Many thanks to Kieran Greig for funding me to write this post and for feedback on it. Thanks as well to Jason Schukraft and Daniela Waldhorn for feedback on the post.
I use the term ‘sentience’ synonymously with ‘phenomenal consciousness.’ I use a definition by example – your experience of savouring a good meal or of stubbing your toe are examples of sentient experience. I’m principally concerned with valenced sentience and as a simplifying assumption I write as though sentience implies valenced sentience even though they are probably separable. I don’t have enough space in this post to fully flesh out my position on consciousness, but they are fairly similar to Luke Muehlhauser’s position as articulated in his Report on Consciousness and Moral Patienthood.
Context for this post
I have been researching invertebrate sentience for approximately the past year and I believe this post is relevant to that subject. For more background on the research that I’ve been doing look at this post here.
Thinking about how and why sentience evolved is relevant for finding out which beings may be sentient. It may enable us to say that most or all animals descending from a certain ancestor are sentient. It’s particularly useful for cases we are most uncertain about such as insects and other invertebrates.[i] This can better inform decisions about helping animals. It can also inform us about how contingent sentience may be, which can be relevant in assessing how likely it is that different types of digita or alien minds are sentient.[ii]
One example of how learning about the evolution of sentience may be relevant is the case of octopuses and other cephalopods. There is some strong reasons to think that octopuses may be conscious. But cephalopods are very distantly evolutionarily related to us. They also likely evolved from quite simple animals possibly resembling snails.
If we believe octopuses are sentient this leaves us with two main options: believe that sentience evolved at least twice, or believe that our common ancestor with them was sentient. The former option commits us to position that probably implies that alien life and to a lesser extent digital minds are more likely to be conscious then we otherwise would think. This is because it indicates that consciousness is more easily multiply realizable then we would otherwise think. The latter option implies that at least some simple minds are conscious. Both of these potential conclusions are interesting and important.
Limitations of this post
I'm not an expert on the subject I’m writing about in this post. This post is quite speculative and preliminary. Much of the post is set up, and I don’t reach any firm conclusions. I may follow this post up with another post that gets deeper into the issue.
One factor that calls into question some of the reasoning in this post is some of the newer forms of genetic and molecular analysis on evolutionary history that push back the common ancestors of many taxonomic groups to much earlier than the standard fossil driven narrative would have it. If these are to be believed then that pushes in favour of a conclusion that the Cambrian explosion is not exceptional and so does not need an exceptional explanation, such as being driven by the evolution of sentience.
In keeping with this, another argument that has been put forward is that the Cambrian explosion may appear to be a greater break with evolutionary continuity than it appears to be only because we lack as many fossils from earlier periods. The discovery of the Ediacaran biota fills in this to some extent and supports the claim that the Cambrian explosion is less of a break with previous evolutionary continuity then once believed.
I will mostly be reasoning using the fossil driven model, because that is the one that I know better, and I believe the one that has been best developed in the literature, but be aware that this other evidence exists that calls it into question.
The Cambrian explosion is a mystery of evolution. It occurred in the prehistoric oceans from Persia around 545 to 520 million years ago. It is called an ‘explosion’ because of the vast number of diverse animal body plans that seem to suddenly emerge out of nowhere. Many of these creatures were extremely alien looking and difficult to classify in existing taxonomic groupings. Stephen Jay Gould, an advocate of the position that evolution often progresses through sudden leaps and bounds, has cited it to support his position.
While examining specimens from the Cambrian explosion Simon Conway Morrison is reported to have said “Oh fuck, not another phylum!” More recent analyses have tended to push against the most radical pictures of what the Cambrian explosion was, but it is still generally agreed that something like the Cambrian explosion took place and requires some sort of explanation.
Many diverse factors have been posited as explanations for the Cambrian explosion. Some of these are environmental or ecological factors such as an increase in the level of oxygen enabling different body plans and others are evolutionary innovations such as animal to animal predation or distance senses.
One thing that seems likely to me is that there was a cascade of different factors. I believe this because many of the features that seem to have evolved seem like they would would inevitably spur future evolution and because a cascade of different factors is a more powerful explanation than any single factor. Here’s an example of how this might go: there may have been an increase in the amount of oxygen, which enabled organisms to have more complex and energetically demanding bodies, which enabled us the evolution of distance senses, which enabled predation, which caused selection pressure for defences to predation, and so on.
What is the functional role of sentience
Now in order to speculate on sentience as a factor in the Cambrian explosion we must speculate on the functional role of sentience. The definition of sentience I gave earlier is intended to leave open what it’s functional role is, so this should not be taken as building more into the definition, only as speculation on top of the already given definition.
To think about this question we must think about the functional roles that sentience might play. I’m not claiming that any entity that has approximately these capabilities is therefore sentient. Indeed, I think it is probably the case that all of these may sometimes be carried out unconsciously in humans. Some of the main roles that it seems to me that it plays are:
1) It allows for different types of reinforcement learning. The conscious experience of pain or pleasure respectively causes us to avoid or approach phenomena or actions associated with that experience. Probably some basic level of associative learning occurs in some organisms without sentience, but for at least more complicated forms of learning on the basis of experience, it is handled in humans through us feeling pain and pleasure.
2) It allows us to rank different alternatives and weigh conflicting options against each other to give it an overall ‘utility function’ that allows us to decide on the best course of action. For example, if you decide to eat a pineapple you make that decision on the basis of various pleasures and pains associated with that decision and determined that it is positive on balance.
3) It allows us to respond in the appropriate way to current positive and negative stimuli. For example, that allows us to avoid predators and identify and eat good food. Some of this process is handled through reflexes in us, which appear to be unconscious, but most of the process, especially those parts involving (2), are handled consciously in us.
4) It facilitates us having different conscious emotions. Different moods or emotions are experienced consciously in us. These states have different respective functional roles. The conscious experience of the emotion seems to facilitate the different behaviour that different mood states tend to lead us to.
5) In addition, non-affective sentience plays the role of allowing the organism to sense objects and have a spatial model of the world. This may be handled unconsciously in the case of blindsight, so it may not necessarily be a conscious process, though it is typically handled consciously in humans.
Of course, it could be the case that there is some additional complexity on top of these factors that needs to be added before an organism that can do these things can rightfully be called ‘sentient.’ This seems likely to me, though I think as the organism comes to look increasingly like it is sentient, the likelihood that it is missing whichever of these factors that would make it ‘truly sentient’ decreases.
I think that we should also resist the position that there is some special single factor that when added to an unconscious organism suddenly makes it conscious. This is because I think the arguments for particular proposed features for this such as mirror self-recognition are poor and because a theory such as this might be imagining something similar to a Cartesian theatre picture of consciousness where some ‘magic spark’ is required.
Did sentience help drive the Cambrian explosion?
If this analysis is correct, sentience appears to be particularly useful for active motile organisms that need to avoid predators. This ties in with some of my previously mentioned proposals of what factors drove the Cambrian explosion. For this reason, it seems to me like sentience may at least be part of the cascade of factors that drove the Cambrian explosion.
It also seems to be the case that some of the animals that evolved in the Cambrian explosion are quite plausibly sentient. For example, Anomalocaris seems to have been the apex predator of the time and was a metre long swimming arthropod with flexible ‘arms’ near its mouth and eyes approximately as powerful as those of dragonflies. This suggests that at least in the perhaps exceptional case of Anomalocaris, sentience did evolve in the Cambrian explosion or an earlier era.
Another Cambrian animal is Haikouichthys, an early fish. Some authors have recently argued to be perhaps the first sentient animal. Some of the reasons they give for this are it’s adequate image forming eyes an additional senses that may have led to more complex neural hierarchies. It’s difficult to imagine active motile animals with distance senses such as Haikouichthys that had to cope with predators such as Anomalocaris not having something that at least resembles sentience. I have written another post here where I give more than reasons why I expect large active animals such as anomalocaris to be sentient.
However, much of the diversification that represents the explosion may have happened in animals that were simple enough that are more likely than not non-sentient. Animal such as Anomalocaris and Haikouichthys may be sentient, but most other animals, including the ancestors of these animals, are more simple and less likely to be conscious.
However, even if they didn’t themselves represent much of the diversity, arthropods have complex active bodies and may have driven diversification in other groups. Other authors have suggested that predation by arthropods may have been significant in driving the explosion.
It's difficult to tell what role sentience would've played in the Cambrian explosion. It could've been a pivotal factor. Or it could have contributed relatively little and have been mostly a byproduct. It's even more difficult to say if it would've been sentience or some sort of proto-sentience with many of the same functional properties that I have suggested, but somehow not quite there to fully count as sentience and so not be morally relevant.
Simona Ginsberg and Eva Jablonka 2010 argue that the evolution of associative learning may have driven the Cambrian explosion. Associative learning may be related to sentience in certain ways, but at least I'd simplest forms it does not seem to require sentience. If they are right then perhaps sentience played relatively little role in explosion and was merely 'along for the ride’ because associative learning would have probably evolved earlier than sentience.[iii]
[i] It is particularly helpful with invertebrate consciousness because invertebrate consciousness is a less tractable and straightforward issue than vertebrate consciousness, and so any tool that can help with the question is
[ii] One of the principle ways that studying the evolution of sentience can help us with questions about which digital beings are sentient is by moving us towards or away from a consciousness is rare versus consciousness is common conclusion. In brief, if consciousness evolves multiple times that indicates that it is more multiple realizable than we might otherwise have expected.
[iii] One oversight in the Ginsburg paper is that they seem not to be aware of evidence of associative learning in plants and protists. If protists do show associative learning, this suggests that associative learning evolved much earlier and so could not explain the Cambrian explosion. Of course, it could be that it was a mechanism that allowed for more complex associative learning then we see in protists that evolved later on and drove the Cambrian explosion. Sentience could also have allowed for this higher level of associative learning, but this is pure speculation.