger delayed reward after a 15-s interval. During this extensive training, the bumble bees learned the colour/reward/timing contingency of the flower, as they began to wait on the DR flowers. Despite this, when the bees were confronted for the first time with both flowers during the preference test, we found no evidence that the training affected their behaviour. More importantly, no learning was observed even during the 10 trials carried out as part of this preference test, and we found no evidence that the preference test affected the bees’ performance in the delay maintenance task. Based on these findings, we suggest that while associative learning may play a role in facilitating self-control, it is unlikely to fully account for the observed behaviours in our study. Nevertheless, the relationship between associative learning and self-control warrants further investigation.

Existing evidence suggests a relationship between intelligence and self-control among a variety of evolutionarily unrelated species25,61,62. This suggests an intriguing concept of convergent evolution, whereby the ability to delay gratification has evolved independently multiple times across different vertebrate and invertebrate taxa. Moreover, it is likely that this phenomenon reflects diverse decision-making processes62. However, whether this convergence extends to eusocial species remains an open question that requires further investigation11,24,63. Taken together, our results suggest that at least a few bumble bees in our sample may possess some ability to endure delayed gratification, similar to what has been observed in other species. Still, due to variability in how individual bees exert self-control, further research is needed to determine whether bees can consistently perform at similar levels. While caution is needed in drawing firm conclusions, our results suggest that the ability to exert some form of inhibition can be achieved, despite differences in the underlying decision-making mechanisms, which may not necessarily be linked to advanced cognition.

Acknowledgements

We acknowledge the assistance of Elisa Pasquini, Federico Ferrante and Giulia Ricciardi for assistance in the Lab.

Author contributions

LB: Conceptualization, Formal analysis, Writing- Original draft preparation, Reviewing and Editing writing. ER: Investigation. GB: Investigation. DB: Conceptualization, Methodology, Formal analysis, Resources, Data curation, Supervision, Writing- Original draft preparation, Reviewing and Editing writing, Project administration, Funding acquisition. All authors gave final approval for publication.

Funding

Funding for this project was provided by the University of Florence. LB was supported by a Marie Skłodowska-Curie Postdoctoral Fellowship (FEAR-BEES—101065873) during writing.

Data availability

The raw data supporting the findings in this paper are available on request from the corresponding authors.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Luigi Baciadonna, Email: luigi.baciadonna@upmc.fr.

David Baracchi, Email: david.baracchi@unifi.it.

References

  • 1.Santos, L. R. & Rosati, A. G. The evolutionary roots of human decision making. Annu. Rev. Psychol.66, 321–347 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Mischel, W. Processes in delay of gratification. Adv. Exp. Soc. Psychol.7, 249–292 (1974). [Google Scholar]
  • 3.Mischel, W., Shoda, Y. & Rodriguez, M. L. Delay gratification in children. Science244, 933–938 (1989). [DOI] [PubMed] [Google Scholar]
  • 4.Nigg, J. T. Annual Research Review: On the relations among self-regulation, self-control, executive functioning, effortful control, cognitive control, impulsivity, risk-taking, and inhibition for developmental psychopathology. J. Child Psychol. Psychiatr.58, 361–383 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Diamond, A. Executive functions. Annu. Rev. Psychol.64, 135–168 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Duckworth, A. L., Quinn, P. D. & Tsukayama, E. What no child left behind leaves behind: The roles of IQ and self-control in predicting standardized achievement test scores and report card grades. J Educ. Psychol104, 439–451 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Boisvert, D., Stadler, W., Vaske, J., Wright, J. P. & Nelson, M. The interconnection between intellectual achievement and self-control. Crim. Justice Behav.40, 80–94 (2013). [Google Scholar]
  • 8.Beran, M. J. Self-Control in Animals and People. London, England: Academic Press (Acedemic Press, 2018). [Google Scholar]
  • 9.Miller, R. et al. Self-control in crows, parrots and nonhuman primates. Wiley Interdiscip. Rev. Cogn. Sci.10, e1504 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Susini, I., Safryghin, A., Hillemann, F. & Washer, C. A. F. NoDelay of gratification in non-human animals: A review of inter-specific variation in performance. BioRxi (2021).
  • 11.Cheng, K. E. N., Peña, J., Porter, M. A. & Irwin, J. D. Self-control in honeybees. Psychon. Bull. Rev.9, 259–263 (2002). [DOI] [PubMed] [Google Scholar]
  • 12.MacLean, E. L. et al. The evolution of self-control. Proc. Natl. Acad. Sci. USA111, E2140–E2148 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Dufour, V., Wascher, C. A. F., Braun, A., Miller, R. & Bugnyar, T. Corvids can decide if a future exchange is worth waiting for. Biol. Lett.8, 201–204 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Auersperg, A. M. I., Laumer, I. B. & Bugnyar, T. Goffin cockatoos wait for qualitative and quantitative gains but prefer ‘better’ to ‘more’. Biol. Lett.9, 20121092 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Hillemann, F., Bugnyar, T., Kotrschal, K. & Wascher, C. A. F. Waiting for better, not for more: Corvids respond to quality in two delay maintenance tasks. Anim. Behav.90, 1–10 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Koepke, A. E., Gray, S. L. & Pepperberg, I. M. Delayed gratification: A grey parrot (Psittacus erithacus) will wait for a better reward. J. Comp. Psychol.129, 339–346 (2015). [DOI] [PubMed] [Google Scholar]
  • 17.Aellen, M., Dufour, V. & Bshary, R. Cleaner fish and other wrasse match primates in their ability to delay gratification. Anim. Behav.176, 125–143 (2021). [Google Scholar]
  • 18.Brucks, D. et al. Intra- and interspecific variation in self-control capacities of parrots in a delay of gratification task. Anim. Cogn.25, 473–491 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Stevens, J. R., Hallinan, E. V. & Hauser, M. D. The ecology and evolution of patience in two New World monkeys. Biol. Lett.1, 223–226 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Stevens, J. R. et al. Will travel for food: spatial discounting in two New World monkeys. Curr. Biol.15, 1855–1860 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Amici, F., Aureli, F. & Call, J. Fission-fusion dynamics, behavioral flexibility, and inhibitory control in primates. Curr. Biol.18, 1415–1419 (2008). [DOI] [PubMed] [Google Scholar]
  • 22.Bond, A. B., Kamil, A. C. & Balda, R. P. Serial reversal learning and the evolution of behavioral flexibility in three species of North American corvids (Gymnorhinus cyanocephalus, Nucifraga columbiana, Aphelocoma californica). J. Comp. Psychol.121, 372–379 (2007). [DOI] [PubMed] [Google Scholar]
  • 23.Tobin, H. & Logue, A. W. Self-control across species (Columba livia, Homo sapiens, and Rattus norvegicus). J. Comp. Psychol.108, 126–133 (1994). [DOI] [PubMed] [Google Scholar]
  • 24.Mayack, C. & Naug, D. Starving honeybees lose self-control. Biol. Lett.11, 20140820 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Schnell, A. K., Boeckle, M., Rivera, M., Clayton, N. S. & Hanlon, R. T. Cuttlefish exert self-control in a delay of gratification task. Proc. R. Soc. B.288, 20203161 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Kacelnik, A. Kacelnik, A. (2003). The evolution of patience. In G. Loewenstein, D. Read, & R. Baumeister (Eds.), Time and decision: Economic and psychological perspectives on intertemporal choice 115–138. Russell Sage Foundation. in 115–138 (2003).
  • 27.Chittka, L. & Thomson, J. D. Cognitive ecology of pollination: Animal behaviour and floral evolution (Cambridge University Press, 2001). [Google Scholar]
  • 28.Jones, P. L. & Agrawal, A. A. Learning in insect pollinators and herbivores. Annu. Rev. Entomol.62, 53–71 (2017). [DOI] [PubMed] [Google Scholar]
  • 29.Baracchi, D. Cognitive ecology of pollinators and the main determinants of foraging plasticity. Curr. Zool.65, 421–424 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Pull, C. D. et al. Ecology dictates the value of memory for foraging bees. Curr. Biol.32, 4279-4285.e4 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Heinrich, B. Energetics of pollination. Annu. Rev. Ecol. Syst.6, 139–170 (1975). [Google Scholar]
  • 32.Detrain, C., Deneubourg, J. L. & Pasteels, J. M. Decision-making in foraging by social insects. In Information processing in social insects 331–354 (Birkhäuser Basel, Basel, 1999).
  • 33.Pyke, G. H. & Starr, C. K. Optimal foraging theory. Encyclopedia of social insects 677–685 (2021).
  • 34.Hills, T. T. Animal foraging and the evolution of goal-directed cognition. Cogn. Sci.30, 3–41 (2006). [DOI] [PubMed] [Google Scholar]
  • 35.Loukola, O. J., Solvi, C., Coscos, L. & Chittka, L. Bumblebees show cognitive flexibility by improving on an observed complex behavior. Science355, 833–836 (2017). [DOI] [PubMed] [Google Scholar]
  • 36.Chittka, L. Bee cognition. Curr. Biol.27, R1049–R1053 (2017). [DOI] [PubMed] [Google Scholar]
  • 37.Calderai, G. et al. Nectar-borne GABA promotes flower fidelity in bumble bees. Entomol. Gen.43, 4 (2023). [Google Scholar]
  • 38.Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw.67(1), 48 (2015). [Google Scholar]
  • 39.Lenth, R. V. emmeans: Estimated marginal Means, aka least-squares means. (2025).
  • 40.Baracchi, D. & Baciadonna, L. Insect sentience and the rise of a new inclusive ethics. Anim. Sentience5, 29 (2020). [Google Scholar]
  • 41.Birch, J. Animal sentience and the precautionary principle. Anim. Sentience2 (2017).
  • 42.Crump, A., Gibbons, M., Barrett, M., Birch, J. & Chittka, L. Is it time for insect researchers to consider their subjects’ welfare?. PLoS Biol.21, e3002138 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Wolf, T. J., Ellington, C. P. & Begley, I. S. Foraging casts in bumblebees: Field conditions cause large individual differences. Insect. Soc.46, 291–295 (1999). [Google Scholar]
  • 44.Wolf, Th. J., Schmid-Hempel, P., Ellington, C. P. & Stevenson, R. D. Physiological correlates of foraging efforts in honey-bees: Oxygen consumption and nectar load. Funct. Ecol.3, 417 (1989). [Google Scholar]
  • 45.Goulson, D. Bumblebees: Behaviour, ecology, and conservation (Oxford University Press, 2010). [Google Scholar]
  • 46.Goulson, D. Foraging strategies of insects for gathering nectar and pollen, and implications for plant ecology and evolution. Perspect. Plant Ecol. Evol. Syst.2, 185–209 (1999). [Google Scholar]
  • 47.Suarez, R. K., Lighton, J. R., Joos, B., Roberts, S. P. & Harrison, J. F. Energy metabolism, enzymatic flux capacities, and metabolic flux rates in flying honeybees. Proc. Natl. Acad. Sci. USA93, 12616–12620 (1996). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Townsend-Mehler, J. M., Dyer, F. C. & Maida, K. Deciding when to explore and when to persist: A comparison of honeybees and bumblebees in their response to downshifts in reward. Behav. Ecol. & Sociobiol.65, 305–312 (2011). [Google Scholar]
  • 49.Katz, K. & Naug, D. Energetic state regulates the exploration-exploitation trade-off in honeybees. Behav. Ecol.26, 1045–1050 (2015). [Google Scholar]
  • 50.Crall, J. D. et al. Spatial fidelity of workers predicts collective response to disturbance in a social insect. Nat. Commun.9, 1201 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Hendriksma, H. P., Toth, A. L. & Shafir, S. Individual and colony level foraging decisions of bumble bees and honey bees in relation to balancing of nutrient needs. Front. Ecol. Evol.7, 177 (2019). [Google Scholar]
  • 52.Heinrich, B. ‘Majoring’ and ‘Minoring’ by foraging bumblebees. Bombus vagans: An Exp. Anal. Ecol.60, 245–255 (1979). [Google Scholar]
  • 53.Heinrich, B. Do bumblebees forage optimally, and does it matter?. Integr. Comp. Biol.23, 273–281 (1983). [Google Scholar]
  • 54.Willmer, P. G. & Stone, G, N. Behavioral, ecological, and physiological determinants of the activity patterns of bees. In Advances in the Study of Behavior (eds. Slater P.J.B et al.) 347–466 (Elsevier Academic Press, 2004).
  • 55.Raine, N. E. & Chittka, L. Comparison of flower contancy and foraging performance in three Bumblebees species (Hymenoptera: Apidae: Bombus). Entomol. Gen.28, 081–089 (2005). [Google Scholar]
  • 56.Grüter, C. & Ratnieks, F. L. W. Flower constancy in insect pollinators: Adaptive foraging behavior or cognitive limitation?. Commun. Integr. Biol.4, 633–636 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Laverty, T. M. & Plowright, R. C. Flower handling by bumblebees: A comparison of specialists and generalists. Anim. Behav.36, 733–740 (1988). [Google Scholar]
  • 58.Lind, J. What can associative learning do for planning?. R. Soc. Open Sci.5, 180778 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Anderson, J. R., Awazu, S. & Fujita, K. Can squirrel monkeys (Saimiri sciureus) learn self-control? A study using array selection tests and reverse-reward contingency. J. Exp. Psychol. Anim. Behav. Process.26, 87–97 (2000). [DOI] [PubMed] [Google Scholar]
  • 60.Beran, M. J. et al. Rotating Tray Task and the Accumulation Task. Behav. Process.129, 68–79 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Beran, M. J. & Hopkins, W. D. Self-Control in chimpanzees relates to general intelligence. Curr. Biol.28, 574-579.e3 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Schnell, A. K., Boeckle, M. & Clayton, N. S. Waiting for a better possibility: Delay of gratification in corvids and its relationship to other cognitive capacities. Philos. Trans. Royal Soc. B377(1866), 20210348 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Wendt, S. & Czaczkes, T. J. Individual ant workers show self-control. Biol. Lett.13, 20170450 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The raw data supporting the findings in this paper are available on request from the corresponding authors.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES