A Free Will Challenge

[Fred H.]

Todd: If you should want to pursue this, Gary Marcus has an excellent book on the differences between rule-based and connectionist models of the mind which would help you see why "strong AI" is a strawman argument.

“Straw man argument?” Come on Todd. From Wikipedia:

Connectionism is an approach in the fields of artificial intelligence, cognitive science, neuroscience, psychology and philosophy of mind. Connectionism models mental or behavioral phenomena as the emergent processes of interconnected networks of simple units. There are many different forms of connectionism, but the most common forms utilize neural network models.

Regarding “emergence,” again from Wikipedia:

Like intelligence in AI, or agents in DAI, it [emergence] is a central concept in complex systems yet is hard to define and very controversial. There is no scientific consensus about what weak and strong forms of emergence are, or about how much emergence should be relied upon as an explanation in general. It seems impossible to unambiguously decide whether a phenomenon should be considered emergent.

Further, "emergent" is not always a deeply explanatory label even when it is agreed on: the more complex the phenomenon is, the more intricate are the underlying processes, and the less effective the word emergence is alone. In fact, calling a phenomenon emergent is sometimes used in lieu of a more meaningful explanation.

So basically your “connectionist models” are “simple units,” that are essentially algorithmic (what else is there?) interconnecting to make larger and large interconnected networks, when at some point, inexplicably, voilà !—consciousness magically “emerges.”

And for your reading enjoyment regarding any supposed “connectionism vs. computationalism” debate, here’s Wikipedia once again:

Connectionism and computationalism need not be at odds per se, but the debate as it was phrased in the late 1980s and early 1990s certainly led to opposition between the two approaches. However, throughout the debate some researchers have argued that connectionism and computationalism are fully compatible, though nothing like a consensus has ever been reached. The differences between the two approaches that are usually cited are the following:

· Computationalists posit symbolic models that do not resemble underlying brain structure at all, whereas connectionists engage in "low level" modeling, trying to ensure that their models resemble neurological structures.

· Computationalists generally focus on the structure of explicit symbols (mental models) and syntactical rules for their internal manipulation, whereas connectionists focus on learning from environmental stimuli and storing this information in a form of connections between neurons.

· Computationalists believe that internal mental activity consists of manipulation of explicit symbols, whereas connectionists believe that the manipulation of explicit symbols is a poor model of mental activity.

Though these differences do exist, they may not be necessary. For example, it is well known that connectionist models can actually implement symbol manipulation systems of the kind used in computationalist models. So, the differences might be a matter of the personal choices that some connectionist researchers make as opposed to anything fundamental to connectionism.

[James Brody]

Sez Todd about Fred: "That's why we find each other such unsatisfying intellectual conversation partners."

Sure coulda fooled me!

Nice going and thanks, both of you!!!!

JB

[ToddStark]

Hi Fred,

I'm not arguing with Wikipedia, I'm trying to figure out what *your* argument is!

Rather than continue to go deeper into technical definitions that I feel are only superficially relevant to the real argument, let's take a step back to concepts.

You seem to be arguing in essence that the mind is not a machine and so cog sci in any form cannot represent it properly.

I see the positions on this issue as falling roughly into these camps (I know I'm oversimplifying, but I think this makes some important distinctions):

1. People who don't think there can ever be a legitimate science of mind because it is too complex, or has properties that are outside of scientific explanation. Cartesian substance dualists and people who insist on souls and supernatural causation or philosophical "free will" (as incompatible with physical causation) fall into this camp. I doubt that one can be a scientist in the sense of a "causalist" scientist (one who thinks theories explain causal relations) and still fall into this category because there has never been any way of reconciling causal effects between separate substances of mind and body.

2. People who think there can be a legitimate science of mind but it must be in mental or intentional terms and cannot possibly be made consistent with or in terms of neural and cognitive models because special properties like subjective experience and free will are in principle beyond those models. Non-cognitive scientists like Roger Penrose seem to fall into this camp. Penrose calls himself a "non-computational functionalist." That is, I think he agrees with me that the mind can probably only be understood in terms of what it does, but speculates that *something* about its functions cannot be translated into terms of computation.

To me, this places Penrose in a similar camp with Chalmers and Searle, in sharing the committment that while we can talk about minds in cognitive terms, something basic will always be left over when we do. This is exactly where philosophers get caught up in talking about whether functional zombies who behave and respond exactly like us but lack a soul and are without "real experience" are possible. My third category thinks that the notion of functional zombies is a compelling illusion; the second category with Penrose, Searle, and Chalmers seems to believe that functional zombies are a real difficulty for philosophy of mind.

I think the second category can in principle make serious contributions to a science of mind (unlike camp 1) because they are at least causalists and are looking to enhance the causal models we use, but personally I suspect they are on the wrong general track and will mostly spawn disconnected speculations like Penrose's quantum microtubule effects.

3. People who think there can be a legitimate science of mind, but that it should best be made in terms of or at least consistent with neural and cognitive models or at least some model that sees a mind in terms of what it does. Here are the functionalists, who for the most part, are rooted in cognitive neuroscience and evolutionary biology because these are the best warranted current science.

The mainstream position of cognitive science, places those of us who are devotees (which includes almost all functionalists) firmly in the third camp. People like Dennett, Crick, Koch, Baars, and the Churchlands are some of the best known folks in this camp. I think they are the ones who are making the most progress right now in understanding how the brain works, and how it relates to the mind.

I want to recommend an interesting volume by Susan Blackmore, "Conversations about Consciousness." It has great interviews on this topic with Penrose, Dennett, and a number of other very bright folks who have thought a lot about this and worked in related fields for a long time. It gives a unique insight into the various theories of these people, I found.

kind regards,

Todd

[Fred H.]

Todd: Rather than continue to go deeper into technical definitions that I feel are only superficially relevant to the real argument, let's take a step back to concepts.

Indeed Todd, “only superficially relevant to the real argument,” although your claim that you’re “trying to figure out what [my] argument is ” strikes me as being less than earnest. Perhaps you missed my post # 37 in this thread:

The available evidence indicates that human consciousness—sentience, sapience, self-awareness—is indeed something real, something that does indeed exist; although it also seems to be beyond the precise explanation of any currently available science. More to the issue, the available evidence also indicates that we humans use our cognitive consciousness to discern objective mathematical truth, and that we then use that objective truth to understand, explain, and, to some extent manage, our physical world and ourselves.

Accordingly, the available evidence overwhelming supports the view that we humans do indeed have some sort of, and some amount of, free will (and also implies that we humans are probably the only creatures that do have it.)

IOW, the human mind is undoubtedly more than “computation,” or algorithms—i.e. your “machine,” whatever currently available science you might utilize to define “machine,” connectionism models and so-called “emergent” processes notwithstanding.

IOW, explaining the human mind—or more specifically consciousness and free will—with the currently available science is roughly as naive as explaining the atom using Newton’s laws of motion and gravity and/or using the solar system as a model.

[ToddStark]

Hi Fred,

I wouldn't waste my time with disingenuous posts. You pretty much get my thoughts as they are, with a little editing.

Here's why I disagree with the non-computation assumption ...

As I think almost any serious student of the human mind will probably agree, none of us knows for sure whether the human mind can be fully described in computational terms, and we choose to make different assumptions.

Several bright people have argued famously that they have very good and scientifically coherent reasons to feel that the mind is more than computation. My favorite examples are Ned Block and Roger Penrose. A computational mind makes sense to cognitive scientists, and most but not all functionalists. I think Penrose and Block are actually (mostly) functionalists, in the sense I use the term. However they both argue persuasively that the mind must be doing things that are beyond what we currently think of as the sort of computation that a computer does.

Eventually, if we persue computational and non-computational lines of modelling, we should be able to discern whether non-computational models are neccessary and useful. We will have more to go on than just our intuition that "there must be more to human consciousness than computation." That is a compelling argument, but not exactly conclusive. We have discovered many compelling illusions about the mind over the years.

I think it makes sense to assume a computational mind order to see how far that takes us, because we have a lot of really powerful tools for modelling things computationally, and as Penrose acknowledges, almost no tools to study non-computational things. Even if we agreed in principle that Godel Incompleteness is relevant to how we represent the mind, and we don't, it still isn't neccessarily relevant to how we can best study it right now in order to make scientific progress.

Personally, if I were a researcher I wouldn't see making my working hypothesis a theory that I cannot use to investigate the mind at this time, I would focus on the tools we have and assume that the "hard problems" of consciousness will either dissolve, as Dennett and the Churchlands believe, or will eventually clarify to become more accessible questions, as Penrose believes.

I think it probably makes sense for some people to persue the non-computational line as Penrose does. I also think it makes sense for people to persue computational models.

My guess is that the computational models will take us farther for quite a while, and eventually we will know with greater confidence whether there are non-computational elements that we need to study. We don't know that yet.

My particular form of compatibilism says that some sort of "free will" is true and that physical determinism is also true (physical events have physical causes). Sophisticated enough systems can have upward, downward, and sideways causal paths. I don't know if this requires non-computational processes or models, and I don't agree with Penrose that Godel Incompletness neccessarily forces the issue. However I do think it is a reasonable thing for him to suggest and persue in principle.

I just think it is perhaps a long way off until we will know if he is right or have any real non-computational models to use to study the mind.

I hope this helps,

Todd

[Fred H.]

Todd: My particular form of compatibilism says that some sort of "free will" is true and that physical determinism is also true (physical events have physical causes). Sophisticated enough systems can have upward, downward, and sideways causal paths. I don't know if this requires non-computational processes or models, and I don't agree with Penrose that Godel Incompletness neccessarily forces the issue. However I do think it is a reasonable thing for him to suggest and persue in principle.

Penrose clearly shows us that the truth of Godel’s incompleteness (and the Turing halting problem) confirms that human consciousness is the result of something more than algorithms—just as quantum behavior/process of protons and electrons, say in atoms, confirms that quantum processes are something more than the result of Newton’s laws of motion and gravity, something more than a miniature solar system. So it doesn’t matter whether you “agree” or not, one simply goes where the available evidence/truth takes us.

It does seem that we kind of agree that “some sort of free will is true,” although I find your “compatibilism”—“that some sort of "free will" is true and that physical determinism is also true”— to be little more than equivocation at best, or cognitive dissonance at worse. Regarding your view that “physical determinism is also true,” I’m inclined to agree except that human “free will” would somehow, ultimately, have to trump such determinism to some extent.

I really do have empathy for your agnosticism regarding some of these complex issues Todd, and indeed agnosticism may well be the judicious view. But I wish you’d be a little more upfront about it—besides irritating me, your equivocations cause people like Tom to mistakenly believe that that you, he, and the other “atheists” have actually reached some sort of meaningful consensus.

[TomJrzk]

Originally Posted by Fred H.

But I wish you’d be a little more upfront about it—besides irritating me, your equivocations cause people like Tom to mistakenly believe that that you, he, and the other “atheists” have actually reached some sort of meaningful consensus.

Ha, that's hilarious because I was saying the same thing to Todd; he's encouraging you to see a difference between his views and mine. As I've said, if his 'free will' is also available to machines (as well as lower animals) then he and I don't differ in the slightest, much to your discomfort. We have reached a very meaningful consensus.

[Fred H.]

TJ: . . . then he and I don't differ in the slightest, much to your discomfort. We have reached a very meaningful consensus.

See Todd? Your equivocations have resulted in Tom’s cognitive dissonance—while Tom insists that he has “proof that there is no Free Will,” and while you alternatively declare that “some sort of free will is true and that physical determinism is also true,” Tom nevertheless concludes that you and he “have reached a very meaningful consensus.” Please Todd, gently explain to Tom that what he perceives as the nice warm feeling of consensus is really nothing more than your release into his swimming pool.

[Margaret McGhee]

Hi. I just got back from a week skiing in Canada - with tired legs but pleased to find this thread still running strong. I'm at the library as my ISP is down at home so my posts may be sporadic for a while.

After catching up on the posts I'd just like to add a gentle plug for my POV on this

I think perhaps the difficulty we have discussing free will - or any aspect of consciousness, is that our real way of knowing and our way of making survival decisions based on what we know, is actually emotional, not intellectual. But consciousness is a product of our intellect - that is only occasionally aware of our emotions (and what we know emotionally).

So our discussions are about concepts, things that our consciousness can deal with, like free will and compatibilism and whatever. And we think that is what knowing is - because that's all we have conscious access to in our minds. (I think our intellect is largely along for the ride )

And since this path is divorced from our emotional knowing we are free to imagine whatever concepts we wish. Although, some of those concepts will create emotional values that will get weighed if we make any decisions based on those concepts.

To summarize, we are free to imagine whatever we wish - and that feels like free will.

But we can only make decisions (like what to believe about free will or what to state in a post like this) that result from a summing of the emotional inputs that we are subconsciously aware of.

We therefore lack free will to make decisions in any other way. The challenge remains - can anyone provide an example where we make a decision that violates this principle? If you can't, then how can you make a credible claim for free will?

Margaret

[Fred H.]

MM: We therefore lack free will to make decisions in any other way. The challenge remains - can anyone provide an example where we make a decision that violates this principle?

I have Margaret, but as I’ve indicated b/f, you may be too locked into your preconceived notions, or perhaps just not really paying attention. Nonetheless, magnanimous guy that I am, I’ll try one last time—

First, I’d agree that the evidence from currently available neuroscience indicates that “survival decisions” are indeed made largely with (subconscious/subcortical) emotional input. However, our cognition/intellect (at least for those of us that aren’t completely slaves to our emotions and/or preconceived notions) can discern objective truth, enabling us to act accordingly, even though it may not always “feel” right emotionally.

I’ll give the same example I gave b/f—

In the Monty Hall problem there’s a prize behind one of three doors and you guess/choose which door—Monty then eliminates one of the two remaining doors (that obviously wouldn’t have the prize) and asks if you now want to switch your choice to the other remaining door. Initially most people will say, and are convinced emotionally, that it doesn’t matter, that it’s a 50/50 chance whether you stay with your first choice or switch to the remaining second door. However, many of those same people will, if they truly attempt to understand the statistics/probability of the problem, eventually see that in fact their odds of winning will increase from 33% to 67% if they switch to the remaining door.

The proof of the Monty Hall problem is a kind of objective (probability/statistical) truth—and I think that most are probably capable of comprehending such objective truth, and acting on it, even if it initially is contrary to their emotional feelings/beliefs and/or presuppositions.

So now it’s up to you: Do you “choose” to stay with your first choice, the lack of free will, which you may still “feel” very strongly about? Or, assuming you now comprehend, cognitively/intellectually, that human intellect is indeed capable of discerning objective truth (with minimal emotional input), enabling you to act on it—enabling the cognitive part of your mental trilogy to engage in “downward causation—do you “choose” to (cognitively/intellectually) accept and acknowledge that we humans do indeed have at least some free will? (If you remain convinced that we lack free will, then I suppose I’ll have to more or less agree that you yourself probably do . . . sort of a nice little win-win for us, albeit somewhat delusional and somewhat akin to Tom’s delusion that he and Todd have reached some sort of consensus.)