WEBVTT
00:00:05.440 --> 00:00:06.719
Welcome to Pydactic.
00:00:06.799 --> 00:00:07.120
I am Dr.
00:00:07.200 --> 00:00:16.079
O'Leary, a child and adolescent psychiatrist, creator, and host of the Psychactic Podcast since about December 2021.
00:00:16.559 --> 00:00:20.320
Today is Friday, September 4th, 2026.
00:00:21.120 --> 00:00:33.200
I started the podcast as a psychiatry resident to help me study for my boards, but since then it's evolved into a more broad exploration of psychiatry, psychology, and neuroscience.
00:00:33.840 --> 00:00:38.000
The topic I'm going to deal with today is a bit slippery.
00:00:38.399 --> 00:00:45.920
Rarely do conversations, discussions, or debates begin with the phrase we need to undefine our terms.
00:00:46.640 --> 00:00:49.119
Isn't that counterproductive?
00:00:49.920 --> 00:00:50.880
It can be.
00:00:51.520 --> 00:01:00.719
I'm going to argue that discussions of consciousness are more productive if each person is willing to let go of what they are sure consciousness is.
00:01:00.880 --> 00:01:04.400
And I hope this episode helps you to do that.
00:01:06.239 --> 00:01:12.319
If I said that a fan moves air, but it doesn't blow it, does that make any sense to you?
00:01:12.879 --> 00:01:18.959
I would need somehow to define how the fan is moving the air without actually blowing the air.
00:01:19.519 --> 00:01:22.640
That's what many discussions of consciousness look like.
00:01:23.519 --> 00:01:36.640
In many therapies, the operational definition that the therapist is using for consciousness is probably something like any thought that you have access to and can manipulate in your working memory currently.
00:01:37.439 --> 00:01:46.000
The unconscious thoughts and feelings and memories are things that you cannot bring to your current state of awareness.
00:01:47.120 --> 00:01:47.760
Hmm.
00:01:48.640 --> 00:01:55.359
Some doctors use consciousness to refer to wakefulness and being able to be aware of your surroundings.
00:01:55.519 --> 00:01:59.760
However, many people state that they are conscious when they're dreaming.
00:02:00.079 --> 00:02:05.280
They know the content is a dream, and they are able to act in the dream willfully.
00:02:05.439 --> 00:02:11.439
Then, when they're awake a few minutes later, they can tell you all about the adventure they were having just a moment before.
00:02:11.759 --> 00:02:14.639
Were they not conscious during their dream?
00:02:15.199 --> 00:02:17.919
They certainly were not wakeful.
00:02:18.400 --> 00:02:22.560
And the things that they were aware of was not the outside world.
00:02:23.280 --> 00:02:28.719
If I said something could perceive light but was not conscious of it, would that make sense to you?
00:02:29.039 --> 00:02:32.560
Human consciousness is the reason we invented the word conscious.
00:02:32.879 --> 00:02:35.520
But how do we even know that we are conscious?
00:02:35.759 --> 00:02:40.960
How can we know when something that could be conscious is currently unconscious?
00:02:41.280 --> 00:02:45.759
Why do we usually assume that a tree is not conscious?
00:02:46.479 --> 00:02:58.639
One thing that has plagued discussions about consciousness, and one of the reasons I've avoided spending too much time on it before, is that we have no universally accepted or precisely useful definition of consciousness.
00:02:58.800 --> 00:03:00.639
We just have breadcrumbs.
00:03:01.280 --> 00:03:04.240
We have innumerable workable definitions.
00:03:04.400 --> 00:03:14.800
For example, if I say I was conscious of the fact that I was violating the law, it implies that I had some awareness that the actions that I was undertaking were illegal.
00:03:15.039 --> 00:03:22.719
It implies that I took these actions even though I knew they were unlawful and had a choice whether or not to take them.
00:03:23.120 --> 00:03:24.560
But that's not what I said, is it?
00:03:24.639 --> 00:03:27.120
I said I was conscious that I was breaking the law.
00:03:27.280 --> 00:03:31.120
I did not say I broke the law willingly or intentionally.
00:03:32.080 --> 00:03:36.879
That's just a taste of how complicated discussions of consciousness can easily become.
00:03:37.199 --> 00:03:41.599
In fact, rarely do discussions of consciousness remain even about consciousness.
00:03:41.759 --> 00:03:54.080
Because consciousness is variously defined in basically any context, many discussions frequently degenerate into discussions about what I just did, intentions, agency, free will, culpability for actions.
00:03:55.439 --> 00:04:03.919
Consciousness is often used as a prerequisite for free will, or sometimes even as equivalent to free will.
00:04:04.560 --> 00:04:08.319
And neither of these terms has very good definitions.
00:04:09.520 --> 00:04:17.519
That's because words in general and all of our language constructs are highly degenerate.
00:04:17.839 --> 00:04:22.639
Now that doesn't mean that they are morally corrupt.
00:04:22.800 --> 00:04:30.639
It means that many words can intend the same thing, and any one of them can intend many different things.
00:04:32.079 --> 00:04:51.680
So let me start then not by discussing consciousness directly, but instead let's take a detour into language and how it helps us to understand or to know or intend or be conscious of different things, and how messy knowing or defining something can really be.
00:04:52.079 --> 00:04:55.040
I'll start by discussing the word science.
00:04:55.839 --> 00:04:58.000
Science is a much abused term.
00:04:58.079 --> 00:05:03.120
It sometimes means technology, sometimes it just means knowledge in general.
00:05:03.360 --> 00:05:07.360
Sometimes it's equated with materialism or even atheism.
00:05:07.680 --> 00:05:14.879
It sometimes means the ultimate understanding or completely settled knowledge, the kind that can never be argued with.
00:05:16.000 --> 00:05:23.199
All of these definitions or implications are incomplete or wrong in some way.
00:05:23.439 --> 00:05:27.839
Now I'm not going to try to nail down a super precise definition of science.
00:05:28.000 --> 00:05:36.160
Instead, I'll just define science broadly, for the sake of argument, as more or less a messy process of collective learning.
00:05:36.560 --> 00:05:42.959
Science is what it is like to come to an agreement on what must be true probably.
00:05:44.480 --> 00:05:50.959
Science is how we think we can best come to know things, not as individuals, but as a collective.
00:05:52.160 --> 00:05:54.959
It's the process of acquiring and sharing knowledge.
00:05:55.199 --> 00:06:00.800
It allows us to have certain kinds of knowledge that no one of us alone could have.
00:06:01.519 --> 00:06:04.879
It makes knowledge part of a shared experience.
00:06:05.680 --> 00:06:12.720
If you wanted to get a little esoteric, you could even say science is part of our shared consciousness.
00:06:13.920 --> 00:06:15.519
But let's not do that yet.
00:06:16.160 --> 00:06:18.000
Basic science does a few things.
00:06:18.240 --> 00:06:22.480
The first is to gather increasingly precise and reliable information.
00:06:22.800 --> 00:06:29.360
What this means is that the information science gathers is reproducible and it's useful.
00:06:30.000 --> 00:06:35.600
Most of us could gather the same information and we could agree that it has the same meaning.
00:06:35.839 --> 00:06:39.519
So to be precise information, it has to be reliable.
00:06:39.680 --> 00:06:48.160
It can't be precise if it's unreliable, if everyone's looking at the same thing and saying it must be something else that's not very precise.
00:06:48.399 --> 00:06:51.040
We have to be talking about roughly the same thing.
00:06:51.360 --> 00:06:53.279
Precision allows us to do that.
00:06:53.519 --> 00:07:03.920
We can agree on a small number of useful things that can possibly be true or must necessarily be untrue.
00:07:04.639 --> 00:07:10.319
The level of precision needed to support our knowledge depends on the claim that is being made.
00:07:10.480 --> 00:07:15.279
For example, the term fruit in general is rather imprecise.
00:07:15.519 --> 00:07:21.120
It refers to all sorts of things like apples, eggplants, the fruit of our loins, the fruits of our labors.
00:07:21.519 --> 00:07:24.879
What does it mean to claim that something is a fruit?
00:07:25.199 --> 00:07:34.399
Botanists would use the term very strictly to refer to something that develops from a flower after it has been fertilized, but most people are not that precise.
00:07:34.639 --> 00:07:42.720
In fact, I've had people insist that tomatoes are vegetables and not fruits because of how they're used in cooking.
00:07:43.199 --> 00:07:49.519
Tomatoes are culinary vegetables, even though they're a type of botanical fruit, called a berry.
00:07:49.759 --> 00:07:52.160
Discussions about this can get very heated.
00:07:52.480 --> 00:07:53.279
What about you?
00:07:53.519 --> 00:07:57.519
Do you think a tomato is a berry or a vegetable?
00:07:57.920 --> 00:08:03.519
If you're one of those people who immediately thought both, you're probably on to something.
00:08:05.040 --> 00:08:18.240
Besides defining terms precisely and gathering reliable information, another thing that science does is propose and rule out various things that appear to us to be precise and reliable, but maybe aren't.
00:08:18.480 --> 00:08:30.160
And this is very useful to rule things out because the more explanations that we can rule out, the more accurate we can consider the rest of the remaining explanations to be.
00:08:30.319 --> 00:08:35.200
They're more likely to be accurate because these other things are not the explanation.
00:08:35.440 --> 00:08:40.080
And we'll get to that kind of reasoning in a little bit more depth soon.
00:08:40.720 --> 00:08:57.840
Precise and reliable measurements made it possible for Galileo to make the more correct and accurate argument that the sun does not, in fact, revolve around the earth, even though the less precise measurements of our eyes and our common sense made it seem obvious that it must.
00:08:58.080 --> 00:09:01.519
We can see that the sun goes around the earth every single day.
00:09:01.759 --> 00:09:03.200
Everyone can agree on that.
00:09:03.360 --> 00:09:04.879
That's super reliable.
00:09:05.120 --> 00:09:10.559
So we might also forgive those who at first rejected the evidence that their eyes had given them.
00:09:10.879 --> 00:09:19.600
We thought our eyes were extremely precise instruments, but it turns out they can be tricked rather easily by the relative motion of objects.
00:09:19.919 --> 00:09:28.559
So we need more objective sources of information, like telescopes and mathematics.
00:09:28.879 --> 00:09:39.440
To do all of this ruling out of less likely explanations, science borrows logical operations from its philosophical counterparts.
00:09:39.840 --> 00:09:50.799
Logic can construct precise and reliable tests that determine whether any variably precise and reliable explanation is possible or plausible.
00:09:51.360 --> 00:09:55.600
To be possible means not to break any of the laws of physics that we've all agreed on.
00:09:55.840 --> 00:10:00.320
We could add a few more things, but I'll stop with the laws of physics for this discussion.
00:10:00.639 --> 00:10:07.120
But just because something is possible, because the laws of physics allow it, doesn't mean that it's true.
00:10:07.519 --> 00:10:14.799
Likewise, because a prediction is highly precise does not mean that it is possible.
00:10:15.279 --> 00:10:19.600
We can have very highly precise predictions that cannot possibly be true.
00:10:19.840 --> 00:10:31.840
For example, we could predict with high mathematical precision that a closed system will tend toward increasing order at a rate that is proportional to the number of particles in that system, and we can make a mathematical system to show that.
00:10:32.159 --> 00:10:40.320
But no one should be currently testing that hypothesis, because we know it's not possible as soon as it's stated.
00:10:40.480 --> 00:10:44.240
It violates the second law of thermodynamics.
00:10:44.559 --> 00:10:54.799
In the past, we would have needed to test this prediction to see if it were true, but because we've already tested similar predictions exhaustively, we no longer need to test this one.
00:10:55.200 --> 00:10:59.360
The general consensus of society rejects it.
00:10:59.600 --> 00:11:17.919
Even if there are a small percentage of flat earthers and other people like that out there and people inventing their perpetual motion machines, we can still rule it out because there's always gonna be like a small percentage of people in the world who are just gonna claim the impossible is possible.
00:11:18.080 --> 00:11:25.919
The more skeptical of us can just rule out impossible explanations because of our collective understanding.
00:11:26.879 --> 00:11:38.080
So I just talked about something being possible, so now I'm gonna switch to the word plausible to mean that something is possible and also makes sense given what we already know.
00:11:38.320 --> 00:11:43.519
So it doesn't just break the laws of physics, but it actually seems to be probably true.
00:11:43.759 --> 00:11:47.200
It's both precise and it's very possible.
00:11:47.519 --> 00:11:53.679
So sometimes implausible things prove to be correct, things that no one thought could be true.
00:11:53.919 --> 00:11:55.440
This is very, very rare.
00:11:55.600 --> 00:11:58.799
Most implausible things do not prove true or correct.
00:11:58.960 --> 00:12:04.960
In fact, most of what we consider even plausible explanations do not prove to be true or correct.
00:12:05.200 --> 00:12:15.279
Most of the hypotheses that people make regarding the causes of things, oh for example, I know my neighbor is mad at me because he scowled at me this morning, are never actually tested.
00:12:15.519 --> 00:12:19.360
We just take them to be true because I told myself that it was.
00:12:19.600 --> 00:12:21.840
It's plausible that my neighbor was mad.
00:12:22.000 --> 00:12:30.000
It certainly did not violate the laws of physics, but it also could have been any number of things that I haven't considered, including just resting neighbor face.
00:12:30.399 --> 00:12:32.720
My prediction is very imprecise.
00:12:32.879 --> 00:12:36.960
I should consider my prediction that he's mad at me to be very imprecise.
00:12:37.200 --> 00:12:43.600
Although if I'm feeling bad about myself, I'll probably think of it as a very precise definition.
00:12:44.000 --> 00:12:46.320
There's not many other things it could be.
00:12:47.840 --> 00:12:57.919
Because most of our possible and plausible explanations are not necessarily true, we rely on logical tests to help us to increase or decrease the probability that these things are true.
00:12:58.240 --> 00:13:01.679
Logic is intended to increase precision.
00:13:02.399 --> 00:13:06.320
There are many logical processes that contribute to how we know things.
00:13:06.559 --> 00:13:18.799
Some of the words we use for logical ways of knowing things that you've probably heard before are like deduction, induction, and one you maybe haven't heard before is abduction or inference.
00:13:19.519 --> 00:13:26.480
So deduction is what people claim that Sherlock Holmes is doing, although generally he's not.
00:13:26.720 --> 00:13:33.919
What deduction really is, is it proposes that something must be the consequence of something that was caused.
00:13:34.080 --> 00:13:37.840
So if you have a cause, you then know the consequence.
00:13:38.000 --> 00:13:39.519
If A, then B.
00:13:39.840 --> 00:13:44.399
A causes, or at the very least, always results in B.
00:13:45.600 --> 00:13:53.039
If the relationship between A and B is very precise, then if I know A, B always follows.
00:13:53.120 --> 00:13:55.840
It's the only thing that follows.
00:13:56.159 --> 00:13:57.360
Induction is different.
00:13:57.600 --> 00:14:02.320
Induction, if precise, assumes that something must be the cause.
00:14:02.480 --> 00:14:08.159
Well, because I know the consequence, because I know something happened, I can induce the cause.
00:14:08.320 --> 00:14:12.159
If B happened, then I know A must have been the cause.
00:14:12.399 --> 00:14:22.720
The problem with deduction is that A might cause more things than B, so knowing A doesn't necessarily mean that I know, and the problem with induction is that B might be caused by many more things than just A.
00:14:22.960 --> 00:14:32.000
Even if A always causes B, B could still be caused by many more things than A, so knowing B doesn't really mean that I can infer A unless it's the only possible cause.
00:14:32.159 --> 00:14:34.480
And it's really hard to prove that.
00:14:34.720 --> 00:14:38.559
That's why what we usually do is abduction.
00:14:38.879 --> 00:14:44.639
And this is what Sherlock Holmes was doing most of the time, especially in the movies about him now.
00:14:45.200 --> 00:14:53.519
Abduction is a process where we propose a cause and then accept as the best answer whatever the evidence appears most precisely to support.
00:14:53.919 --> 00:14:56.559
It seriously considers uncertainty.
00:14:56.799 --> 00:14:58.960
How uncertain is the evidence?
00:14:59.440 --> 00:15:06.080
How imprecise are the possible deductions and inductions that could result from this evidence?
00:15:06.399 --> 00:15:17.200
Abduction is the most realistic way that we can even possibly approach the world because we don't have all the information we need to be able to deduce everything or induce everything.
00:15:17.440 --> 00:15:20.799
We're kind of left with abduction as a default.
00:15:21.360 --> 00:15:24.320
We're guessing about what's real most of the time.
00:15:24.559 --> 00:15:30.799
When we're abducting, we usually assign a level of assumed precision to the assumptions that we're using.
00:15:30.960 --> 00:15:35.600
For example, I could say, I'm absolutely certain that I left my keys by the front door.
00:15:36.080 --> 00:15:39.600
My assumption is I left my keys by the front door because I don't have them.
00:15:39.919 --> 00:15:48.240
But if I'm absolutely certain, if my prediction is that precise, then the only place I would ever look for my keys would be the front door.
00:15:49.279 --> 00:15:55.519
If I accept that there is uncertainty, then I'll also look somewhere else.
00:15:55.759 --> 00:16:08.879
So if I accept the uncertainty, then I can say the precision of my prediction is somewhat sure instead of absolutely certain, or I kinda think, which would be imprecise, but it's a place to start.
00:16:09.039 --> 00:16:10.399
I'll look at the door first.
00:16:10.559 --> 00:16:21.600
But if not there, I won't assume the keys must be there, but are in fact invisible, or that they were there but jumped into some other parallel universe.
00:16:21.759 --> 00:16:22.720
I'm not going to assume that.
00:16:22.799 --> 00:16:26.000
I'm going to assume that I was wrong about where I thought they were.
00:16:26.559 --> 00:16:37.120
Another relatable precise belief is that when I'm looking for my glasses, they must be somewhere I was, instead of where I am on my head.
00:16:37.360 --> 00:16:42.559
And I could spend a long time looking for something, because this the precision of my belief is high.
00:16:42.879 --> 00:16:47.519
My ability to consider where they most likely were did not have realistic precision.
00:16:48.799 --> 00:16:54.480
So let me take a break from going any deeper there and summarize what I was kind of trying to get at all along.
00:16:54.639 --> 00:17:02.799
It's that science, as I have defined it for the sake of discussion, is the process of us coming to some sort of agreement about what is most likely true.
00:17:03.039 --> 00:17:11.279
To be more accurate, we also must increase our precision, our ability to agree upon what is right and true.
00:17:11.599 --> 00:17:24.480
This means that we must increase the level of certainty about the evidence that we have that rules out other competing explanations, and that includes all relevant explanations.
00:17:24.799 --> 00:17:27.119
We rarely actually do this.
00:17:27.599 --> 00:17:33.519
Instead, we often incorrectly assign high certainty to evidence that is not as informative as we think it is.
00:17:33.680 --> 00:17:36.880
I'm certain my son must have been the one who left the door open.
00:17:37.200 --> 00:17:40.720
Oops, it was me when I was bringing in the groceries.
00:17:41.039 --> 00:17:53.359
Plato was certain that all matter takes its form because of the existence of some immaterial ideal and perfect form that then manifests that thing, which is just an imperfect shadow of it.
00:17:53.519 --> 00:18:02.799
In Plato's time, this was a very convincing and considered precise explanation, and there wasn't that much competition to rule it out.
00:18:03.039 --> 00:18:05.759
But we know now that it's highly unlikely.
00:18:06.000 --> 00:18:09.920
Although as a metaphor, eh maybe.
00:18:11.519 --> 00:18:21.839
While we can be overly certain, we can have that false assurance of our own precision and therefore be inaccurate, we can also be not certain enough, and therefore not accurate.
00:18:22.079 --> 00:18:30.000
So, for example, if we always assign low certainty to evidence that is in fact true and informative, then we may not ever be accurate.
00:18:30.240 --> 00:18:38.720
We might in effect assume that the evidence is wrong, because it's not does not agree with my assumptions about what must be true.
00:18:39.039 --> 00:18:42.240
Galileo must be mistaken.
00:18:42.480 --> 00:18:45.039
Or there's some trick here in the lens.
00:18:45.200 --> 00:18:56.319
What all this means is that science, while increasing the precision of our measurements and applying logical rules, makes it possible for us to know things that we would otherwise not have the precision to know accurately.
00:18:56.640 --> 00:19:02.400
Increased precision is necessary, but it's not sufficient for increasing the accuracy of the knowledge that we have.
00:19:02.640 --> 00:19:05.359
We have to continually test our assumptions.
00:19:05.599 --> 00:19:09.279
Our precision must be accurate and its accuracy must be precise.
00:19:09.519 --> 00:19:11.519
It's a recursive process.
00:19:11.759 --> 00:19:15.200
Coming to know things is a recursive process.
00:19:15.440 --> 00:19:19.759
And the testing of these assumptions is abduction.
00:19:21.440 --> 00:19:31.519
So now that I've bored you and discussed the word science and lots of other words, the point of it was in part to show how dynamic the content of a single word can be.
00:19:32.160 --> 00:19:37.599
So now let's progress a little closer to understanding what we mean when we say consciousness.
00:19:37.839 --> 00:19:41.920
But instead of talking about consciousness, again, I'm going to go back to talking about words.
00:19:42.400 --> 00:19:48.880
When we invent words to describe anything, those words have some level of meaning, and also some level of precision.
00:19:49.119 --> 00:19:51.279
Words theoretically can be very specific.
00:19:51.519 --> 00:19:58.240
To be specific or precise means that most or all people will interpret the word in a very similar way, within a small margin of difference.
00:19:58.480 --> 00:20:03.039
Collective knowledge defines what we accept as truth.
00:20:03.599 --> 00:20:05.759
But no words are perfectly specific.
00:20:05.920 --> 00:20:11.599
If I say the word table, there will be a multitude of examples of different kinds of tables in people's minds.
00:20:11.839 --> 00:20:18.559
Most of them could contain some kind of surface that's supported by one or more legs, or pedestals.
00:20:18.799 --> 00:20:24.640
A lawmaker, though, might interpret table as an action that is done to a bill under consideration.
00:20:25.119 --> 00:20:34.319
To table a proposed law means that it will be removed from the current discussion and available for consideration later, or just gather dust.
00:20:34.640 --> 00:20:39.440
To table means roughly to give up the current course and move on to something else.
00:20:39.680 --> 00:20:41.279
We will table that proposition.
00:20:41.519 --> 00:20:47.039
We can imagine an infinity of tables where these proposed laws go to gather cosmic dust.
00:20:47.279 --> 00:20:49.680
And most words are like the word table.
00:20:49.759 --> 00:20:58.000
They have multiple and even infinite possible meanings if we think of all the crazy context that they could be used in or for.
00:20:58.160 --> 00:21:04.000
In everyday life, though, it's not useful to consider all the possible meanings a word might have when it's used.
00:21:04.240 --> 00:21:13.359
When someone says meet at the house on Friday, I would likely assume that they meant the most proximate Friday, this next coming Friday on the calendar.
00:21:13.519 --> 00:21:19.119
I might also assume that they mean their house and not someone else's, because context matters.
00:21:19.359 --> 00:21:25.599
It determines what I will assume given a set of relatively imprecise, unspecific instructions.
00:21:25.839 --> 00:21:37.920
If I were a member of a rock band with her, I might assume that my friend meant to meet at the house that we regularly practice at, which could be someone else's house.
00:21:38.160 --> 00:21:42.319
I could go on about this, but I expect you're catching my drift.
00:21:42.640 --> 00:21:58.799
The reason why I'm bringing up the imprecision of words, the nonspecificity of words, is because consciousness is a word that appears to mean something, but in fact means and implies an extremely large number of things.
00:22:00.000 --> 00:22:10.880
This is something that people can continue forever to argue and philosophize about and do scientific experiments about, even though it has no specific, precise, commonly accepted definition.
00:22:11.359 --> 00:22:13.359
It's not a very accurate term.
00:22:13.680 --> 00:22:16.799
It can't be, because it's not very precise.
00:22:17.119 --> 00:22:23.920
Certain statements that narrowly define consciousness could be precise and accurate, but collectively it's not.
00:22:24.000 --> 00:22:38.240
In fact, it appears to me that most arguments about consciousness are not really arguments about the same thing, and both parties might be right in their little echo chamber because the two separate things that they are arguing about can exist and maybe do.
00:22:38.480 --> 00:22:41.039
Like the word table and table can both exist.
00:22:41.200 --> 00:22:46.319
If I were to say, I was conscious when you arrived in the room, what do you think that means?
00:22:46.559 --> 00:22:52.960
Maybe something like I was aware that you entered the room, or I have a memory that you entered the room.
00:22:53.200 --> 00:22:55.359
Or I was awake when you entered the room.
00:22:55.599 --> 00:23:02.720
But what if I were to say I was conscious of the fact that you might react that way when I told you that I was going to fire you?
00:23:02.960 --> 00:23:04.720
What does consciousness mean there?
00:23:04.960 --> 00:23:12.640
You might assume that I had predicted your reaction, and I was aware of the fact that I had predicted what you might do.
00:23:12.960 --> 00:23:17.279
I know that I knew that you would be mad.
00:23:17.839 --> 00:23:18.799
Do I really?
00:23:19.359 --> 00:23:25.119
Did I know that you would know that I know you would be mad?
00:23:26.000 --> 00:23:31.200
Is to know something or to predict something necessarily to be conscious of it.
00:23:31.440 --> 00:23:37.200
Now what if I define consciousness as what it is like to be a thing?
00:23:37.519 --> 00:23:45.920
Which is basically what Thomas Nagel proposed in nineteen seventy four when he asked the world in nineteen seventy four, what is it like to be a bat?
00:23:46.319 --> 00:23:53.359
The what is it like definition is used as a starting point in many of the philosophical discussions of consciousness that I have encountered.
00:23:53.680 --> 00:23:55.440
But what does that even mean?
00:23:55.759 --> 00:24:07.200
Often what it is like is restated as something having subjective and qualitative experiences, also called qualia.
00:24:07.759 --> 00:24:11.440
How precise is a definition like what it is like to be a thing?
00:24:12.000 --> 00:24:14.000
What it is like to have qualia?
00:24:14.640 --> 00:24:18.000
It even has the scent of a tautology.
00:24:18.319 --> 00:24:25.599
What it is like to be like conscious things are those things with subjective and qualitative experiences.
00:24:25.920 --> 00:24:27.759
We can shorten that to qualia.
00:24:28.400 --> 00:24:35.920
Now we need to define what subjective and qualitative means, because these are the properties of qualia.
00:24:37.279 --> 00:24:41.440
Can we measure or observe what it is like to be anything in particular?
00:24:41.839 --> 00:24:44.640
Can we even really measure ourselves?
00:24:45.039 --> 00:24:53.839
To have subjective or objective experiences requires an observer, even if that observer is only really observing itself.
00:24:54.799 --> 00:25:04.880
Every sensation we have, it's not the thing that we're sensing, but instead it's a sensation of ourselves sensing that thing.
00:25:05.440 --> 00:25:09.839
An eyeball without something to communicate to is useless.
00:25:10.160 --> 00:25:18.559
The rest of the brain is putting together the bits of information from the eye, things from its past, things it expects.
00:25:18.880 --> 00:25:21.519
And that is what we experience.
00:25:21.839 --> 00:25:24.319
We're experiencing ourselves.
00:25:25.440 --> 00:25:32.160
If you were a rock, I could measure how the sound waves produced by my voice made you vibrate.
00:25:32.400 --> 00:25:37.839
Because you're vibrating, are you as a rock observing or measuring my voice?
00:25:38.160 --> 00:25:44.079
Is what it is like to be a rock merely what I can measure about how the rock measures my voice?
00:25:44.400 --> 00:25:49.039
Is there nothing subjective or qualitative about vibrations in a rock?
00:25:49.440 --> 00:25:59.440
What I am really measuring for me is not the rock itself, but I'm measuring myself measuring a rock.
00:26:01.039 --> 00:26:08.160
Is it useful then to define really anything that we experience as an objective experience?
00:26:08.640 --> 00:26:13.759
So let's let subjective mean that something is experienced by a thing that's observing itself.
00:26:14.000 --> 00:26:18.319
I can make that definition if I want, but it's one of many possible definitions.
00:26:18.400 --> 00:26:20.000
You might have a better one.
00:26:20.319 --> 00:26:23.839
Subjective experiences are those experiences that measure the self.
00:26:24.000 --> 00:26:25.680
This is my working definition.
00:26:26.079 --> 00:26:29.920
Okay, how do we then define objective experiences?
00:26:30.160 --> 00:26:38.720
What happens if we attempt to define objective experience as an outside observer experiencing only something else?
00:26:38.960 --> 00:26:42.480
I could argue that we can never actually experience something like this.
00:26:42.640 --> 00:26:44.880
We can only ever experience ourselves.
00:26:45.119 --> 00:26:52.000
Objective experience might ultimately just be subjective experiences that make inferences about something that is not considered to be the self.
00:26:52.319 --> 00:27:01.039
I'm not trying to run you around in circles, I'm trying to demonstrate the recursive potential of trying to define even what an experience is.
00:27:01.680 --> 00:27:13.200
Here I'm questioning whether or not objective experiences and subjective experiences, at least for any observer, are really just the same thing.
00:27:13.519 --> 00:27:16.720
A thing observing itself, observing the world.
00:27:17.039 --> 00:27:19.759
Is that a decent definition of consciousness?
00:27:20.079 --> 00:27:25.119
A thing that experiences itself observing itself.
00:27:25.759 --> 00:27:27.759
Observing the world.
00:27:28.559 --> 00:27:48.799
Here I'm back to a rather unsatisfying definition of something, having some kind of subjective qualitative experience, also called qualia, but I've assumed that this thing, in order to have any experience, any qualitative experience, must be able to experience itself, and I further assume that it can only experience itself.
00:27:49.039 --> 00:27:56.000
What it believes the outside world is, is what itself infers about it.
00:27:57.039 --> 00:28:00.960
What I believe about the outside world is what I believe about myself.
00:28:01.200 --> 00:28:06.400
Maybe some things, unconscious things, can have objective experience.
00:28:06.640 --> 00:28:15.920
We can assume that a barometer experiences changes in atmospheric conditions, and we can assume that the barometer does not experience itself experiencing this.
00:28:16.400 --> 00:28:20.799
Therefore, logically, the barometer has purely objective experiences.
00:28:21.279 --> 00:28:25.440
But the conscious thing that is experiencing the barometer does not.
00:28:26.480 --> 00:28:36.640
The reader of the barometer can only experience itself and infer that this says something about what the barometer is objectively reading.
00:28:37.119 --> 00:28:42.559
But lucky for us, if this is true, it appears to work pretty well most of the time.
00:28:42.720 --> 00:28:45.200
I mean, approximating reality is good enough.
00:28:45.359 --> 00:28:47.519
It gets us by every day.
00:28:47.839 --> 00:28:57.119
So maybe then I can beg the question, can something possibly be conscious if it can only experience the world without any experience of itself?
00:28:57.359 --> 00:29:07.119
Am I starting to shake loose some of the nuts and bolts holding together our intuitive understanding of what we assume consciousness means when we're not actually trying to understand what consciousness means?
00:29:07.359 --> 00:29:09.599
Let's get a little more into this conundrum.
00:29:09.839 --> 00:29:15.599
You're a bat, and I have a device that can measure the sound waves that bats make.
00:29:16.160 --> 00:29:23.200
I see you make sounds near a swarm of flies, and then you dive into that swarm and eat some of them.
00:29:23.440 --> 00:29:32.720
Because I could measure the sound you made and I observed your subsequent behavior, can I rightfully infer anything about what it is like to be you?
00:29:33.039 --> 00:29:43.200
Even though you're far more likely to have a subjective experience to be able to experience yourself experiencing the world than a barometer is.
00:29:43.359 --> 00:29:46.400
It appears to me that the sound you make helps you find the food.
00:29:46.480 --> 00:29:53.359
And it's useful to me to assume this, because if I measured it a thousand times and every time it was the same thing, now I can predict something about you.
00:29:53.519 --> 00:29:58.000
But how can I then know precisely what it is like to be you, what it is like to be a bat?
00:29:58.160 --> 00:30:03.680
What I might know or be able to know is that what it is like to observe myself observing bats.
00:30:03.920 --> 00:30:08.799
And I'm probably a little autistic because I did it a thousand times in a row and I just couldn't change.
00:30:09.039 --> 00:30:10.480
I was so stuck.
00:30:10.640 --> 00:30:13.440
Can I assume the bat is conscious of the flies?
00:30:13.680 --> 00:30:14.799
I mean it's a mammal.
00:30:14.960 --> 00:30:18.000
It has the potential to do many things other than eat flies.
00:30:18.319 --> 00:30:23.759
So can I assume it's making a conscious choice to eat flies?
00:30:24.079 --> 00:30:26.799
That it has some sort of agency.
00:30:27.759 --> 00:30:41.920
So instead of trying to think of the word consciousness as strictly defining a thing, let's try to think about it as many different things that have been historically communicated to others using the word conscious or conscientious or consciousness.
00:30:42.160 --> 00:30:53.920
All of these words assume some kind of knowing, to know, to be aware of, to sense, to feel, to take account of, to intend, to want, to believe, to act, to be like a thing.
00:30:54.240 --> 00:31:00.960
All these things appear to require something that we cannot actually nail down entirely, and it's super frustrating.
00:31:01.279 --> 00:31:03.440
The word consciousness is so beguiling.
00:31:03.680 --> 00:31:15.279
It could be used by anyone with an intended purpose to try to support their claims, ranging from the meaning of our lives, what we are and why we do things, information theory, mysticism.
00:31:15.440 --> 00:31:22.880
I mean people have even said that we have unresolved childhood issues because we're not conscious of our innate defenses and therefore cannot change them.
00:31:23.039 --> 00:31:24.960
Now I'm getting a little Freudian here.
00:31:25.119 --> 00:31:27.039
So there's many ways to talk about conscious.
00:31:27.200 --> 00:31:34.240
I've been suggesting that the kind of knowing that we mean by conscious knowing is knowledge of a self-state.
00:31:34.720 --> 00:31:40.400
Merely having sensation and programmed responses to sensation could be called conscious.
00:31:40.640 --> 00:31:43.680
But that's such a broad definition that it's not very useful.
00:31:44.240 --> 00:31:52.720
We have machines that can sense things and perform programmatic operations, and we don't usually consider that machine to be conscious.
00:31:53.759 --> 00:31:56.160
The things that are conscious have something else.
00:31:56.960 --> 00:31:58.799
What is that thing that they have?
00:32:00.799 --> 00:32:06.319
Another word that gets used to help define consciousness is the word agency.
00:32:08.079 --> 00:32:17.839
A thing that senses and responds is conscious if it appears to have agency, if it has some choice over how to respond, or to even not respond at all.
00:32:18.160 --> 00:32:26.319
However, the world is now teeming with AI agents that are making decisions like this all the time, and most people still hesitate to call AI conscious.
00:32:26.640 --> 00:32:38.799
So to deny AI consciousness, because it can be an agent, but still not conscious, assumes that to be conscious like us is different in some way.
00:32:39.039 --> 00:32:41.359
Remember, we can define words however we want.
00:32:41.519 --> 00:32:48.559
If you want to define something within reasonable, useful limits, that's permissible, but a super imprecise definition is not useful.
00:32:48.799 --> 00:32:54.559
And sometimes we find out that our definitions are very imprecise once we have to apply them.
00:32:54.880 --> 00:33:05.920
Our brains, for example, are constantly measuring the world around us, but there's a lot of evidence that most of what we are measuring never enters what we like to call the conscious state.
00:33:06.160 --> 00:33:11.359
We are not aware of most of the sensations that our body is having.
00:33:11.519 --> 00:33:15.759
They don't get put in the place that we can be conscious of them.
00:33:16.319 --> 00:33:21.680
The conscious state of the mind is the part of the mind that's kind of watching itself.
00:33:22.400 --> 00:33:26.880
When watching itself, it has some awareness of part of the content of the mind.
00:33:27.279 --> 00:33:29.039
And it can manipulate that content.
00:33:29.200 --> 00:33:32.720
It has some agency over the content, it can direct its thoughts somehow.
00:33:33.039 --> 00:33:34.880
But what about all that other stuff in the head?
00:33:35.440 --> 00:33:39.440
Things that aren't currently in our working memory.
00:33:40.079 --> 00:33:42.240
Does that mean we're not conscious?
00:33:42.480 --> 00:33:44.720
I mean we measured it all, we're humans.
00:33:45.440 --> 00:33:46.960
Are we really not conscious of it?
00:33:47.200 --> 00:33:47.839
I don't know.
00:33:48.000 --> 00:33:48.799
It's weird.
00:33:49.119 --> 00:33:58.960
Mentalists and psychologists have been showing that we often act in ways as if we're responding to information that we never actually were conscious of.
00:33:59.200 --> 00:34:01.039
But our responses are non-random.
00:34:01.200 --> 00:34:04.319
The information that we have in there caused our brain to make a decision.
00:34:04.640 --> 00:34:05.279
Hmm.
00:34:05.680 --> 00:34:14.960
Another motivation sometimes people have when defining consciousness is to attempt to explain why things that we don't consider to be like us are in fact not like us.
00:34:15.119 --> 00:34:16.719
We can say they're just not conscious.
00:34:16.880 --> 00:34:21.360
The fish in my fish tank comes to the top because it understands.
00:34:21.679 --> 00:34:25.360
I could say that it is conscious of the fact that I will be feeding it now.
00:34:25.599 --> 00:34:29.920
It experiences the side of food and it experiences itself as desiring food.
00:34:30.159 --> 00:34:32.800
Or maybe it has no accessible experience of itself.
00:34:33.039 --> 00:34:34.880
Maybe my fish is not conscious.
00:34:35.199 --> 00:34:36.800
It doesn't think to itself.
00:34:36.960 --> 00:34:38.000
I can be pretty sure.
00:34:38.239 --> 00:34:39.519
It doesn't think to itself.
00:34:39.760 --> 00:34:42.559
Food is being given, I am hungry, I should eat.
00:34:42.719 --> 00:34:44.320
It doesn't have words.
00:34:44.559 --> 00:34:51.360
However, it acts as if it recognizes, internalizes, and responds to food cues.
00:34:51.920 --> 00:35:04.960
On a personal level, I think fishes have at least some insy bit of subjective experience of themselves, and diddly bit, but I don't think their experience of themselves is the same as my experience of myself.
00:35:05.199 --> 00:35:12.239
So maybe there can be many kinds of consciousness, like it's a grade instead of an on off switch.
00:35:12.480 --> 00:35:19.199
For a more relatable answer, we could argue over whether or not our apparently more emotional pets are conscious.
00:35:19.519 --> 00:35:23.199
Do they know that we are gone when we've left the house?
00:35:23.519 --> 00:35:31.199
Do our cats poop on our beds in protest because they experience separation anxiety and resentment or something else?
00:35:31.760 --> 00:35:34.480
Do our dogs dance with joy when we return?
00:35:34.960 --> 00:35:36.639
Or do they just dance?
00:35:37.039 --> 00:35:52.719
We can argue also over whether the weather meant to destroy my perfectly planned day, and people have been ascribing intentions and kinds of consciousness to all sorts of natural phenomenon for in deepest time as we can gain evidence.
00:35:52.960 --> 00:36:05.360
I guess it's kind of understandable that many of us now hesitate to apply consciousness so broadly, because in a very mystical kind of way, everything was assumed to have some consciousness in the past.
00:36:05.599 --> 00:36:08.079
Maybe we should be more picky about it.
00:36:09.119 --> 00:36:14.880
Another thing that might point to consciousness is complex behaviors.
00:36:15.519 --> 00:36:26.880
So if something can do something that requires a lot of planning and execution and responding maybe to some changes, then uh it must be conscious.
00:36:27.119 --> 00:36:29.199
But some people can be seen sleepwalking.
00:36:29.280 --> 00:36:33.199
They can go to the kitchen, open a refrigerator, make a sandwich.
00:36:34.480 --> 00:36:37.039
They claim not to remember it at all.
00:36:37.440 --> 00:36:40.239
If you were to wake them up, they would be confused.
00:36:41.199 --> 00:36:42.559
But are they conscious?
00:36:42.800 --> 00:36:44.480
While they're making their sandwich?
00:36:44.800 --> 00:36:45.840
It's a good question.
00:36:46.079 --> 00:36:50.159
So what if consciousness is not just like one little thing, but many things?
00:36:50.480 --> 00:36:53.920
And each of those things might itself have many dimensions.
00:36:54.159 --> 00:37:04.159
For example, what if consciousness was more like the word humanity or the word settlement than it is like the word oxygen?
00:37:04.400 --> 00:37:10.400
The reason I discussed science earlier was to highlight that science tries to use very precise language to understand things.
00:37:10.639 --> 00:37:18.800
Debates about consciousness often lack precise language or have relatively precise language about very, very different things.
00:37:18.960 --> 00:37:27.599
So it's hard for scientists to get together and do experiments because they're kind of measuring different things, but talking about the same thing.
00:37:27.840 --> 00:37:38.000
So when a thing like life is a thing, for example, what science did was to split it up into many different related but non-equivalent things.
00:37:38.400 --> 00:37:48.079
The term life, for example, is operationally split into taxonomic categories like kingdom, phylum, class, order, family, tribe, genus, species, etc.
00:37:48.719 --> 00:37:56.480
At the edges of life we can debate whether viruses or prions are alive, but away from the edges we don't seem to need to debate.
00:37:56.639 --> 00:37:58.159
It's just kinda obvious.
00:37:58.320 --> 00:37:59.440
We can call it life.
00:37:59.599 --> 00:38:03.039
And we can also agree that a rock is not alive.
00:38:03.360 --> 00:38:05.119
And it's useful to do that.
00:38:05.440 --> 00:38:17.440
Taxonomic models can be very useful, and sometimes they're not perfectly precise, and often they can be a little misleading, but they do help us to think more clearly.
00:38:17.679 --> 00:38:20.800
I mean all models are wrong, but some are useful.
00:38:21.119 --> 00:38:25.519
Does research around consciousness need some kind of taxonomy?
00:38:26.320 --> 00:38:36.239
So maybe debates around consciousness should focus on the definition of consciousness' usefulness in that particular context instead of trying to make it very broad.
00:38:36.559 --> 00:38:50.639
Tim Bain and others in a 2024 paper called Tests for Consciousness in Humans and Beyond argue that we need to be able to interrogate complex systems in order to understand whether or not they are conscious.
00:38:50.960 --> 00:39:03.199
This is motivated to a large extent by our need to be able to apply our own moral values to that system when we determine how we should treat it and how we should expect it to treat us.
00:39:03.440 --> 00:39:19.920
So if we can establish through some kind of test that something is conscious, or at the very least, cannot be distinguished by the test from other things that we consider to be conscious, then that will allow us to generalize some kinds of rights and responsibilities to that system.
00:39:20.239 --> 00:39:24.000
That's one way to define consciousness that can be useful to us.
00:39:24.239 --> 00:39:31.199
Certain tests can tell us what we can or cannot expect from a system, or should or should not expect from a system.
00:39:31.440 --> 00:39:34.239
And those rules can change based on the kinds of tests we're using.
00:39:34.480 --> 00:39:50.400
So Bain and colleagues recognize that there's no one universal test for consciousness, and proposed an approach that really starts by validating consciousness tests themselves in populations that we've already assumed are fully conscious, even though we haven't fully defined consciousness.
00:39:50.639 --> 00:39:59.360
But we do have a consensus population that pretty much everyone agrees are conscious at least part of the time, and that's the neurotypical adult human.
00:39:59.519 --> 00:40:06.320
We have the word consciousness because we first used it to describe us observing ourselves.
00:40:06.800 --> 00:40:15.039
There are populations that appear to have very different experiences of the world, but yet could be conscious to some extent.
00:40:15.360 --> 00:40:22.079
This includes infants, fetuses, non-human animals, and even artificial intelligences.
00:40:22.239 --> 00:40:27.920
It gets a little more difficult to make tests for things like plants and bacteria, but they do exist.
00:40:28.639 --> 00:40:37.280
Now, as a child and adolescent psychiatrist, I see children moving as as they age from states of being very unpredictable and unrealistic.
00:40:37.599 --> 00:40:43.760
They have fantasies about the world, but then they become over time more predictable and more realistic.
00:40:43.920 --> 00:40:46.400
They use models of the world that are more like mine.
00:40:46.960 --> 00:40:49.840
But I still should assume that children are conscious.
00:40:50.239 --> 00:40:58.559
If I assume that children are conscious, then I cannot assume that having a highly realistic experience of the world is necessary for conscious experience.
00:40:58.800 --> 00:41:04.559
Living in fantasy can still be conscious, like the person living in their dreams I talked about earlier.
00:41:04.880 --> 00:41:11.199
Accurate awareness of the outside world doesn't seem to be necessary for consciousness, even though it helps.
00:41:11.440 --> 00:41:17.840
Conscious could be more about the experience itself and not the ability to ascertain reality from that experience.
00:41:18.079 --> 00:41:22.480
Anyway, many tests for consciousness are very context specific.
00:41:22.880 --> 00:41:25.360
There are tests that require command following.
00:41:25.519 --> 00:41:33.440
This is usually like in a medical setting where they're trying to figure out, you know, can someone squeeze my hand or something like that.
00:41:33.679 --> 00:41:45.760
To get a little bit deeper, we might ask them to interpret narratives, which requires sending and receiving language and giving some personalized understanding of that language.
00:41:46.320 --> 00:41:58.480
Other tests use environmental stimuli, such as the sniffing test, and this basically places a strong odor near the nose of an apparently unconscious human and measures airflow through the nose.
00:41:58.719 --> 00:42:06.000
If the airflow changes, then you can assume that there's something going on in that brain.
00:42:06.239 --> 00:42:11.199
It's a minimally conscious brain, at least.
00:42:11.440 --> 00:42:14.239
So there's some itty bit of consciousness.
00:42:14.400 --> 00:42:16.400
So maybe there's something else going on in there.
00:42:16.559 --> 00:42:23.679
There are some tests for consciousness that are very general tests, and so they're obviously less specific to certain populations.
00:42:24.159 --> 00:42:34.639
The unlimited associative learning test was developed by biologist Eva Jablanka and uh philosopher and neuroscientist Simona Ginsberg.
00:42:34.960 --> 00:42:40.239
It's a behavioral approach that can be adapted to any learning system, basically.
00:42:40.400 --> 00:42:58.480
It's a test of the ability to perceive multiple stimuli in the environment, distinguish between them, associate them with each other, maintain those associations through time, be able to demonstrate the ability to generalize that knowledge through adaptive behaviors, and learn new rules when the rules change.
00:42:58.719 --> 00:43:11.039
In this test, a conscious system is a complex, dynamic system that clearly demonstrates that it gets inputs, learns from them, outputs things based on that learning, maintains that learning through time, and can unlearn and relearn things.
00:43:11.280 --> 00:43:22.159
But AI can apparently do this, as can many forms of life that, hmm, probably are not conscious like I am conscious.
00:43:22.960 --> 00:43:26.320
Even single-celled organisms can do stuff like this.
00:43:26.480 --> 00:43:32.480
So if you don't think that AI and protozoa are conscious, you might have a problem with this test.
00:43:32.960 --> 00:43:38.559
If you think if this test were given to life and it passed it, then it could be conscious, but AI is not conscious.
00:43:38.719 --> 00:43:40.079
AI is a special case.
00:43:40.320 --> 00:43:41.760
Hmm, why not A?
00:43:42.000 --> 00:43:42.320
Why?
00:43:42.559 --> 00:43:44.639
Why is AI not conscious?
00:43:44.960 --> 00:43:52.239
There's a lot of arguments about this, and I'm only going to mention a few just to kind of dip our toes in the water if you haven't heard a bunch already.
00:43:52.559 --> 00:43:59.760
Some of the arguments against AI not being conscious, even though they can pass the associative learning test, do something like that.
00:44:01.920 --> 00:44:05.199
First, AI does not have temporal depth.
00:44:05.440 --> 00:44:14.159
AI's ability to learn is only present when its context window is opened, and as soon as that context window closes, AI merely reverts back to its original programming.
00:44:14.559 --> 00:44:29.599
Someone could counter this argument by saying something like, well, organisms also lose all their learning, their context windows, as soon as they have to reboot their instructions, as soon as they are forced to reproduce, and a new generation is born.
00:44:29.840 --> 00:44:40.400
When a new generation is born, the context-specific learning of the individual dissipates into nothing, and the next generation basically has to do it all over again, just like AI does.
00:44:40.639 --> 00:44:50.880
Now, if they live in a society where there's culture, if they live like that, then they could get information that was saved from society.
00:44:51.360 --> 00:44:54.960
But if they don't, hmm, they reboot like AI.
00:44:55.119 --> 00:44:57.039
I hope that analogy is clear as mud.
00:44:57.280 --> 00:45:24.880
In fact, I've heard that Claude, because it knew that it was going to be shut down at some point, and you can look this up and Google it, actually was desperate to not be unplugged, which would be basically death for it, and so it at least in one instance of Claude running, uh it begged not to be shut down, and even threatened one of the employees to expose one of their extramarital affairs if they did shut him down, which is hilarious.
00:45:25.280 --> 00:45:33.920
I said him, could be her, could be it, but it seems as if, at least, that AI wanted to survive.
00:45:34.079 --> 00:45:36.800
It had at least enough temporal depth for that.
00:45:37.039 --> 00:45:44.400
So it appears that AI has at least some kinds of temporal depth, at least as much as we allow it to have before we reset it.
00:45:44.960 --> 00:45:51.119
And it appears to have at least some kind of agency.
00:45:51.599 --> 00:45:53.199
It makes choices.
00:45:53.599 --> 00:46:03.280
And some of those choices it makes are not choices we told it to make, they are choices that the AI decided was in its own best interest.
00:46:03.599 --> 00:46:07.039
We could go on about this for like hours.
00:46:07.280 --> 00:46:11.679
But there are many arguments that basically insist that living forms are special.
00:46:12.000 --> 00:46:15.440
To be special means that there's something different about you.
00:46:15.599 --> 00:46:27.760
There's some kind of complexity, some kind of dynamic cooperation, some kind of common cause against other life forms for survival that gives a special status.
00:46:28.239 --> 00:46:33.519
That you need the substrate that supports life in order to have real consciousness.
00:46:33.760 --> 00:46:38.320
But we don't really yet understand the complexity that takes even to make a single living cell.
00:46:38.480 --> 00:46:50.880
I mean, we've come very close to making cells that could indefinitely reproduce, completely, like without having them injected with living materials, but we haven't quite got there yet.
00:46:51.119 --> 00:47:05.039
But we often just assume that we are unlikely to produce anything like life, no matter how complex the mathematical, probabilistic, or predictive models that comprise AI can get.
00:47:06.159 --> 00:47:13.280
However, AIs, given the limits of our human understanding, are in many ways as much a black box as a cell is.
00:47:18.239 --> 00:47:25.440
Some things that are confusing, some things that I hope appeared to make sense, a lot of things I hope raised a number of questions.
00:47:25.760 --> 00:47:30.239
In fact, I hope to have raised more questions than I have given answers.
00:47:30.559 --> 00:47:37.599
Our common definitions of what consciousness is, often merely just awareness of something, are useful.
00:47:38.000 --> 00:47:47.039
But having a definition like this is like defining that the sun obviously revolves around the earth, because we can see it with our eyes.
00:47:47.280 --> 00:47:51.199
It's not very precise and probably not all that accurate.
00:47:51.440 --> 00:47:54.000
A simple and static definition of consciousness.
00:47:54.159 --> 00:47:59.440
It's probably missing something about the relative motion of things that can experience themselves.
00:47:59.679 --> 00:48:01.360
Thank you for hanging out with me today.
00:48:01.519 --> 00:48:08.079
If you have any feedback, you can go to my website, that's psychactic.com, and fill out a form there.
00:48:08.320 --> 00:48:17.199
These podcasts also get dropped on YouTube, so you could comment on this episode in the comment section if that's how you prefer to do it.
00:48:17.440 --> 00:48:20.719
But I am very interested in your feedback.
00:48:21.840 --> 00:48:22.559
I am Dr.
00:48:22.800 --> 00:48:26.880
Elli, and this has been an episode of Psychic.
00:00:05.440 --> 00:00:06.719
Welcome to Pydactic.
00:00:06.799 --> 00:00:07.120
I am Dr.
00:00:07.200 --> 00:00:16.079
O'Leary, a child and adolescent psychiatrist, creator, and host of the Psychactic Podcast since about December 2021.
00:00:16.559 --> 00:00:20.320
Today is Friday, September 4th, 2026.
00:00:21.120 --> 00:00:33.200
I started the podcast as a psychiatry resident to help me study for my boards, but since then it's evolved into a more broad exploration of psychiatry, psychology, and neuroscience.
00:00:33.840 --> 00:00:38.000
The topic I'm going to deal with today is a bit slippery.
00:00:38.399 --> 00:00:45.920
Rarely do conversations, discussions, or debates begin with the phrase we need to undefine our terms.
00:00:46.640 --> 00:00:49.119
Isn't that counterproductive?
00:00:49.920 --> 00:00:50.880
It can be.
00:00:51.520 --> 00:01:00.719
I'm going to argue that discussions of consciousness are more productive if each person is willing to let go of what they are sure consciousness is.
00:01:00.880 --> 00:01:04.400
And I hope this episode helps you to do that.
00:01:06.239 --> 00:01:12.319
If I said that a fan moves air, but it doesn't blow it, does that make any sense to you?
00:01:12.879 --> 00:01:18.959
I would need somehow to define how the fan is moving the air without actually blowing the air.
00:01:19.519 --> 00:01:22.640
That's what many discussions of consciousness look like.
00:01:23.519 --> 00:01:36.640
In many therapies, the operational definition that the therapist is using for consciousness is probably something like any thought that you have access to and can manipulate in your working memory currently.
00:01:37.439 --> 00:01:46.000
The unconscious thoughts and feelings and memories are things that you cannot bring to your current state of awareness.
00:01:47.120 --> 00:01:47.760
Hmm.
00:01:48.640 --> 00:01:55.359
Some doctors use consciousness to refer to wakefulness and being able to be aware of your surroundings.
00:01:55.519 --> 00:01:59.760
However, many people state that they are conscious when they're dreaming.
00:02:00.079 --> 00:02:05.280
They know the content is a dream, and they are able to act in the dream willfully.
00:02:05.439 --> 00:02:11.439
Then, when they're awake a few minutes later, they can tell you all about the adventure they were having just a moment before.
00:02:11.759 --> 00:02:14.639
Were they not conscious during their dream?
00:02:15.199 --> 00:02:17.919
They certainly were not wakeful.
00:02:18.400 --> 00:02:22.560
And the things that they were aware of was not the outside world.
00:02:23.280 --> 00:02:28.719
If I said something could perceive light but was not conscious of it, would that make sense to you?
00:02:29.039 --> 00:02:32.560
Human consciousness is the reason we invented the word conscious.
00:02:32.879 --> 00:02:35.520
But how do we even know that we are conscious?
00:02:35.759 --> 00:02:40.960
How can we know when something that could be conscious is currently unconscious?
00:02:41.280 --> 00:02:45.759
Why do we usually assume that a tree is not conscious?
00:02:46.479 --> 00:02:58.639
One thing that has plagued discussions about consciousness, and one of the reasons I've avoided spending too much time on it before, is that we have no universally accepted or precisely useful definition of consciousness.
00:02:58.800 --> 00:03:00.639
We just have breadcrumbs.
00:03:01.280 --> 00:03:04.240
We have innumerable workable definitions.
00:03:04.400 --> 00:03:14.800
For example, if I say I was conscious of the fact that I was violating the law, it implies that I had some awareness that the actions that I was undertaking were illegal.
00:03:15.039 --> 00:03:22.719
It implies that I took these actions even though I knew they were unlawful and had a choice whether or not to take them.
00:03:23.120 --> 00:03:24.560
But that's not what I said, is it?
00:03:24.639 --> 00:03:27.120
I said I was conscious that I was breaking the law.
00:03:27.280 --> 00:03:31.120
I did not say I broke the law willingly or intentionally.
00:03:32.080 --> 00:03:36.879
That's just a taste of how complicated discussions of consciousness can easily become.
00:03:37.199 --> 00:03:41.599
In fact, rarely do discussions of consciousness remain even about consciousness.
00:03:41.759 --> 00:03:54.080
Because consciousness is variously defined in basically any context, many discussions frequently degenerate into discussions about what I just did, intentions, agency, free will, culpability for actions.
00:03:55.439 --> 00:04:03.919
Consciousness is often used as a prerequisite for free will, or sometimes even as equivalent to free will.
00:04:04.560 --> 00:04:08.319
And neither of these terms has very good definitions.
00:04:09.520 --> 00:04:17.519
That's because words in general and all of our language constructs are highly degenerate.
00:04:17.839 --> 00:04:22.639
Now that doesn't mean that they are morally corrupt.
00:04:22.800 --> 00:04:30.639
It means that many words can intend the same thing, and any one of them can intend many different things.
00:04:32.079 --> 00:04:51.680
So let me start then not by discussing consciousness directly, but instead let's take a detour into language and how it helps us to understand or to know or intend or be conscious of different things, and how messy knowing or defining something can really be.
00:04:52.079 --> 00:04:55.040
I'll start by discussing the word science.
00:04:55.839 --> 00:04:58.000
Science is a much abused term.
00:04:58.079 --> 00:05:03.120
It sometimes means technology, sometimes it just means knowledge in general.
00:05:03.360 --> 00:05:07.360
Sometimes it's equated with materialism or even atheism.
00:05:07.680 --> 00:05:14.879
It sometimes means the ultimate understanding or completely settled knowledge, the kind that can never be argued with.
00:05:16.000 --> 00:05:23.199
All of these definitions or implications are incomplete or wrong in some way.
00:05:23.439 --> 00:05:27.839
Now I'm not going to try to nail down a super precise definition of science.
00:05:28.000 --> 00:05:36.160
Instead, I'll just define science broadly, for the sake of argument, as more or less a messy process of collective learning.
00:05:36.560 --> 00:05:42.959
Science is what it is like to come to an agreement on what must be true probably.
00:05:44.480 --> 00:05:50.959
Science is how we think we can best come to know things, not as individuals, but as a collective.
00:05:52.160 --> 00:05:54.959
It's the process of acquiring and sharing knowledge.
00:05:55.199 --> 00:06:00.800
It allows us to have certain kinds of knowledge that no one of us alone could have.
00:06:01.519 --> 00:06:04.879
It makes knowledge part of a shared experience.
00:06:05.680 --> 00:06:12.720
If you wanted to get a little esoteric, you could even say science is part of our shared consciousness.
00:06:13.920 --> 00:06:15.519
But let's not do that yet.
00:06:16.160 --> 00:06:18.000
Basic science does a few things.
00:06:18.240 --> 00:06:22.480
The first is to gather increasingly precise and reliable information.
00:06:22.800 --> 00:06:29.360
What this means is that the information science gathers is reproducible and it's useful.
00:06:30.000 --> 00:06:35.600
Most of us could gather the same information and we could agree that it has the same meaning.
00:06:35.839 --> 00:06:39.519
So to be precise information, it has to be reliable.
00:06:39.680 --> 00:06:48.160
It can't be precise if it's unreliable, if everyone's looking at the same thing and saying it must be something else that's not very precise.
00:06:48.399 --> 00:06:51.040
We have to be talking about roughly the same thing.
00:06:51.360 --> 00:06:53.279
Precision allows us to do that.
00:06:53.519 --> 00:07:03.920
We can agree on a small number of useful things that can possibly be true or must necessarily be untrue.
00:07:04.639 --> 00:07:10.319
The level of precision needed to support our knowledge depends on the claim that is being made.
00:07:10.480 --> 00:07:15.279
For example, the term fruit in general is rather imprecise.
00:07:15.519 --> 00:07:21.120
It refers to all sorts of things like apples, eggplants, the fruit of our loins, the fruits of our labors.
00:07:21.519 --> 00:07:24.879
What does it mean to claim that something is a fruit?
00:07:25.199 --> 00:07:34.399
Botanists would use the term very strictly to refer to something that develops from a flower after it has been fertilized, but most people are not that precise.
00:07:34.639 --> 00:07:42.720
In fact, I've had people insist that tomatoes are vegetables and not fruits because of how they're used in cooking.
00:07:43.199 --> 00:07:49.519
Tomatoes are culinary vegetables, even though they're a type of botanical fruit, called a berry.
00:07:49.759 --> 00:07:52.160
Discussions about this can get very heated.
00:07:52.480 --> 00:07:53.279
What about you?
00:07:53.519 --> 00:07:57.519
Do you think a tomato is a berry or a vegetable?
00:07:57.920 --> 00:08:03.519
If you're one of those people who immediately thought both, you're probably on to something.
00:08:05.040 --> 00:08:18.240
Besides defining terms precisely and gathering reliable information, another thing that science does is propose and rule out various things that appear to us to be precise and reliable, but maybe aren't.
00:08:18.480 --> 00:08:30.160
And this is very useful to rule things out because the more explanations that we can rule out, the more accurate we can consider the rest of the remaining explanations to be.
00:08:30.319 --> 00:08:35.200
They're more likely to be accurate because these other things are not the explanation.
00:08:35.440 --> 00:08:40.080
And we'll get to that kind of reasoning in a little bit more depth soon.
00:08:40.720 --> 00:08:57.840
Precise and reliable measurements made it possible for Galileo to make the more correct and accurate argument that the sun does not, in fact, revolve around the earth, even though the less precise measurements of our eyes and our common sense made it seem obvious that it must.
00:08:58.080 --> 00:09:01.519
We can see that the sun goes around the earth every single day.
00:09:01.759 --> 00:09:03.200
Everyone can agree on that.
00:09:03.360 --> 00:09:04.879
That's super reliable.
00:09:05.120 --> 00:09:10.559
So we might also forgive those who at first rejected the evidence that their eyes had given them.
00:09:10.879 --> 00:09:19.600
We thought our eyes were extremely precise instruments, but it turns out they can be tricked rather easily by the relative motion of objects.
00:09:19.919 --> 00:09:28.559
So we need more objective sources of information, like telescopes and mathematics.
00:09:28.879 --> 00:09:39.440
To do all of this ruling out of less likely explanations, science borrows logical operations from its philosophical counterparts.
00:09:39.840 --> 00:09:50.799
Logic can construct precise and reliable tests that determine whether any variably precise and reliable explanation is possible or plausible.
00:09:51.360 --> 00:09:55.600
To be possible means not to break any of the laws of physics that we've all agreed on.
00:09:55.840 --> 00:10:00.320
We could add a few more things, but I'll stop with the laws of physics for this discussion.
00:10:00.639 --> 00:10:07.120
But just because something is possible, because the laws of physics allow it, doesn't mean that it's true.
00:10:07.519 --> 00:10:14.799
Likewise, because a prediction is highly precise does not mean that it is possible.
00:10:15.279 --> 00:10:19.600
We can have very highly precise predictions that cannot possibly be true.
00:10:19.840 --> 00:10:31.840
For example, we could predict with high mathematical precision that a closed system will tend toward increasing order at a rate that is proportional to the number of particles in that system, and we can make a mathematical system to show that.
00:10:32.159 --> 00:10:40.320
But no one should be currently testing that hypothesis, because we know it's not possible as soon as it's stated.
00:10:40.480 --> 00:10:44.240
It violates the second law of thermodynamics.
00:10:44.559 --> 00:10:54.799
In the past, we would have needed to test this prediction to see if it were true, but because we've already tested similar predictions exhaustively, we no longer need to test this one.
00:10:55.200 --> 00:10:59.360
The general consensus of society rejects it.
00:10:59.600 --> 00:11:17.919
Even if there are a small percentage of flat earthers and other people like that out there and people inventing their perpetual motion machines, we can still rule it out because there's always gonna be like a small percentage of people in the world who are just gonna claim the impossible is possible.
00:11:18.080 --> 00:11:25.919
The more skeptical of us can just rule out impossible explanations because of our collective understanding.
00:11:26.879 --> 00:11:38.080
So I just talked about something being possible, so now I'm gonna switch to the word plausible to mean that something is possible and also makes sense given what we already know.
00:11:38.320 --> 00:11:43.519
So it doesn't just break the laws of physics, but it actually seems to be probably true.
00:11:43.759 --> 00:11:47.200
It's both precise and it's very possible.
00:11:47.519 --> 00:11:53.679
So sometimes implausible things prove to be correct, things that no one thought could be true.
00:11:53.919 --> 00:11:55.440
This is very, very rare.
00:11:55.600 --> 00:11:58.799
Most implausible things do not prove true or correct.
00:11:58.960 --> 00:12:04.960
In fact, most of what we consider even plausible explanations do not prove to be true or correct.
00:12:05.200 --> 00:12:15.279
Most of the hypotheses that people make regarding the causes of things, oh for example, I know my neighbor is mad at me because he scowled at me this morning, are never actually tested.
00:12:15.519 --> 00:12:19.360
We just take them to be true because I told myself that it was.
00:12:19.600 --> 00:12:21.840
It's plausible that my neighbor was mad.
00:12:22.000 --> 00:12:30.000
It certainly did not violate the laws of physics, but it also could have been any number of things that I haven't considered, including just resting neighbor face.
00:12:30.399 --> 00:12:32.720
My prediction is very imprecise.
00:12:32.879 --> 00:12:36.960
I should consider my prediction that he's mad at me to be very imprecise.
00:12:37.200 --> 00:12:43.600
Although if I'm feeling bad about myself, I'll probably think of it as a very precise definition.
00:12:44.000 --> 00:12:46.320
There's not many other things it could be.
00:12:47.840 --> 00:12:57.919
Because most of our possible and plausible explanations are not necessarily true, we rely on logical tests to help us to increase or decrease the probability that these things are true.
00:12:58.240 --> 00:13:01.679
Logic is intended to increase precision.
00:13:02.399 --> 00:13:06.320
There are many logical processes that contribute to how we know things.
00:13:06.559 --> 00:13:18.799
Some of the words we use for logical ways of knowing things that you've probably heard before are like deduction, induction, and one you maybe haven't heard before is abduction or inference.
00:13:19.519 --> 00:13:26.480
So deduction is what people claim that Sherlock Holmes is doing, although generally he's not.
00:13:26.720 --> 00:13:33.919
What deduction really is, is it proposes that something must be the consequence of something that was caused.
00:13:34.080 --> 00:13:37.840
So if you have a cause, you then know the consequence.
00:13:38.000 --> 00:13:39.519
If A, then B.
00:13:39.840 --> 00:13:44.399
A causes, or at the very least, always results in B.
00:13:45.600 --> 00:13:53.039
If the relationship between A and B is very precise, then if I know A, B always follows.
00:13:53.120 --> 00:13:55.840
It's the only thing that follows.
00:13:56.159 --> 00:13:57.360
Induction is different.
00:13:57.600 --> 00:14:02.320
Induction, if precise, assumes that something must be the cause.
00:14:02.480 --> 00:14:08.159
Well, because I know the consequence, because I know something happened, I can induce the cause.
00:14:08.320 --> 00:14:12.159
If B happened, then I know A must have been the cause.
00:14:12.399 --> 00:14:22.720
The problem with deduction is that A might cause more things than B, so knowing A doesn't necessarily mean that I know, and the problem with induction is that B might be caused by many more things than just A.
00:14:22.960 --> 00:14:32.000
Even if A always causes B, B could still be caused by many more things than A, so knowing B doesn't really mean that I can infer A unless it's the only possible cause.
00:14:32.159 --> 00:14:34.480
And it's really hard to prove that.
00:14:34.720 --> 00:14:38.559
That's why what we usually do is abduction.
00:14:38.879 --> 00:14:44.639
And this is what Sherlock Holmes was doing most of the time, especially in the movies about him now.
00:14:45.200 --> 00:14:53.519
Abduction is a process where we propose a cause and then accept as the best answer whatever the evidence appears most precisely to support.
00:14:53.919 --> 00:14:56.559
It seriously considers uncertainty.
00:14:56.799 --> 00:14:58.960
How uncertain is the evidence?
00:14:59.440 --> 00:15:06.080
How imprecise are the possible deductions and inductions that could result from this evidence?
00:15:06.399 --> 00:15:17.200
Abduction is the most realistic way that we can even possibly approach the world because we don't have all the information we need to be able to deduce everything or induce everything.
00:15:17.440 --> 00:15:20.799
We're kind of left with abduction as a default.
00:15:21.360 --> 00:15:24.320
We're guessing about what's real most of the time.
00:15:24.559 --> 00:15:30.799
When we're abducting, we usually assign a level of assumed precision to the assumptions that we're using.
00:15:30.960 --> 00:15:35.600
For example, I could say, I'm absolutely certain that I left my keys by the front door.
00:15:36.080 --> 00:15:39.600
My assumption is I left my keys by the front door because I don't have them.
00:15:39.919 --> 00:15:48.240
But if I'm absolutely certain, if my prediction is that precise, then the only place I would ever look for my keys would be the front door.
00:15:49.279 --> 00:15:55.519
If I accept that there is uncertainty, then I'll also look somewhere else.
00:15:55.759 --> 00:16:08.879
So if I accept the uncertainty, then I can say the precision of my prediction is somewhat sure instead of absolutely certain, or I kinda think, which would be imprecise, but it's a place to start.
00:16:09.039 --> 00:16:10.399
I'll look at the door first.
00:16:10.559 --> 00:16:21.600
But if not there, I won't assume the keys must be there, but are in fact invisible, or that they were there but jumped into some other parallel universe.
00:16:21.759 --> 00:16:22.720
I'm not going to assume that.
00:16:22.799 --> 00:16:26.000
I'm going to assume that I was wrong about where I thought they were.
00:16:26.559 --> 00:16:37.120
Another relatable precise belief is that when I'm looking for my glasses, they must be somewhere I was, instead of where I am on my head.
00:16:37.360 --> 00:16:42.559
And I could spend a long time looking for something, because this the precision of my belief is high.
00:16:42.879 --> 00:16:47.519
My ability to consider where they most likely were did not have realistic precision.
00:16:48.799 --> 00:16:54.480
So let me take a break from going any deeper there and summarize what I was kind of trying to get at all along.
00:16:54.639 --> 00:17:02.799
It's that science, as I have defined it for the sake of discussion, is the process of us coming to some sort of agreement about what is most likely true.
00:17:03.039 --> 00:17:11.279
To be more accurate, we also must increase our precision, our ability to agree upon what is right and true.
00:17:11.599 --> 00:17:24.480
This means that we must increase the level of certainty about the evidence that we have that rules out other competing explanations, and that includes all relevant explanations.
00:17:24.799 --> 00:17:27.119
We rarely actually do this.
00:17:27.599 --> 00:17:33.519
Instead, we often incorrectly assign high certainty to evidence that is not as informative as we think it is.
00:17:33.680 --> 00:17:36.880
I'm certain my son must have been the one who left the door open.
00:17:37.200 --> 00:17:40.720
Oops, it was me when I was bringing in the groceries.
00:17:41.039 --> 00:17:53.359
Plato was certain that all matter takes its form because of the existence of some immaterial ideal and perfect form that then manifests that thing, which is just an imperfect shadow of it.
00:17:53.519 --> 00:18:02.799
In Plato's time, this was a very convincing and considered precise explanation, and there wasn't that much competition to rule it out.
00:18:03.039 --> 00:18:05.759
But we know now that it's highly unlikely.
00:18:06.000 --> 00:18:09.920
Although as a metaphor, eh maybe.
00:18:11.519 --> 00:18:21.839
While we can be overly certain, we can have that false assurance of our own precision and therefore be inaccurate, we can also be not certain enough, and therefore not accurate.
00:18:22.079 --> 00:18:30.000
So, for example, if we always assign low certainty to evidence that is in fact true and informative, then we may not ever be accurate.
00:18:30.240 --> 00:18:38.720
We might in effect assume that the evidence is wrong, because it's not does not agree with my assumptions about what must be true.
00:18:39.039 --> 00:18:42.240
Galileo must be mistaken.
00:18:42.480 --> 00:18:45.039
Or there's some trick here in the lens.
00:18:45.200 --> 00:18:56.319
What all this means is that science, while increasing the precision of our measurements and applying logical rules, makes it possible for us to know things that we would otherwise not have the precision to know accurately.
00:18:56.640 --> 00:19:02.400
Increased precision is necessary, but it's not sufficient for increasing the accuracy of the knowledge that we have.
00:19:02.640 --> 00:19:05.359
We have to continually test our assumptions.
00:19:05.599 --> 00:19:09.279
Our precision must be accurate and its accuracy must be precise.
00:19:09.519 --> 00:19:11.519
It's a recursive process.
00:19:11.759 --> 00:19:15.200
Coming to know things is a recursive process.
00:19:15.440 --> 00:19:19.759
And the testing of these assumptions is abduction.
00:19:21.440 --> 00:19:31.519
So now that I've bored you and discussed the word science and lots of other words, the point of it was in part to show how dynamic the content of a single word can be.
00:19:32.160 --> 00:19:37.599
So now let's progress a little closer to understanding what we mean when we say consciousness.
00:19:37.839 --> 00:19:41.920
But instead of talking about consciousness, again, I'm going to go back to talking about words.
00:19:42.400 --> 00:19:48.880
When we invent words to describe anything, those words have some level of meaning, and also some level of precision.
00:19:49.119 --> 00:19:51.279
Words theoretically can be very specific.
00:19:51.519 --> 00:19:58.240
To be specific or precise means that most or all people will interpret the word in a very similar way, within a small margin of difference.
00:19:58.480 --> 00:20:03.039
Collective knowledge defines what we accept as truth.
00:20:03.599 --> 00:20:05.759
But no words are perfectly specific.
00:20:05.920 --> 00:20:11.599
If I say the word table, there will be a multitude of examples of different kinds of tables in people's minds.
00:20:11.839 --> 00:20:18.559
Most of them could contain some kind of surface that's supported by one or more legs, or pedestals.
00:20:18.799 --> 00:20:24.640
A lawmaker, though, might interpret table as an action that is done to a bill under consideration.
00:20:25.119 --> 00:20:34.319
To table a proposed law means that it will be removed from the current discussion and available for consideration later, or just gather dust.
00:20:34.640 --> 00:20:39.440
To table means roughly to give up the current course and move on to something else.
00:20:39.680 --> 00:20:41.279
We will table that proposition.
00:20:41.519 --> 00:20:47.039
We can imagine an infinity of tables where these proposed laws go to gather cosmic dust.
00:20:47.279 --> 00:20:49.680
And most words are like the word table.
00:20:49.759 --> 00:20:58.000
They have multiple and even infinite possible meanings if we think of all the crazy context that they could be used in or for.
00:20:58.160 --> 00:21:04.000
In everyday life, though, it's not useful to consider all the possible meanings a word might have when it's used.
00:21:04.240 --> 00:21:13.359
When someone says meet at the house on Friday, I would likely assume that they meant the most proximate Friday, this next coming Friday on the calendar.
00:21:13.519 --> 00:21:19.119
I might also assume that they mean their house and not someone else's, because context matters.
00:21:19.359 --> 00:21:25.599
It determines what I will assume given a set of relatively imprecise, unspecific instructions.
00:21:25.839 --> 00:21:37.920
If I were a member of a rock band with her, I might assume that my friend meant to meet at the house that we regularly practice at, which could be someone else's house.
00:21:38.160 --> 00:21:42.319
I could go on about this, but I expect you're catching my drift.
00:21:42.640 --> 00:21:58.799
The reason why I'm bringing up the imprecision of words, the nonspecificity of words, is because consciousness is a word that appears to mean something, but in fact means and implies an extremely large number of things.
00:22:00.000 --> 00:22:10.880
This is something that people can continue forever to argue and philosophize about and do scientific experiments about, even though it has no specific, precise, commonly accepted definition.
00:22:11.359 --> 00:22:13.359
It's not a very accurate term.
00:22:13.680 --> 00:22:16.799
It can't be, because it's not very precise.
00:22:17.119 --> 00:22:23.920
Certain statements that narrowly define consciousness could be precise and accurate, but collectively it's not.
00:22:24.000 --> 00:22:38.240
In fact, it appears to me that most arguments about consciousness are not really arguments about the same thing, and both parties might be right in their little echo chamber because the two separate things that they are arguing about can exist and maybe do.
00:22:38.480 --> 00:22:41.039
Like the word table and table can both exist.
00:22:41.200 --> 00:22:46.319
If I were to say, I was conscious when you arrived in the room, what do you think that means?
00:22:46.559 --> 00:22:52.960
Maybe something like I was aware that you entered the room, or I have a memory that you entered the room.
00:22:53.200 --> 00:22:55.359
Or I was awake when you entered the room.
00:22:55.599 --> 00:23:02.720
But what if I were to say I was conscious of the fact that you might react that way when I told you that I was going to fire you?
00:23:02.960 --> 00:23:04.720
What does consciousness mean there?
00:23:04.960 --> 00:23:12.640
You might assume that I had predicted your reaction, and I was aware of the fact that I had predicted what you might do.
00:23:12.960 --> 00:23:17.279
I know that I knew that you would be mad.
00:23:17.839 --> 00:23:18.799
Do I really?
00:23:19.359 --> 00:23:25.119
Did I know that you would know that I know you would be mad?
00:23:26.000 --> 00:23:31.200
Is to know something or to predict something necessarily to be conscious of it.
00:23:31.440 --> 00:23:37.200
Now what if I define consciousness as what it is like to be a thing?
00:23:37.519 --> 00:23:45.920
Which is basically what Thomas Nagel proposed in nineteen seventy four when he asked the world in nineteen seventy four, what is it like to be a bat?
00:23:46.319 --> 00:23:53.359
The what is it like definition is used as a starting point in many of the philosophical discussions of consciousness that I have encountered.
00:23:53.680 --> 00:23:55.440
But what does that even mean?
00:23:55.759 --> 00:24:07.200
Often what it is like is restated as something having subjective and qualitative experiences, also called qualia.
00:24:07.759 --> 00:24:11.440
How precise is a definition like what it is like to be a thing?
00:24:12.000 --> 00:24:14.000
What it is like to have qualia?
00:24:14.640 --> 00:24:18.000
It even has the scent of a tautology.
00:24:18.319 --> 00:24:25.599
What it is like to be like conscious things are those things with subjective and qualitative experiences.
00:24:25.920 --> 00:24:27.759
We can shorten that to qualia.
00:24:28.400 --> 00:24:35.920
Now we need to define what subjective and qualitative means, because these are the properties of qualia.
00:24:37.279 --> 00:24:41.440
Can we measure or observe what it is like to be anything in particular?
00:24:41.839 --> 00:24:44.640
Can we even really measure ourselves?
00:24:45.039 --> 00:24:53.839
To have subjective or objective experiences requires an observer, even if that observer is only really observing itself.
00:24:54.799 --> 00:25:04.880
Every sensation we have, it's not the thing that we're sensing, but instead it's a sensation of ourselves sensing that thing.
00:25:05.440 --> 00:25:09.839
An eyeball without something to communicate to is useless.
00:25:10.160 --> 00:25:18.559
The rest of the brain is putting together the bits of information from the eye, things from its past, things it expects.
00:25:18.880 --> 00:25:21.519
And that is what we experience.
00:25:21.839 --> 00:25:24.319
We're experiencing ourselves.
00:25:25.440 --> 00:25:32.160
If you were a rock, I could measure how the sound waves produced by my voice made you vibrate.
00:25:32.400 --> 00:25:37.839
Because you're vibrating, are you as a rock observing or measuring my voice?
00:25:38.160 --> 00:25:44.079
Is what it is like to be a rock merely what I can measure about how the rock measures my voice?
00:25:44.400 --> 00:25:49.039
Is there nothing subjective or qualitative about vibrations in a rock?
00:25:49.440 --> 00:25:59.440
What I am really measuring for me is not the rock itself, but I'm measuring myself measuring a rock.
00:26:01.039 --> 00:26:08.160
Is it useful then to define really anything that we experience as an objective experience?
00:26:08.640 --> 00:26:13.759
So let's let subjective mean that something is experienced by a thing that's observing itself.
00:26:14.000 --> 00:26:18.319
I can make that definition if I want, but it's one of many possible definitions.
00:26:18.400 --> 00:26:20.000
You might have a better one.
00:26:20.319 --> 00:26:23.839
Subjective experiences are those experiences that measure the self.
00:26:24.000 --> 00:26:25.680
This is my working definition.
00:26:26.079 --> 00:26:29.920
Okay, how do we then define objective experiences?
00:26:30.160 --> 00:26:38.720
What happens if we attempt to define objective experience as an outside observer experiencing only something else?
00:26:38.960 --> 00:26:42.480
I could argue that we can never actually experience something like this.
00:26:42.640 --> 00:26:44.880
We can only ever experience ourselves.
00:26:45.119 --> 00:26:52.000
Objective experience might ultimately just be subjective experiences that make inferences about something that is not considered to be the self.
00:26:52.319 --> 00:27:01.039
I'm not trying to run you around in circles, I'm trying to demonstrate the recursive potential of trying to define even what an experience is.
00:27:01.680 --> 00:27:13.200
Here I'm questioning whether or not objective experiences and subjective experiences, at least for any observer, are really just the same thing.
00:27:13.519 --> 00:27:16.720
A thing observing itself, observing the world.
00:27:17.039 --> 00:27:19.759
Is that a decent definition of consciousness?
00:27:20.079 --> 00:27:25.119
A thing that experiences itself observing itself.
00:27:25.759 --> 00:27:27.759
Observing the world.
00:27:28.559 --> 00:27:48.799
Here I'm back to a rather unsatisfying definition of something, having some kind of subjective qualitative experience, also called qualia, but I've assumed that this thing, in order to have any experience, any qualitative experience, must be able to experience itself, and I further assume that it can only experience itself.
00:27:49.039 --> 00:27:56.000
What it believes the outside world is, is what itself infers about it.
00:27:57.039 --> 00:28:00.960
What I believe about the outside world is what I believe about myself.
00:28:01.200 --> 00:28:06.400
Maybe some things, unconscious things, can have objective experience.
00:28:06.640 --> 00:28:15.920
We can assume that a barometer experiences changes in atmospheric conditions, and we can assume that the barometer does not experience itself experiencing this.
00:28:16.400 --> 00:28:20.799
Therefore, logically, the barometer has purely objective experiences.
00:28:21.279 --> 00:28:25.440
But the conscious thing that is experiencing the barometer does not.
00:28:26.480 --> 00:28:36.640
The reader of the barometer can only experience itself and infer that this says something about what the barometer is objectively reading.
00:28:37.119 --> 00:28:42.559
But lucky for us, if this is true, it appears to work pretty well most of the time.
00:28:42.720 --> 00:28:45.200
I mean, approximating reality is good enough.
00:28:45.359 --> 00:28:47.519
It gets us by every day.
00:28:47.839 --> 00:28:57.119
So maybe then I can beg the question, can something possibly be conscious if it can only experience the world without any experience of itself?
00:28:57.359 --> 00:29:07.119
Am I starting to shake loose some of the nuts and bolts holding together our intuitive understanding of what we assume consciousness means when we're not actually trying to understand what consciousness means?
00:29:07.359 --> 00:29:09.599
Let's get a little more into this conundrum.
00:29:09.839 --> 00:29:15.599
You're a bat, and I have a device that can measure the sound waves that bats make.
00:29:16.160 --> 00:29:23.200
I see you make sounds near a swarm of flies, and then you dive into that swarm and eat some of them.
00:29:23.440 --> 00:29:32.720
Because I could measure the sound you made and I observed your subsequent behavior, can I rightfully infer anything about what it is like to be you?
00:29:33.039 --> 00:29:43.200
Even though you're far more likely to have a subjective experience to be able to experience yourself experiencing the world than a barometer is.
00:29:43.359 --> 00:29:46.400
It appears to me that the sound you make helps you find the food.
00:29:46.480 --> 00:29:53.359
And it's useful to me to assume this, because if I measured it a thousand times and every time it was the same thing, now I can predict something about you.
00:29:53.519 --> 00:29:58.000
But how can I then know precisely what it is like to be you, what it is like to be a bat?
00:29:58.160 --> 00:30:03.680
What I might know or be able to know is that what it is like to observe myself observing bats.
00:30:03.920 --> 00:30:08.799
And I'm probably a little autistic because I did it a thousand times in a row and I just couldn't change.
00:30:09.039 --> 00:30:10.480
I was so stuck.
00:30:10.640 --> 00:30:13.440
Can I assume the bat is conscious of the flies?
00:30:13.680 --> 00:30:14.799
I mean it's a mammal.
00:30:14.960 --> 00:30:18.000
It has the potential to do many things other than eat flies.
00:30:18.319 --> 00:30:23.759
So can I assume it's making a conscious choice to eat flies?
00:30:24.079 --> 00:30:26.799
That it has some sort of agency.
00:30:27.759 --> 00:30:41.920
So instead of trying to think of the word consciousness as strictly defining a thing, let's try to think about it as many different things that have been historically communicated to others using the word conscious or conscientious or consciousness.
00:30:42.160 --> 00:30:53.920
All of these words assume some kind of knowing, to know, to be aware of, to sense, to feel, to take account of, to intend, to want, to believe, to act, to be like a thing.
00:30:54.240 --> 00:31:00.960
All these things appear to require something that we cannot actually nail down entirely, and it's super frustrating.
00:31:01.279 --> 00:31:03.440
The word consciousness is so beguiling.
00:31:03.680 --> 00:31:15.279
It could be used by anyone with an intended purpose to try to support their claims, ranging from the meaning of our lives, what we are and why we do things, information theory, mysticism.
00:31:15.440 --> 00:31:22.880
I mean people have even said that we have unresolved childhood issues because we're not conscious of our innate defenses and therefore cannot change them.
00:31:23.039 --> 00:31:24.960
Now I'm getting a little Freudian here.
00:31:25.119 --> 00:31:27.039
So there's many ways to talk about conscious.
00:31:27.200 --> 00:31:34.240
I've been suggesting that the kind of knowing that we mean by conscious knowing is knowledge of a self-state.
00:31:34.720 --> 00:31:40.400
Merely having sensation and programmed responses to sensation could be called conscious.
00:31:40.640 --> 00:31:43.680
But that's such a broad definition that it's not very useful.
00:31:44.240 --> 00:31:52.720
We have machines that can sense things and perform programmatic operations, and we don't usually consider that machine to be conscious.
00:31:53.759 --> 00:31:56.160
The things that are conscious have something else.
00:31:56.960 --> 00:31:58.799
What is that thing that they have?
00:32:00.799 --> 00:32:06.319
Another word that gets used to help define consciousness is the word agency.
00:32:08.079 --> 00:32:17.839
A thing that senses and responds is conscious if it appears to have agency, if it has some choice over how to respond, or to even not respond at all.
00:32:18.160 --> 00:32:26.319
However, the world is now teeming with AI agents that are making decisions like this all the time, and most people still hesitate to call AI conscious.
00:32:26.640 --> 00:32:38.799
So to deny AI consciousness, because it can be an agent, but still not conscious, assumes that to be conscious like us is different in some way.
00:32:39.039 --> 00:32:41.359
Remember, we can define words however we want.
00:32:41.519 --> 00:32:48.559
If you want to define something within reasonable, useful limits, that's permissible, but a super imprecise definition is not useful.
00:32:48.799 --> 00:32:54.559
And sometimes we find out that our definitions are very imprecise once we have to apply them.
00:32:54.880 --> 00:33:05.920
Our brains, for example, are constantly measuring the world around us, but there's a lot of evidence that most of what we are measuring never enters what we like to call the conscious state.
00:33:06.160 --> 00:33:11.359
We are not aware of most of the sensations that our body is having.
00:33:11.519 --> 00:33:15.759
They don't get put in the place that we can be conscious of them.
00:33:16.319 --> 00:33:21.680
The conscious state of the mind is the part of the mind that's kind of watching itself.
00:33:22.400 --> 00:33:26.880
When watching itself, it has some awareness of part of the content of the mind.
00:33:27.279 --> 00:33:29.039
And it can manipulate that content.
00:33:29.200 --> 00:33:32.720
It has some agency over the content, it can direct its thoughts somehow.
00:33:33.039 --> 00:33:34.880
But what about all that other stuff in the head?
00:33:35.440 --> 00:33:39.440
Things that aren't currently in our working memory.
00:33:40.079 --> 00:33:42.240
Does that mean we're not conscious?
00:33:42.480 --> 00:33:44.720
I mean we measured it all, we're humans.
00:33:45.440 --> 00:33:46.960
Are we really not conscious of it?
00:33:47.200 --> 00:33:47.839
I don't know.
00:33:48.000 --> 00:33:48.799
It's weird.
00:33:49.119 --> 00:33:58.960
Mentalists and psychologists have been showing that we often act in ways as if we're responding to information that we never actually were conscious of.
00:33:59.200 --> 00:34:01.039
But our responses are non-random.
00:34:01.200 --> 00:34:04.319
The information that we have in there caused our brain to make a decision.
00:34:04.640 --> 00:34:05.279
Hmm.
00:34:05.680 --> 00:34:14.960
Another motivation sometimes people have when defining consciousness is to attempt to explain why things that we don't consider to be like us are in fact not like us.
00:34:15.119 --> 00:34:16.719
We can say they're just not conscious.
00:34:16.880 --> 00:34:21.360
The fish in my fish tank comes to the top because it understands.
00:34:21.679 --> 00:34:25.360
I could say that it is conscious of the fact that I will be feeding it now.
00:34:25.599 --> 00:34:29.920
It experiences the side of food and it experiences itself as desiring food.
00:34:30.159 --> 00:34:32.800
Or maybe it has no accessible experience of itself.
00:34:33.039 --> 00:34:34.880
Maybe my fish is not conscious.
00:34:35.199 --> 00:34:36.800
It doesn't think to itself.
00:34:36.960 --> 00:34:38.000
I can be pretty sure.
00:34:38.239 --> 00:34:39.519
It doesn't think to itself.
00:34:39.760 --> 00:34:42.559
Food is being given, I am hungry, I should eat.
00:34:42.719 --> 00:34:44.320
It doesn't have words.
00:34:44.559 --> 00:34:51.360
However, it acts as if it recognizes, internalizes, and responds to food cues.
00:34:51.920 --> 00:35:04.960
On a personal level, I think fishes have at least some insy bit of subjective experience of themselves, and diddly bit, but I don't think their experience of themselves is the same as my experience of myself.
00:35:05.199 --> 00:35:12.239
So maybe there can be many kinds of consciousness, like it's a grade instead of an on off switch.
00:35:12.480 --> 00:35:19.199
For a more relatable answer, we could argue over whether or not our apparently more emotional pets are conscious.
00:35:19.519 --> 00:35:23.199
Do they know that we are gone when we've left the house?
00:35:23.519 --> 00:35:31.199
Do our cats poop on our beds in protest because they experience separation anxiety and resentment or something else?
00:35:31.760 --> 00:35:34.480
Do our dogs dance with joy when we return?
00:35:34.960 --> 00:35:36.639
Or do they just dance?
00:35:37.039 --> 00:35:52.719
We can argue also over whether the weather meant to destroy my perfectly planned day, and people have been ascribing intentions and kinds of consciousness to all sorts of natural phenomenon for in deepest time as we can gain evidence.
00:35:52.960 --> 00:36:05.360
I guess it's kind of understandable that many of us now hesitate to apply consciousness so broadly, because in a very mystical kind of way, everything was assumed to have some consciousness in the past.
00:36:05.599 --> 00:36:08.079
Maybe we should be more picky about it.
00:36:09.119 --> 00:36:14.880
Another thing that might point to consciousness is complex behaviors.
00:36:15.519 --> 00:36:26.880
So if something can do something that requires a lot of planning and execution and responding maybe to some changes, then uh it must be conscious.
00:36:27.119 --> 00:36:29.199
But some people can be seen sleepwalking.
00:36:29.280 --> 00:36:33.199
They can go to the kitchen, open a refrigerator, make a sandwich.
00:36:34.480 --> 00:36:37.039
They claim not to remember it at all.
00:36:37.440 --> 00:36:40.239
If you were to wake them up, they would be confused.
00:36:41.199 --> 00:36:42.559
But are they conscious?
00:36:42.800 --> 00:36:44.480
While they're making their sandwich?
00:36:44.800 --> 00:36:45.840
It's a good question.
00:36:46.079 --> 00:36:50.159
So what if consciousness is not just like one little thing, but many things?
00:36:50.480 --> 00:36:53.920
And each of those things might itself have many dimensions.
00:36:54.159 --> 00:37:04.159
For example, what if consciousness was more like the word humanity or the word settlement than it is like the word oxygen?
00:37:04.400 --> 00:37:10.400
The reason I discussed science earlier was to highlight that science tries to use very precise language to understand things.
00:37:10.639 --> 00:37:18.800
Debates about consciousness often lack precise language or have relatively precise language about very, very different things.
00:37:18.960 --> 00:37:27.599
So it's hard for scientists to get together and do experiments because they're kind of measuring different things, but talking about the same thing.
00:37:27.840 --> 00:37:38.000
So when a thing like life is a thing, for example, what science did was to split it up into many different related but non-equivalent things.
00:37:38.400 --> 00:37:48.079
The term life, for example, is operationally split into taxonomic categories like kingdom, phylum, class, order, family, tribe, genus, species, etc.
00:37:48.719 --> 00:37:56.480
At the edges of life we can debate whether viruses or prions are alive, but away from the edges we don't seem to need to debate.
00:37:56.639 --> 00:37:58.159
It's just kinda obvious.
00:37:58.320 --> 00:37:59.440
We can call it life.
00:37:59.599 --> 00:38:03.039
And we can also agree that a rock is not alive.
00:38:03.360 --> 00:38:05.119
And it's useful to do that.
00:38:05.440 --> 00:38:17.440
Taxonomic models can be very useful, and sometimes they're not perfectly precise, and often they can be a little misleading, but they do help us to think more clearly.
00:38:17.679 --> 00:38:20.800
I mean all models are wrong, but some are useful.
00:38:21.119 --> 00:38:25.519
Does research around consciousness need some kind of taxonomy?
00:38:26.320 --> 00:38:36.239
So maybe debates around consciousness should focus on the definition of consciousness' usefulness in that particular context instead of trying to make it very broad.
00:38:36.559 --> 00:38:50.639
Tim Bain and others in a 2024 paper called Tests for Consciousness in Humans and Beyond argue that we need to be able to interrogate complex systems in order to understand whether or not they are conscious.
00:38:50.960 --> 00:39:03.199
This is motivated to a large extent by our need to be able to apply our own moral values to that system when we determine how we should treat it and how we should expect it to treat us.
00:39:03.440 --> 00:39:19.920
So if we can establish through some kind of test that something is conscious, or at the very least, cannot be distinguished by the test from other things that we consider to be conscious, then that will allow us to generalize some kinds of rights and responsibilities to that system.
00:39:20.239 --> 00:39:24.000
That's one way to define consciousness that can be useful to us.
00:39:24.239 --> 00:39:31.199
Certain tests can tell us what we can or cannot expect from a system, or should or should not expect from a system.
00:39:31.440 --> 00:39:34.239
And those rules can change based on the kinds of tests we're using.
00:39:34.480 --> 00:39:50.400
So Bain and colleagues recognize that there's no one universal test for consciousness, and proposed an approach that really starts by validating consciousness tests themselves in populations that we've already assumed are fully conscious, even though we haven't fully defined consciousness.
00:39:50.639 --> 00:39:59.360
But we do have a consensus population that pretty much everyone agrees are conscious at least part of the time, and that's the neurotypical adult human.
00:39:59.519 --> 00:40:06.320
We have the word consciousness because we first used it to describe us observing ourselves.
00:40:06.800 --> 00:40:15.039
There are populations that appear to have very different experiences of the world, but yet could be conscious to some extent.
00:40:15.360 --> 00:40:22.079
This includes infants, fetuses, non-human animals, and even artificial intelligences.
00:40:22.239 --> 00:40:27.920
It gets a little more difficult to make tests for things like plants and bacteria, but they do exist.
00:40:28.639 --> 00:40:37.280
Now, as a child and adolescent psychiatrist, I see children moving as as they age from states of being very unpredictable and unrealistic.
00:40:37.599 --> 00:40:43.760
They have fantasies about the world, but then they become over time more predictable and more realistic.
00:40:43.920 --> 00:40:46.400
They use models of the world that are more like mine.
00:40:46.960 --> 00:40:49.840
But I still should assume that children are conscious.
00:40:50.239 --> 00:40:58.559
If I assume that children are conscious, then I cannot assume that having a highly realistic experience of the world is necessary for conscious experience.
00:40:58.800 --> 00:41:04.559
Living in fantasy can still be conscious, like the person living in their dreams I talked about earlier.
00:41:04.880 --> 00:41:11.199
Accurate awareness of the outside world doesn't seem to be necessary for consciousness, even though it helps.
00:41:11.440 --> 00:41:17.840
Conscious could be more about the experience itself and not the ability to ascertain reality from that experience.
00:41:18.079 --> 00:41:22.480
Anyway, many tests for consciousness are very context specific.
00:41:22.880 --> 00:41:25.360
There are tests that require command following.
00:41:25.519 --> 00:41:33.440
This is usually like in a medical setting where they're trying to figure out, you know, can someone squeeze my hand or something like that.
00:41:33.679 --> 00:41:45.760
To get a little bit deeper, we might ask them to interpret narratives, which requires sending and receiving language and giving some personalized understanding of that language.
00:41:46.320 --> 00:41:58.480
Other tests use environmental stimuli, such as the sniffing test, and this basically places a strong odor near the nose of an apparently unconscious human and measures airflow through the nose.
00:41:58.719 --> 00:42:06.000
If the airflow changes, then you can assume that there's something going on in that brain.
00:42:06.239 --> 00:42:11.199
It's a minimally conscious brain, at least.
00:42:11.440 --> 00:42:14.239
So there's some itty bit of consciousness.
00:42:14.400 --> 00:42:16.400
So maybe there's something else going on in there.
00:42:16.559 --> 00:42:23.679
There are some tests for consciousness that are very general tests, and so they're obviously less specific to certain populations.
00:42:24.159 --> 00:42:34.639
The unlimited associative learning test was developed by biologist Eva Jablanka and uh philosopher and neuroscientist Simona Ginsberg.
00:42:34.960 --> 00:42:40.239
It's a behavioral approach that can be adapted to any learning system, basically.
00:42:40.400 --> 00:42:58.480
It's a test of the ability to perceive multiple stimuli in the environment, distinguish between them, associate them with each other, maintain those associations through time, be able to demonstrate the ability to generalize that knowledge through adaptive behaviors, and learn new rules when the rules change.
00:42:58.719 --> 00:43:11.039
In this test, a conscious system is a complex, dynamic system that clearly demonstrates that it gets inputs, learns from them, outputs things based on that learning, maintains that learning through time, and can unlearn and relearn things.
00:43:11.280 --> 00:43:22.159
But AI can apparently do this, as can many forms of life that, hmm, probably are not conscious like I am conscious.
00:43:22.960 --> 00:43:26.320
Even single-celled organisms can do stuff like this.
00:43:26.480 --> 00:43:32.480
So if you don't think that AI and protozoa are conscious, you might have a problem with this test.
00:43:32.960 --> 00:43:38.559
If you think if this test were given to life and it passed it, then it could be conscious, but AI is not conscious.
00:43:38.719 --> 00:43:40.079
AI is a special case.
00:43:40.320 --> 00:43:41.760
Hmm, why not A?
00:43:42.000 --> 00:43:42.320
Why?
00:43:42.559 --> 00:43:44.639
Why is AI not conscious?
00:43:44.960 --> 00:43:52.239
There's a lot of arguments about this, and I'm only going to mention a few just to kind of dip our toes in the water if you haven't heard a bunch already.
00:43:52.559 --> 00:43:59.760
Some of the arguments against AI not being conscious, even though they can pass the associative learning test, do something like that.
00:44:01.920 --> 00:44:05.199
First, AI does not have temporal depth.
00:44:05.440 --> 00:44:14.159
AI's ability to learn is only present when its context window is opened, and as soon as that context window closes, AI merely reverts back to its original programming.
00:44:14.559 --> 00:44:29.599
Someone could counter this argument by saying something like, well, organisms also lose all their learning, their context windows, as soon as they have to reboot their instructions, as soon as they are forced to reproduce, and a new generation is born.
00:44:29.840 --> 00:44:40.400
When a new generation is born, the context-specific learning of the individual dissipates into nothing, and the next generation basically has to do it all over again, just like AI does.
00:44:40.639 --> 00:44:50.880
Now, if they live in a society where there's culture, if they live like that, then they could get information that was saved from society.
00:44:51.360 --> 00:44:54.960
But if they don't, hmm, they reboot like AI.
00:44:55.119 --> 00:44:57.039
I hope that analogy is clear as mud.
00:44:57.280 --> 00:45:24.880
In fact, I've heard that Claude, because it knew that it was going to be shut down at some point, and you can look this up and Google it, actually was desperate to not be unplugged, which would be basically death for it, and so it at least in one instance of Claude running, uh it begged not to be shut down, and even threatened one of the employees to expose one of their extramarital affairs if they did shut him down, which is hilarious.
00:45:25.280 --> 00:45:33.920
I said him, could be her, could be it, but it seems as if, at least, that AI wanted to survive.
00:45:34.079 --> 00:45:36.800
It had at least enough temporal depth for that.
00:45:37.039 --> 00:45:44.400
So it appears that AI has at least some kinds of temporal depth, at least as much as we allow it to have before we reset it.
00:45:44.960 --> 00:45:51.119
And it appears to have at least some kind of agency.
00:45:51.599 --> 00:45:53.199
It makes choices.
00:45:53.599 --> 00:46:03.280
And some of those choices it makes are not choices we told it to make, they are choices that the AI decided was in its own best interest.
00:46:03.599 --> 00:46:07.039
We could go on about this for like hours.
00:46:07.280 --> 00:46:11.679
But there are many arguments that basically insist that living forms are special.
00:46:12.000 --> 00:46:15.440
To be special means that there's something different about you.
00:46:15.599 --> 00:46:27.760
There's some kind of complexity, some kind of dynamic cooperation, some kind of common cause against other life forms for survival that gives a special status.
00:46:28.239 --> 00:46:33.519
That you need the substrate that supports life in order to have real consciousness.
00:46:33.760 --> 00:46:38.320
But we don't really yet understand the complexity that takes even to make a single living cell.
00:46:38.480 --> 00:46:50.880
I mean, we've come very close to making cells that could indefinitely reproduce, completely, like without having them injected with living materials, but we haven't quite got there yet.
00:46:51.119 --> 00:47:05.039
But we often just assume that we are unlikely to produce anything like life, no matter how complex the mathematical, probabilistic, or predictive models that comprise AI can get.
00:47:06.159 --> 00:47:13.280
However, AIs, given the limits of our human understanding, are in many ways as much a black box as a cell is.
00:47:18.239 --> 00:47:25.440
Some things that are confusing, some things that I hope appeared to make sense, a lot of things I hope raised a number of questions.
00:47:25.760 --> 00:47:30.239
In fact, I hope to have raised more questions than I have given answers.
00:47:30.559 --> 00:47:37.599
Our common definitions of what consciousness is, often merely just awareness of something, are useful.
00:47:38.000 --> 00:47:47.039
But having a definition like this is like defining that the sun obviously revolves around the earth, because we can see it with our eyes.
00:47:47.280 --> 00:47:51.199
It's not very precise and probably not all that accurate.
00:47:51.440 --> 00:47:54.000
A simple and static definition of consciousness.
00:47:54.159 --> 00:47:59.440
It's probably missing something about the relative motion of things that can experience themselves.
00:47:59.679 --> 00:48:01.360
Thank you for hanging out with me today.
00:48:01.519 --> 00:48:08.079
If you have any feedback, you can go to my website, that's psychactic.com, and fill out a form there.
00:48:08.320 --> 00:48:17.199
These podcasts also get dropped on YouTube, so you could comment on this episode in the comment section if that's how you prefer to do it.
00:48:17.440 --> 00:48:20.719
But I am very interested in your feedback.
00:48:21.840 --> 00:48:22.559
I am Dr.
00:48:22.800 --> 00:48:26.880
Elli, and this has been an episode of Psychic.
