# Causal Systems

Recently, I’ve been involved in some discussions about causality, and some of those discussions have struggled to find any resolution, which I find frustrating. I don’t think people need to agree on ideas, but my experience is that usually people can agree on how to frame and talk about those ideas.

I sometimes get the feeling people are so set on disagreeing that they don’t always engage on what the other party is saying. I never know if it’s a lack of comprehension, a lack of willingness, or (on my part) a lack of communication skill or sufficient explanation.

So here are some things I think (I hope) are uncontroversial.

I’d like to consider five different physical causal systems.

What I mean is that these systems exist as fully identifiable physical entities that work according to well understood physical principles. There are no components or behaviors in these systems we don’t understand.

Per a recent post, we can both completely define and completely describe these systems.

§

It’s amazing how many pictures show the phones supposedly being used while the string is completely slack. I’m not sure everyone gets the idea. String must be taut!

The first causal system is the old two tin cans and string “phones” we made as kids in a bygone era.

The system consists of two empty tin cans with lids removed. Each can has a small hole in the center of its bottom. A string is threaded through from the inside — a knot at the end keeps it anchored to the can’s bottom.

The string, which is usually very long, is stretched and held taut.

If Alice speaks into one can, her vocal chords vibrate the air which vibrates the can which vibrates the string which vibrates the other can which vibrates the air in that other can which vibrates Bob’s eardrums, and he can hear what Alice said.

Bob can speak into his can, and a reverse chain of causes happens, and Alice can hear his voice.

The system isn’t efficient — voices sound faint and “tinny” — but it does work. A voice at one end, through a chain of physical causes, is transmitted to the other end.

But the fact of a chain of physical causes isn’t the point!

I can’t stress this enough. All systems that actually do something are physical; they all have physical causal chains.

The point is what is going on in those causal chains.

In this case, what is going on is physical vibrations being transferred from vocal chord to air to can to string to other can to other air to eardrum. And at every point along the way, these vibrations directly represent the speaker’s voice.

That’s the crucial point! If Alice or Bob speak loudly, the vibrations are strong. If they speak softly, the vibrations are weak. When they speak low or high pitched the vibrations directly reflect that.

We can “see” (i.e. in some way directly sense) the system’s information (the vibrations) at any point along the way, and that information looks like the vibrations.

A technical way to put this is that the state of the system at any point in the system directly and proportionally represents the primary information content of the system.

§

Close up of a phonograph stylus following a record groove. Those wiggles in the walls of the grooves are the sound vibrations. (Looks like Bach.)

The second causal system is similar, and I’ve discussed it in detail before — analog music recording, from musical instruments and voices to reproduced sound coming from speakers at a later time.

This system also deals with sound vibrations, but here the chain is more complex, requires external electrical power, and allows the vibrations to be stored and recalled later.

In this case the causal chain involves things like microphone diaphragms, electrical wires and components, magnetic tape and vinyl records, and (most importantly) speakers or headphones to recreate the vibrating air.

Even so, the state of the system at any point directly and proportionally represents the primary information (the music) the system is processing.

It might require an oscilloscope to see the electrical signal or an audio amplifier with a speaker to hear it, but the (electrical) vibrations of the system are the music.

With the tin can phones it’s easy to see how one part of the system physically pushes on another part of the system in order to transmit physical vibrations. Very much the same thing happens electronically — one part electronically “pushes” on the next part.

The crucial point is that the pushes directly and proportionally represent the original sound — the states of the system directly reflect primary information content.

One difference between sound recording and tin can phones involves the need for external electrical power. The latter were powered by the vibrations themselves. (Which demonstrates of the physicality of the causal chain!)

Electricity in sound recording both powers the system — that is, it enables the system to function — and at many points represents the sound — that is, it’s the signal.

At other points the signal is represented by magnetized domains on a strip of tape or by wiggles in the groove of a vinyl record. In all cases, electrical signal, magnetic signal, or wiggle signal, the original sound vibrations are clear and proportional.

§

A very early one-bit system that couldn’t really compute anything, but it did manage to shed a great deal of light on things.

The third causal system is a simple light switch circuit, such as found in most homes. This system includes the switch, the wiring, the light bulb.

Electric current plays a role here, too. In fact, it’s kind of the point of this system — electrical power to the bulb creates light for a dark room.

The chain of causality in this case is that someone flips the switch which closes a circuit which allows current to flow which heats up the filament in the bulb which emits lots of photons, and let there be light.

The primary information content, such as it is, is whether the switch is closed and current is flowing, or not. The physical state of the system, once again, clearly shows that content.

The switch has physical states, current flow is easy to detect (in fact, one needs to be careful detecting it!), and, of course, the bulb glows or not.

This system is a bit like the tin cans in that the power source is the signal, but in this case what controls the information content is whoever flips the switch.

[I’ve been saying “primary” information. The light bulb system, for example, has information about how hot the bulb gets, how much current flows, what the voltage is, etc. This is all secondary information because it is not conditional to providing light.]

§

One big complex metal system of gears, pistons, cams, belts, linkages, and spinning things. Lots of things causing things! (And 0 to 60 in 4 seconds!)

The fourth causal system is a conventional car engine.

There are a lot of causal chains going on in a car engine: pistons and valves reciprocating, shafts cranking, spark plugs sparking, pumps pumping, and so on. The causality involved mechanics, electronics, fluid dynamics, and combustion.

An engine has a network of separate causal chains that connect together to generate output. For example, moving gas from the tank through the carburetor and into the piston chambers is a distinct chain. Getting spark to the spark plugs is another.

For a car the primary information is the output revolutions and torque. (Which, like the light switch, is controlled by the operator.)

And, as in the previous cases, the states of the system directly and proportionally reflect that primary information content. When the car goes fast, the engine goes fast. When the car goes slow, so does the engine. If the car labors on a hill, the engine struggles. Downshift; it recovers.

§

Systems using lots and lots of light switches have really improved their capabilities, but they aren’t much help for reading in the dark.

The fifth causal system is a conventional (digital) computer.

This, like sound recording, is a system that uses electrical power both to enable the system to function and to represent the information the system handles.

On the small scale, there is similarity to the light switch in that individual signals (“bits”) are either on or off. The system treats ordered groups of many such signals as comprising unique symbols (“words”) based on the on-off patterns within each group.

By combining many bits, words can have many possible patterns. At a basic level of interpretation, those patterns are interpreted as numbers in base two.

But note that at the physical level, these are just a bunch of on-off switches. The causality of the computer is the causality of the switches. What the computer knows how to do — what its system states reflect it doing — is manipulate those switches according to built in rules.

I want to stress that the computer is fully defined, and fully described, in terms of turning switches on and off. That’s its physical causality. That’s all it does.

The operations the computer can do with symbols involve moving them around, directly manipulating the bits, and doing simple math on the number the bits are interpreted to represent.

As with previous systems, we can inspect the computer states and see that information in real time. For instance, as with the light switch, we can check the voltage. And, also likewise, those states reflect the information content.

A computer logic scope displaying a bunch of on-off switches.

Crucially, no description or definition of the computer refers to the software being run. A computer is defined as a machine capable of running software written for it. (A computer that never runs an application is still a fully functioning computer.)

Finally, the computer has some resemblance to the car engine in having many parts and a complicated architecture. In particular, a network of separate causal chains link to implement the machine’s operation.

§ §

A very tiny piece of “Spider Fingers” by Bruce Hornsby (you probably recognized it).

In all cases, these systems transfer information, and in all cases, the system states directly and proportionally represent that information.

The main point is that when we inspect that information flow at any point in the system, it always resembles itself. Loud looks loud, soft looks soft, slow looks slow, fast looks fast. If a signal has low frequency, or high frequency, this is directly observable in the signal itself.

The signal means what it appears to mean.

§ §

This has gotten long, so I’m going to leave here for now.

Hopefully, nothing here is controversial (but we shall see).

Stay physically causal, my friends!

The canonical fool on the hill watching the sunset and the rotation of the planet and thinking what he imagines are large thoughts. View all posts by Wyrd Smythe

#### 8 responses to “Causal Systems”

• SelfAwarePatterns

I fear this will end similar to our previous discussions, but since you really seem to want to discuss it, I’ll go for a bit.

“I want to stress that the computer is fully defined, and fully described, in terms of turning switches on and off. That’s its physical causality. That’s all it does.”

I think my issue is you seem to be talking about this at a particular level of abstraction as though that’s the only one. A hard core reductionist might insist that there’s nothing happening here except elementary particles and their interactions. A harder core reductionist might insist on talking about quantum fields. All the rest, to these guys, would just be higher level abstractions.

At some point in the layers, we get to the one you’re talking about. You seem to want to privilege that layer, to draw a line between it and any above it. But I can’t see any reason for such a privileging.

But if we do decide to stop things at that layer, then what stops us from just insisting that all brains do is fire neural spikes, that they are entirely defined by nothing but those spikes? At a certain level of abstraction, it’s true, but I can’t see it as productive for understanding what’s going on.

You mentioned that with records the chain is more complex than the string cups. I think the same thing can be said for digital information. The chain is more complex, but ultimately there is a causal chain from, say, someone playing a guitar, and a digital recording of that playing. It is more complex than an analog recording, but it’s ultimately just a matter of how many transformation are involved.

Maybe a way forward is for you to elaborate on why you think this distinction is so special? What about it is different from the other layers of abstraction?

• Wyrd Smythe

“I fear this will end similar to our previous discussions,”

I’m trying to find consensus or common ground. Don’t stack the deck against us before we even begin, dude!

Most of your comment orients on what you perceive as me being selective about a descriptive level, and I think I can put that to rest, but if that’s your only quibble then there isn’t much daylight between us on this one.

“I think my issue is you seem to be talking about this at a particular level of abstraction as though that’s the only one.”

Nowhere did I suggest only. As for a particular level of abstraction, I’m talking about the top one — the one that we use to define things.

The reductionist views are fine, I’m not excluding them. But no one designs or defines these things in terms of elementary particles or quantum fields. (For one, at that level, there’s not much distinction between cans of soup, trees, elephants, or asteroids. For another, it’s a pain in the ass.)

“You seem to want to privilege that layer, to draw a line between it and any above it.”

You don’t think the top level description of a system has some privilege? What’s above the description of tin cans as a phone? Or a sound system or light circuit or car engine or computer?

“But if we do decide to stop things at that layer, then what stops us from just insisting that all brains do is fire neural spikes, that they are entirely defined by nothing but those spikes?”

If we stop things at the top level of description, a brain is a network of neurons connected with synapses and surrounded by a lot of other probably important stuff (glial cells, myelin, etc).

You can insist “that all brains do is fire neural spikes” but that seems a pretty ignorant view in light of what we know about the brain.

BTW: Where did brains come from? They aren’t mentioned in this post. This post, in fact, has nothing to do with brains or consciousness. This is about causal physical systems. (“Computationalism” and “consciousness” aren’t in the tags, nor is this posted in Computers.)

“At a certain level of abstraction, it’s true, but I can’t see it as productive for understanding what’s going on.”

Well I quite agree! 🙂 As I said, it seems an ignorant view if the goal is a top-level description of the system.

As you say, it’s a valid abstraction of an aspect of the brain, and I believe some researchers are investigating the properties spiking networks, so it’s potentially a useful abstraction.

But it’s not a top-level full description of the system. That’s all I’ve done here. Describe five physical systems. And highlight the nature of their primary information — a crucial point for what comes next.

“I think the same thing can be said for digital information.”

Absolutely.

I also completely agree a digital music system has a causal chain. Parts of it are the same causal chain I discussed with regard to the computer system, parts of it (like guitar & speakers) are what I discussed with regard to sound recording.

“…but it’s ultimately just a matter of how many transformation are involved.”

No, there’s more to it than just how many, but since that also isn’t any part of this post, and since it is what the next post is about, how about we defer that aspect until then?

“What about it is different from the other layers of abstraction?”

It’s the top layer for the systems in question. It’s what we normally use to describe them, and it’s what use use to define or design them.

• Wyrd Smythe

Something just hit me… where do you see me using abstraction? I’m describing five different physical systems… do you consider that abstraction?

• SelfAwarePatterns

So the question that comes to my mind is, who defines what the one and only “top level of the system” is? I can’t see that there’s any such thing. There is only a level that is convenient for certain purposes. You’re focusing on a particular level, declaring it the top level, then saying everything above it is some other kind of thing. (You’ve used the word “abstract” before but for some reason seem to be backing away from it.)

So, we can talk about a system that turns switches on or off, or we can talk about a computational system, a phone, a navigation device, a music player, and many other things. Or we can talk in terms of atoms. Which one we use depends on our current purposes. But none are the one and true “top” one.

Put another way, to a particle physicist, all the “physical” systems you discussed are abstractions, at least while they’re wearing their particle physicist hat. That line is always relative, never absolute.

Put yet another way, navigation is an application of my phone, which is applied computation, which is applied electricity and physics. Sociology is applied psychology, which is applied neuroscience, which is applied biology, which is applied chemistry, which is applied physics. You can call these various things: levels of emergence, levels of abstraction, levels of organization, etc, but the concept is the same.

None are the absolute top level. At least unless you can identify a reason why one should be?

• Wyrd Smythe

“None are the absolute top level. At least unless you can identify a reason why one should be?”

I’ll repeat the question you didn’t answer: What’s above the description of tin cans as a phone? Or a sound system or light circuit or car engine or computer?

How would you describe those physical systems?

“I can’t see that there’s any such thing [as the top level].”

What’s above the description of tin cans as a phone? Or a sound system or light circuit or car engine or computer?

“You’ve used the word ‘abstract’ before but for some reason seem to be backing away from it.”

Because I don’t see how I’m dealing with abstractions here. I also asked you: Where do you see me using abstraction? Why do you consider my descriptions of physical systems as what they are “abstractions”?

“Put another way, to a particle physicist, all the ‘physical’ systems you discussed are abstractions”

I very much doubt they would consider things made from atoms to be “abstractions.” I’ve never in my life ran into anyone who thought something physical made from other physical things was an “abstraction.” Would you call a pencil an abstraction of atoms?

“Put yet another way, navigation is an application of my phone, which is applied computation, which is applied electricity and physics.”

Indeed, and if you’re using it and someone asks you what it is, how do you answer? Do you say it’s a navigation app? (I bet you do.)

“You can call these various things: levels of emergence, levels of abstraction, levels of organization, etc, but the concept is the same.”

Sure, the things you’re talking about. But I’m not talking about levels of organization or emergent levels or abstract levels. I’m talking about five physical systems and how they behave as those physicals systems.

What’s the alternative if I want to do that? Are you really suggesting I should be talking about atoms? If you’re saying I’m wrong to even talk this way, how should I be talking?

• Lee Roetcisoender

“I sometimes get the feeling people are so set on disagreeing that they don’t always engage on what the other party is saying. I never know if it’s a lack of comprehension, a lack of willingness, or (on my part) a lack of communication skill or sufficient explanation.”

There is an underlying dynamic at play that gets overlooked. Arthur Schopenhauer says it best:

“The discovery of truth is prevented more effectively, not by false appearance things present and which mislead into error, not directly by weakness of reasoning powers, but by preconceived opinion, by prejudice.”

For what it’s worth, I thought you did a fine job articulating your points in this post. I fail to see any underlying hidden agenda…

Peace

• Wyrd Smythe

Hi Lee. Thanks!

Schopenhauer’s quite right (and I am aware of that dynamic). Unfortunately, if you gore someone’s pet ox too effectively, that can seem an act of war. (Far better to just roast that ox, wash it down with some hearty ale and brown bread, and find a healthier ox. I’ve eaten ox and I’ve eaten crow — the former is better. But I digress. Or do I mean digest?)

Along general social lines, I’ve always had a lot of faith in the dialectic, but it does require some degree of training in thinking and arguing. (Which is why it’s good to study one or more of: logic, math, or philosophy. They all train in precise and factual thinking.) If people don’t admit to reasonable arguments, the dialectic is impossible.

But the social climate these days (talk about climate change)… I have a note tacked up near my desk to remind me where we are:

Opinion = Truth
Assertion = Fact
Emotion = Reality
Hyperbole = Normal
Tribalism = Everything

Back in 2009, Charlie Pierce wrote Idiot America: How Stupidity Became a Virtue in the Land of the Free which is about how we used to be better at ignoring the bat poop bonkers while still embracing that 1% of original useful ideas they came up with. (In some ways, that book is prescient about how we went and elected Twitler.)

• Information Systems | Logos con carne

[…] Last time I explored five physical systems. This time I want to implement those five systems as information systems, by which I mean numeric versions of those five systems. The requirement is that everything has to be done with numbers and simple manipulations of numbers. […]