Sentience is a fact. We recognize the sentience in others because of their resemblance to us. But how can we know if other objects very different from us are sentient? How to prioritize limited resources aimed at reducing future suffering if we do not know who is suffering are and how much they suffer?
The situation could be even worse than it seems. Not only we need to know who the suffering beings are and how much they suffer. We also need a better understanding about what sentience is. For example, according to some theories, the binary idea of “sentient beings” apart from everything else (completely separated from other beings that feel, and from things that do not feel) could be wrong. That is, we not only ignore the answers, but we could be asking some wrong questions.
In this document I present a project that I believe can be used to understand and evaluate the different theories and metaphysical hypotheses in relation to sentience, through simulations. In this way we could have a more precise idea of how much sentience there is and where, which would allow us to be more effective in reducing suffering.
This is an example of what I am referring to with “simulations”, in which the basic elements of these simulations are subject to evolution (variability / mutation, copy / recombination, and selection / scarcity):
Clarifications
- This project does not try to demonstrate in particular if our world is a simulation. The simulation hypothesis is just one more of the theories that this project tries to assess.
- This project is not about creating sentience in the simulations (which, if possible, might not be ethical), but creating different simulated worlds in which we simulate that there is a certain way of sentience, to observe its viability and consequences. In many cases it would not even be necessary to run the simulations; it would be enough to design the simulations.
- This project does not try to simulate sentience, but theories about sentience.
Objectives of the project
- Create tools to better understand the different hypotheses, theories, paradigms and worldviews on sentience, as well as their implications, providing clarity in the debate about the details and alternatives within each of the theories.
- Find good arguments to assess the plausibility of the different theories, so that we can have better criteria to allocate more or less resources to investigate or take into account one or the other within a plan to reduce future suffering. In particular, assess theories evolutionarily to check if they are evolutionarily viable.
This should allow real advances in the understanding of sentience, as well as a more adequate distribution of the attention received by the different theories. In this way we can be more effective in reducing suffering, and we will minimize the risk of neglecting some types of suffering.
Project assumptions
- Many people, and among them, many of those who focus their work on reducing suffering, believe in a single theory of sentience, discarding all others. This would be a great position if that theory, with all its details, were correct. But we can not be sure of this, and if the hypothesis is not true, the resources allocated to reduce suffering could be very inefficient and in some cases even ineffective.
- Therefore, I believe that it is justified to take into account more than one theory about the ability to feel and investigate its implications. For example:
- I believe that it is justified to listen to those who say that the suffering of insects does not exist, or if it existed, would be so small that it would not be relevant, but also to those who argue that the suffering of insects is not only real, but also grouped, it is much more relevant than the suffering of the rest of the animals.
- I believe that it is justified to listen to the voices of those who defend the theory that artificial machines such as robots will never be able to feel, but also of those who claim that robots, under certain conditions, could feel as much or even more than human beings.
- I believe that communities (scientific and others), in general, could be incoherent with some of the consequences of the paradigms that they themselves, intuitively, assume to be true; as well as discarding other paradigms perhaps simply because they are difficult to understand.
- Specifically, I believe that a large part of the scientific community assumes an evolutionary emergentist paradigm as true, but rejects mainly, probably by intuition, the idea of sentience in machines, being in our opinion both incompatible ideas. That is to say, the evolutionary emergentist paradigm should be incoherent, at least under certain conditions, with the rejection of the idea of sentience in machines.
- I also believe that a large part of the community without scientific specialization and that does not base the sentience on some religious criteria or of divine nature, assumes as true, at least implicitly, a paradigm based on biological aspects of natural neurons, but rejects the moral consideration of non-human animals even, in some cases, rejects their ability to feel. We think that these simulations can be very useful to provide clarity in the debate on these types of issues.
- I believe that computer simulations, besides serving to solve problems or perform calculations, have an enormous didactic and explanatory capacity, allowing to communicate complex ideas, as is the case of philosophical theories about sentience, and therefore can serve as tools for make progress in this field.
How are we going to do it?
- Identifying different theories or hypotheses about sentience, as well as various aspects of them, whose implications are relevant in order to prevent future suffering.
- Proposing different simulations that can help us understand these different theories and aspects.
- Describing in detail such simulations. In certain cases, this could be sufficient to achieve the objectives pursued.
- In the cases in which it were necessary, programming the simulations and executing them to check their results.
Can you give an example?
I have created an explanatory presentation and a video of 1h duration that explains through concrete examples of concrete simulations, the type of aspects of the theories of sentience that we can represent, understand and value better through these simulations. The first people who have seen it have told me that it is not entirely clear, so I am preparing a second version. I have summarized one of the examples in the following draft:
How to gamify it?
We can consider the use of collaborative and competitive video game environments to make research on sentience fun. For example, we can think of one of these environments:
- Video game “dialectical tournaments”: The passionate debates about metaphysics that take place on internet forums could be moved to an imaginary epic world in which players play against each other or against a GPT-3 machine in philosophical debates, being forced to represent characters who must defend different theories or philosophical frameworks about the nature of reality (including the matter of sentience) such as monism, dualism, emergentism or epiphenomenalism, in an imaginary world in which there is a system of tournaments, rewards and digital coins, in games that can last for months or never end. For example, each time a game is recovered, the video game forces the player to advance dialectically with a specific theory (championed by a specific character), facing other theories defended by other player-characters.
- Video game “configuration of worlds”. In this challenge, the players must define some initial parameters (as in the “Ants and Plants” program) that define the behavior of a certain world, which tries to resemble reality. Depending on the values of these parameters, worlds can be configured that reflect the different theories or philosophical frameworks on the nature of reality (including the matter of sentience) such as monism, dualism, emergentism or epiphenomenalism. Once defined, the “world” is allowed to run, to see what happens. The challenge of the competition is to create coherent worlds, which exist longer than others, since we can suppose, for example, that the world disintegrates if it encounters incoherent rules or if it begins to behave in a way that is incoherent with the world that we consider real.
Do you have a Power Point presentation of the project?
Objections to the project
- It is not necessary to compare different theories about sentience because there is only one correct theory (mine). The others are wrong.
- The same can can be said for those who believe in other theories. This is reminiscent of discussions about religion. The simulations can help explain the different theories and check if they have incoherent aspects. If your theory is correct, it should bear the test of being simulated, and that will make it stronger.
- It is not possible to create sentience in a simulation.
- This project does not try to create sentience in a simulation.
- It is not possible to simulate the sentience because the sentience is not material.
- This project does not try to simulate sentience in particular, but theories about sentience.
- In any case, yes, we can simulate sentience on a screen, just as we can simulate gravity on a screen, even if the pixels do not attract each other. In this project what we pretend to simulate are different possibilities on how the sentience can be. In the analogy with gravity, it’s not about simulating gravity as we know it on planet Earth, but to simulate different theories about gravity to understand them better, communicate them better, and check if they are coherent and compatible with other available evidence.
9 Comments