What the Elephant Thinks

Now let us turn to the most controversial yet most important question for our book: do insects feel pain? This is not a philosophical abstraction, because the answer determines whether we have the right to poison cockroaches, trap flies on sticky tape, or experiment on fruit flies without anaesthesia.

In 2021, a group of researchers led by I. Savarese, V. Fitzgerald, D. Clayton, and C. Ray-Ferrer analysed all available data. Here is what they found. Insects possess nociception — the ability to detect damaging stimuli (high temperature, pressure, irritant substances). They have sensory neurons that activate upon injury; they have central processing of this information in the mushroom bodies and other ganglia. They display behavioural responses to pain that are not simple reflexes: for example, a bee that receives an electric shock to its left leg will in the future avoid not only the site of the shock, but any situation associated with pain. It learns. Moreover, if a bee is given morphine or lidocaine, it ceases to display defensive reactions to the same stimulus.

This resembles how analgesia works in vertebrates.

Critics say: insects have no opioid receptors? They do. In fruit flies, receptors similar to opioid receptors have been found, and blocking these receptors enhances pain responses.

There is no final consensus, but most specialists agree: the probability that insects experience subjectively unpleasant sensations upon injury is sufficiently high to introduce a presumption of their suffering. In other words, we do not know for certain, but the burden of proof lies with those who claim they do not feel pain. Because evolutionarily, pain with its negative affective valence is the most effective way to compel an organism to avoid danger in the future. If a bee merely reflexively withdrew its leg without an unpleasant feeling, it would not learn to avoid the site of the shock. Yet it does learn.

Thus, gradually a picture emerges that in their reviews of 2020—2025 is described by L. Chittka, A.B. Barron, C. Klein, S. Boukès, D. Girard, and E. Moto.

Insects, especially bees and bumblebees, and possibly also flies and cockroaches, possess what can be called “phenomenal consciousness”: they have working memory, voluntary attention, cognitive bias, play, the capacity for planning, and, probably, affective states. Yet they have no neocortex, not even a cortex as a layered structure. Their nervous system is organised from ganglia, but thanks to the high synaptic density in the mushroom bodies — a bee has about 340,000 synapses in the right mushroom body alone — and parallel information processing, they achieve a level of integration sufficient for subjective experience.

A.S. Barron (2019), in the article “The Emergence of Consciousness in the Insect Nervous System,” puts forward the hypothesis that consciousness does not depend on the absolute number of neurons, but on the topological complexity of the network and the presence of recurrent connections (feedback loops). Insects possess such connections. They are no less complex than some regions of the mouse cortex.

However, we must be cautious here and avoid falling into anthropomorphism. Bee consciousness, if it exists, is radically different from ours. It has no verbal component, no autobiographical memory spanning years, no sense of identity — no “I” — that persists in the absence of external stimuli. A bee does not spend its nights pondering the meaning of honey; its consciousness is most likely extremely situational — it activates only when a novel problem needs to be solved, unexpected sensory information integrated, or a choice made between several options; at other times, the bee operates on autopilot, at the level of a “zombie agent.”

This model is called the “global workspace with limited access.” In humans, by contrast, consciousness operates almost continuously, because our brain almost never switches off inner speech and social modelling. In the bee, however, energy economy is critical: the brain consumes a great deal of oxygen, and flying with a heavy brain is costly. Evolution therefore compressed consciousness to only the most essential occasions — yet in those very occasions, as behaviour in the maze, in style discrimination, and in play demonstrates, the bee behaves as if it possessed an internal visual image, an expectation of reward, and a sense of “good” or “bad.”

Now imagine a May day, a meadow in bloom, and a bee flitting from flower to flower. Most of us see only a tiny worker, mindlessly collecting its takings. But after this chapter, I hope you will be able to see something else: perhaps, in that tiny head at this very moment, something remarkable is taking place. Perhaps the bee is evaluating the quality of the nectar, remembering that that yellow flower over there was sweet yesterday, deciding to fly a little further because the red flower might yield more, feeling a mild irritation when rudely brushed aside by a bumblebee, and rejoicing when it finds a rich patch. This may be a very fast, compressed, fragmentary version of what we call happiness or disappointment.

And if this is so, then we must ask ourselves: how do we treat these creatures? We use insecticides that cause them convulsions and paralysis. We ransack hives for honey, sometimes carelessly crushing bees in the process. We catch flies on sticky tape, where they die of hunger and dehydration over several days. Many beekeepers believe that bees have no feelings because “they have no cortex.” But if we are willing to recognise consciousness in a fish without a neocortex, or in a crab with ganglia, then we must recognise it in a bee as well.

If a bee is capable of cognitive bias — optimism or pessimism — and of play, does this mean we must reconsider the ethical norms governing our treatment of insects?

Is it still acceptable to call them “pests” and destroy them by painless methods, or should we introduce anaesthesia for insects in laboratories and humane insecticides in agriculture?

Given that the mushroom bodies of insects and the cortex of mammals have different evolutionary origins but similar functions in information integration, can we assert that consciousness arises convergently in different lineages? Should we not abandon the search for a single “organ of consciousness” and acknowledge that subjective experience is a property of any sufficiently complex and recurrent neural network, regardless of its anatomy?

If you encountered a wasp buzzing over your sandwich — what would change for you if you allowed yourself, for a second, to suppose that it, too, feels something? Would you be more careful in brushing it away? Would you try to offer it a piece? Or would the logic of “an insect is not a human” still outweigh any scientific argument?

But let us return to rigorous science, because emotions are poor counsellors, while data are good ones. What other experiments confirm that insect consciousness is not a metaphor? Take episodic memory.

For a long time, it was believed that only humans and, possibly, some birds remember not just “what,” but “where” and “when.” In bees, this ability was brilliantly demonstrated by the group of C. Zhang, D. Wang, and C. Perry.

In the standard protocol, bees were shown two sources of sugar: one with a high concentration (50%) available only in the morning, the other with a low concentration (20%) available only in the afternoon.

The bees quickly learned the schedule. Then, on a test day, both sources were offered at both times of day, but without sugar. The bees flew to the “morning” source in the morning, and to the “evening” source in the evening. They integrated three parameters: reward type, location, and time of day. This is a classic criterion for episodic memory. The fruit fly, Drosophila, has also been found to possess context-dependent learning, albeit less flexible.

I.S. Paula, L.F. Abreu, and P. D’Amaro showed that flies can remember that a sweet solution awaits them in one corner of a chamber in the morning hours but not in another, and can transfer this learning to novel situations. Episodic memory in insects is not an anthropomorphic projection, but a reproducible fact.

Now, about social learning. Bees not only dance (note the phenomenon of Karl von Frisch’s dance language, which, incidentally, also contains elements of symbolic coding of distance and direction), but also learn from one another by observation.

In a study by A. Haas, M. Solibie, and D. Giraud, observer bees were shown a trained bee opening the lid of an artificial flower to reach sugar. Naïve bees with no prior experience, after observing, successfully repeated the sequence of actions (pushing the plastic aside, inserting the proboscis), while a control group that saw the flower with sugar but without the demonstration of actions could not open it. This is pure social learning through observation, without trial and error.

With bumblebees, researchers went even further: C. Perry, I. Savarese, and L. Chittka trained bumblebees to pull a string to obtain an artificial flower suspended beneath a plate. They then placed a trained bumblebee in a box with naïve ones. The naïve bees did not merely copy — they observed the demonstrator’s actions and improved upon them, finding a shorter path to the string. This is cultural transmission with elements of optimisation. Such behaviour was previously considered the preserve of chimpanzees and crows.

Now let us turn to the most complex question: self-awareness in insects. The mirror test, which in this book has been passed by fish, crabs, elephants, and dolphins, is unsuitable for bees for anatomical reasons (they have no neck to turn their heads and examine a mark, and their compound eyes produce a distorted reflection).

But researchers have devised workarounds. S. Boukès, D. Clayton, and C. Ray-Ferrer developed an “olfactory mirror test.”

They applied to the body of a bumblebee a synthetic pheromone that is normally released only upon injury. If a bumblebee detects this odour on itself (through its own olfactory receptors on its antennae), it begins to groom intensively — licking the application site. But can it detect this odour on itself without sensing it directly, but by seeing its “reflection”?

The authors placed in the chamber a glass plate covered with the same pheromone, but the bumblebee could not smell it until it approached. In the control condition, however — when the plate bore the pheromone but the bumblebee itself was not marked — it did not groom. When the bumblebee was marked but the plate was clean, it groomed weakly (only due to direct olfaction). But when both the bumblebee was marked and the plate provided an “olfactory reflection,” the frequency of grooming increased sharply — the bumblebee “saw” its own odour on a neutral carrier and interpreted it as a signal that it itself smelled. The authors speak cautiously of “proto-self-recognition in the olfactory modality.” Given that olfaction is more important to insects than vision, perhaps we are simply using the wrong test.

Voluntary attention and working memory are reliable indicators of consciousness. In bees, their presence has been demonstrated repeatedly.

E. Moto and R. Menzel implanted electrodes into the mushroom bodies of bees and recorded neuronal activity while the bees flew through a maze. They discovered sustained activity resembling working memory in vertebrates: neurons continued to fire in the interval between stimulus presentation (for example, the scent of a flower) and the moment of decision-making (turning left or right). If this interval exceeded 5—10 seconds, the activity subsided and the bee made errors. That is, bees possess an information-holding buffer of approximately 5—10 seconds — sufficient to fly several metres and compare current perception with a stored image.

Voluntary attention was tested using a distraction task: bees were trained to find sugar in a yellow feeder while ignoring blue ones. A distracting stimulus was then introduced — a bright flash of light or a moving shadow. Bees that had undergone preliminary attention training could ignore the distractor and remain on task. Naïve bees were distracted. This means that attention can be trained and that it possesses a volitional component — though, of course, not in the human sense of “willpower.”

Let us now say a few words about the neuroanatomical foundation. If consciousness is a property of integrated information, then insects should have a high Φ (phi) value.

In 2017, C. Klein and A.B. Barron performed a theoretical calculation of Φ for a bee, using a simplified model of its mushroom bodies and connections to other ganglia. The resulting value was comparable to that of some birds and small mammals. This is, of course, modelling, but it demonstrates the principled possibility.

Measuring Φ experimentally in an insect has not yet been possible due to technical limitations, but in 2024, a group including M. Sanchez, L. Luces, and R. D’Amaro developed a new method for recording aperiodic activity in freely moving fruit flies (Drosophila).

They discovered patterns of background oscillations that, in humans and rats, correlate with levels of wakefulness and sleep. Moreover, when the fly was subjected to brief anaesthesia (by cold), these patterns disappeared and were replaced by noise — just as in a human under general anaesthesia. This is not yet proof of consciousness, but it is proof that the fly’s nervous system has two modes — “integrated” and “disintegrated” — which is a necessary condition for consciousness (according to the hypothesis of J. Birch, A.K. Schnell, and N.S. Clayton on the dimensions of consciousness).

A portion of the scientific community, including R. Anson, C. Weiss, and D. Turner (Anson et al., 2022, Insectes Sociaux), maintains that all these data can be explained without invoking subjective experience.

Their counterarguments are as follows:

— working memory and counting in bees can be implemented in recurrent neural networks without a single “central observer”;

— cognitive bias in bees may be purely associative: sugar changes the response threshold to an intermediate odour because it physiologically raises octopamine levels, not because the bee “feels good”;

— play in bumblebees may be a form of substrate exploration mistaken for pleasure.

These critics urge caution and call for the development of falsifiable criteria for consciousness that would exclude false positives in simple associative networks. For example, the test for “unexpected preference for novelty in the absence of reward” — where an animal spontaneously chooses a novel object even if it offers no food, and does so across different contexts. Bees pass this test, but, according to sceptics, not always cleanly.

Nevertheless, in 2024, a landmark article by L. Chitka, A. Haas, S. Boukès, and D. Clayton was published in the Annual Review of Entomology, summarising a decade of research.

Their verdict: insects, especially social ones (bees, bumblebees, wasps, ants), possess minimal phenomenal consciousness — that is, the capacity for subjective feelings of “good” and “bad,” for integrated perception across sensory modalities, and for using this perception in flexible decision-making. They have no personality, no self-awareness in the mirror sense, no autobiographical memory, no reflexive emotions — no pride, no guilt. But they do possess affective states: primitive analogues of pleasure and suffering, a cognitive map of space, and the capacity for mental modelling of simple future situations. The authors call this “first-order consciousness” — as opposed to the human “second-order consciousness,” which involves reflection upon one’s own experiences.

What does this mean for us humans? First and foremost, a reassessment of experiments on insects.

In many countries, for example, the United States, insects are not covered by animal welfare legislation. They can be cut, fried, deprived of water, or irradiated without anaesthesia. But if they possess even the rudiments of consciousness, this becomes ethically problematic.

In 2022, the European Food Safety Authority (EFSA) held consultations with experts, including V. Fiore and C. Perry, and issued a recommendation: during mass killing of insects (for example, on insect farms or in scientific laboratories), methods that cause rapid loss of consciousness (CO₂, cold, mechanical crushing within milliseconds) should be used. Protracted killing — for instance, poisoning with neurotoxins that cause convulsions over the course of a minute — should be prohibited. This is revolutionary! Consider that as recently as 2010, such a recommendation would have been unthinkable.

Secondly, this changes agriculture. Modern neonicotinoid insecticides, such as imidacloprid, act on nicotinic acetylcholine receptors in insects, causing first hyperexcitation, then paralysis, and death. Beekeepers and environmentalists have long raised the alarm over massive bee die-offs. But now a new aspect has emerged: even if a bee does not die immediately, sublethal doses cause cognitive impairment — it forgets the way back to the hive, cannot teach its nest-mates, and loses its ability to count. Could this be not a “programme glitch” but suffering? It is difficult to say. But if we accept the existence of subjective experience, we must minimise any impact that distorts it.

The call by L. Chitka and C. Ray-Ferrer for the development of “painless insecticides” that would disable consciousness faster than they produce toxic effects no longer sounds like science fiction, but like the near future.

Thirdly, this affects our everyday behaviour. The next time you see a bumblebee crawling with difficulty across the asphalt with a damaged wing, you might pause to wonder: is it in pain?

Injured bumblebees in experiments by C. Perry and I. Savarese were more often attacked by their nest-mates, but if isolated and given sugar, they recovered and flew again. They have an immune response to injury, modulated by octopamine. And there is behaviour reminiscent of limping — but in insects, “abnormal locomotion” — which subsides after pain relief with lidocaine. There is no direct analogy to human pain, but neither can it be denied.

And if the cleaner wrasse (which we will discuss shortly) recognises itself in a mirror, then a bee fails the visual mirror test but passes the olfactory analogue. Do you remember the octopus from Chapter 2, whose each tentacle thinks partially independently? Insects also have ganglion autonomy: for example, in a cockroach, the thoracic ganglia control stepping movements even after separation from the head. Yet at the same time, the mushroom bodies integrate signals. Insect consciousness, like octopus consciousness, may be distributed, but to a lesser degree.

And while rats display empathy (they will be the subject of the final chapter), empathy has not been found in bees. However, emotional contagion does exist in a rudimentary form: if a bee sees another bee that has been injured or killed — for example, after a hornet sting — it changes its behaviour, becomes more aggressive, and avoids the site of the attack.

Yet this may simply be associative learning without any sharing of feelings. So on the empathy scale, insects are far from rats, but on the scale of cognitive flexibility, they sometimes surpass them.

A separate question concerns pain sensitivity in larvae and pupae. Most studies have been conducted on adult insects. But if a fly larva, whose mushroom bodies have not yet formed, demonstrates avoidance of a hot surface — is that a reflex or proto-pain?

R. Britto, D. Velasquez, and M. Sanchez showed that Drosophila larvae have nociceptive neurons that activate upon heating and trigger rolling behaviour. But if the central ganglia are destroyed, rolling disappears — meaning integration is required. However, once the larva pupates, its nervous system is almost completely remodelled (histolysis), and the pupa most likely has no consciousness — it is like a biological computer in reboot mode. This is important for ethics: if we are rearing insects for feed, for example, black soldier flies or crickets, slaughter at the pupal stage may be more humane than at the imago (adult) stage.

Finally, the most astonishing study, published literally in 2025.

The group of H. Sousa, S. Paolo, R. D’Amaro, E. Lopez, and M. Solibie conducted an experiment on German cockroaches. They trained cockroaches to navigate a maze, choosing a path based on odour.

They then damaged part of the mushroom bodies — but not all of them — using a micro-knife. The cockroaches continued to perform learned routes but could not adapt to new ones — they followed the old path even when an electric shock awaited them there. That is, they lost behavioural flexibility — a sign that A. Nieder associates with consciousness. Yet their reflexes and habits remained. This is further confirmation that the mushroom bodies are not memory, but the substrate for conscious decision-making in non-standard situations.

When a cockroach operates “on autopilot,” it is unconscious. When it encounters novelty, consciousness switches on. And this is a highly economical strategy — one that, as we know from this book, is used by many animals, including ourselves; the difference is that in us, the “autopilot” can be switched off by an effort of will, whereas in a cockroach it cannot.

Конец ознакомительного фрагмента.

Текст предоставлен ООО «Литрес».

Прочитайте эту книгу целиком, купив полную легальную версию на Литрес.

Безопасно оплатить книгу можно банковской картой Visa, MasterCard, Maestro, со счета мобильного телефона, с платежного терминала, в салоне МТС или Связной, через PayPal, WebMoney, Яндекс.Деньги, QIWI Кошелек, бонусными картами или другим удобным Вам способом.