“No amount of knowledge about the hardware of a computer will tell you anything serious about the nature of the software that the computer runs. In the same way, no facts about the activity of the brain could be used to confirm or refute some information-processing model of cognition.” ~ Max Coltheart
By Paul Almond
I don't even know what the quote means. Software is merely an emergent property of what the hardware is doing.
Let's take it an example: a conventional home computer running Tetris. We might say that the machine is hardware and Tetris is software, but if we look closely enough at the machine we will see a pattern of electrons corresponding to Tetris. Is there anything that makes those electrons software and the microchips themselves hardware? Are the electrons somehow "soft" of "fluffy"? I don't think so.
Some people might even make an argument that software is a purely invented concept -- that the only thing that exists is really the hardware -- and I could understand that. I prefer to accept that software is an emergent property of the underlying hardware in the same way that tables and chairs are emergent properties of the underlying atoms and I don't think that distinguishing between "software" and "hardware" is really just a human convenience. This does not mean that I think Tetris is "just electrons." I think that Tetris exists in its own right, as something caused by the electrons, as an emergent property of them; but there is no philosophically "hard" reason for calling this thing software just because it is based on a pattern of electrons. We could equally well say that a chair is software because it is just based on a pattern of atoms.
This means, incidentally, that I do not agree with philosophers who claim that consciousness or the human mind is not real -- that it is just an illusion caused by atoms. I think it is as real as a table or chair and we could make the same argument that tables and chairs are illusions caused by atoms: if we did that the word "real" would become useless to us.
Whether we think software programs are "real" or not, the fact remains that the software is not something that just "sits on top of" the hardware. If we accept that it exists at all it is because it is caused to exist by the way in which the hardware is organized.
Another way of looking at this:
Imagine that someone said:
“No amount of knowledge about the physical structure of a Harry Potter book will tell you anything serious about the nature of the story that the book tells us. In the same way, no facts about the activity of the brain could be used to confirm or refute some information-processing model of cognition.”
This would be invalid because it is clear that the story does not just sit on top of the book somehow. It is caused by the physical structure of the book -- by the pattern of ink molecules on the pages. It is obvious that by examining the pattern of ink molecules enough we could find out everything we wanted to about the story. Admittedly, our analysis may need to be sophisticated to do this. We would clearly have to gain a very deep understanding of how the ink molecules are arranged to extract plot and characters from them -- but humans do that sort of thing every day from things like the myriad of impulses reaching us through our optic nerves.
Even though I accept the reality of software programs as an emergent property, I don't think that there are any clear semantics that tell us what to call software and what to call hardware. I think it is more about which aspects of a machine can easily be changed. We call something "software" if it is commercially practical to change it quickly in response to consumer demands. Tetris is "software" because if consumers want to play Space Invaders then it can be changed in 2 minutes. The processor chip is "hardware" because it takes longer to change and you don't buy changes to the processor chip as casually: this involves very subjective semantics.
Suppose we have an interpreted program in BASIC. We might call that "software," and what is underneath "hardware," but it is really based on machine code software underneath. We could go further though: the machine code software relies on microcode in the computer's processor. Maybe we have finally reached the true "hardware" under that? Not really. Why couldn't we just say that the "hard" parts of the computer are really software based on the laws of physics. This might seem a strange view, but if a pattern of electrons in a machine is considered to give rise to "software" then it is not doing anything much different from how a table or chair manifests itself to us. I am not saying this to argue, seriously, that chairs and tables are "software," but rather to show that the distinction does not help outside commercial activity.
There is another possibility here. Maybe the semantics of the quote are a bit vague? Maybe the author really meant that no "general facts" about the hardware would be of any use, e.g., maybe he felt that you can't just discuss things at a certain level of generalization and expect to capture what the brain is doing. I can't really respond to what [Max Coltheart] might have been intending to say, though, only on what he actually said.
Visit Paul Almonds website here.
Riding the same hobby horse...
Hi Paul,
Couldn't agree more!
Well, actually I suppose in a way I could agree more. I possibly take an even more hard-line (soft-line?) view on these things and assert that ALL hardware is software, right down to subatomic particles. They, like a chair or a mind, are really self-preserving spatiotemporal *patterns* in something, much like a tornado is a self-preserving pattern in the atmosphere. Even matter itself is an emergent phenomenon.
The "hardware" of a computer is doubly so (as I think you are also saying). It's a common mistake to think that "hardware" is referring to the Silicon and other elements that make up the chips (or even the electrons that pulse through them). It can't - it must refer to the *organization* of those elements. A different organization of the same materials might be a transistor radio instead of a computer, so the "hardware" being referred to can't be the materials themselves, and thus even the "hardware" of a computer is really software.
So I agree with you that these are all just different levels of emergence. And therein perhaps lies the issue. For four hundred years, science has behaved as if the properties of the whole are immanent in its parts - that all systems are reducible. This has led to an analytical standpoint, in which emergence is anathema. It's what makes people (especially physicists) search for the basis of mind in some magic ingredient of matter (quantum weirdness or whatever) - it's the only way they can conceive of anything because they don't have a systems perspective, even though they're perfectly familiar with the idea that you wouldn't look for bits of a radio station inside a single transistor.
The analytical tools of science are close to useless for understanding emergent phenomena, which just vaporize under the microscope. Emergence can only be understood through synthesis. Analysis comes first, for sure - you have to know what the parts are - but then you have to put them together again. It's the relationships that count. This is the primary tenet of Artificial Life (the use of computer simulation to synthesize life-like phenomena) in contrast to biology, which tries to understand life by taking it apart.
If one makes inductive leaps, inspired by observation, then plays around with them by making models, then one has the tools needed to understand the whole in relation to its parts. But if one expects to deduce the behavior of the whole from the parts then one is going to be disappointed.
Conway's Life is a good example. NO amount of deductive analysis of the rules will predict the startling behavior of the overall system. The only way to work out what will happen is to try it, either in the system itself or a model of it. In the case of Life there are no models simpler than the game itself - the system is its own best model. So you have no choice but to run it to figure out how it will behave. There's no way to reduce the problem.
It remains to be seen how much we can simplify our models of the brain and expect to get behavior comparable to the mind. My hunch is that we can make quite sweeping simplifications but the models are still going to look very much like brains (and not a bit like algorithms, even if an algorithm is what's producing the model). But to understand brain imaging we HAVE to use induction and modeling, and I think maybe this is the paradigm problem that underlies Coltheart's gloomy assertion.
Anyway, I'm ranting. It's just nice to find other people such as yourself who appreciate the emergent "levels of being" of which the universe is composed! On the whole it's a viewpoint more common among artists than scientists, I find.
The Distinction Between Hardware and Software
Hi Steve
Thank you for your interesting comments. I am glad you were able to make sense of what I wrote, particularly as I had a huge number of typos in that reply that I had given to the Editor by email.
I think we agree on a lot, and differences are largely semantics. I understand you would regard a table as "software" - it is merely an emergent property of atoms being arranged in particular way. I can understand that fully. I would be happy to describe a table as "software". The only reluctance I have is that I think we could also describe computer programs as "hardware", using the same sort of reasoning about emergent properties, etc. It seems to me that the ontology of how tables exist and how software (or computational states) exists is the same - as an emergent property of the way something "lower level" is arranged, so what the parts of a table are doing to make a table exist, and how they are doing it, is not profoundly different to how a computational state of a computer exists. For this reason - because it seems to me that anything that is called "hardware" could equally well be called "software" and anything that is called "software" could equally well be called "hardware" - while I might say that tables are really "software", if I am being pedantic I might rather say that the words "software" and "hardware" are both useless in serious philosophy and are only useful for convenience in areas like commerce and engineering.
I am not sure to what extent you will agree with this. I feel that you may agree with it to some extent and, if you don't agree with it fully, it will at least be a position that you understand and can readily identify with.
There is one respect in which we might be able to say that there is a meaningful distinction between "software" and "hardware". If there is a "bottom level" of physics, with something that just is, rather than being an emergent property of anything else, e.g. if there are fundamental particles, or if there are fundamental strings, or fundamental quantum wavefunctions (if Hugh Everett was right), then we might feel justifed in distinguishing that bottom level with the term "hardware" and describing everything else as "software" - because it exists only by virtue of being logically "implied" in some way by the way in which things on the lower level of hardware are arranged. In my view, we have little reason to think that any such bottom level exists. I suspect (and I could justify it with an argument if I had more time here) that nothing is fundamental and that reality is nothing more than logically implied, emergent properties, built from other emergent properties and so on - like a computer program running without any hardware. I do not see any need for any ultimate hardware. If my table is real then emergent properties in themselves should be sufficient. If my mental states are real then that should be evidence that mental states have some sort of reality - and I experience those mental states so I can be assured of their reality. If the "software" can be "real" in its own right then it does not need any hardware to support it - just more software - making the distinction fairly irrelevant. This is not the best argument that could be used for such a view: as I said I would need to take more time if I were to properly argue it here. Someone might say that surely I must think that things like space and time are fundamental and maybe they are the "hardware" level of reality - not at all. I have no reason for presuming the space and time that humans perceive to be any less constructed objects than cats and tax returns. Humans have often encountered things in the past that they thought fundamental - they often thought they were staring straight at the basement of reality - and they have always been wrong. So it is now, in my opinion.
Anyway, that might make me view even more extreme than yours. :)
Turtles all the way down?
Hi Paul,
Yes, I'd be quite happy to accept your assertion that the hardware/software distinction is meaningless. I suppose the most one can say is that one level's hardware is the next one down's software, in much the same relative sense that we use the words "yesterday" and "tomorrow".
As for whether there is a basement level to reality, that's a very interesting question that has vexed me a lot, so I'd be very interested to hear your longer argument some time when you have the opportunity. I'm sure you're right that we haven't found it yet, but is there such a thing in principle, or does everything in the universe wrap around on itself and pull itself up by its own bootstraps?
Certainly the world of space and time is a pretty emergent thing in some senses. If we presume that the universe started in a big bang then "before" that moment there must have been a singularity. A singularity simply doesn't exist in space or time (it can have no direction or size or speed or position), because there is nothing else against which those properties can be measured. Space and time must therefore be emergent properties too, coming into existence at precisely the same moment (!?!?) the big bang started, since that was the first moment they could have any meaning whatsoever (although I confess I can't understand how a singularity can even be said to have split into two without some space to occupy the discontinuity and some time to permit a separation between "before and after the split").
Actually I can't fathom ANY of that. It really makes my brain hurt. But I can see that even some of the apparently most fundamental properties of the universe are actually consequences of it. So I'd be grateful for any insights into all this. Is is really turtles all the way down?