SGU Episode 123
Transcriber's notes: This is a pretty liberal transcription. I leave out most "you know" and "I think", and half-formed sentence fragments. I take liberties to convert actual utterances into grammatical sentences.
Introduction
S: Hello and welcome to the Skeptic's guide to the universe. Today is Wednesday November 28th 2007, and this is your host, Steven Novella, president of the New England Skeptical Society. Joining me this evening are Bob Novella
B: Hello everybody
S: Rebecca Watson
R: Hello everyone
S: Jay Novella
J: Hey guys
S: And evan bernstein
E: Good evening, everybody
S: How is everyone tonight?
? Good, Steve
R: Couldn't be better
? Very good but apparently, you're not good, Steve?
S: Oh, we'll get to that a bit later. (laughter) Jay's referring to a recent Skeptiko podcast which talks about the Skeptic's Guide specifically. We'll be getting into that in the beginning of the email section of our show. But first let's do some news items.
Science and faith
S: Several of our listeners referred us to this New York Times editorial by Paul Davies - this past Saturday's edition of the New york Times - in which he claims that science is based upon faith. Have you guys had a chance to read this?
Yes
S: This is a claim that we hear frequently. David, for example, writes "the problem with this neat separation into non-overlapping magicisteria – as Steve Jay Gould describes science and religion – is that science has its own faith-based belief system. All science proceeds on the assumption that nature is ordered in a rational and intelligible way. You couldn't be a scientist if you thought the universe was a meaningless jumble of odds and ends juxtaposed. When physicists probe to a deeper level of subatomic structure or astronomers extend the reach of their instruments, they expect to encounter additional elegant mathematical order. And so far, this faith has been justified. So, you know, this is a claim we hear frequently and I think Davies is making the really common mistake of confusing methodological naturalism with philosophical naturalism. What he's saying is that science assumes that the laws of the universe are stable, and that they make sense. And he says that science requires faith in that. And that is absolutely not correct - that is a complete misunderstanding of science. Science doesn't really require anything, because science is just a system of methodology. It assumes methodological naturalism, the idea that effects have causes, that the system internally functions together and makes sense - the system meaning nature - because it has to assume that. It takes that as a premise only because the methods of science only work within that framework. So it's actually not an assumption about reality, it's not faith in any particular metaphysica ultimatel reality; it's just saying "these are the methods that work, and therefore these are the methods that science is going to use", because it's the only ones in which you can proceed with empirical hypothesis testing. It actually is agnostic towards the ultimate metaphysical realities of the universe. So his entire premise is false.
? So, Steve, would you say that the following statement is wrong: "I have faith in the scientific method"
S: Well, it depends what you mean by that. I think we use the term "faith" differently. If that means that it has worked so far and therefore I think it's highly probable that it will continue to work in the future, then I think that that's a legitimate statement. But if you're saying that you believe something that's a choice without justification, then I think that that doesn't apply - the term "faith" doesn't apply.
?: Ok, because I say that, "I have faith in the scientific method" because from my perspective I'm saying that I'm banking on the proof upon proof that science has delivered over the years...
S: Right, so we hear this a lot from the intelligent design crowd and I'm sure they love this kind of editorials, because this is their mantra: the notion that you have to have faith in science or faith in evolution, and that they've been complaining endlessly. And this is Phillip Johnson who basically started the Intelligent Design movement, this was his core premise: that science should not be based upon the assumption of naturalism, because that's rigging the game. That's rigging the game against supernatural or spiritual explanations. And they're continuing to make that case. In fact, in preparation for our show tonight I was listening to an episode of Skeptico - the podcast Skeptico - from a few episodes ago where he interviewed an Intelligent Design proponent, and that's what it was all about. It was all about "scientists are assuming philosophical naturalism and they're not following the evidence where it goes, they're only restricting their enquiry to naturalistic explanations and that's not fair, that's rigging the game". What that misses is that methodological naturalism is not a choice, it's a necessity. Without limiting the answers that we're willing to consider to the ones that fit our paradigm we're limiting the questions to ones that can be answered scientifically. If you can't formulate your hypothesis in a way that can be tested, it can be falsified, then it doesn't meet the minimum criteria for being considered as science. They totally do not get that, and that's true at the spiritual end, like the Intelligent Design proponents, and it's true on the New Age end, like Alex from Skeptico. Because they were in complete agreement on this point, that skeptics and scientists are feeding into their own assumption of philosophical naturalism and it's completely untrue.
J: What I don't understand is they're going back to what Carl Sagan said so eloquently: science delivers the goods. Science in and of itself is a system that has been proven over and over and over again to work.
S: That's a good point, Jay, and I often refer to that as the meta-experiment of science. If methodological naturalism didn't work because our universe was hopelessly not rational, or not naturalistic...
?: Acausal
S: ...it was acausal or retro-causal, or the rules, the laws of the universe changed frequently or could be suspended at random, or by the whim of some being. If these things were true, or if our universe were part of a larger universe that we could not access but that influenced our universe - whatever - if any of those situations, then science wouldn't work very well. We would be constantly running up against enduring anomalies that we could never resolve, we couldn't make sense out of. Things that we thought were well-established would be overturned chaotically and at random, and that's just not the case: science has been working quite well over the last few hundred years: slowly, methodically building an ever-improving model that is very, very powerful in its ability to predict the future, to predict what's going to happen. That is the only criterion by which science is really judged - how well does it predict the outcome of future observations. That doesn't prove philosophical naturalism just like you can't really "prove" anything in science, right? Nothing is proven metaphysically in science. All we can say is that so far, all the evidence is pointing in that direction.
Computer Brain
S: Bob, you sent me the next item on a new computer brain. Why don't you tell us about that?
B: Yeah, pretty interesting. One of the holy grails of neuroscientists, I think, since its inception is the creation of a simulation of a human brain. Of course with the advent of computers it's obvious that the best way to embody that simulation would be in a digital computer. Scientists in Switzerland working with IBM researchers have shown that a computer simulation of a part of a rat brain called the neo-cortical column, which is arguably one of the most complex parts of the mammalian brain, it appears to behave like its biological counterpart, which they're calling a pretty big milestone. Now, up until yesterday I'd never even heard of neo-cortical columns, not specifically. And apparently, they're the basic building blocks of the cortex - the outer part of the brain, specifically the neo-cortex, which is the most recently evolved outer folded part of the brain. They consist of, in humans, about 60,000 neurons and they're pretty small. They're about half a millimetre wide and about two millimetres long, so they're pretty tiny. But they're the fundamental functional unit of the brain, and they're extremely complex. They've figured that if you have a goal of duplicating or simulating the brain, that's the one thing you really need to nail, so that's what they've been trying to do. Also, another thing about the neo-cortical columns: apparently that was a milestone in human brain development about two hundred million years ago, Steve, when mammals split from reptiles and the neo cortex started to grow at that point.
S: Yeah, that's true, the reptillian brain is what in mammals you actually call the reptilian brain, that's the deep primitive brain structures. And reptiles have only a very minimal cortical ribbon on top of that. In mammals that became the bulk of the brain.
B: Right, I think it's about 80% of the brain, now, the neo-cortex is about 80% of the brain.
S: Yeah, and it's the more primitive, reptilian part of it. It functions for more automatic kinds of reflexes as well as basic emotions, things like that. But the thinking part of the brain is the cortex, is the mammalian-
B: The so-called higher cognitive functions. Well, it's these columns that have been multiplying for millions of years. And as they've multiplied, our brains became more and more sophisticated. And they're actually responsible for the folds you see in the brain because there's obviously selective pressure to have more of these columns in the cortex. In order to make room, they kind of just expanded any way they could, and that's actually why we have all the folds in the outer part of the brain, because it's the classic-
S: It's not just the volume, it's the surface area, because this vertical organisation – the bulk of the connections in the cortex run vertically, of course there's also horizontal cross-connections, but the primary processing unit, if you will, is this vertical column – so to get more vertical columns in, you need more surface area, and therefore you need more wrinkled folds of the brain, not just an expanding balloon, for example. So, this is really cool. I like [how] in the title it says "a computer simulation could eventually allow neuroscience to be carried in silico". I like that term "in silico".
B: Yeah
S: I've been thinking about this a lot, actually, because as we reverse-engineer the brain, we start to actually see that this piece of the brain is doing this, and connects to this other piece which has this other specific function: this is how they interact. You know, we're definitely going to be moving in the direction where this line of technological development – actually computer modelling the brain – is then going to start working alongside of the neuroscience which is modelling the pieces of the brain. So, imagine in five, ten, twenty, thirty years, when these computer models are actually not just duplicating the raw structure of the brain, but actually in greater and greater detail, and we can actually test our models of how the brain works, by simulating them in a computer. And then do things like say "well what happens if we turn off this nucleus or this part of the brain" and then see what the net result is. I think that these two parallel lines of research are going to play off each other in very, very interesting ways, over the next couple of decades.
B: Now, Henry Markram, the co-director of the Brain-Mind Institute at the Ecole Polytechnique Fédérale de Lausanne in Switzerland – I probably mangled that pronunciation – he said what you said: "what we're doing is reverse-engineering the brain".
S: Yeah
B: But he had another quote. He said "we're not trying to build a copy of the human brain or some magical, artificial intelligence device". I wonder why he threw, I don't like how he put "magical" in there like you need magic to get AI. "This is really a discovering of how the brain works" is how he sees it. And I agree with him, that this is what they're doing, but I don't think he can ignore the potential applications because they're at the point right now, this milestone they've passed, where the output of this simulation matches what they're seeing biologically. Now they're taking the next step and they want to start actually – extend it beyond just this one neo-cortical column and extend it into many others, and eventually the whole brain of at least the mice that they're trying to simulate here. And eventually after that, a human, obviously. But their time scale is three years for a rat brain, and actually, ten years for a human brain. But of course, some scientists scoff at that idea. They think ten years is way too soon to even be talking about that.
S: We'll see
B: Well, we'll see, and if it's actually ten or twenty or thirty years, eventually, I don't think it's inherently impossible, that we won't have this type of simulation. And imagine what can come from that. He's kind of poo-pooing that, trying to stay away from the whole AI thing, and the whole "copy the human brain" [thing] but I think definitely those are extensions of his research that will inevitably come.
S: Yeah, I agree, I think there's no question about that.
J: So, when you say they make a computer model of the brain. The first thing I think of is, is it possible that if they make a complex computer model and if it's complete enough, could it obtain any kind of consciousness, just on its own? And then you just brought the whole AI thing, but you would imagine that if they could simulate the functionality of the human brain, why wouldn't it become conscious?
B: I agree. I think that if they have a sufficient level of detail and I'm sure many other factors. I mean, I don't think neurons are some magical substrate that allows consciousness and nothing else can. I think that if you have many other different types of substrates, as long as they're connected properly and lots of other things, I think there's no reason why you can't have this in software.
S: Right. The interesting thing will be, in whatever timeframe – twenty, thirty, forty years – we have powerful enough computers and we have adequately modelled the brain. We could create a virtual human brain in silico - or whatever computers are made of. And the result of that behaves as if it's conscious and aware and artifically intelligent. Of course, I predict that will lead to the philosophical debate about whether or not it's really conscious and self-aware, or whether it just acts like it does. I don't know how we would be able to resolve that. That's interesting in terms of the dualism versus materialism debate. Of course the dualists think that the physical brain is not adequate to explain the phenomenon of mind, of consciousness. And I maintain, a lot of neuroscientists maintain that the materialist model's doing just fine, thank you. And we do not need dualism to explain any neurological phenomena. But if the dualists are right, we should run into problems. If we make a purely materialistic model of the brain, that should not result in something that acts like consciousness. There should be some mysteries, something spiritual, something missing.
B? Missing, right, a missing ingredient
S: Missing from our attempt to simulate the brain. I'm sure, you know, the dualists will have some rationalisation if and when that occurs. And if it does turn out that there is something missing, then we'll definitely have to reconsider our materialistic assumptions.
?: Steve, does the human brain have software-like programming?
S: Sure, I mean it's more of an analogue type of programming in the way our neurons are connected together. Sure.
R: My brain runs Unix.
?: A child is born, and genetically [17:18]