Can bees feel pain?

Stephan Balancy

Queen Mary University London

School of Biological and Behavioural Science

2022

Abstract

Bumblebees are not covered by animal welfare laws that encompass many other animals. Due to the ease, cost-effectiveness, and environmental benefits of maintaining, mass producing, and using insects to feed the growing human population, there is consideration about how insect farming could be used to substitute conventional livestock. The attention falls on the ethical status of the insect and whether they have a conscious experience of the world that differs from humans. Pain alerts the sufferer to actual or potential tissue damage and can be used to test the ethical status of the insect. A motivational trade-off experiment was set up using feeders and heated feeders. The bee would choose non-heated feeders, and the sucrose concentration available at the other heaters would affect the bee’s decision. The hypothesis in this study was that the bees’ decision relied on using memory-based positional and colour cues, that the bee does feel pain and would choose non-heated feeders. The results show that the proportion of feeding events at the heated feeders was significantly affected by the sucrose concentration available at the alternative feeder. The proportion of feeding events at the heated feeders was significantly lower than at the unheated feeders when the sucrose concentrations at both feeders were equal. There was also no significant difference between the proportion of feeding events at the heated feeders and at the test feeders. The bumblebee may be considered sentient, it does feel pain, and would rather suffer painful stimuli to receive larger amounts of sugar than to receive less sugar.

Introduction

Pain can be described as "an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage", according to the International Association for the Study of Pain (IASP)(2). As this description states, pain is more complex than nociception. It involves higher centres to process the affective components which contribute to an alteration in behaviour through learning, to avoid further pain. However, we understand pain through the human experience through verbal reports, but is the sensation of pain as we know it the same in other species?

This question raises many implications on how humans treat animals and ethical concerns with experimentation and welfare laws. The ability to feel pain does not require a verbal report. This would exclude neonates and adults with an inability to communicate, so to prove animal experiences of pain, more vigorous testing of various criteria must be performed (3). Evidence used to prove animal experiences of pain includes central processing of nociception, alterations in motivation that demonstrate learning and fear, physiological changes, protective behaviour of the injured site, and that all these changes can be modified through analgesia(3). Furthermore, changes in motivation can be assessed through learned self-administration of analgesia, prioritising noxious stimuli over other stimuli, avoidance learning, and motivational trade-offs (3).

The subjective experience of pain has been studied across many species. It is generally accepted that vertebrates experience pain to a certain extent, which warrants welfare laws in favour of the vertebrates. However, the experience in invertebrates such as insects has had fewer studies surrounding it and is therefore poorly understood, and these animals are not protected by welfare laws addressing infliction of pain (4).

Arguments against Insect Consciousness

The topic is a hot debate as it can never be proven that insects do not feel pain in the same sense humans do as it is impossible to know what it is like from their perspective and verbal report is impossible (4). Experimentation shows that many insects, such as caterpillars, will continue feeding while being eaten by tachinid larvae; many do not show protective responses to injured body parts and continue as if unaffected, and male mantids will continue mating while being cannibalised (4, 5). These responses differ significantly from human or mammalian responses. They would suggest a lack of affective processing of pain or that stimuli such as mating have greater priority to insects than their survival.

A further argument has been made that insects are comparable to robots. Many robots, such as practice dummies for dentistry exams, are programmed to flinch or cry out when the pre-programmed nociceptors are triggered (4). Furthermore, some robots can be programmed with a gate control mechanism and learn avoidance behaviours when exposed to noxious stimuli that will alter behaviour for some time, as demonstrated in the artificial intelligence (AI) used in video games (4). Robots can be programmed to verbalise pain, demonstrate learning behaviour, and even learn patterns to beat human opponents in games such as chess, which goes far beyond what insects are capable of in terms of a subjective experience of consciousness and experiencing pain (4, 6) Suppose robots are capable of all the criteria required for experiencing pain and insects do not. In that case, it would be more ethical to develop welfare for robots than insects or not provide insects with ethical treatment as they are no more than simple biological robots.

One function of the affective part of pain is behavioural plasticity to allow animals to avoid stimuli which previously resulted in pain. However, this function is limited in insects as they lack the plasticity to respond to pain and instead have "pre-programmed" responses to pain, providing predictable avoidance and escape patterns (5). Furthermore, analysis of the organisation of the human nervous system when compared to an insect leads shows a vast difference in complexity. Human nociceptors are shown to project to the central nervous system (CNS) into two broad systems to aid in sensory-discriminative aspects and affective functions.

However, given that insect nervous systems are comprised of about 10^5 neurones compared to mammals which exceed 10^9, it is reasonable to assume higher functions such as emotional processing of pain may not be present in insects (5). Additionally, given the difference in lifespans and size, the practical portion of the pain that causes learning and motivation changes may not be used in an insect because it is not as essential to preserving longevity as completing other tasks such as mating or retrieving food for a hive (4).

Within this simplified structural organisation of insect brains, various elements that make humans capable of higher-order thinking are absent in the insect brain. One significant anatomical difference is the lack of a cerebral cortex in the insect brain, which holds relevance when assessing the ethics of an insect because the cortex plays a prominent role in emotions and motivation(7). Patients with damage to their orbital prefrontal cortex or monkeys with lesions in this area demonstrate an impairment of their emotional and motivational functions (7). Therefore, when any insect makes a motivational or perceived emotionally driven decision, it is not processed like larger-brained mammals which use a cerebral cortex and cannot be considered the same.

The presence of endogenous opioids within insects could be argued for the modulation of what would be a pain state in an insect. However, opioids such as a synthetic enkephalin analogue were shown to have binding sites in the cerebral ganglia of Leucophaeae Madera. However, they were shown to have roles modulating the reproductive cycle of the insect and, therefore, although present, do not demonstrate a descending pain modulation mechanism (8). This suggests against insects feeling pain because the proteins mammals use to modulate pain have evolved for different purposes in insects. After all, the ability to modulate pain is not as valuable for insects as it is for mammals.

Ultimately, the perception of pain and its link to ethics depends on whether insects such as bees are conscious creatures. Therefore, this relies on the answer to the complex question of where consciousness exists in the body. If localised, comparison to similar or missing structures in the insect minds would answer this. However, there are different states of consciousness, such as higher-order consciousness, which relies on the cortex and allows a self-awareness of one's self and mental state (1). Subjective consciousness relies more on midbrain structures and has comparable carry-over to insect brains (See Fig. 1). Although it can be considered consciousness, it may not be deemed higher-order thinking. It would be deemed ethical to inflict pain upon creatures only capable of subjective consciousness as pain states which alter the mental state may not be possible in such simplistic structures (1).

Insect brain schematic (Not to scale)

Fig. 1 shows an insect brain schematic adapted from (1). The antennal lobes process olfaction, and the lamina, medulla, and lobula process visual stimulus(1). These structures process environmental awareness. The mushroom bodies facilitate learning and memory (1). The central structure, comprised of the upper and lower central bodies and noduli, organises spatial information, although this varies between species (1). Finally, the protocerebrum contains information about the physiological state of the insect and integrates sensory information (1).

Arguments for Insect Consciousness

While it is impossible to tell if an insect truly has consciousness, various criteria can be used to identify an affective component of pain. For example, an experiment on hermit crabs demonstrated they could make motivational trade decisions by providing different shells (9). The crabs given preferred shells were willing to endure the electric shocks produced by wires installed into the shells significantly longer than crabs given a less preferred shell (9). This experiment demonstrated a motivation trade-off, implying that affective processing of pain is possible in invertebrates. The crabs also spent less time investigating the preferred shells before entering, demonstrating an inclination to reduce caution when finding a sufficient reward; proof of motivational trade-off is one of the signs of higher-order pain processing (9).

When working, forager bumblebees have also been demonstrated to have emotional states when working (10). When a pretest sugar reward, bees become more 'optimistic' and will make decisions while foraging differently than if they had not received a reward. Bees were trained to identify a reward at a blue feeder but not green and were then additionally presented with an ambiguous green/blue coloured feeder (11). The bees that received a 60% sugar drop reward were much quicker to try the ambiguously coloured feeder than the controls, demonstrating a positive cognitive bias when dopamine from the reward is higher (10, 11).

Although endogenous opioids may not have a pain-modulating function in insects, it does not mean no descending modulatory pathways are available. Descending pain modulation has been demonstrated through behavioural, molecular, and anatomical evidence in insects (12). Behavioural evidence has been shown through the sting of a parasitic jewel wasp on the American cockroach, which modulates pain in the cockroach and increases the threshold pain needed to trigger the defensive escape response (12, 13). The sting releases neurotoxins aimed at the cockroach's sub-oesophagal ganglion, a part of the cerebral nervous system which triggers a walking response. The toxins prevent descending signals from the ganglion to prevent a walking response from being initiated, which demonstrates that descending modulation of nocifensive escape responses is achievable in the insect brain (13).

Modulating defensive behaviours can also occur through stimuli such as feeding. This modulation is like that demonstrated in the experiment conducted in this paper. The fruit fly has been shown to decrease avoidance of noxious stimuli when presented with the taste or smell of sugar, demonstrating a motivational trade-off that can be explained by modulating pain in exchange for a reward, which may also explain the absent defensive behaviours of some insects show when in danger (12). An experiment on Drosophila found that hunger also triggered a suppressed response to nociceptive stimuli, showing the insect's motivation for food will increase, thus decreasing its necessity to preserve health (14).

Additionally, when decapitated, the Drosophila no longer had these suppressed behavioural responses, demonstrating that the change necessary to modulate the pain was in the brain(14). The neuropeptide responsible for these modulatory changes is Leucokinin and the Leucokinin receptor (14). This neuropeptide was silenced using CRISPR/Cas9 to knock out the receptor and Leucokinin-expressing neurones, resulting in an inability to suppress defensive responses to heat (14). Therefore, insects do have endogenous proteins responsible for pain modulation.

Anatomical evidence for descending pain modulation has been demonstrated through long axons descending from the brain of Drosophila to Basin and Goro neurons, which triggers the defensive rolling response in the presence of a noxious stimulus (15). These neurones depend on sensory pathway-specific modulation during development, allowing Drosophila larvae to become accustomed to the stimuli they would come into contact with within their environment (16). The nociceptors in Drosophila are connected to second-order neurones that can be multimodality (touch, vibration etc.) or nociceptor-specific, allowing for the development of input-specific modulation of various pathways that change the adult Drosophila's response to various stimuli (16). These changes to nociceptive processing are caused by serotonergic interneurons that can cause feedback inhibition to the presynaptic nociceptors to allow long-term functional plasticity to the insect nervous system, which is rather sophisticated and beyond that of a programmed robot (16).

Additionally, grooming and protective behaviours have been observed in some invertebrates, such as glass prawns (17). Glass prawns have been shown to rub and protect an antenna exposed to sodium hydroxide. Shore crabs with sodium hydroxide rubbed against their mouth will repeatedly scratch at their mouth with their claw (17). Additionally, a declawed crab (Cancer Pagurus) will protect the wounded site by holding the remaining claw over the wound during a confrontation. Compared to a crab that autotomised their own claw, they have a lower motivation to compete for mates, demonstrating that the behaviour change is due to the injury rather than the lack of a claw (17).

Another example of the pain experience is the long-term motivation change that comes from pain, which shows learning and adaptation due to a pain stimulus and is one of the functions of the affective component of pain. This experiment was demonstrated by Elwood & Appels' experiment in hermit crabs, where they were shocked within their shell at a lower intensity than that which caused them to evacuate (18). The crabs were offered a new similar shell, and the group that experienced the shock were more likely to move into the new shell and did so quicker and with less investigating than the non-shocked group (18). Therefore, the crabs demonstrated long-term motivational change that lasted up to a day, showing the long-term change in invertebrates.

Avoidance learning is another example of an affective experience of pain that causes learned behaviour changes to avoid repeating a course of actions that caused pain and damage. This experiment was demonstrated in crayfish (Orconectes rusticus), which were placed in a tank with four areas, two of which delivered shocks, and compared to a control which received shocks everywhere in the tank(19). The crayfish in the experimental tank rapidly moved to the safe areas and, if returned to the danger area, would get shocked and retreat to the safe area where it slowed and remained, compared to the control group, which used the whole tank(19). This experiment is a demonstration of learned avoidance behaviour in invertebrates.

Bumble bees have previously been shown to experience cross-modal object recognition using visual and tactile discrimination in an experiment conducted by (20). This experiment gives insight into how bees process the environment around them and shows that their perception of their environment comes as an integrated package of the sensory modalities they receive, unlike a simple machine which acts on a particular sense. This experiment suggests an advanced, globally accessible perception of the world more akin to large-brained animals such as humans, linking the conscious state of an insect to something similar that we would experience.

Additionally, it can be suggested that insects can experience consciousness because they can also enter states where consciousness is absent, such as sleep (21). Drosophila have been shown to sleep, and sleep patterns can be pharmacologically altered using caffeine, similar to humans (21). Additionally, when deprived of sleep, the flies would experience rebound sleep, where they spent more time sleeping after being deprived, and the amount of rebound sleep needed was dependent on how much sleep was lost (21).

When deprived of sleep, the flies would also require an increasing amount of arousal to keep them awake longer, suggesting that, like humans, insects get tired when staying awake for prolonged hours, which is a shift in the state of arousal (21). These alterations in arousal states demonstrate shifts in consciousness, which can only be achievable if the animal could experience a state of consciousness, as opposed to a robot which would not require sleep so long as an energy source can be provided to them.

The dopaminergic system in many mammals plays essential roles in motivational states, reward, and motivational behaviour, and in invertebrates such as bumblebees, it is not so different. An experiment on bees found that an unexpected sugar reward given to bees before foraging altered their behaviour (11). The administration of an unexpected pretest reward reduced the time to initiate foraging in an ambiguous stimulus compared to control bees. When exposed to a pretest reward after a simulated predator attack, the rewarded bees did not experience an aversive behavioural response and returned to normal foraging sooner than control bees (11). These changes were reversed when the rewarded bee was given a topical dopamine antagonist, fluphenazine, suggesting the behaviour change is attributed to the dopamine system, which is not so different from humans or other large-brain animals and suggests bees are capable of experiencing emotional states (11).

When focussing on anatomical comparisons, the insect brain has certain parallels that can be drawn; for example, the superior colliculus of the human brain performs a similar role to the central bodies and noduli of the insect brain in organising environmental spatial information and providing input and outputs to and from the protocerebrum and mushroom bodies, which are analogous to several midbrain structures in the vertebrate brain such as the hypothalamus, to filter information and drive motivational states (22). Although the vertebrate mind is larger and has a more significant number of neurons, it is more the brain's functional organisation that matters when comparing functions. The human brain is more precise and has a greater storage capacity, but as functioning goes, the same process that may involve thousands of neurons in the human brain can be achieved with the more economical insect brain with fewer neurons (22).

In the ability to experience consciousness at the lowest level, a being should be able to sense its internal and external environment in relation to time and space to understand the interrelations between these elements. For example, a nematode (Caenorhabditis elegans) can sense its internal environment using specialised receptors aimed at multiple modalities in relation to time, but cannot sense its position in space (22). Therefore, it acts only on what it is immediately feeling at the time, unlike a bee, which will act when hungry by navigating to places it recognises as food (22). The bee can, therefore, also identify itself relative to time and space, so it can be considered subjectively conscious, unlike a roundworm. Therefore the ethical topic of subjecting it to pain would be different to inflicting pain on a nematode or plant.

The human population is rapidly increasing and is expected to reach 10 billion by 2050. With a significant rise in population, so comes the difficulty in feeding. To do so sustainably is increasingly difficult due to livestock farming causing a considerable contribution to climate change. As such, mass-producing insects as a food source is one method suggested to combat this issue; however, ethical implications must be considered. Animal welfare laws should be re-evaluated before commencing operations of such a scale. Insects are not protected by animal welfare protection, and therefore, questions such as conscious awareness, pain, and suffering in insects should be considered.

The United Nations suggested the conversion of livestock to insects in anticipation of the challenge of feeding the ever-growing human population by sustainable means (23). Insects provide good nutrition and are a source of protein, fat, fibre, vitamins, and minerals and can therefore provide a large quantity of sustainable food. Additionally, the mass production of insects would provide job opportunities through gathering, farming, and selling products, which could improve the diets and livelihoods of people in less economically developed countries (23). The environmental impact of switching to insect farming would cause reduced land space and water usage compared to conventional livestock farming and reduce greenhouse gas and ammonia production (23).

Furthermore, when consumed more than livestock, insect farms may have a reduced risk of passing on zoonotic infections. Farmed insects can also feed livestock such as aquaculture creatures and poultry, which reduces the resources needed to conventionally feed (23). The idea has been implemented in countries such as Thailand and Vietnam, where cricket farming for human and zoo animal consumption is used, although never to an industrial scale required to feed large populations (23).

Additionally, using insects for animal models would allow easier access to models for treating chronic pain conditions. By proving insects can perceive pain on a scale beyond reflexive nociception, further insight into the levels of consciousness experienced by insects would further progress the search for defining consciousness, and potential analgesics could find use in preclinical trials using insects are easier to obtain and maintain in accordance with the 3 R principles.

This question will also take a step forward in answering the complex philosophical question of the nature of consciousness and whether lesser creatures such as insects are capable of it. In humans, consciousness can be identified through having an experience and feeding back to the observer. Perceptual awareness scales can be used to test a human feedback on a conscious experience of a sensory modality such as a presented image, and the method of feedback can range from verbal feedback to subtle signs to answer 'yes' or 'no' for patients incapable of speech or little movement (24).

Anatomically, the human mind contains identified structures believed to be the minimum required for humans to have conscious experiences called the neural correlates of consciousness (NCC) (24). These can be subdivided based on specific conscious experiences called content-specific NCC, which in the context of seeing a face would be the neurones and their mechanisms that activate when seeing or dreaming a face (24). The full NCC are the general structure required for conscious experiences. However, they do not include the background conditions required, such as glucose and oxygen required for essential brain activity for consciousness to occur (24).

To identify the full NCC, conscious states are used and compared to states where consciousness is absent, such as dreamless sleep or anaesthesia. Haemodynamic analysis of the brain when someone is awake with their eyes closed vs someone under rapid-eye-movement sleep (REM), in which dreams occur vs. dreamless sleep, can help identify the picture of the entire NCC (24). The brainstem, when damaged, results in a coma, but this is due to projections from the brainstem projecting to the thalamus, which then projects globally, thus removing the background conditions required for consciousness (24). On the other hand, severe cortex damage can cause a vegetative state in which the patient is conscious but unable to move. This means the brainstem facilitates consciousness by allowing the necessary interactions among cortical areas (24).

The basal ganglia, claustrum, and the primary visual cortex have all been demonstrated to affect awareness and consciousness when stimulated or damaged. However, they alone do not encapsulate the complete picture of consciousness (24). Electrostimulation of the posterior cortex has been shown to stimulate experience in patients undergoing neurosurgery. Functional magnetic resonance imaging (fMRI) has shown that the posterior cortical hot zone is more active in REM sleep than the prefrontal cortex. These tests provide evidence that conscious experience may lie in these cortical areas (24).

When observing the brain and behaviour of other animals, the conscious experience can be observed through an animal's ability to dream and respond to stimuli. Macaque monkeys can signal that a visual stimulus is present under blindsight conditions, implicating a form of a report of consciousness without vision (24). It is likely that mammals have conscious experience, and similarly evolved animals, such as monkeys with similar brain organisation, can relate closely to the human experience. However, bumblebees have a far less complex nervous system and lack the ability to make complex decisions. While certain paradigms show choices made by individual insects, it is difficult to prove whether this decision was made with the conscious intention or by following an unconscious pre-programmed decision.

Compared to the structure of a bee brain, where consciousness is more experienced as a processed integration of sensory information rather than a state of awareness of experience. In this sense, the bee is only conscious in the sense that it can feel the stimuli of its environment but does not have the awareness that it is experiencing it. This subjective experience can be shown in the context of the understanding of pain, where an unpleasant sensory stimulus will cause larger creatures to receive a negative experience of consciousness and would therefore avoid it. However, an insect, driven more by the desire to survive, would not need a subjective positive experience of consciousness if it is not capable of experiencing consciousness in the first place.

This experiment will test the motivational trade-off capabilities of bumblebees (Bombus terrestris). We hypothesise that the bees will learn that the greater reward comes at the cost of pain and will be able to weigh up the risks and benefits to make motivational trade-off decisions and avoid the noxious stimulus. The motivational trade-off will provide a key piece of evidence for proving an insect's ability to feel and process pain and support the idea that consciousness can be felt by invertebrates (3). This experiment hypothesises that bees rely on spatial and colour cues to form memory-based decisions to guide feeding behaviours. Additionally, the bumblebee will be able to perform motivational trade decisions based on its ability to feel pain.

Methods

Materials

The experiment was designed and run in a similar way to Gibbons et al experiment on motivational trade-offs(25).

The experimental design tested 12 forager bumblebees (Bombus terrestris) all taken from the same hive obtained from Biobest, Belgium. The bees were transferred to 28*16*11cm two-part wooden boxes with one part used as a nest, covered completely to ensure dark conditions, and the other had a Perspex top to allow for a 12:12 hour light-dark cycle and a 1cm-deep gravel substrate (see Fig. 2). A 1cm hole connected the two sides of the box and externally, four 2cm diameter exterior holes with wire mesh covering to prevent escapes were included to allow sufficient ventilation.

A Perspex tunnel 25cm in length and 3.5 x 3.5 cross-section connected the nesting box to the testing arena which was 56 x 56cm in area and roofed with a Perspex covering. The tunnel had slots along it to allow for strips of cut-up cards to be slotted along the tunnel and control access to the arena. The arena had a cut-out flap on one side to allow human access for cleaning and feeding purposes and makeshift Lego brick stairs leading up to the entrance of the Perspex tunnel to allow the bees easier access to the nest. Within the arena four feeders were set up 30cm from the entrance and 15cm from each other, arranged in a semi-circle (See Fig. 2). The feeders had a 50 x 50cm heat pad on the bottom and four 25 x 25cm coloured Perspex squares arranged as shown in figure 2 in either pink or yellow. The feeders were placed in alternating colours and the pattern was switched as necessary for the experiment, with one colour representing heat or concentration of sugar found at the heater.

The heat pads themselves were custom-made using a single side 70micrometer thick copper-clad standard printed circuit board laminate (made from woven fibre-glass and epoxy resin, 1.57mm thick and 50 x 50mm each). Each with an etched resistive heating element consisting of 18 loops with a path width of 0.6mm. The heat pads are each fitted with LM35 temperature sensors with thermally conductive adhesive (AG TermoGlue, TermoPasty, Poland). The power to the heat pads was delivered via a 12V DC output regulated by TIP122 transistors, driven with Arduino Uno Rev3. The readouts for temperature were smoothed with a Kalman filter (SimpleKalmanFilter library).

Arena and feeder setup

Fig. 2 shows the testing arena setup, image and setup remain the same as the experiment used in Gibbons et al experiment(15). The dimensions and the feeders’ orientation are shown and a bee is shown in the feeding position.

Training and testing

The bees are left with cotton soaked in sugar water of matching concentration to the feeder orientation that will be used on the test the following day. Forages are identified as bees that left the nest to feed more than four times in 30 minutes and were marked and were trained with the rest of the hive to recognise what the colour of the feeder represented. The marking involved capturing the bees and attaching an Opalith number tag on the dorsal thorax using Loctite Super Glue or by using a coloured Uni POSCA marker pen with varying colours.

Before testing begins, the arena is cleared by individually capturing the bees using a plastic cup and Perspex square and releasing them back to their nest or by luring them by turning off the lights and luring them out of the arena towards the Perspex tunnel using a torch. Once the arena has no more bees, the cotton is removed, and the feeders are filled with either 40% or 10% sugar water based on what the test requires, and the arena and feeders are cleaned with a 70% ethanol solution. The chosen forager would then be required to complete a pretraining to demonstrate she can feed from all four feeders without the cotton and so that she gets a taste of the reward at each feeder. To encourage the chosen forager to feed from all, the feeders the bee finished were not refilled so it would have no choice but to try the others.

During testing, the forager would be let into the arena and would feed from the feeders, and the feeders would be refilled while the forager fed at a different feeder. A bout would be completed when the bee decided to return to the nest and if the bee spent longer than 15 minutes in the arena the forager would be returned to the nest. After 10 bouts the forager would be returned to the nest and the heat pads would take about 45 minutes to heat up before the next stage of testing began. The temperature of both heat pads would be set to 55 degrees Celsius because honeybees have thermal nociceptors that respond to temperatures over 34 degrees (26). Honeybees can solve a place learning paradigm best when measured by time to locate a cool spot when the temperature is 46 to 48 degrees, and anything above that could be lethal if exposed for a prolonged time (27).

With the heat pads installed in the arena, the chosen forager would be let into the arena with the feeders filled and would again have to complete the pretraining to experience all four feeders. Once the pretraining was completed, the forager would again be let out to complete 10 bouts. Heat pad temperature was recorded using an infrared camera (FLIR One Infrared Camera, USA) and an infrared thermometer (CASON CA380 Infrared Thermometer, TMS Europe). The feeder orientation as shown in Fig.2 was assigned ABCD with A being the left of the bees’ entrance and D being the right where the opening flap is. A and C were pink while B and D were yellow making it a ‘pink, yellow, pink, yellow’ orientation for 6 subjects, and the opposite orientation was used for the other 6.

After completing the 10 bouts on the heated stage of testing, the bee was removed from the arena and the area was wiped with ethanol. The heated pads would be taken out and replaced with fake feeders of the same colour and the pipettes in the feeders would all be removed and replaced with pipettes filled with normal water. The bee would once again be let into the arena and left to make its choices for 5 minutes before concluding the test.

Analysis

Our hypothesis is that the proportion of feeding at the heated feeders (i.e. feeders that are heated and contain 40% sucrose solution) will not be significantly different from the proportion of feeding at the test feeders (i.e. feeders that are unheated and contain water).

To analyze the results, a generalized linear model was used on the basis that a constant change in the fixed effects, which in this case is sucrose concentration or temperature, would cause a constant change in the response variable which is the proportion of feeding. Additionally, this model allows us to consider the random effects as well. In this model ‘subject’ nested into temperature was included because each bee experienced all the available temperatures, therefore, there are variations within temperatures that must be caused by the bee, not by the temperature.

The statistical analysis was done using R (R Core Team, Cran-r-project, Vienna, Austria, version 1.3.1093), using generalized linear mixed effect models (GLMMs; packages: “lme4”, “car”.) We identified the most parsimonious models through the ANOVA function and stepwise backwards elimination. As our data were proportions, and thus binary, we used a binomial distribution and logit link function. We checked model assumptions using histograms and Q-Q plots, and considered p < 0.05 significant. One GLMM was fitted, the response variable was the proportion of feeds on the heat-pad feeder; the fixed effects were temperature condition (heated, unheated, test) and sucrose condition (10% or 40%); and the random effects were bee ID nested in temperature. Model simplification removed the location and colour of the heat-pad feeder and bout number.

Results

The videos of the bees' choices were analysed, and the bees' choices were split into five distinctive decisions. The bee could perform a 'feed' which involved landing at the feeder, inserting the proboscis, and performing abdomen pumping for over 3 seconds into the pipette containing the sugar water. A 'taste' was classified as the bee landing at the feeder and inserting the proboscis for less than 3 seconds before moving on. The 'feeding area' choice involved landing at the feeder but not inserting the proboscis into the pipette. An 'alternative feed' involved the bee landing at the feeder, inserting its proboscis into the pipette, and feeding there for 3 seconds while having at least two or more legs in the air. Finally, an 'upside-down feed' was when the bee would not land on the feeder but instead crawl over the back or side of the structure, insert the proboscis into the pipette and feed for more than 3 seconds without touching the floor of the feeder.

The decisions were noted on an Excel spreadsheet after carefully analysing the videos taken from each bout. The testing stage would be analysed based on which feeders the bee landed on in order, regardless of whether it attempted to feed from it. The only decisions used in the analysis were 'feed' behaviours because all others could be considered a rejection of the feeder or could mean feeding occurred without the heat stimulus being present.

The previous study demonstrated that the proportion of feeding events on the 40% feeders decreased as the concentration of sugar water at the other feeders increased (25). Additionally, it was shown that the bees feed at the unheated feeders more than the heated ones when they all contain the same concentration of sugar water.

The results of this experiment showed that the proportion of feeding events at the heated feeders was significantly affected by the sucrose solution concentration available at the alternative feeder (N=12, z= -38.167, p < 0.005) (25). The proportion of feeding events at the heated feeders was significantly lower than at the unheated feeders when the sucrose concentrations at both feeders were equal (N=12, z= 3.280, p<0.005) (25). There was also no significant difference between the proportion of feeding events at the heated feeders and at the test feeders (N=12, z = -0.294).

Fig. 3 shows the proportion of feeding at high-quality feeders that are either heated, unheated, or test feeders containing water against alternative non-heated feeders with either 10% or 40% sugar water available.

Discussion

The results of the study indicate that bumblebees can also learn and use visual and spatial cues to guide their feeding behaviour, as shown by the fact that temperature was the only consistent variable that could differ feeding patterns and that colour and positioning were the only ways to identify it.

The lack of difference between the proportion of feeding at the heated and test feeders demonstrates that colour and positioning were used to guide feeding behaviour rather than temperature or smell. This is because the guidance they have that there is sugar available is based on memories formed on the colour and location. This would confirm the aspect of consciousness described by Klein and Barron in which subjective consciousness that relies on more midbrain-like structures can be described as an awareness of spatial information used in conjunction with a temporal scale of internal physiological functions (22).

Bumblebees are aware of pain since temperature would cause a change in feeding behaviour, and they would learn the associated cues that cause it and choose the option that does not inflict it. This has implications for the handling of insects since they can feel pain and remember the stimulus that caused it; this is typically a feature of the affective component of pain working. In the human brain, pain activates a cascade of events that result in neural plasticity and memory formation (28). The structures involved in the pain pathway for humans are the anterior cortex, amygdala, and hippocampus, which are typically involved in memory formation (28).

The amygdala is a structure responsible for emotions and is highly active in moments of fear and stress and has substantial implications in the emotional experience of pain in mammals and an analogue is present in birds and reptiles (29). When experiencing the pain of the heat-pad, the bumblebee would also trigger a pain cascade resulting in the formation of a memory to avoid the spatial and colour cues associated with the pain. This comparative system by which pain input results in a similar output would implicate a pathway where an unpleasant sensation of pain must have been experienced, suggesting that the bee would feel what humans can relate to as pain. Therefore, a similar structure responsible for unpleasant emotional processing, such as the amygdala, may be activated, which would cause an unpleasant emotional state of consciousness. As such, insects, too, would warrant a similar treatment to mammals, and a reconsideration of certain welfare policies regarding insects should be reconsidered.

As for the project of mass farming insects for food, the implication that insects feel pain may not be as relevant since they can be kept and raised in conditions that would not result in pain, and some animals such as bees are relatively compact in how they usually live (23). Large livestock animals such as cows or pigs are often given less space than they would naturally receive, making non-free-range farming practices unethical. Insects such as bumblebees, however, can be raised in a compact box since they live in small underground spaces in the wild.

Additionally, insects are more environmentally friendly than livestock due to less greenhouse gas emission, space, and food required to farm. Insects can be raised and fed using pig, poultry, cow, fish, and other livestock waste; turning this into biomass requires very low energy input due to the poikilothermic nature of insects, thus producing a very efficient source of food (30). The farming of insects would allow for a greener, more readily available, and affordable protein source that could reduce the large carbon footprint left by the meat industry and promote more ethical farming practices.

The only downside is if one were to consider each insect an individual life, then more deaths are required to produce the same amount of food, so the question goes to what defines the value of life. Defining what gives a life value regarding decisions such as taking a life is a hot topic, especially regarding the topic of abortion of foetuses, who lack sentience but will develop it later if left alone, or comatose patients who may never awaken. Defining sentience could provide an insight into whether another biological creature has a moral implication to be treated as living and could be considered any creature with the ability to feel multiple states of conscious experience defined as feeling or emotion, or will develop into a creature capable of sentience (31).

With regards to sentience, a bumblebee could be considered sentient as it is a conscious being, given that it can identify itself in space and time and can feel different states of emotion, given that painful stimulus triggers a state in which the bee wants to avoid in the future. Furthermore, the bee has been shown to enter states of positive or negative cognitive biases based on whether an unexpected reward or a predatory attack has occurred (11). This state that causes cognitive could be seen as optimistic and pessimistic states and perhaps relate to happy or sad feelings. With multiple emotional states and consciousness, the bumblebee fits this criterion and can be deemed sentient, so should be valued as a life.

However, this argument could also be used to argue that larger mammals such as humans or dolphins, which have been shown to demonstrate self-identification, morality, and other pieces of evidence that demonstrate higher-order thinking, would warrant a different experience of the reality we live in (32). The ability to recognise oneself in the mirror allows the differentiation among self and others just like us and results in the formation of complex self-identities that go far beyond the mind of a bumblebee whose "purpose" as a member of the hive could be considered part of a collective consciousness (32). In this stance, the value of the life of a livestock animal such as a cow could be compared to a whole hive of bumblebees which make a collective consciousness, rather than the life of an individual bee.

Another criterion that could assess the moral implications regarding a life could be the humane functions associated with conscious beings. One such aspect would be that of a moral code in the animal or evidence of displayed empathy. Empathy can be considered a higher-order function, where emotions displayed by others through verbal, body language, or facial expressions can project emotions which are automatically picked up and trigger a pathway in the brain triggering an emotional response (33). Metacognitive processing of the input stimulus can alter the automatic or independent control of the outcome response to the empathised emotion; however, this cognitively processed empathy is variable between individuals and is more of a human function (33).

Some aspects of empathy have been observed in animals such as dogs, rats, and mice demonstrating it is a trait observed in other mammals (33). In these animals, vocal cues produce the most profound empathetic responses because these cues are the easiest to learn. Empathy may have an innately programmed aspect due to the pre-set structures in the brain that fire when empathy can be expressed. However, no genetic finding suggests that associated genes result in the development of empathy and that empathy may be a learned behaviour (33). Such examples of altruistic behaviour would be considered conscious creatures capable of morality.

Invertebrates have been studied to display acts of what can only be described as altruistic behaviour when observing ants (34). Ants, when distressed, have been observed to signal for help, and other ants from the same colony will attempt to help the distressed ant. Some rescue behaviours that the rescuers perform include digging to try and dig out a trapped ant, limb pulling, and biting at snares that have entrapped a fellow ant (34).

Bumblebees are very protective of their hive and will attack and sacrifice their lives to prevent potential threats from invading their nest (35). This protective behaviour includes invaders of the same species but from different colonies and seems to be orchestrated by scent identification (35). These observations would show that the bee is willing to sacrifice its own longevity to protect the food, structure, or lives of fellow hive members, demonstrating an altruistic motivational trade-off decision for its own kind. On observation, these actions would indicate morally made decisions, which like humans, are societal features that are learned and not genetic, suggesting an aspect of consciousness that can be shared with insects.

However, it could be argued that such behaviours, like in the robot argument, are pre-programmed responses to a distress pheromone and do not encapsulate the human perspective of empathy where a shared emotional state with another being is possible (34). In this regard, the animal attempts to help its fellow species out of a programmed desire to preserve its 'self' by aiding workers of the same colony. However, there is no way to confirm how an animal perceives distress signals through chemical pheromones signalling or vocal cries. Furthermore, a lack of empathy can occur in humans through psychological disorders such as psychopathy (36). In these conditions, empathy and guilt are absent, and psychopaths often partake in antisocial behaviours such as violence and manipulation, yet these people can be identified as conscious humans, too (36).

Furthermore, the bee may feel pain and wish to avert it, but this does not necessarily mean inflicting pain has the same moral implications as it would inflicting pain on a human. This is because the unpleasant component of pain felt in a bee may only be used to trigger simple nocifensive actions, such as the observed flapping of wings to cool off after standing on the heat pad, rather than forming traumatic memories which in humans can result in psychological comorbidities such as depression or chronic pain conditions to occur (37). The ethics of pain in animals of higher-order thinking are in place because suffering pain alters the conscious perception of the sufferer's world by forcing them into a state of unpleasant experience which can result in mindsets only observed in such creatures.

There is no evidence that an insect can feel saddened or upset as we know it or form memories of events that cause post-traumatic stress disorder symptoms and anxiety for years to come. Like humans, the bumblebee wishes to avoid pain but for reasons that could be deemed relative to its objective to retrieve food for the hive, not a decision based on emotional well-being.

The bees can make motivational trade-off decisions, as the results demonstrate that when the heat pads contained a higher concentration of sugar water than the non-heated pads, the bees would still choose the heat. Since the bee's decision to feed at the heat pad depended on the sugar concentration at the other feeders, and they would prefer not to feed at the heated feeder, it demonstrates a trade-off between competing motivating stimuli: sugar or heat aversion. The bees' decisions to feed from the chosen feeder were dependent on memory rather than sensory input, which meant the bee had to recall memories of the competing stimuli and use this mental representation to guide its choices, demonstrating the ability to make memory-based motivational trade-offs in invertebrates.

When using results from Gibbons's previous experiment, the proportion of feeds at the hot feeders would decrease as the concentration of sugar available at the alternative feeder increased, with the preference for the non-heated feeders taking over somewhere between sugar concentrations of 30-40% at the other feeders (25). This is reflected in the results where feeding at the high-quality feeder was almost always preferred when the alternative was 10%, even when heated.

This prioritisation of obtaining more sugar, even at the risk of harming oneself, could be explained in the terms that a bee does not feel an affective component of pain. Although they are aware of the suffering that awaits at this feeder, the bee is willing to suffer non-fatal injuries to forage more food, suggesting that food stimulus is a higher priority to the bee than its own longevity. This confirms that the bee's nociceptive response can be considered elastic, centrally processed, and integrates information about competing motivators. The bee can inhibit the nocifensive response and avert the heat when it integrates the information about the large sugar stimulus associated with the feeder, demonstrating a system akin to humans' descending modulation pain pathway.

The descending pain modulation pathway in humans can be triggered by various factors, such as pharmacological intervention, the gate control theory, and deep brain stimulation have been found to activate these pathways in humans (38). This pathway is present in insects, as shown by the Jewel wasps' ability to control the American cockroach with a sting. However, the ability to make motivational trade-offs demonstrates a cognitive ability to overcome the pain stimulus, knowing a more significant stimulus is present. This would suggest a perception of pain that can be understood in human terms.

Another explanation is, as mentioned above, the idea of the hive mind, where the individual's life is only as valuable as its purpose of serving its hive, and therefore, pain should not defer the bee from making a choice that will provide more for its hive. Honeybee foragers have been shown to demonstrate an understanding of the colony's need through 'eagerness' and speed of the food uptake by food receivers and use this information to share additional information about spots that may contain a greater amount of nectar or nectar of higher concentrations (39).

This information is spread to other foragers using 'dance' and is suggestive of the foragers' initiative when understanding individual purpose relative to the status and needs of the hive (39). Since the bee acts with prioritisation of the hive but also understands itself as an individual, it stands to reason that the motivational trade-offs may be based on the individual bee's ideas on what would be most valuable at the given time given the food status of the hive.

The results do demonstrate that overall, bumblebees will choose the heat pads over the alternative 10% sugar, but one bee, 'White', made a different choice. The pretesting and unheated bouts demonstrated an understanding of what was available at the feeders, and White chose the 40% feeders every time. However, when the heat pads turned on, White chose to feed only from the alternative 10% feeders. Statistically, this would be deemed an outlier, but it was no accident that in the remaining bouts and test bouts, the 10% feeder was chosen despite knowing what was available. The reasoning for this could come down to the bee being an individual mind who decided that retrieving less sugar for the hive was not as crucial as self-preservation or that this bee was more sensitive to pain and, therefore, the trade-off placed different values on the competing stimuli.

Limitations

Sensitivity to pain is complex and varies on an individual basis in humans, with implications on psychology, experiences, and underlying medical conditions such as tension headaches or migraines (40). It is reasonable to assume bumblebees may have similar individual differences in their experience of pain, especially given that after careful prolonged observation when working with them, individual temperaments can be observed.

For example, bumblebees may play dead when faced with a threat from a predator or may try to attack it, but the response will vary from bee to bee. In the Perspex tunnel, removing the slips of cards acting as doors would sometimes cause the bees to anger, play dead, or they would simply ignore it. Providing an unexpected pretest reward caused a dopamine increase which in turn increased motivation to forage; however, the inverse can also occur, with rough handling of the tunnel doors acting as a predator-like attack before foraging may affect motivation by setting the forager in a different emotional state (11).

The bees in the experiment would often swarm the doors to the arena, which led to difficulties in letting the chosen bee through. As such, the doors would inevitably close on legs or trap one under a door, but occasionally this would happen on the forager. As such, their motivational decisions may change from having experienced a 'predator attack' just prior. Since the bumblebee forager has been demonstrated to carry a cognitive bias based on its emotional state, the results could be affected by this (10, 11).

Therefore, the bee's response may be due to a change in colony needs. Although it was attempted to be prevented by limiting food intake the previous day, running the experiment with the same food deficit each time is hard. Bumblebees communicate through a mix of chemical pheromones and directly communicated cues such as movement patterns (41). The ability to communicate the hive's needs through various communication methods would imply that the motivation for the foragers to retrieve food may fluctuate daily depending on the hive's food status. This could have impacted the results of the experiments because the hive's status could have influenced the bees' decisions.

Additionally, bumblebees have a diurnal circadian rhythm that affects their motivation to forage. This was demonstrated in an experiment of bumblebees in Finland, where the sun does not set for weeks in the summer, which would disrupt a bee's circadian rhythm based on day-light cycles, which is how the lab bees were kept (42). The experiment was always run in the "day" part of the 12-hour cycle. However, the motivation to forage fluctuates during day hours, too, being low in the early morning and increasing towards midday, where it maximises, then drops again towards evening (42). Therefore, our experiment may have been influenced as some of the experiments were run in the morning and some in the afternoon/evening, where the change in the motivation of the bee could affect the results.

The foragers selected could have been optimised better for the task by choosing larger foragers to run the experiment in the morning and using smaller ones past midday (43). This is because in the early morning when light levels are lowest, the larger bee with more enormous eyes can see better in low light conditions, while the smaller, faster bees can see better in higher light conditions (43). However, this allocation of forager roles may be learned behaviour in the wild and may not affect lab bees who are exposed to a constant level of non-ultraviolet light.

Bees come in different shapes and sizes, and a smaller forager may have a greater surface area-to-volume ratio than a larger bee. Therefore, thermoregulation may be easier for the smaller bee than the larger bee, which could result in the larger bee having to make a trade-off with heat aversion 'weighing' more since the same temperature was used.

The light source itself may not impact this fluctuation in motivation as it may be caused by the changes in the spectral compositions of light, particularly in the ultraviolet range (42). Temperature and light intensity are other external cues postulated to regulate the circadian rhythm, and these were kept constant (42). However, the lighting system themselves often distracted the foragers in the arena. Once they took notice of the lights above, they would continuously attempt to fly into the arena's glass ceiling and break the 'feeding rhythm'. Artificial light sources have been shown to disrupt feeding, pollination, and chemical communication between insects (44). The use of artificial lighting may result in changes in feeding behaviours by distracting them during the task and altering their natural feeding and circadian rhythms (44).

Conclusion

In conclusion, the question, 'do bumblebees feel pain?' has multiple implications regarding the status of invertebrates with regards to animal rights, particularly welfare laws concerned with ethical treatment for experimentation. Additionally, it sheds some light on the consideration of using invertebrates as a more renewable source of farming, which would provide economic, environmental, and safety benefits. Finally, the problematic question of consciousness's origin and nature can be taken a step further by understanding how more superficial brain structures of invertebrates are capable of thought and awareness.

The experiment assessed the bee's ability to seek food using memory of spatial and colour cues, assessed their ability to feel nociceptive pain and link it to the memory of the feeder, and finally made motivational trade-off decisions using two competing stimuli: heat aversion and sugar rewards.

The results showed that the proportion of feeds occurring at the heated feeders and the test feeders containing water demonstrated the bee feeds using spatial and colour cues to guide its decisions, rather than sensory stimuli such as temperature or smell. Bees feed more at feeders with higher sucrose concentrations, demonstrating that feeding is driven by memory based on spatial and colour cues. Additionally, when the larger sugar reward was present at both heated and non-heated feeders, the bee would feed more from the unheated feeder. This demonstrates that pain is felt, and a memory of it is associated with spatial and colour cues. Finally, when a lower sugar reward was offered for the non-heated pads, the bee would choose the heated pad more. This showed that the bee could assess two competing stimuli and make a motivational-trade decision.

This experiment provides data consistent with other research that a bee can feel pain and use it as a deterring stimulus to feed. The implications of these results suggest that bumblebees are sentient creatures capable of a subjective conscious experience but lack the ability to make higher-order thinking decisions and are programmed to serve as part of a collective hive. This opens for discussion as to what extent a life should be valued, based on whether sentient, conscious, or capable of higher-order thinking, and whether invertebrates are more akin to our perception of a conscious experience as we think.