Thinking machines

Is "thinking" a "computation"? If so, making a computer "think" should be a simple problem of engineering. Almost all popular movies featuring "intelligent machines" make this assumption. Sooner or later, machines with "out-think" us.

What would be another way of seeing this?

As a starting point, let me create an image of a "mind" as local disturbances in meme space. I visualize this "disturbance" as typically not confined to a single brain, but more typically a "wave" spreading through connected minds.

This is an evocative image but the word "meme" is itself a "meme" that suffers greatly from over-use and lack of rigor.  To make the idea more rigorous, I recommend "Surfaces and Essences" by Sandler and Hofstadter.  This gives a more close-up view of how humans "think" and it hints that the "computation" underlying thinking is the analogy. The book is massively documented (perhaps over-documented). Although the authors make no claim to base their observations on some theory of how the brain works (it is a theory of how language works), there is a strong reason to think that analogy plays a large role, as we see this in our "deep learning" machine models that are basically seeking analogies between their inputs and the training set.

What makes human "thinking" difficult to model this way is the fact that ultimately humans are basing analogies on human experience - having things like mothers, ice cream, sex, Mac attacks, stubbed toes and so forth. We have sections of our brains that seem to be dedicated to special cases of analogy seeking, such as vision, language, and recognition of human faces. What's more, all of this is cross-wired and mutually enforcing using an electro-chemical processor with 100 trillion synapses, each of which behaves in a chaotic, quantum-uncertain way, impossible to model in a deterministic way. To make things harder, the whole system behaves in a self-modifying way so that the way a synapse behaves in time t is a function of how the other 100 trillion synapses behaved at t-1. I see this as a massive Schrodinger equation that gives a hint of the process without being much use at predicting anything or even describing a single state that would satisfy the equation at any one time.

But that's not all. Really, the "computation" performed by this wonder between our ears is being done as part of a social network. We cannot utter a single word or understand a single sentence without drawing on thousands of years of forgotten human experience built into the language. This is perhaps a more precise image of our minds being a "disturbance in a meme field". In fact, seen this way, "thinking" is a social process that takes place partly in our heads, but significantly among the minds of connected individuals. You might say that our words are little packages of thinking that have been off-loaded and pre-processed by anonymous individuals long dead. Think of the massive amount of pre-processing packed into the word "set".

The fact that we experience "thinking" as something to do with the "self" is an accident of the way we are "wired up" along with the very useful ability of our minds to hide the details of where our thoughts come from. Wonderful though it is, language is a clunky and error-prone way to share an experience.  I have always thought that dolphins act and share experience much more efficiently using sound. I think a dolphin can say "this is what I see" to another dolphin, or perhaps a dolphin can experience what a group of dolphins is seeing. This impression comes from personal experience with dolphins which is another topic for another day ...

To my mind, the role of the machine in this situation is to greatly enhance the ability of groups of humans to "think" together more rapidly and to "think" about matters that would otherwise be beyond the ability of even very large numbers of well-connected humans*. Our concerns that machines may "replace" us needs to be re-framed as concern that the future version of machine-mediated humanity will be something unrecognizable in terms of today's ideas and assumptions. I believe this has always been taking place but it's happening exponentially faster now. Now, as in the past, the "machine" of society has had a tendency to have little regard to the fate of individual human beings. This doesn't have a lot to do with actual, physical machines. For example, the impersonal machine of the marketplace has a tendency to grind up those with nothing to sell and no money to buy another day of life.

This line of thought goes back to 1969 and comes up again and again in my blog. Most recently, I have been charmed with the image of Indra's Net as a way of picturing the field of the mind. This way of thinking leads me to imagine the self as something not entirely caught up in the body of an individual human. It is tempting to see something of Buddhism in this idea but I think it's much more of a 21st-century concept -- something that would only occur to people who interact intensely all day with "social networks".

----------------------

* "thinking", "attention", "intention" and "action" are all subject to the approach outlined here, but that is perhaps another topic for later.

Mind and Body

Rene Descartes 1596-1650

Descartes took it for granted that "mind" or "soul" existed in different universes. He wondered about how these universes communicated. Specifically, he wondered how "spirit" can move "matter".

There are many who still take the idea of the spirit world seriously. They imagine the "soul" can be separated from the body and migrates on to some other universe (heaven) at death or somehow migrate to another body. Let's put this belief system to one side. From here on, I will speak to those who don't take this idea seriously.

These days, those of us who are scientifically inclined imagine the "mind" to be a phenomenon that somehow happens in the Universe that is governed by physical law (and perhaps other laws so far undiscovered). Exactly how this happens is a live debate and a key area of investigation in the new university departments of "Cognitive Studies". These same departments work on another hot subject: machine learning and "artificial intelligence". Naturally, these two issues tend to overlap, producing "computational" theories of "mind". It's quite respectable these days to consider "mind" to be some kind of computation that could, in principle, happen on a computer.

I'm not too impressed by "computational" theories. As far as I can see, they tend to side-step the core question and propose some kind of hypothetical process that bears no obvious relationship to "mind" as we experience it nor, for that matter, much to do with actual brains.

The debate over the mechanism of consciousness draws a lot of attention away from the core issue, whatever the mechanism may be. These debates also tend to assume that consciousness is something that happens in the brain: an assumption that has become so "obvious" that it is rarely questioned. This is partly due to the "computational" theories, which picture the brain as a kind of computer. What else could it be? It is an uphill fight to wrench people away from this metaphor, but the clues are easy to find if you examine what human consciousness is actually like. It also helps to have a few decades of experience with computers. My experience with simulated people like Alexis and Siri reinforces my impression that "there is nobody home".

The more you learn about the brain (and we are just starting to learn about this), the more we learn about how the brain plays a central role in creating "mind", but the picture changes when we ask What the brain is for? What is it doing? No organ in the human body exists for its own sake. The brain is one of the largest and most energetically expensive organs in the body. What is it doing?

Suppose Rip Van Winkle came to life in 2019 after a 40-year sleep. We hand him an iPhone and ask him about it. After a few hours of careful investigation, he might guess that it's a tiny computer operated with a touch screen, which is kinda true. What he might miss is that the essence of the thing is its ability to communicate with the outside world. It's a phone (which Rip may understand) but it's "aware" of its location and its orientation. It's constantly chatting with the owner and the outside world, warning of upcoming radar, suggesting places to eat, keeping the user up to date on the news and assisting the owner to navigate the real world. The essential role of an iPhone is to "know" the status of the owner and to connect relevantly to the outside world. It's like a little "mind". More precisely, it's a tool that extends the power of the mind.  It gives us a clue about what a mind is.

My claim is that an iPhone may be a computer, but this fact hardly describes its role in the world or the role it plays in the owner's life. Similarly, whatever the brain may be doing, and whatever neuroscience may tell you about how it "works", you will not know what it is for, how it works and what benefit it bestows on the owner. In fact, the role of the brain resembles the role of the iPhone. It "knows" about its internal state, the location of the owner. It handles inbound and outbound messages. Unbidden, it alerts the owner to relevant local information and opportunities and so forth. Subjectively, all these things add up to what we call "mind"*. The brain's role in creating mind is undisputed but, like the iPhone, that role makes no sense if we ignore the role of the outside world, the needs of the owner and the interests of the owner in participating in society by personal messages and general information. You may say that the brain creates consciousness but the consciousness is not "in" the brain. The brain strikes a match which lights a fire that joins in the conflagration that is the shared experience of all human society. Participation in this conflagration is what it feels like to be conscious.

The way that all that happens in the brain is still a huge mystery. It may not even make sense to ask why it feels the way it does (the "hard problem"). It is, however, quite clear to me that we will never understand consciousness by ignoring everything that goes on outside our brain. In my view, we can learn a great deal about "mind" by exploring the parts of "mind" that are wide open for investigation, such as language, memory, perception and decision making. I see the mind in two ways: mind in the world and world in the mind. By studying the world itself, we discover the mind that creates and "models" the world. We can also study the "mind" (especially the brain) as an object in the world.

It may turn out that the brain is, after all, doing a kind of "computation". It seems to be doing other things that computers do, such as retaining memory, sensing the body and the world, controlling a "puppet" body and so forth. All this is appropriate to any kind of brain - the brain of a worm, for example. But we don't want to stop there. We are interested in the mind of human beings. Human minds do things like write symphonies and build space ships. That's the part of "mind" that interests us. That's the phenomenon of mind that is interested in itself.

My dog is very smart. She has a brain that follows the same mammalian architecture of my own. Her brain does all the kinds of "brain things" that I mentioned above. The "computational" power of her brain is probably not orders of magnitude different from my own. But she doesn't know she's a dog, nor is she curious about it. It would certainly be interesting to know all about how her brain works and what kind of a "mind" she has. This would certainly tell us a lot about how my own brain works. But it would hardly tell me much about my mind, which seems to participate in all human experience.

It's natural to think of "mind" as something that arises in the brain due to the complex physical connections between neurons. That's kinda like a computer, isn't it? Perhaps we need a different metaphor. Maybe something based on 21st-century computers, not the computers of 1960. The Internet is created by connections between multiple "hosts" - usually computers. These connections need not be physical (think cell phones and WIFI). The way they are connected is irrelevant to the way they work and what they are for. Most of these hosts are in the business of communicating with other hosts, usually for the purpose of sharing information. Similarly, our "minds" are connected to other minds through speech (sound waves and marks on paper) and many other indirect ways. Breaking these non-physical connections produces a serious degradation in the function of "mind", like turning your computer into an old fashioned "stand-alone" machine with no Google, no Email etc. A brain that is never "connected" to the outside world or other minds is a baby without much of a mind at all. You may choose to call it a "mind", but it is hardly the kind of "mind" we wish to know about. Not the kind of mind you and I have.

------------------------------------------------
* Of course, the "mind" is more than what is hinted here. "Consciousness" is a faculty of mind but the mind is far more than that. In fact, it seems that most of what "mind" does operates on an automatic and unconscious level. When considered in detail, much of this operation slips out of the "computational" model, but that's a topic for another day.

Wolfram's "New Kind of Science"

The Kindle preview of "A New Kind of Science" is very generous - over 100 pages. Definitely enough to give an overall flavor of the book. The hard copy of the book is virtually unreadable due to tiny text fonts. The Kindle version is very affordable compared to the $40 cost (down from $70) of the phone-book sized original.

I would normally not review a book I have not read from cover to cover, but a good friend presented this book to me as one of the most interesting in his collection. For the sake of discussion, I felt the need to dive in.

I didn't find the first 100 pages to be particularly difficult to understand, but there wasn't much that I found interesting or original. My impression was that the next 700 pages would milk the basic cellular automaton idea (not original with Wolfram) to construct "original" ways of looking at virtually everything, allowing Wolfram to insert himself as offering the key insight - the turning point - in virtually every field of knowledge. This his explicit claim in the first chapter and it's made not in the name of this "new science" but on behalf of Wolfram himself.

Wolfram places himself at the center of focus, mentioning his own unique insight and achievements in virtually every paragraph. Whatever his discoveries may be, by drawing the spotlight to himself, he forces the reader to draw conclusions about Wolfram as a person. He does not present himself as an honest or reliable guide to his subject. "Unreliability" may go as far as to question his sanity.

He is a clear example of a Narcissistic personality (one of the many things we find alarming in Trump). He consistently underplays and dismisses the work of others and praises his own "discoveries" with laughable hyperbole.  Again and again, we see that what is amazing to Wolfram is his own genius. It's not difficult to compare his "genius" to a parallel work of Mandelbrot. Nobody could fail to be astonished by a deep dive into a Mandelbrot set, but when Mandelbrot himself discusses this phenomenon (much more interesting than Wolfram's examples), he explains it clearly and moves on to other fundamental issues. I may have missed it, but Wolfram Alpha seems to be the perfect tool for visualizing Mandelbrot sets, yet Wolfram doesn't mention Mandelbrot in his "origin story" of his "new science".

One quick check of his supposed "revolution" in biology may be obtained by Googling "Computational Biology" - a thriving field that's the explicit focus of many University departments. The Wikipedia article on this subject does not mention Wolfram. Yet, to read the book, nothing is going on here except for Wolfram's Nobel-class work on cellular automata (a field he did not invent). Complexity arising from simple formulae is not new - the field is called "Chaos Theory". If one read Wolfram's book in isolation, one would get the impression that Wolfram has more or less invented these fields (or at least "revolutionized" them). Outside of Wolfram's own mind, this seminal influence is hard to find. 

The other thing I note is "manic syndrome". His feeling that everything in the Universe depends on "laws" he has discovered is classic mania. Everything is connected to everything else and only he can see the connections. The massive size of the book, the tiny fonts, the endless examples of fascinating patterns all speak of a manic frame of mind. I would be very surprised if Wolfram has not been under psychiatric care from time to time. This would not conflict with his obvious business success and real achievement (such as Mathematica). Insights gained on manic "trips" can actually be totally valid. It's the affective aspect of it (the illusion of cosmic significance) that creates problems in daily life and eventually the need for treatment. The manic (paranoid) illusion that only you can see the truth is evident on every page of this book.

Narcissism and mania grow in the same soil as psychosis - living in a fantasy world that is detached from what the rest of us call "reality". Wolfram's ranting, rambling style reminds me of nothing better than I have seen on the psych ward. Another way of saying this is that Wolfram seems to be "lost in his own head".

So, if we grant for the moment that Wolfram is bat-shit crazy, do we ignore the book? I'm reminded of Newton, who was anything but a picture sanity and stability. He spent most of his time doing alchemy and looking for hidden meanings in the Hebrew Bible. His ego was an obstacle to his ability to share his results. Yet he was clearly a great genius. He was certainly "on to something". I'm sure Wolfram would welcome the comparison. 

It seems to me that, for the most part, the geniuses that do make history are perhaps a bit odd but generally sane and ordinary. I'm thinking of Einstein, Feynman, Bor, Faraday, Wilczek ... 

So, I may want to return to this book at some point. I found it quite easy to skim (by ignoring self-congratulation and claims that are close to tea leaf reading). The Kindle version solves the problem with tiny text. There are a few early observations that seem quite valid but are far from original, such as the need to use simple algorithms and not simple continuous formulae as tools to "explain" nature. On the other hand, I tend to steer away from authors who misrepresent or contradict what I already know, especially when they fail to connect their work to the work of others.

--

Wayne Brown

Evolution of Mind - Evolution of Machine

I'm wary of the term "evolution" - especially when it is used in a vaguely Darwinian sense to sprinkle scientific fairy dust over wild speculation. I think it makes sense to use "evolution" in a fairly narrow sense, in which we can see that some object, "thing" or situation exhibits the following qualities:

• It's in some "state" of current interest. The analysis of the "state" has some non-trivial explanatory power over the thing in question even when it is considered as a current, static object.
• The "thing" has obviously, necessarily undergone change over time - it wasn't always the same as it is now. It must have been in some other state, usually a simpler state, at some time in the past.
• There is strong evidence that the thing has undergone versions, where one version incorporates design features of previous versions.
• It is legitimate to ask what changes have taken place over time to get from the "simpler" state in the past to the current state.
• Some kind of specific process should be proposed, even if somewhat tentatively. For example, Darwin started with the observation that species are related, asked what we mean by "species" and only gradually came to theorize that species must have emerged over time.  He could only speculate about the mechanism (He liked, but did not use, what we call "survival of the fittest". He didn't know about genetics or DNA). 
• The specific timeline of change will inevitably be speculative. It's nice to establish milestones and approximate dates but this must regarded as secondary to analysis of the change itself. Darwin was famously off by orders of magnitude with his time line but this turned out to be irrelevant to his core ideas.

With those provisos, there are a few questions in the "evolution" of humans that are interesting:

• Speech
• Language
• Writing
• Money
• Social Structure (small bands to cities, craft to corporation, tribe to Nation State)
• Religion
• Nation States
• War

It is useful to speak of the evolution of computation or perhaps the wider and broader term "information processing" rather than the "evolution" of the computer. The other aspect of this evolution is characterized by "information storage". Clay tablets and fingers have their place in this history. It doesn't begin with Babbage or Turing. I like to avoid terms such as "information processing", which would be perfectly acceptable except that virtually all readers will conjure up an image of modern computers when this phrase is used. It is, in fact, still fashionable to call the Computer Department at a University the "Information Technology" department, even though they don't teach about clay tablets or even money. The common thread seems to be progress in externalizing or delegating the task of computing and remembering to machines that grow in complexity and power over the years. Questions can be asked about the progress of mechanical information processing and they are woven in to the above questions about human society:

• Writing
• Counting
• Money
• Computation
• Stored programs ("software as data")
• Networks
• Virtualization
• Class libraries (divorce of design from platform)
• Social networks - close integration of human and computer "societies"

An important thread of this evolution is the way that we gradually shift attention to aspects of reality that can be computed and stored while pushing other aspects of reality aside. This is perhaps the fundamental way that human evolution is linked to the evolution of computation. Wittgenstein's famous saying pops into mind: What can't be said must be passed over in silence.

Several common themes emerge when discussing the parallel evolution of humans and machines:

• Information processing and the evolving nature of what counts as "information"
• Compression
• Virtualization (platform independence)
• Information embedded in language
• Communication
• Formalization and standardization 
• Efficiency

We gain special insight into these common themes by comparing and contrasting their history in the two streams of evolution. In the case of computation, many of the "landmarks" are very clear and have firm dates. Others, such as the appearance of money, have a more complex and debatable history but there is clearly a time when money did not exist and a time when it did. 

In some cases, important innovations appear in one stream and they are not obvious in the other, at least at first glance. For example, "virtualization" is a very clear and specific engineering concept in the computational field. What, if anything, can it mean to speak of a virtual "mind"? One clue is in the fact that philosophers have always felt free to describe the inner workings of the mind without reference to how the brain "works". When they do make reference, their ideas are remarkably sketchy and dated. Tellingly, their ideas about how the physical mind works are not much affected by the specifics. The ideas are "fairy dust". This leaves us free to propose that the "hardware" doesn't matter at all, leadinging in the extreme case to ideas such as Kursweil's - "uploading the mind to silicon" seem perfectly reasonable. This is an argument firmly based on ignorance, like the idea that humans were created out of nothing by God, which made sense until the facts were known.

The evolution of "society" in computation also exhibits some fascination and non-obvious parallels. One precondition of "society" is standardization of components and protocols. Such standards in human history have lead to sweeping changes in human society, including capitalism, cultural genocide and waves of economic crisis when policy is about "jobs" rather than human well being. Standardization of weapons, protocols and military "jobs" has played a key role in the evolution of modern warfare to the point where a single instruction can now spell the destruction of civilization through a "fan out" through standard protocols to standardized weapons and human beings turned into standard components of the military "machine". In the case of the military "economy" (legalized mass murder), there are many milestones on the road to standardization, along with a rich set of parallels between the kind of society humans can form compared to the kind of structure a computation can take.

Society can be usefully regarded as a machine than as collections of living things. In this sense, our two streams have been steadily converging over recorded history. The evolution of this aspect of society is worth discussion in its own right, as I have done in "The Programmable Ape". By regarding machines and humans as co-evolving along very similar paths, it becomes possible for me to describe the human machine (the "dragon") in more precise and familiar terms. This can possibly overcome the common perception of what a machine is. 

So what is a "machine"?

• Machine "language" is tiny, formal and inflexible. 
• Machines are formed of components which exist for no other purpose than to serve the purpose of the machine.
• Very few machines are "unique". They are themselves "cogs" in a larger system that demands precise standardization of components. 

In contrast, in a human ...

• Language is vast, informal, open-ended and general
• Humans are famously difficult to organize into coherent groups with coherent purpose. In practice, this involves stripping them of what makes them human in the first place.
• Humans are unique. This uniqueness is built in by the way DNA works and the inherent limitations of their nervous system. 

Over time, these differences become blurred:

• Humans become "standardized" into "jobs" in order to fit into the larger economic machine. Unique human qualities are suppressed or even punished.
• Human "machine languages", especially the language of economics, become more and more rich and complex. To many, it starts to sound like talk about the "real world" and not "machine language".
• Machines are more and more frequently referred to in anthropomorphic terms, such as what "Wall Street fears" or what "Germany wants".
• Individuals are increasingly "programmed" by tremendous pressures of living in a man-made world of things and information. Thinking "outside the box" is difficult if all of language - including your inner voice - speaks only about boxes.

You can get a lot of fun out of taking these analogies seriously at face value. There is usually a kernel of truth - a way of seeing that machines and men are really the same in important ways. For example, if someone is talking about "efficiency", they are talking about a machine. Think of the situation they describe in terms of machines, computation and the history of such things. When they talk about "profit", think about computer games. When they talk about what is "best" for a corporation, take them at their word and imagine the corporation to be alive and part of human history. Ask how such an abstract machine can possibly acquire a desire to survive and thrive.

The Brain As an Amazing Symmetry Computer

I have a coffee cup in front of me (I usually do). If I rotate it or move or see it in different light it my brain automatically makes me experience it as the same cup. If you have ever tried to make a computer figure this out, you will see how astonishing this is. What's more, the cup will be seen as the same cup tomorrow and the same as the cup that I washed a week ago (or is it two weeks? The point is, that doesn't matter). This is symmetry under a transformation in time and space. It takes place so automatically that most of us will go through our entire lives without thinking of it at all.

There is another, related, trick that the brain does. It makes me "see" the cup as a form. A form is a bunch of "stuff" that is more or less arbitrarily treated as the "same thing". My dog is quite capable of seeing a flying tennis ball as a "thing" that can be snatched out of the air, but she seems to be uninterested in the "things" shown on our TV screen. This is just to point out that selection and recognition of "things" is a brain function and not "out there" in the real world.

Brains which make an isomorphic transformation (or I suppose a "reverse" transformation) are able to manipulate the "things" of the mind as if they are "things" of the real world. Brains are good at detecting relevant things and picking them out of the hurricane of sensation that presents itself to working memory in every second. You could say this allows working memory to "compress" the world into a very small set of "things". I don't expect the cup to turn into a cat, but I would quickly notice if it did. Otherwise, it can just sit on my desk and play a small part in what I perceive as my surroundings.

From Bacteria To Bach and Back (Dennett) provides a detailed exploration of how our minds come to recognize "things" and how the "thinginess" of the world arises from relevance rather than raw reality. An ant somehow recognizes certain things but presumably doesn't "know" it's doing it or even that it's an ant. "Consciousness" is not required to recognize "things". The brain has inherited this very ancient capability from the dawn of life and has had billions of years to perfect it. That's why it's so good at it. In Dennett's terminology, symmetry computations are a "competence" of brains: a competence without comprehension or, one might say, a pre-condition of comprehension. Dennet makes another subtle point: this competence doesn't depend on any kind of representation of the world in the ant's head or our head. In other words, we need not look for the set of neurons or synapses that "represent" my coffee cup in my brain. I picture the cup as a result of a result of a fantastically nested and fractal computation that the brain does "on the fly", a computation this is best described in Hofstader's"Surfaces and Essences"/ This computation relies mainly on analogy which is a special case of symmetry. This is why I refer to the brain as a "symmetry computer". It uses symmetry to "compress" all experience into "merely" a few billion connected neurons.