Thinking About "Dual Descriptions"

I want to look at the broadly Davidsonian idea of “dual descriptions” of mental and physical phenomena, especially the idea that “reason” and “cause” represent different functions of a token-identical item. To do this, I’ll use a musical analogy.

In Western music theory, chords are often identified and described in terms of what they do in a musical context. The same chord may be give two different descriptions, depending on function. A typical example is the four-note chord consisting of the intervals of a major 3rd, minor 3rd, and another minor 3rd(or augmented 2nd; I’ll explain the (non)difference in a minute), stacked one atop the other with the top note of the first interval forming the bottom note of the next.

This four-note chord can be described as either a dominant 7th chord or an augmented 6th chord (the so-called “German 6th”), depending upon the musical context. The details don’t much matter, except to say that notes, and intervals, can be “spelled” different ways using our Western alphabetic system of designating notes. A particular spelling will usually be chosen to help the musician understand the functional role of the musical item. So in this case, if the chord is being described as a dominant 7th, the top interval will be spelled as a minor 3rd; if it’s an augmented 6th, that same interval will be spelled as an augmented 2nd. This is the same principle – known as “enharmonic spelling” – that is demonstrated in the more familiar example that the notes C# and Db are identical, just spelled differently.

The important thing is that both these descriptions of the four-note chord are at the same level – they are both functional descriptions, intended to shed light on what the chord does in a particular context. (Again, not to get too far into the weeds, but the chord can resolve in two different “correct” ways, according to 19th century theory, and the description of the chord will show which way it’s going to resolve.) Both descriptions assume the same subject matter, so to speak: four notes arranged in a particular structural relation, as described. (For that too is a description; but not a functional one.)

Bear with me; the end of the music lesson is in sight!

So, is there a different level of description, when it comes to a musical chord? Sure: the acoustic properties of the notes themselves. These are expressed in frequencies, usually in terms of cycles per second of vibration (Hertz). So another way to describe the note “middle C”, for instance, would be to call it “the note that vibrates at 261.63 Hz”. And we can expand this description to include the other three notes comprising our chord, all of which have precise Hz descriptions. Combining all four notes, finally, we could describe the total chord as a simultaneous sounding of the four frequencies, described in Hz. (There’s another whole story about what has to happen for a frequency to be considered a pitch in music, such as middle C, but that’s a side issue.)

An interesting point to keep in mind is that, in order to use this description, we have to pick a particular note for the chord to be built on, whereas the functional description is abstract; it applies to any chord that’s organized in the required way. Musicians pinpoint particular instances of abstract chordal shapes by using the alphabetical name of one of the notes, usually the bottom note.

Now we can start to see a glimmer of an analogy with the physical/mental problem. There’s good reason to say that the Hz description of notes is the foundational or basic description – the “scientific” description, if you like. It is the underlying “something” upon which our functional descriptions are premised. We can even invoke a sort of causal arrow: An underlying acoustical reality is required for there to be any functions, but we do not need the functions in order for there to be acoustics.

Similarly, many people want to say that brain states or neuronal firings are the physical reality upon which thoughts must supervene. We’re not quite at the point where we can give a “brain description” of any given thought, in the same way we can give the frequency of a note, but we’re getting close, and there seems no reason in principle why it wouldn’t be possible. So, if we grant this, then in theory we could re-describe any thought in the “brain description” language.

What would this be like?

We can return to the musical analogy for some insight. Why don’t musicians talk to each other using frequency descriptions, despite the fact that these descriptions are readily available and based on a fairly simple system of nomenclature? Well, we do, sometimes; in recording, we often need to discuss musical tones in terms of Hz, because we need to know the acoustic properties in order to get a piece of recorded music to sound better. (Musicians will know that I’ve vastly oversimplified how frequencies appear in music; all natural tones have overtones, also expressed in Hz, which dramatically affect how a note sounds and combines, though not its basic pitch.)

But when performers talk to each other, they use the “higher” level of description. They call a C a C, and not 261.63 Hz. Why? Several reasons. First of all, “261.63 Hz” is too precise; it refers only to middle C, as I’ve explained, and often musicians think in terms of abstract pitch in general, not a specific octave. But more basically, it’s just too cumbersome. We came up with “middle C” and the rest of the alphabetic system to (more or less) describe certain frequencies, and it prospered because it’s so easy to use.

But there’s another reason as well: We didn’t have access to the Hz vocabulary when we first began talking about, and playing, music. We didn’t know we were describing frequencies. Going back to Pythagoras, at least, musicians knew that notes stood in semi-mathematical relationship to each other, and that these relationships did not depend on which notes were involved – the relationships were abstract. What they didn’t have was an acoustic language in which to describe this. In other words, they knew that the ratio of root to octave was 2:1, based on listening to different lengths of strings vibrate, but they didn’t know of what it was a ratio. So by the time the science of acoustics came along, the musical vocabulary at the functional-harmonic level was very firmly established, and switching would have been difficult even if it weren’t so cumbersome.

Again, notice the similarities with the physical/mental conundrum. Humans have, for thousands of years, described thoughts in terms of their “subject” or “content” or “intension” or “meaning,” and developed elaborate ways to describe how thoughts do or don’t go together to get certain jobs done. What I called “brain descriptions” didn’t exist. Instead we had “logic,” or “rhetoric,” or “poetry,” or “politeness.”

So far, so good, as an analogy. But here’s where I think it breaks down, and the physical/mental problem assumes a unique configuration.

We could imagine a Davidsonian “anomalous monism of music,” in which the chord “C dominant 7th”, played at a certain time and place, is token-identical to its acoustical substrates. But that chord would not be type-identical to the four Hz descriptions beginning (if the chord starts on middle C) with “261.63 Hz”. Rather, we would have two different descriptions of the same thing, just as we allow reasons and causes to serve as different descriptions of . . . but here’s the problem.

Different descriptions of what? What does that innocent phrase “the same thing” refer to here?

We know how to answer that question when it comes to music: “C dominant 7th ” and “261.63 Hz . . . et al.” both describe, or identify, an acoustical event – the first in terms of function, the second in terms of acoustical composition. But what is the analogous way to answer the question about thoughts and neurons? Should we just say that neuronal firing is what happens, and that thoughts are a “functional description” like “dominant 7th”? But what is the function that requires the re-description? What are thoughts doing that neuron-firings are not doing?

The thought “My cup is yellow,” when thought by me at a specified time and place, is token-identical to whatever pattern of neuron-firing takes place simultaneously, if we adopt the Davidsonian scheme. But what we usually call the propositional content (or however we want to name the feature which could allow that thought to function as a reason rather than a cause) is not type-identical.* If “My cup is yellow” were a premise in an argument, it would not be possible to substitute the “brain description” language, in terms of neurons firing at a particular time and place, and still have the argument go through. If we could make such a substitution for all the statements in the argument, we would still not have a valid conclusion. At best, we might have a blueprint showing why a certain brain event causes another brain event (whether in a lawlike fashion or not; I’m not aware that Davidson talked much about brain events causing other brain events), but this isn’t equivalent to providing reasons why a certain proposition entails another.

[* Another possibility: Are “My cup is yellow” (thought by me, now) and the accompanying neural firings extensionally equivalent? I would welcome others’ opinions, but I don’t think they are. I find it hard to accept that “My cup is yellow” has an extension, so I don’t think the concept of extensional equivalence is going to help us here.]

This is all either very apparent or quite mysterious, depending on your philosophical commitments. I believe it’s apparent, if we allow that something like Sellars’ Space of Reasons exists at all. (And that, I freely admit, isn’t obvious, but I operate on the premise that it’s likely true.) Asking a pattern of neuronal firing to explain an argument is a category mistake, on this view. The type-non-identity is crucial.

So why doesn’t this continue to conform to the musical analogy? This is where – no doubt to the despair and impatience of many – we have to invoke the Hard Problem. We understand how the translation from an acoustic description of our four-note chord to its functional re-descriptions works. We know what the functions are that we want explained, and we understand why a description like “C dominant 7th chord” can supervene on the acoustic phenomenon which is also describable, in a much more cumbersome fashion, in terms of four Hz numbers. All this is possible because we agree that the acoustic-level description is the more basic. The functional description “C dominant 7th” just is a way of showing what the acoustical-level phenomenon can be used for.

Could we ever say that about neuron-firings and thoughts? Does Davidson’s monism ask us to make a similar agreement about the physical basis of thoughts? Ontologically (not necessarily descriptively), we’re asked to agree that the physical is indeed basic – physical stuff is all there is, if we’re inventorying the items that may be found in the world. But Davidson, as I read him, doesn’t insist that the physical description must be basic. In fact, he denies it. It would back him into the position of saying something like “Well, if you want the most basic description of my thought ‛The cup is yellow,’ you’ll need to discover the corresponding pattern of neuron firing.” He’s not saying that – though that is exactly what we would say, using the appropriate vocabulary, about a musical event like a chord.

The problem, then, is: Is there a “description-neutral” way of describing what the “thing” is in the phrase “the same thing”? I don’t see how there can be, when “the same thing” is meant to refer to whatever is being described in both physical and mental language. We could use either description – physical or mental – to name the event in question, but since neither description is basic, we don’t know what it would mean to compare them along the axis of “sameness”.

How do we get from “p describes X” and “q describes Y” to “X and Y are the same thing”? The missing premise, presumably, is “p and q describe the same thing”. But how would we know this, without a previous description of X/Y which is neither the physical nor the mental description? This is either viciously circular, or else just incoherent.

In short, it seems that we need a stipulated “basic description” before the concept of “same description” can really get off the ground. This usually isn’t problematic – most items have a basic description – but I’m arguing that it becomes so in the mental/physical conundrum.

To conclude: We would need to be more physicalist than Davidson if we wanted to claim that the ontological primacy of the physical results in a “basic description” that is also primary. Davidson doesn’t say that, and nor should we. Neither “cause” (physical) nor “reason” (mental) is a basic description in the way that the Hz language is a basic description of musical tones. So we’re left with a puzzle about how to talk about the relation between these two types of description. Which in the end is another way of saying that the Hard Problem is still opaque to us.

3 Likes

Of course not? I thought Davidson must be considering a limit case where “tokens” counted as the same type only because of a sci-fi duplication of the whole person?

Brain patterns clearly aren’t subjected to a classification into types, since we don’t have anything like the access normally required for such a classification.

This is true, but do you see any reason to suppose this won’t happen eventually? We can already say with confidence which parts of the brain “light up” under certain stimuli. I think a precise “neurons-to-thought” map is well within theoretical possibility. (And won’t that be fun world to live in . . . mind-reading via brain-scanning. Thought Police etc.)

But let’s say it isn’t possible. Should that be our reply to a physicalist who wants to see the brain-level events as the “basic description”? Should we say, “We’ll never get such a description, so it can’t be the basic one?” I don’t think that’s strong enough. I think we should concede the possibility of such a description and then argue against its being basic.

I suppose it’s conceivable that we could read brain-patterns from brains, and write them into brains, and were able to equate each pattern with consciousness of a particular sentence, or chord. Yes.

I agree it seems to miss the point. Hmm.

And then I agree that this scenario matches the current situation with computers, except for the addition of consciousness. So I see what you mean about it returning us to the hard problem. In which case I resort to my usual solution, which is that consciousness requires Searle’s “proper semantics”: the machine must play a social game of connecting the words (or music) to bits of the world.

Which is how the very idea of mental machine code misses the point. Thoughts aren’t symbols. They are the brain playing a game with symbols and other bits of the (external) world.

We don’t need to waste time trying to force a shared vocabulary when we can just reason this out directly. Look at how simulation theory cuts straight through the whole mess. If anything you can define as a rule, function, or relation is just code running a simulation, then the entire debate vanishes. Inside a simulation, a character can never use the local physics or biological senses to achieve true understanding, because you can’t debug the simulation using the tools the simulation itself has made available for you.

​You don’t need to invent some clunky physical bridge to tie a neural cause to a mental reason. That whole struggle assumes they are two separate things trapped on the same floor, desperately needing a translation to begin with. They aren’t. The neuron firing and the subsequent thought are just different executions running off the exact same upstream within the simulation. No neutral baseline required.

What’s missing here is theology. That would be you outside the simulation. The bridge would be Platos Allegory Of The Cave.

(Much like in one of my favorite movies interstellar, the scientists spent a lot of their time trying to figure out their hard problem which was not possible from outside the singularity or black hole. It took a higher life form to do it. I simply reason, much like the main character in the movie, that the higher life form isn’t a deity per say, it’s us and our current models are simply wrong including some of our shared vocabulary. It would take a talented philosopher to provide cohension which starts at redefining our base perspectives. )

OK. So the question would be, is “the exact same upstream within the simulation” the basic description?

I think we would need multiple levels of description to address the topic, but as food for thought…

A step above neuron firings are networks of neurons which have been trained in various ways. For example a person might have a neural network that has learned to discriminate decimal digits in the person’s visual field. (Or a dominant 7th heard.) We might talk about the intentionality/aboutness of the output states of such neural networks while ignoring the underlying details of neural firings. Then we would need to consider how the brain integrates a great number of intentional neural network outputs into a conscious experience.

Yes, this is where the language really gets tricky. Sounds like we both agree that neural firings don’t have “aboutness” or intentionality. But if I’m understanding you, you’re suggesting that the output states of neural networks could be so described. Could you say more about that? Does the intentionality reside in the output state as interpreted by us, or may the neural networks themselves be described intentionally? Or neither . . . but please say more.

I do not follow the music theory, but it sounds like most of the issue lies in the limitations of human language. Humans use language (as in, syllables, words and sentences) to describe things, also in science. A more advanced science and humanity would be “no-wording” content.

You are also touching upon description of things (sensu lato) involving different subject-parts, often described as different “levels”. Hearing a specific deep tone and hearing a 50 Hz sine-wave is the exact same if the specific deep tone is the tone that is heard when a 50 Hz sine-wave is generated and presented (“heard”). The description is different, because the latter description here introduces waves and Hertz.

And that is not even introducing that, just because a 50 Hz sine wave is touching your eardrum does not mean it is heard the same.

If you wanna discuss cause, reason and effect… I think that is something else that you are trying to talk about.

Perhaps we can get a shorter example that does not involve music theory?

Brains patterns can be categorized, and is.

For example: Decoding Patterns of Human Brain Activity - PMC

Yes. And “specific deep tone” is a description too, albeit not a very helpful one.

Quite right.

That’s fair. I picked music because it’s one of my fields and I’m fairly sure I won’t get it wrong!

But OK, another example might be using an object for practical versus artistic purposes. The vessel in front of me can be described in cooking-related terms (volume, durability, etc.) if it’s being used in the kitchen. Or it can be described in aesthetic terms (color, texture, form) if it’s in a museum. Both descriptions are functional; they’re designed to help us pick out features that will be relevant to a particular use or context. But there is also an underlying physical description which both functional descriptions assume and supervene upon . . . and here I’m outside my expertise so I don’t know the terminology. Something about chemical composition and dimensions, I would imagine – whatever instructions one would need to reproduce the object exactly, with no reference to possible uses.

I recommend this video as a brief but well done introduction to neural net information processing.

In light of what the video explains, it seems reasonable to say that the output of the trained neural network is about numerals presented to the input. So inasmuch as intentionality is understood as “aboutness”, we can say that intentionality is an emergent property of the trained neural network.
Now how other parts of the brain make use of the outputs of many such subnets is still quite mysterious, but to me this seems a useful step towards carving nature at its joints.

Thank you! I’ll watch it ASAP.

I understand the intend now. My answer is that, the vessel (container of some sort?) you mention does not have description in physical form whatsover. To say so would be to “languify” observations, and even worse, “languify” physical phenomena. You can describe what we observe (physical tokens as perceived), granted, but that is just language.

This “languification” touches relativism and non-realism, I would say. But if you want to go this way, you would be postulating the existence of an intrinsic/sui generis physics-description “thing”. It would be combining language and matter/energy in some way.

Next steps would be to investigate whether this description/language is a physical thing or a metaphysical thing. From reading what you write, it seems that it is (postulated to be) a physical thing.

What was the question again?

No, there isn’t a basic description you can hunt down inside the simulation. It doesn’t live on the same plane as us.

What about sleep? ​Look at how the math breaks down if you map it against a average physical timespan, if you live to be 99 years old, roughly a third of that time,33 years,is spent completely decoupled in sleep or unconsciousness. What happens to you during that chunk of time? Dismissing it as absolute nothingness is just a lazy cop-out when “nothingness” doesn’t actually answer the mechanics for anyone paying attention. You’re outside the simulation. That’s when. By most of our rules, you’re not dead or alive.

Epistemic or ontological emergence? Wetness is an epistemic emergence of a ~1000 H2O molecules but wetness is just the Cliff notes version of hundreds and thousands and tens of thousands of molecules are doing. We actually don’t (theoretically, at least) need to refer to the Cliff notes since all of causality dealing with the wetness of water can be accounted for without referring to wetness. Can the same be said about intentionality?

Epistemic (or weak) emergence.

Yes, the same can be said about the conception of intentionality that I am suggesting.

1 Like

This is a good opportunity to point out that a “basic description” as I’m using the term is indeed made of linguistic elements – words, numbers, etc. I suppose some philosophers might go on to say that there is nothing beyond such elements, but that would indeed be to “languify” physical phenomena. Rather, I’m maintaining the distinction between basic ontology and basic description, as I believe Davidson does. If one wanted to say that a basic description is “less basic” than ontology, I would agree. But it’s the question of dual description that occupies us here.

OK. So, applying the analogy to our world, can there ever be a truly basic description we can devise for anything? Do the things I’ve claimed do have basic descriptions – musical tones, for instance – have them only in a relative, “inside-the-simulation” sense?

Look at how science operates. Our entire existence is spent reverse-engineering a black-box environment because we don’t natively understand the underlying laws governing it. We grab pieces of the mechanics and stitch them together, with physicists treating a “Theory of Everything” as the ultimate holy grail.

​That reverse-engineering process is our makeshift bridge for understanding, but it hits a wall by design. Hard limits like the Planck scale, the speed of light, or quantum indeterminacy aren’t just accidental gaps in our toolset; they are hardcoded rendering boundaries built into our existence. They represent the exact threshold where the local laws breaks down because you are trying to use the simulations own tools to look past itself. You can’t dive any deeper, because the simulation’s architecture intentionally stops you from seeing behind the curtain by design. You need to re-invent the wheel, which many philosophers are trying to do, I’ve ran into two or three in this community who are attempting that very thing.

Most of us are hobbyist and are not on the front lines of usually the celebrity world who makeup the standards of where we are at in that arena. The exception to this are the shadow leaders if you believe in that kind of thing. Without Locke, Newton may have never had released anything to the world.