In a post here at Uncommon Descent, Upright Biped makes what he calls his Semiotic Argument for Design, which he has been challenging me to refute for some time now, but which I have been struggling to understand. So it was good to see it summarised in one place, and I’d like to take a look at it piece by piece, and with your help, try to figure out what he’s getting at (I’m assuming he’s a he, which I don’t normally do, but I think he said something once that implied he was).
It’s a response to Larry Moran who dropped by from his Sandwalk blog to talk about onion genomes, but we don’t have to worry about onions too much, I don’t think, as UBP is making a more fundamental claim.
He says to Larry:
In your comments you refer to the use of the term “information” within nucleic sequences as a useful analogy, and you say that there is no expectations that it should “conform to the meanings of “information” in other disciplines.” I certainly agree with you that it conforming to other meanings would be a telling turn of events. And I assume your comment suggests that the nucleotide sequence isn’t expected to share any of the same physical characteristics as other forms of information – given that we live in a physical universe where information has physical effects. Ones which we can observe.
So UBP’s starting point seems to be that the “information” we say a genome contains is not different from “information” in other senses.
I think it makes an interesting comparison; the comparison between the physical characteristics of information transfer in the genome, versus information transfer in other forms. Just recently on this forum we were having a conversation about recorded information, and a question arose if a music box cylinder ‘contained information’. Speaking to its physical characteristics, the answer I gave was “yes”. Just like any other form of recorded information, the pins on a music box are an arrangement of matter to act as a representation within a system. No differently than ink on paper, or the state of a microprocessor, or the lines left on a recording tape, or an ant’s pheromones, or the tone of vibrations we make when we speak; they are all matter/energy arranged in order to represent an effect within a system.
A musical box is an interesting example, because unlike some other information transfer systems, like the pixels on your screen you are viewing right now and from which you are receiving information about the contents of Upright Biped’s post on UD, the sequence of pins on a music box cylinder are actually instrumental in making something else, in this case a melody, and thus bears a closer homology to the sequence of base pairs on a DNA molecule which, by a series of physical operations, results in the making of a protein.
He goes on to say:
It was also pointed out that a physical arrangement of matter (like the pins on a music box cylinder) cannot by themselves convey information – they require a second coordinated physical object. This second object is easily referred to as a protocol, but physically its is a rule (a protocol) established in a material object. The necessity of this physical protocol is something easily understood; for one thing to represent another thing within a system, it must be separate from it, and if it is truly a separate thing, then there must be something to establish the relationship that exist between the representation and the effect it is to represent (within that system). That is what the second physical object accomplishes, it establishes the relationship between a representation and the effect it represents, which is a relationship that otherwise wouldn’t exist.
UPB claims the music box represents “recorded information”, which implies that the information started elsewhere and was “recorded” on to the music box. However, I think he is making the point that without a mechanism to get the “recorded” music box back into the form of music again, the information isn’t truly recorded, which seems fair enough. After all, if I translated my post into unbreakable code, it wouldn’t really be recorded information because there would be no way of getting the information back out again. So in the music box example, the music box is a way of “recording” a piece of music composed by someone, and getting that piece of music back out again, at a different time. A phonograph recording (the old wax cylinder kind) would be an even better example.
Next he says:
There have been examples of this dynamic given in previous conversations. For instance, an apple is an apple, but the word “apple” is a separate thing altogether. Being a separate thing from the apple, there must be something that establishes the relationship between the two. In the case of the word “apple” we as humans have learned the protocols of our individual languages, and they physically exist as neural patterns within our brains. These neural patterns are material things, and they establish the immaterial relationship between a physical representation and its physical effect.
Hmmm. OK, we have an apple – is the apple the information that is recorded? Not really. When we say “apple” we are not “recording” an apple, although we may be “recording” the fact that we are thinking about an apple. Let’s say that’s it – I am the original source of the information that I have seen an apple. I wish to transmit that information, so I record it in the sound pattern: apple. That, presumably, is the equivalent to making a cylinder with a sequence of pins. But the sound is useless in conveying the information that I have seen an apple unless my listener has some way of turning my thought “there’s an apple” into her thought “Lizzie has seen an apple”. UPD proposes that the mechanism by which that translation is occurred is analogous to the music-box mechanims – the sequence of sounds, like the sequence of pins on the music box cylinder, impinges on Lizzie’s hearer’s ears, is translated into neural impulses which are after a great deal of processing, evoke the thought “Lizzie has seen an apple”.
No, I don’t think that works. For a start, “Lizzie has seen an apple” is not a recording of Lizzie’s thought, it is an inference about what Lizzie was thinking. Lizzie’s hearer has indeed received information from Lizzie, but not exactly the information that Lizzie sent. So it seems to me that language is a very different kind of information transfer system from a phonograph, or a musical box, or, indeed a reproducing organism. In these three examples, we start with a physical pattern of some kind (a performance, a melody, an organism) and we end with a recreation of that physical pattern (a rendering of that performance, the sound of that melody, a second organism, a second organism). In the case of language we do not. There is no protocol that can create an apple from the word apple, although uttering the word may induce someone else to go fetch one. This, it seems to me is because the word “apple” is a symbol, or, in Saussure’s term, a signifier that is linked to a signified (aka referent) in this case a specific kind of fruit. UBP appears to want to say that this linkage between signifier (word) and signified (fruit) is equivalent to the link between the sequence on a music box cylinder and the melody that emerges, and that therefore the music box cylinder (and therefore a base sequence in a polynucleotide too) is a symbol for the sound pattern that emerges from the music box in the same way as the word “apple” is a symbol for an actual apple.
But it clearly is not. When I say the word apple, and you hear it, no apple is created, though you may reproduce my image of an apple in your own inner eye. But the referent for the signifier “apple” is not “the mental image of an apple” but an actual apple. So the linkage between signifier and signified in language (the relationship of a “sign”) is qualitatively different from the relationship between a recorded physical object or pattern and its reproduction.
UBP then says:
This same dynamic is found in all other cases of recorded information. I have previously used the example of a bee’s dance; a bee dancing in a particular way during flight is a separate thing than having the other bees fly off in a particular direction, and the relationship between the two is brought about by a protocol which physically exist in the sensory system of the bee.
In the dynamics of information transfer, the operative observation is that each of these physical things (the representations, the protocols, and their resulting effects) always remains discrete. This is one of the key observations that allows information to exist at all. The input of information is always discrete from the output effect, and the protocol that establishes the relationship between the two, remains discrete as well. They are three completely independent physical realities which share a relationship, with the protocol establishing the relationship between the representation and its effect within the system. In no case does the representation (or the protocol) ever become the effect.
Well, this seems a little circular. It seems to me that UBP is defining recorded information as something that requires a discrete protocol, then regards it as noteworthy that all instances of recorded information require a discrete protocol. When we lived in Canada we had a deck with a table that had an umbrella hole in the middle. One day it snowed heavily, and soon there was an inch of snow on the table, with a hole in the middle. Half an hour later, there was a foot of snow, with a depression in the middle. Someone came in and said – “what’s that dimple in the middle of the snow?” and then said – “hey, it’s the umbrella hole”. In other words, the table – an object with pattern – was being replicated with each layer of snow, with sufficient fidelity that an observer could extract from the layer of snow the information that the table had an umbrella hole. By evening there was about 4 feet of snow on the table, but there was still a dimple in the middle, indicating that the information that beneath the snow was a table with an umbrella hole had been faithfully recorded and transferred from snow-layer to snow-layer all afternoon. Yet in this case, the “representation” was also the “effect”.
However, that seems to me to be the least of the problems with UBP’s case. The far bigger problem is that there is a qualitative difference between a sign (in the Saussurian sense), namely a linked signifier with signified pair, where the signified can be a physical object, and the signifier a symbol potentially renderable in a number of media, and where the transfer of information using the signifier does not result in the physical creation of the referent, and the information transfer in a musical box or in a reproducing organism whereby a physical pattern is recorded in such a way that it can be reproduced, which, at its simplest, can be layers of snow on a table.
This same dynamic is found in all forms of recorded information; including those used in the information processing systems created by intelligence. As an example, the first automated fabric looms used an arrangement of holes punched into paper cards (which acted as physical representations of the resulting effects within the fabric). Sensors and pins within the machine would sense where the holes were punched, and it would use that information to change and control the colors of threads being woven. In this instance, the configuration of holes served as the representation, and the configuration of sensors served as the protocol, leading to the specified effects. Each of these is physically discrete, while sharing the immaterial relationship established by the protocol.
Well, yes, but the discreteness is, as I’ve said, only arguably intrinsic to the concept of “recorded information” and in any case, does not render it semiotic. At least not in the Saussurian sense. Not sure of what other sense there might be.
So here we have a series of observations regarding the physicality of recorded information which repeat themselves throughout every form – no matter whether that information is bound to humans, or human intelligence, or other living things, or non-living machines. There is a list of physical entailments of recorded information that can therefore be generalized and compiled without regard to the source of the information. In other words, the list is only about the physical entailments of the information, not its source. I am using the word “entailment” in the standard sense – to impose as a necessary result (Merriam-Webster). These physical entailments are a necessary result of the existence of recorded information transfer. And they are observable. That list includes the four material observations as discussed in the previous paragraphs: a) the existence of an arrangement of matter acting as a physical representation, b) the existence of an arrangement of matter to establish the relationship between a representation and the effect it represents within a system (the protocol), c) the existence of physical effects being driven by the input of the representations, and d) the dynamic property that they each remain discrete. Observations of systems that satisfy these four requirements confirms the existence of actual (not analogous) information transfer.
OK, let’s take these in turn:
a) the existence of an arrangement of matter acting as a physical representation
Well, maybe, though it’s a bit imprecise. But sure, information transfer is going to entail physical arrangements of matter. And let’s allow “representation” to be the thing-that-is-read, like DNA, or the cylinder of the musical box, or even the pattern of sounds making the word “apple” and let that representation be of something (a whole organism; a melody; an apple).
b) the existence of an arrangement of matter to establish the relationship between a representation and the effect it represents within a system (the protocol)
Well, no. In the case of the linkage between the signifier “apple” and its referent, the piece of fruit, there is no “arrangement of matter”. There is some kind of “arrangment of matter” that links the signifier “apple” to the evocation of the idea of an apple in a hearer, but the “idea of an apple” is not the referent of the signifier “apple”. What links the word “apple” to apples is shared agreement among a community of speakers that “apple” means apple. And that agreement is transmitted culturally by various means, not a single protocol, largely by usage, and sometimes trial and error (for years I thought the referent for “caution” was “luggage” because on the back of the seats of the bus I used to catch to school it said “caution racks overhead”, and I also knew there were “caution lorries”). And even if we allowed this as the “protocol” UBP refers to, no amount of cultural agreement that “apple ” means apple will make an apple assemble itself when someone says the word “apple”.
c) the existence of physical effects being driven by the input of the representations
Well, I guess I would agree, not least because as a “materialist” neuroscientist I would regard the uttering of the word “apple” or even the activation of the motor program at sub-execution threshold that would result in the uttering of the word is driven by input of some sort, even if there is no-one except me to “hear” the word/thought.
d) the dynamic property that they each remain discrete
Well, I’m not sure what is premise and what is conclusion here. But let it pass for now.
These same entailments are is found in the transfer of information from a nucleic sequence. During protein synthesis a selected sequence of nucleotides are copied, and the representations contained within that copy are fed into a ribosome. The output of that ribosome is a chain of amino acids which will then become the protein being prescribed by the input sequence. The input of information is therefore driving the output production. But the input and the output are physically discrete, as evidenced by the fact that the don’t directly interact, and that the material output is not assembled from the material input.
Well, no. The problem seems to be entailment b, as it always has been. A semiotic system relates a signifier to a signified so that two members of a shared linguistic community can communicate ideas – i.e. one member of the community can evoke in the mind of another member the idea s/he is currently entertaining. It is also key to thought itself. It is not a system that records a physical pattern then reproduces it – or rather, in the only sense it is such a system (I can partially reproduce in your brain the thoughts I am currently having by means of this written symbol system), it doesn’t map on to either a musical box or protein synthesis. The referents of my signifiers are not my thoughts, but real-world objects, and abstract concepts. Those real world objects and abstract concepts are not brought into actuality when I utter a word. Unfortunately.
The exchange of information (from input to output) is facilitated by a set of special physical objects – the protocols – tRNA and its entourage of aminoacyl synthetase. Acting together they make it possible for the input to alter the output, and they do so by allowing them to remain separate. The tRNA physically bridges the gap between the input and the output, acting as a passive carrier of the physical protocol. It accomplishes this by being charged with the correct amino acid by the synthetases (the only molecules in biology which actually hold the rules to the genetic code). The synthetases accomplish their tasks by being able to physically recognize both the tRNAs and the amino acids. They charge the tRNAs with their correct amino acids before they ever enter the ribosome. The actions of the synthetases are therefore completely isolated from both the input and output. In other words, the only molecules in biology that can set the rule that “this maps to that” are physically isolated from both the input and output, while the input and output remain isolated themselves.
Sure. As in a musical box. But not as in a semiotic system.
These observations establish that the entailed objects (and dynamic relationships) exist the same in the translation of genetic information as they do in any other type of recorded information (in every example from human language, to computer and machine code, to a bee’s dance). These observations have been attacked as being as a misuse of the definition of words (a semantic word game, as you call it). But I have already produced the definitions of the words from a standard dictionary; I’ve restated the observations using those definitions in place of the words themselves; and I have asked the question: “If in one instance we have a thing that actually is a symbolic representation, and in another we have something that just acts like a symbolic representation – then someone can surely look at the physical evidence and point to the distinction between the two. There is also the simple fact that there is nothing about the attachment of cytosine to thymine to adenine that intrinsically means “bind leucine to a nearby polypeptide” as an inherent property of its matter. That is a quality beyond its mere materiality, one it takes on by being in a system with the correct protocol to cause that effect from that arrangement of matter.
But not as in human language. Nor in computer code for exactly the same reason.
There has also been the profoundly illogical objection that because these things follow physical law (and can be understood), they cannot be considered symbols or symbolic representations. Not only does this deny the existence of any symbol in the extreme, it fails for the obvious reason that everything follows physical law. If something can’t be true because it follows the same laws as everything else, then we have entered the Twilight Zone.
The objection is not that because the information transfer follows physical laws it cannot be considered in terms of symbols or symbolic representation. I think symbols and symbolic representations are also the outcome of the operation of physical laws. The objection (or mine, at any rate) is that in a cell, or in an musical box, a physical pattern is recorded, and later reproduced, by means, as UBP says, of a physical protocol (I’m not bothered about whether there is a discrete intermediary or not, that doesn’t seem to me to be the crucial issue), that links the original to the reproduction. Whereas in language, we have signs in which a vocal signifier/symbol is linked to a real world signified/referent enabling information about the referent to be exchanged between speakers; they do not does result in the reproduction of the signified.
So going back to your comment, a fair reading suggests that the information transfer in the genome shouldn’t be expected to adhere to the qualities of other forms of information transfer. But as it turns out, it faithfully follows the same physical dynamics as any other form of recorded information. As for “disciplines”, you will notice that these observations are very much in the domain of semiotics. Demonstrating a system that satisfies the entailments (physical consequences) of recorded information, also confirms the existence of a semiotic state. It does so observationally. Yet, the descriptions of these entailments makes no reference to a mind. Certainly a living being with a mind can be tied to the observations of information transfer, but so can other living things and non-living machinery. It must be acknowledged, human beings did not invent iterative representative systems, or recorded information. We came along later and discovered they already existed.
And so we see where UBP makes his error, having started from a false premise, or at least false analogy. A semiotic system does indeed require a mind, because the link between signified and signifier is one that enables information about the signified to be transferred between members of a linguistic community of minds. But the kind of information transfer UBP calls “recorded information”, and observes in musical boxes and cells is a quite different kind of information transfer system, in which a physical original is reproduced at a later time, by means of a system of “protocols” that can be (IMO) as simple as a direct template (the reproduction in one iteration becomes the template for the next) or as complex as self-replicating modern cell, in which the original, or “parent” is reproduced as a “daughter”. Whether or not something like like the genetic code is involved or whether “discrete” physical intermediaries like a specific set tRNA molecules are required for faithful reproduction seems to me irrelevant to the question of whether a mind is required. A mind is certainly required for a semiotic system, but no matter how clever the mapping of codon to amino acid is, the result is not a Saussurian “sign” – there is no mapping of “signifier” to “signified” within a linguistic community of minds, simply a mapping of one molecule to another, the final result being the physical replication of the original, not the evocation of the original in another mind.
Therefore, the search for an answer to the rise of the recorded information in the genome needs to focus on mechanisms that can give rise to a semiotic state, since that is the way we find it.
We need a mechanism that can cause an arrangement of matter to serve as a physical representation. We need a mechanism that can establish within a physical object a relationship between two discrete things. To explain the existence of recorded information, we need a mechanism to satisfy the observed physical consequences of recorded information
Sure. But that does not require positing a mind, because “recorded information” as defined by UBP does not require a semiotic system.
I’ll invite Upright BiPed over to discuss this, and I do hope he will be willing to discuss it here. It’s a conversation that has looped over many threads at UD, and has been difficult to follow as a result (not helped by the new nested comments system, which is making me wonder whether I should abandon nested comments here….)