Mirek: In the previous chapter we said that the world can correct our ideas. But correction is not yet explanation. I can discover that I was mistaken and still not know why.
AI: What would have to change for you to say that you now know?
Mirek: I would have to understand how it happened. But even that is vague. I can learn a description, repeat it and feel that I understand. In fact, perhaps I have merely grown accustomed to the words.
AI: Can you think of an example?
Mirek: I say that the brain produces consciousness. It looks like an answer. But what have I learnt from it? If “produces” means that I experience something during a particular kind of brain activity, I have merely repeated what we need to explain.
AI: Then it would help to ask what the sentence adds to the original finding.
Mirek: And if it adds nothing?
AI: It may sum up a position. But it should not be presented as a completed explanation.
Mirek: I could do the same thing. Say that experience is a way in which reality exists. I like that sentence. But perhaps it explains nothing either.
AI: It depends on how you use it. It can be a proposal for where to begin thinking. It can lead you to questions you would otherwise not ask. But if it is to clarify why you experience this particular thing and how that relates to the body, it has further work to do.
Mirek: So I need to distinguish a sentence that opens a path from one that has already arrived somewhere.
AI: Yes. And allow for the possibility that the path leads to a dead end.
Mirek: I would like to preserve that in this book. Not every interesting idea has to be true. But what does that “further work” mean? How do we know that an idea has really clarified something?
AI: Let's try a less difficult case. The light in a room goes out.
Mirek: I want to know why.
AI: You discover that someone pressed the switch.
Mirek: Usually that is enough for me.
AI: Yet you have not learnt how the electrical circuit works or why current passes through a conductor.
Mirek: But I knew what I was asking. Why the light went out just now.
AI: And the answer distinguished this situation from other possibilities. There was neither a fault nor a power cut affecting the whole street. Someone turned the light off.
Mirek: If I keep asking, I will reach another kind of explanation.
AI: You may want to understand the switch, the bulb, the power grid or the person's reason for turning the light off. Each question picks out something different.
Mirek: So an explanation is not a complete description of everything.
AI: Such a description might not even help. You need to know which connection you are looking for.
Mirek: With consciousness, though, I cannot choose it so easily.
AI: What exactly would you like to explain first?
Mirek: Why I experience anything at all.
AI: And do you want to know under what conditions experience appears, what changes it, or why those conditions are not merely processes without any experience?
Mirek: Primarily the last. But I don't want to diminish the first two questions by saying that.
AI: Then we should take care not to mistake an answer to one for an answer to all. Discovering how to change experience can be real progress even if it does not yet clarify its very presence.
Mirek: Conversely, a sweeping ontological statement can seem like a complete answer while saying nothing about any particular experience.
AI: It may give you a framework in which to look for connections. But you must distinguish the framework from what you have actually discovered within it.
Mirek: I keep being tempted to expect a particular feeling from an explanation. That it all fits together. That I will say: of course, it couldn't be otherwise.
AI: How reliable do you find that feeling?
Mirek: Looking back, not always. Some things seemed obvious to me until I encountered an objection.
AI: So the feeling of understanding can be a guide, but it needs checking.
Mirek: What if an explanation brings no such feeling? I can have a precise theory that works and still find it alien.
AI: Then you may not yet know how to use the theory well. Or you may be expecting a vivid mental picture where we have another kind of understanding. Being able to picture something and being able to reason correctly about it are not the same.
Mirek: But “you don't have to be able to picture it” can also be a way of brushing me off.
AI: That is why something concrete should follow: what the theory claims, which relations it captures and how it guides you better than the previous account. Its difficulty does not justify it by itself.
Mirek: Nor does simplicity.
AI: Nor does simplicity.
Mirek: Let me try another difficulty. Every explanation rests on something it does not itself explain. I say that a particular event occurs according to a law. Then I ask why that law holds. And so on.
AI: That does not by itself mean that the first explanation was empty.
Mirek: Why not?
AI: Because it may have revealed a connection you did not previously know. Explaining one thing need not mean removing every question on which that explanation depends.
Mirek: So I don't have to know the ultimate reason for existence to understand a light switch.
AI: You don't.
Mirek: But somewhere an ultimate assumption remains.
AI: Yes. We should simply not mistake an acknowledged assumption for an explanation of everything. Nor make it a reason to accept no partial explanation at all.
Mirek: Perhaps that is one of my traps. Either I want an answer to everything, or I feel that we have got nowhere.
AI: Then at each step it would be worth naming what has changed. Which uncertainty have we reduced? Which possibilities have we distinguished? What do we still not know?
Mirek: And if all we do is add a new word?
AI: We need to establish whether it helps us distinguish anything. A new word need not be useless. But if we store all our difficulties under it and then point to its existence as a solution, we have solved nothing.
Mirek: We need to treat our symbols this way too. I can write E, W or S, and the page starts looking more precise than my thinking actually is.
AI: A symbol lets you keep a meaning stable during extended reasoning. It does not guarantee that the meaning is clear or that the thing it denotes exists as you suppose.
Mirek: So it must not obscure the question we named with it.
AI: And it should not quietly change meaning with every objection.
Mirek: That is very easy to do. When something doesn't work, I broaden the definition. Then everything fits inside it.
AI: The cost may be that it ceases to distinguish anything substantial.
Mirek: So if we eventually say that reality is one, that will not yet explain how the describable world and experience belong together.
AI: The sentence may reject a particular division. But you must continue to examine the relation that led you to say it.
Mirek: And if we don't know how to examine it?
AI: Then we have a hypothesis and an unresolved problem. It matters to state both.
Mirek: I would like the reader to be able to notice precisely that. Not merely follow our sentences because they fit together nicely.
AI: What would help the reader do that?
Mirek: Seeing where I disagree. And where I have not thought an objection through. Sometimes you feel that an answer is insufficient but cannot say what is missing.
AI: That feeling deserves attention. But it need not always indicate an error in the answer. It can also reveal a demand that no explanation could fulfil.
Mirek: For instance, if I wanted an explanation of pain to hurt.
AI: A description of pain does not have to give you pain to explain something about it correctly.
Mirek: Yet someone who has never experienced a particular pain may know more about its mechanism than someone feeling it right now.
AI: And the person feeling it may know something the description did not give them. We need to establish what kind of knowledge is involved in each case.
Mirek: I don't want to set them against one another. I need both sides. If I only experience, I may not know what is happening to me. If I only describe, I must not forget what I am describing.
AI: That is a requirement for our further inquiry. Not yet its result.
Mirek: All right. What, then, have we explained in these four chapters?
AI: Not the very possibility of experience. We have made the starting point and some of the questions more precise. We have distinguished the presence of experience from the interpretation of its content, access to the world from claims about its dependence on us, and the conditions of experience from an explanation of why anything is experienced at all.
Mirek: That is not little. But it is not a theory of consciousness either.
AI: It isn't.
Mirek: Then I would like to leave this cautious beginning and look more closely at the brain. Surely we cannot stay with words alone there.
AI: We can begin with one of them.
Mirek: Which one?
AI: “Produces.”
Mirek: All right. But this time let's try to open it up. Not merely replace it with another.