So–this is going to be a hash through post–a kind of processing exercise in the virtual realm..
I want to talk about–or at least start to talk about the ideas, notions, meanings, etc. that go into the conceptions and processes that one might describe as signal processing and symbol processing.
Now–I came across this juxtaposition while reading Lakoff’s Women, Fire, and Dangerous Things that I’ve mentioned in a few posts prior to this one–and the contrast really struck me. As I understand it and/or would define it, the difference between these two things can be somewhat subtle–but also very profound. Let me try and flesh this out here…
1. Symbol processing is the kind of activity that computers do very, very, well, and which also plays a kind of a role in how humans think. It also has tons to do with things like mathematics, which involves moving symbols around and processing them according to certain defined rules to produce other symbols. This kind of activity is most easily seen, at least for me, in the way that a computer program written in a particular language works. In any such language–Perl, Java, even HTML, etc–there are certain symbols or strings of symbols that are governed by various rules. If you combine such strings of symbols according to the defined ways–then other symbols or strings of symbols can be generated or outputted.
2. Signal processing is a somewhat broader term that encompasses the more general processing of information that is perceived by some sort of sensory apparatus (eyes, motion detectors, a microphone) and is then transformed or modified in various ways to produce an output that may or may not closely resemble the input. This kind of processing can be analog or digital in nature–and the outputs vary greatly in nature. For example, signal processing by human eyes transforms various wavelengths of visible light into the perception of colors. Another example might be how a microphone input can be altered to produce distorted vocals in a speaker. The transformations can also involve symbols or symbolic representations of elements of reality. The input of certain sounds into a person’s ear may result in that person writing out a particular set of words (symbols) or in drawing a particular kind of object.
3. Now–this starts to become interesting to me in that I think there are some very important differences buried within the different kinds of processing, and these differences often center around the understandings of truth and meaning. In particular, one should note that the “truth” that goes into symbol processing has everything to do with whether the symbols are “properly” arranged in ways that are specified by the rules of the system. In other words–if you write a computer program–it doesn’t work unless you write your code in the proper way–if you forget that semi-colon at the end of that line, you can be totally boned even if you do everything else correct. Even more so–this kind of “truth” doesn’t have any necessary connection to what most humans would consider to be meaning or understanding. Staying with our computer program metaphor–it is entirely possible to write computer programs correctly that do nothing but produce gibberish. This program is still generating “correct” or “true” strings of symbols based on the rules it possesses even if the strings of symbols are totally meaningless to anyone else looking at them. Now–most computer programs are obviously constructed to do something we consider “valuable”–but that has nothing to do with the facts or structures that are inherent to the computer language itself. Meaningful output from these programs only occurs because people use these structures in particular ways. This becomes very obvious when you get early versions of computer programs that still contain “bugs” in them–and you get output that is non-sensical.. until the bug is removed and new code is added in to exclude these kinds of output.
4. This example brings up the crucial aspect that in symbol processing systems, almost everything has to be defined from the outset if we are going to make things work. The example of programs that go through more and more versions to gradually winnow away “errors” comes about in a way that goes about excluding more and more possibilities. The applicability of these kinds of symbol processing systems are thus always a kind of closed domain or closed system. They exist in a realm where everything is defined or must be defined–thus–they form a very interesting relationship to concepts like creativity or novelty. While it is not impossible to be creative with these kinds of systems–they are much more like legos than clay when it comes to mimicking reality–they are dependent upon the forms and rules of their basic components and thus their “re-creation”/simulation of reality will never attain the kinds of accuracy that other kinds of systems will..
5. Now.. I’ve spent a lot of time on symbol processing–what about signal processing…Well, in certain ways, the signal vs symbol processing might be a false equivalence in that symbol processing seems to be greatly subsumed under signal processing overall. I say this in the sense that an input stream of symbols can be interpreted as just a kind of signal and then any kind of symbol processing is just a particular subset of signal processing.
However–one important way that I see them as different is that signal processing overall is different in that the source of signals includes (and actually, I might argue is overwhelmingly) a phenomenon that deals with open-domain kinds of input that do not (mainly) come from human constructions. In other words–while symbol processing requires and is founded upon the construction of a set of symbols from the outset by a consciousness in order to work–signal processing originated as an evolutionary adaptation to environmental stimuli and it preceded any kind of consciousness by many, many millions, if not billions of years.
In this sense–signal processing may actually be formative of consciousness whereas symbol processing is formed by consciousness–which makes them very different kinds of things. Another way of phrasing/analogizing it is that signal processing is part of the basic foundation upon which consciousness was built–in the sense that without signal processing, nothing like consciousness would probably exist… On the other hand, symbol processing–i.e. mathematics, aspects of human language, computer programs–only seem to exist after conscsiousness has already been created–and additionally–only a while after consciousness seems to have formed.
6. Now–what is again interesting to me is the fact that there seem to be a lot of people who now–because we live at a point in time when symbol processing has become an extremely important facet of many people’s daily lives–posit that the very definition of what thinking and consciousness is–is symbol processing. They do this when they talk about the mind being like a computer–or when they claim that all thinking and knowledge is based on human language.
This gets the entire history of human evolution backwards, though. It also gets really problematic when people–who adhere to this kind of assumption that symbol processing is equal to thinking–try to explain just what the hell meaning is. In symbol processing, the symbols themselves are assumed to be content less–they are just markers that obey certain rules laid down by axioms or initial guidelines. I think a decent analogy is again to look at a computer program that randomly spits out strings of letters. While you can create a program that works perfectly at generating a random string of letters–or even non-random strings–but strings trying to obey certain rules–it is very unlikely that the string of letters being generated is the same thing as how humans think and construct languages. Words are not content-less strings of symbols–but instead are conscious constructs that are meant to express meaningful ideas. Furthermore–a program generating strings of symbols that are complete gibberish would obviously not be making meaningful statements–it would still be creating entirely truthful statements within it’s own closed domain in the sense that the gibberish strings would still be obeying all the rules of the original program language.
7. Before I close, one thing I want to mention is that although I think it is obvious that I do not think symbol processing is a valid model for what human thought actually entails–I also want to note that I don’t think signal processing by itself would function as a sufficient model of human thought either. While I think that signal processing–which helps create the foundation of perception for higher lifeforms (and if you consider chemical “signals”–for lower lifeforms also!)–I think that real consciousness requires more than just processing of external stimuli. It requires things like memory and kinds of internal processing that are driven without further external input as well as other things that I’m not remembering right now. In this sense, you can see “kinds” of consciousness when you look at other animals–say like a cat–that make them far more similar to the idea of consciousness that free-willed hairless monkeys seem to employ–than to wind up machines following a particular program.
8. There is no 8.
9. I’m sorry this has been rambling–but it is part of my process of trying to grok consciousness at this point in time. So go and process the symbols I’ve written down and see if it helps you understand the signals that you are constantly perceiving all the time…