the hot new programming language

If you can't read, I can't help you understand your error.

Expound.

A tautology would be "thought is impossible, therefore no thought can be taking place." This argument is a positive one, just a bit too conclusive for the comfort of mechanists (of whom I used to be one).

Claiming what exists? You're the one making the metaphysical claims here, Cliff. You seem to think that your metaphysics is somehow privileged, exempt from elementary philosophical investigation. Yet you have no problems putting words in my mouth and attributing all sorts of views to me that I'm not advancing.

I'm not arguing for ghosts in machines, or the nonexistence of matter, or anything else of the kind. I'm merely claiming to demonstrate that your position is too simple, in that it is self contradictory.

Aristotle knew that, and he's nobody's idea of a fundy. Kant knew it too. So did every thinking person until about 1955ish, when the fashion for computers and their programs seems to have made everybody forget how to add and subtract.

and I'm just pointing out the psychological motivations for

I said nothing of the sort. I just said that the mind wasn't merely software. I don't know what it is, but it logically cannot be merely an epiphenomenon of the brain.

I doubt you could defend that statement.

Cheers

Phil Hobbs

high

to

I saw Toyota has just come up with a new class of shares aimed at financing longer term development programs, you can't sell them only swap to common shares for a short period each year, and for the first 5 years the dividend will increase for each year held

-Lasse

I think you will have to elaborate on why you think that a simple mechanistic materialism logically entails the consequence that logical thought cannot exist. Our individual reasoning may be flawed but there is always peer review to find errors in any mathematical proofs.

What we perceive as consciousness and thought is likely to be an emergent behaviour on a sufficiently large network of interconnected simple computing nodes. It looks complex from the outside but the individual elements are nothing more than a pattern of clicks on neurons.

We are not all that far off being able to build dedicated hardware with comparable complexity to the human brain now so it will soon be a viable experiment. Though a cat brain seems to be first choice.

In much the same way as the simple rules for Conway's life by sheer luck happen to produce a computationally complete Turing machine.

Not a program as we would understand it, but a pattern of interconnects and signals that represent things we have learnt in a physical medium.

I don't see any compelling reason to invoke magic of any kind.

When we can do the full computer simulations on suitable hardware then it will be possible to decide this question by experiments.

At the moment because we humans can use pattern matching tricks and parallelism that are beyond our ability to program into robots.

The robots are getting better though. I have seen a few now that can do tricky variable jobs that previously only humans could manage. Sorting Smarties (M&Ms) from a tray being one of the more impressive examples.

It is likely that the computational singularity exists somewhere in the future of computer hardware. When I was an undergraduate the idea that a computer could beat me at chess was risible. Now any one of a dozen chess engines is stronger than our best human world chess champion.

David Levy only just won his bet. Turns out though that computer chess was a false dawn and is an easier problem than we first thought.

Go requires much deeper pattern matching skills to play at any kind of serious competitive level and computers are way behind human masters.

More impressive still are the self driving cars which are just into road trials now. Even with some bugs they may be safer than humans.

The madman's brain *is* almost as good as Einstein's assuming that the madman can see, walk and run. It takes insane processing power to handle the visual inputs from the eyes and run a 3D world model.

The amount of the brain dedicated to abstract thought, reasoning and imagination is tiny by comparison.

It is only silly if you choose not to understand it.

New age weirdo thinking. Presently advocated by Penrose in his various popular science books. I remain unconvinced. The crux of the complexity of a human brain is a huge number of tiny simple computing elements and the insanely large number of permutations of possible interconnects.

A few things like highway collision avoidance, maybe warnings and applied braking, might be good. I can't imagine a self-driving car being feasible in a dense city.

Single-cell and few-cell brainless critters do impressive things, like hunting and hiding and defending themselves and finding mates. Why would neurons be limited to acting like slow majority logic gates, dumber than a bacteria? The Neural Network model is popular because people don't understand how cells actually work; it's cargo cult science. What might the image recognition processing time be for a trillion element neural net computer with millisecond element prop delay? It wouldn't win many tennis matches.

Why wouldn't neurons use quantum computing principles inside? If it's possible, evolution would have taken advantage of it. So you are saying that it's not merely a weird idea, but it's impossible. Pretty strong statement.

And yet we have done both those things already with hearing, and the operation to implant the bionic ear is approaching routine. The bionic eye project is active and will succeed within a decade. Will you then move the goalposts again, to defend your magical thinking?

Occam's razor. There's no need to consider it, because nothing appears to be going on that actually needs it. Contrary to your assertion, no-one claimed it was impossible that quantum effects play a part. Just that it seems unnecessary (so far).

Neural nets - the kind you refer to - are not remotely structured on the actual neural structure as it's currently understood. In particular, they do not have temporal feedback (whereas all cognitive thinking occurs in a mess of massively-interconnected *oscillators*). They also lack the hierarchical layering and reinforcement structure of the neural cortex. They are constructed in modules whose learning time grows exponentially with size, so are completely unworkable at scale.

All these things have been fixed in the HTM research I referred to. The breakthrough demonstrations occurred around 2010, and have grown since then, and as I said, it as recently adopted by IBM's Cortical Learning Center. They would not be committing resources to wafer-scale fabrication without some pretty compelling demonstrations.

But hey, why not just continue spouting old criticisms of some earlier technology instead of doing some reading. That's much easier, right?

Clifford Heath, CTO, Infinuendo.

Paul Davies is a proponent of this magical thinking too. It's all an unnecessary attempt to salvage free will (and hence, significance, and the ancestral spirit world) from the machine.

David Pennington (who's no slouch himself; studied under Dirac, worl-class mathematician and also an Anglican priest!) unwittingly showed me why that salvage attempt is unnecessary, and put the clincher on my personal non-theism.

He said, more or less:

"Consider the molecules of air in this room, which collide elastically (like billiard balls, except for the shape), such that in a tenth of a nanosecond each molecule will experience an average of 50 collisions. How much do we need to know about a molecule and its environment in order to be able to predict its direction and momentum after those

The astounding answer is that if you omit from your calculations the gravitational attraction (the weakest of the known physical forces) of a single electron (the smallest known stable particle) at the opposite end of the universe (the most distant known location), then after 50 collisions, a tenth of a nanosecond, you know *nothing*, you cannot predict *at all*, the direction or momentum of that molecule."

In other words, without even invoking any quantum effects, everything in the universe is inextricably linked to everything else. Just because there is some non-local causation happening is no reason to disclaim personal responsibility, or to regard your actions as non-free. They're just free in a different context, and that's all that matters.

Clifford Heath, CTO, Infinuendo.

I work every day with neurophysiologists who are trying to understand how memory, decision-making, visual recognition, and other amazing "ordinary" brain activities occur. I have yet to hear of any talk of quantum mechanics being necessary for any of these.

Looking at single neurons or small sets of neurons (and perhaps glia), there's been no need to drag quantum mechanics into analytical descriptions of cell behavior. Moderately complicated computational tasks involving sensory and motor tasks are moderately well understood relatively near the periphery based on these models.

We certainly can't categorically exclude quantum mechanical effects in brain activity at this point, but there is no need to include it, either. Such a suggestion adds a certain fanciful noise to the discussion without really helping to understand or predict anything.

Your argument is barely any better. You're firstly defining thought as something which cannot exist within a machine, and then showing that it cannot exist within a machine. Well duh.

The idea you attacked in your first post on this subject is that "'modern science' has proven that the human mind is just the pure physical operation of the human brain under physical causation and nothing else."

So either you're attacking the "physical operation... under causation" or you're saying that the brain is linked to something more than itself. Furthermore, you're saying that Aristotle proved your point. You say "there would be no way to even notice that the next step went wrong", but there *is*: the real experiential world allows us to discount flights of fancy and determine which things are real.

Claiming that thought exists, and requires something outside the machine that's "not just software".

Your demonstration of self-contradiction was a failure.

Kant himself departed from Plato and aligned himself with Hume in claiming that thought alone (without empirical observation) cannot lead to truth.

Even if we believe that the brain is somehow linked to the rest of the universe in a cosmic consciousness (via quantum gravity or any other kind of woo), and that therefore the entire universe is a "thinking machine", your argument would still claim that it cannot think. To make that argument requires redefining thought as "that which cannot exist inside a machine". That's just folly and sophistry.

Clifford Heath.

It's only a question of degree, not fundamental differences. No quantum wooo is required here.

On the other hand, quite a lot is known about how to interpret the output of the cochlear. They've tapped into individual nerves and can figure out what the ear is hearing. How else would they have known what stimulae to send? The brain is adaptable, learning to understand messages that are off-target, but being on-target reduces the training needs.

If you want more detail, I can put you in touch with a buddy who's a senior communications engineer at Cochlear Inc, just across town from here. But I suspect that blind assertion is more palatable to you than research.

Clifford Heath, CTO, Infinuendo.

Penrose made much the same claim in "The Emperor's New Mind"

Nobody took him seriously, and he's got a rather better track record than you have.

Penrose made much the same claim in "The Emperor's New Mind"

Nobody took him seriously, and he's got a rather better track record than you have.

OK, explain how those things actually work.

For very small values of "moderately." When you can tap into an optic nerve and project the image on a computer screen, or go the other way and fix blindness, I'll be impressed thet you understand how this stuff works.

Since nobody knows how nerves or brains work, why exclude any possibility? And why would evolution discard any useful phemomena?

So many people don't actually believe in evolution.

On Wed, 8 Jul 2015 14:16:56 -0700 (PDT), Bill Sloman Gave us:

Folks tried to ridicule Tiny Tim too, but the guy was a genius.

Have they? Cochlear implants just stimulate local sections of the cochlea; they don't drive the auditory nerves and are hardly a bionic ear. Similarly, people can tickle spots on the retina and produce crude images, sensations of light, but nobody can decode or encode data in the optic nerve.

The bionic

Tapping into the optic nerve would demonstrate serious understanding of how the nervous system works. Zapping spots on the retina is relatively easy, really just a mechanical problem. A real bionic eye would *replace* the eye, not just poke it.

The retina is arguable part of the brain - it does a ton of preprocessing - so the optic nerve is internal to the brain. I don't think anybody understands the encoding, or how images are processed on the other end. But how does a finger transmit its various sensations up into the brain?

Actually they are getting close to being able to fake signals from various external sensors at modest resolutions and the brain network adjusts to use the crude data presented to it. They have also got a lot better at decoding signals sent to muscle groups for bionic limbs.

That we don't fully understand everything about the brain is a natural part of bleeding edge scientific research and in no way invalidates what has been done so far. Current understanding is incomplete it might even be wrong but until there is a better predictive scientific theory there is no point in replacing it with new age handwaving magyck.

It ever there was a proponent of cargo cult science it is John Larkin.

You keep moving the goalposts so as to invoke magyck quantum weirdness for everything that *you* don't understand without bothering to look at what is already being done in the field.

As a series of nerve impulses - faking them is a serious research project for providing feedback from prosthetic limbs. BBC take on it:

Quantum effects are typically limited to relatively small scales at the molecular level. Cells do use every quentum trick available to them to convert light or food into energy available for electron transfer. But at the higher level of the cellular scale differential equations suffice to describe the system.

It is much the same as with thermodynamics - you don't need to know the exact details of every particle motion to work out the bulk properties.

Neural nets are a very crude approximation to how the brain works. That they work as well as they do in practice is somewhat surprising.

That is the problem. John assumes that anything he can imagine is reality and that experienced researchers in the field and all the peer reviewed literature is wrong just because he says so.

It might well work best there since the average speeds are low and predictable vehicles can convoy together much closer together under machine control than with human reaction times 100-1000ms.

Too many drivers these days are using cell phones or worse texting.

Complex apparent behaviour can emerge from the interaction of a few very simple rules. Conways simple 2D automaton Life is Turing complete.

Yours is the cargo cult science. Anything you presently don't understand you put down to handwaving quantum mysticism.

There is no compelling reason to invoke anything more sophisticated than a lot of non-linear differential equations to model neurons.

I am saying that on the scale of a typical cell quantum effects are largely limited to the individual molecules. There is no cellular aura of new age quantum mysticism needed to explain what is observed.

This could change but at the moment it looks exceedingly unlikely that anything other than the huge network combinatorial factors are relevant.

By fooling around.

It was observed that the cochlea senses different frequencies at different points along its spiral. It's trivial to stimulate single points electrically and have subjects report what they hear. Then it's mechanically difficult but logically trivial to insert multiple electrodes, and fiddle with algorithms until the subject reports sorta intelligible sounds. No deep understanding is required to do any of that. Fiddle until it works.

The brain is adaptable, learning to understand

I bet that buddy couldn't cut out the cochlea and stimulate the auditory nerve to any useful effect.