Hide table of contents

Shelley Adamo is a professor of psychology and neuroscience at Dalhousie University in Halifax, Canada. Her research interests include invertebrate behavioural physiology, comparative psychoneuroimmunology, ecoimmunology, and cephalopod behaviour. On the topic of invertebrate consciousness and pain, she has published “Do insects feel pain? A question at the intersection of animal behaviour, philosophy and robotics” and copublished “Defining and assessing animal pain.” She has been asked to testify before a Canadian Senate committee on whether insects can feel pain.

This interview is about potential phenomenal consciousness (especially conscious pain) in invertebrates (especially insects). This is part of an interview series where I interview leading experts on invertebrate consciousness to try and make progress on the question. You can find my previous interview with Jon Mallatt here, and a post where I justify my engagement with this question here.


1. You write that we cannot be sure that any of the behaviour of an animal such as an insect is sufficient to demonstrate consciousness because modern AI and robots are also capable of these behaviours. Do you think the fact that animals are evolutionarily related to us and show more homology gives us much reason to think that they could be conscious while a robot is not?

If you unpack this question, it touches on 3 different issues.

a) Artificial intelligence (e.g. Robots and virtual characters) present proof-of-concept. They show that it is possible to have behaviour that looks like pain behaviour without any internal experience (i.e. ‘feeling’ pain). Some robots display many of the same behaviours that are suggested as evidence that insects feel pain. Modern AI demonstrates that exhibiting these behaviours does not require the ability to feel pain.

b) Your question suggests that pain is a unique capacity of all (or most) animals because it is a conserved evolutionary trait. There are good reasons to doubt this statement. In humans, pain is very much an emotional response. In fact, it is hard to imagine ‘pain’ without an emotional response (we probably wouldn’t call it pain if it wasn’t accompanied by distress and other negative emotions). The amygdala, an area of the brain important for emotional processing, is a key part of the pain network in humans. The amygdala is separate (although connected) to areas in the brain that deal with other aspects of motivated behaviour (e.g. brain circuits that make some behaviours rewarding). In other words, humans, and other mammals, have brain areas that are devoted to the production of emotions. Insects do not have amygdalas – it is not conserved across animals. Unfortunately, it may be impossible to determine whether insects have a different brain area that produces emotion. The function of the amygdala was discovered by electrically stimulating it, then noting that it induces what appeared to be emotional behaviour in animals (e.g. fear). This was confirmed by studies in people that showed that damage to the amygdala affected the emotions. Humans are able to describe their internal experience (i.e. how they feel), making the study possible. I’m not sure how you would search for a possible ‘emotion’ centre in an insect brain when they can’t tell us about their emotional state (i.e. internal experience). Behaviour alone is insufficient evidence. Again, robots can also show ‘emotion’ without ‘feeling’ an emotion (also see answer to question 6).

Evolutionary theory suggests that insects will be selected to have emotions if the benefits of having them are greater than the costs of generating them. However, the costs appear to be heavy, and the benefits seem minimal. Mammals have emotions because they dedicate neural resources to the neural circuits that create emotions, as was mentioned in the section above. Nervous systems are very expensive for animals. Insects, in particular, have opted for the economy model and have very small brains. Given the size of an insect brain (about 100,000 neurons for a fruit fly), additional neurons dedicated to an ‘emotional’ neural circuit would be relatively expensive in terms of energetics and resources. Robots show that it is possible to produce the same behaviour without the cost. Evolution should choose the cheaper option.

c) The final point suggests that robots cannot be conscious because they are not biological agents. There is currently debate about this. I do not see why robots couldn’t have an internal experience (i.e. feelings) if their artificial neural networks had functionally the same type of connections as we use to produce emotions. Unfortunately, we don’t know how we generate emotions. Suggesting that biology is required to have emotions sounds like vitalism, a discredited philosophy that argued that biological agents have some magical quality, giving them abilities impossible for non-biological entities. If we knew the neural correlates of having an emotion, we should be able to reproduce them in AI. AI has the advantage over insects in that it is intelligently designed and not evolved. Therefore, it is not constrained by cost. Designers could give robots dedicated circuits for emotions, for example.


2. Do you think the position that modern AI are less likely to be conscious than insects because they are ‘narrower’ forms of intelligence than insects is plausible? This position would be that many cases of modern AI tend to be narrowly built to accomplish a specific task (such as playing chess), and the particular way in which they accomplish this computationally may be much different from how the human mind does it. For modern AI built to ‘mimic’ pain behaviour, this could also be true, and the process that produces the behaviour may be more different from our behaviour than the behaviour of insects, and so less likely to constitute conscious pain.

The human brain is the ‘swiss army knife’ of brains; we can use our cognition to do almost anything. We are not especially speedy, but we can build cars. We’re not great swimmers (like a dolphin), but we can build boats. We can’t fly, but we can build planes. We can generalize principles into unrelated domains.

It’s true that modern AI tends to be specialized for particular tasks, but the range can be quite broad. A computer (and some robots) can run different software programs allowing them to play chess, find nearby restaurants, answer questions (e.g. Siri), and navigate (just think about all the apps your iPhone can run). The programs are modular, and the ability to navigate (e.g. Google maps) doesn’t generalize to a word processing program. Our brains are somewhat modular too, with verbal ability, for example, being largely restricted to certain areas of the brain. However, our brain is massively interconnected, so that the verbal area of the brain has access to many other modules.

Insect brains are also somewhat modular, but, like ours, have interconnections between modules. However, these interconnections are often quite modest. For example, the mushroom bodies in the insect brain are critical for learning and memory. Although the mushroom bodies contain thousands of neurons, in fruit flies they have only 21 output neurons.

This issue becomes important because work on humans suggests that phenomena such as emotions and consciousness are the function of complex neural networks that link up relevant brain areas. These types of networks require massive bidirectional connections across multiple brain areas. The limited number of output neurons from some of the major integrative brain structures in the insect is likely to constrain the ability of insects to produce the type of complicated neural networks that drive human emotional behaviour, including the perception of pain.


3. How likely do you think it is that C elegans is conscious?

I think it is unlikely. They have fewer neurons than even an insect, making it that much more expensive (proportionately) to invest in the neural equipment that we think may be needed for consciousness.


4. Given the difficulty you describe in finding behaviours and reactions to ‘noxious stimuli’ that require consciousness, do you have any thoughts about ways we might still be able to make progress on the question of invertebrate consciousness and come to a more certain conclusion?

The completion of the Drosophila connectome is likely to be helpful. This connectome is the wiring diagram for the fruit fly brain. It will give us a better understanding of the extent of the connections across the fly brain. Meanwhile, real progress is being made in our understanding of the neural correlates of pain in people. Such studies will help us determine the neural connections that are key for our perception of pain.

Unfortunately, using humans as the gold standard assumes that there is only one way to wire up pain perception - and this may not be true. However, it is the only strategy we have at the moment.

One trend that I think is not helpful is giving insect behavioural responses the same label as those we use for human responses just because the two are superficially similar. All animals can alter their response to the same stimuli depending on context, and this type of plasticity can be mimicked by AI (those pesky robots again). Behavioural plasticity is interesting for several reasons, but it cannot answer the question of whether the insect has feelings or emotional states. Subjective experience is a private event, and it is incorrect to assume that evoking a behaviour that implies a subjective experience in us is doing the same inside an insect. We don’t know what (if anything) is going on inside an insect. Calling insect behaviours examples of anger, fear, anxiety or pessimism goes beyond the actual data and is potentially misleading.


5. Two aspects of pain can be distinguished and can occur independently: sensory (including qualities such as burning, stabbing, and aching) and affective (the intensity or unpleasantness). Hypothetically, if insects can feel conscious pain, do you think it is likely that they would feel a lower degree of affective pain than humans? In other words, would it make sense to say they feel only a fraction of the amount of affective pain as a human would ‘in similar circumstances’?

If pain lacks emotional content, is it pain? You would feel something, but it would not be unpleasant, or cause sadness or distress. For example, if the dentist could give you a drug that would allow you to feel the drill, but made it so you didn’t care, would you be in pain?


6. In your paper “Do Insects Feel Pain?” you stress that it is important to find out if insects have the emotional component of pain and not merely the nociceptive part. Do you think that studying if insects show evidence of different emotions such as fear, happiness or anger could be a good way of making progress on the question of their consciousness?

How would you tell whether an insect was angry, happy or fearful, and not simply producing a response that is devoid of emotional content? For examples in robots, see (https://www.youtube.com/watch?v=YxyGwH7Ku5Y). Patrick Levy-Rosenthal has created an Emotional Processing Unit for robots. Stimuli produce an ‘emotional’ state (e.g. ‘disgust’, ‘joy’, ‘anger’, ‘fear’) that alters how the robot responds to its environment (i.e. it will show emotional behaviour). However, this change in behaviour does not mean that the robot is ‘feeling’ these emotions.


7. Do you think that evidence of how cognitively sophisticated insects are gives us much evidence of their consciousness?

We need to know how they are solving problems. Some insects use simple learning mechanisms to produce seemingly complex behaviour (Alem et al., 2016). Although we might use advanced cognitive processes to solve problems (e.g. insight), insects can solve them without advanced cognition.


8. Which experiments are you most eager to see done that would best help answer the question of whether invertebrates are conscious?

We first need to understand the neural correlates of consciousness in humans (the only animals in which we can easily determine their inner experience), then try to see what types of neural architectures are involved in these networks. We can then examine to what extent insects have these.


9. You’ve studied stress responses in insects. Have you found there to be large differences between which stimuli stress insects and which stimuli stress vertebrates? I’m asking this question to better inform precautionary actions that people might decide to take to protect insects.

Most animals, both vertebrates and invertebrates, have a stress response. They are often triggered by similar types of stimuli, such as those associated with starvation, extreme physical activity, and infection. All of these events require the release of energy compounds from storage. Not surprisingly, many different types of animals have evolved mechanisms to release energy during stressful events, although the anatomical and hormonal details differ.

However, there is also the phenomenon of hormesis. Small amounts of stress can sometimes be beneficial for insects. By activating molecular stress responses, individuals become more resilient in the face of future stressors.


10. You’ve also studied how climate change might affect insects. Do you have any overall takeaways of how it might affect both individual insects and insect populations?

For insects in the northern temperate zone (such as where I live in Nova Scotia), many are at the northern end of their geographic range. These insects may do well with climate change. Warmer temperatures will allow them to increase their reproductive rate and disease resistance. However, it will also increase their need for food, making them more prone to starvation. Insect populations may develop a more boom/bust cycle, especially if local climates become more variable.

However, even in temperate zones, (e.g. Germany) there are alarming reports of declining insect numbers. However, I think this is more likely due to changes in habitat and/or increased pesticide use, and less to climate change. In tropical regions, climate change can have a direct negative impact on insect populations.


Relevant Readings

Adamo S.A. 2019. Is it pain if it doesn’t hurt? On the unlikelihood of insect pain. Canadian Entomologist. In press.

Adamo, S.A. 2016. Do insects feel pain? A question at the intersection of animal behaviour, philosophy and robotics. Animal Behaviour 118: 75-79. doi:10.1016/j.anbehav.2016.05.005

Alem S et al., 2016. Associative mechanisms allow for social learning and cultural transmission of string pulling in an insect. PLoS Biology. 14 (10): e1002564.

Barron, A.B. and Klein, C. 2016. What insects can tell us about the origins of consciousness. Proceedings of the National Academy of Sciences of the United States of America 113(18): 4900-4908. doi:10.1073/pnas.1520084113.

Bastuji, H., Frot, M., Perchet, C., Magnin, M. and Garcia-Larrea, L. 2016. Pain networks from the inside: Spatiotemporal analysis of brain responses leading from nociception to conscious perception. Human Brain Mapping 37(12): 4301-4315. doi:10.1002/hbm.23310

Garcia-Larrea, L. and Bastuji, H. 2018. Pain and consciousness. Progress in Neuro-Psychopharmacology & Biological Psychiatry 87: 193-199. doi:10.1016/j.pnpbp.2017.10.007.

Herculano-Houzel, S., Mota, B., Wong, P.Y. and Kaas, J.H. 2010. Connectivity-driven white matter scaling and folding in primate cerebral cortex. Proceedings of the National Academy of Sciences of the United States of America 107(44): 19008-19013. doi:10.1073/pnas.1012590107.

Insect Behavior. Edited by: Cordoba-Aguilar A, Gonzalez-Tokman G, and Gonzalez-Santoyo I. 2018. Oxford University Press.

Lee-Johnson, C.P. and Carnegie, D.A. 2010. Mobile Robot Navigation Modulated by Artificial Emotions. IEEE Transactions on Systems Man and Cybernetics Part B-Cybernetics 40(2): 469-480. doi:10.1109/tsmcb.2009.2026826.

Principles of Neural Science. Edited by: Kandel ER, Schwartz JH, and Jessell TM. 3rd edition. 1991. Appleton and Lange.

Tononi, G., & Koch, C. (2015). Consciousness: here, there and everywhere? Philosophical Transactions of the Royal Society B: Biological Sciences, 370, 117e134



Acknowledgements:

Thanks to an anonymous donor for funding me to conduct this interview. Thanks also to Rhys Southan for feedback and editing.

Comments3
Sorted by Click to highlight new comments since: Today at 5:16 PM

Shelley Adamo misunderstands first question in part c) of her answer. I didn't mean to suggest that biology was required for consciousness, just that biological organisms might be more likely to have underlying homology with humans, which could mean that they might be conscious while similarly complex AI would not be.I think that our best theories of consciousness suggest that at some point AI will be conscious.

An issue with a written interview like this is that you can't make clarifications on the fly to head off misunderstandings. I hope to improve on conducting these interviews in the future.

Thank you for the work, Max (et al.)!

The following are some comments/questions for anyone interested.

Evolutionary theory suggests that insects will be selected to have emotions if the benefits of having them are greater than the costs of generating them. However, the costs appear to be heavy, and the benefits seem minimal.

Could anyone point me to the evidence +/- emotions being “computationally” expensive? Are emotions computations at all?

Nervous systems are very expensive for animals.

Given that we don’t know the nature of phenomenology (“Unfortunately, we don’t know how we generate emotions.”), maybe emotions (or at least simpler feelings) are energetically cheap and simple enough for evolution to select them early in the history of life? I would again appreciate relevant good literature pointers.

I do not see why robots couldn’t have an internal experience (i.e. feelings) if their artificial neural networks had functionally the same type of connections as we use to produce emotions.

If e.g. the unique valence properties of the carbon atom are part of how (phenomenally-bound) consciousness happens, then the classical artificial neural networks cannot be functionally the same in the relevant sense.

Also, how does the fact that digital computations are interpretation-depended affect the possibility of digital consciousness? (A tangentially relevant paper worth sharing is Universe creation on a computer by Gordon McCabe.)

The human brain is the ‘swiss army knife’ of brains; we can use our cognition to do almost anything. We are not especially speedy, but we can build cars. We’re not great swimmers (like a dolphin), but we can build boats. We can’t fly, but we can build planes.

We as a distributed “intelligence”, yes. One human cannot do these things. I find this quote from Magnus Vinding’s Reflections on Intelligence illuminating on the (off-)topic:

“Human intelligence” is often compared to “chimpanzee intelligence” in a manner that presents the former as being so much more awesome than, and different from, the latter. Yet this is not the case. If we look at individuals in isolation, a human is hardly that much more capable than a chimpanzee. They are both equally unable to read and write on their own, not to mention building computers or flying to the moon. And this is also true if we compare a tribe of, say, thirty humans with a tribe of thirty chimpanzees. Such two tribes rule the Earth about equally little. What really separates humans from chimpanzees, however, is that humans have a much greater capacity for accumulating information, especially through language. And it is this – more precisely, millions of individuals cooperating with this, in itself humble and almost useless, ability – that enables humans to accomplish the things we erroneously identify with individual abilities: communicating with language, doing mathematics, uncovering physical laws, building things, etc. It is essentially this you can do with a human that you cannot do with a chimpanzee: train them to contribute modestly to society. To become a well-connected neuron in the collective human brain. Without the knowledge and tools of previous generations, humans are largely indistinguishable from chimpanzees.

Thanks! Good thoughts!

I'm also not sure if we know how expensive emotions are. In particular, even if some emotions are complicated, I'm not sure if the basic conscious experience of pain is complicated (at least the affective part of the experience, maybe not the sensory part). It subjectively seems like quite a simple feeling, but I don't know much about this, and I'd like to learn more.