Matter Thinks

Autoteleology July 16, 2011

Autoteleology is the property of having a self-referential purpose: one’s existence is one’s goal. Usually “goal” refers to a mental state, because that’s the form that human goals take. But here I mean goal in a broader sense: a goal is a state of affairs that a thing — organism, cybernetic system, whatever — by its nature tends toward.

Autoteleology is a signature of life; living things are virtually always autoteleological,[1] while nonliving things very seldom are.[2] Darwinian theory offers a good explanation for this prevalence: effectual autoteleology will always be rewarded. But in truth Darwinian evolution owes more to autoteleology than vice versa. Evolution, and life itself, depends on autoteleology in its simplest and purest form — the drive to live. Without that, or without the intrinsically autoteleological goal of reproduction, evolution as described by Darwin could not occur. This point is one of three significant claims I make concerning autoteleology: autoteleology is more fundamental than evolution, and necessarily precedes it.

The second claim is that autoteleology is physical but universal — any physical system, regardless of the material that constitutes it or the forces that drive it, can be autoteleological, if it is so arranged that it tends to further its own existence.

The third claim is that autoteleology is antientropic — it counters the statistical probabilities that lead to entropy increase, diminishing that increase or even (locally at least) reversing it. The third claim is perhaps the boldest, so I will elaborate.

To begin with, note that the second law of thermodynamics, which states that entropy always increases, is not a fundamental physical law along the lines of Einstein’s laws of relativity or Newton’s laws of motion. Rather, it’s a statistical law, like Boltzmann’s law relating temperature and pressure in gases: if follows from the aggregate properties of large collections of randomly interacting molecules. If the interactions cease to be random, the law may lose its rationale.

Consider Maxwell’s demon. Maxwell’s demon is a mechanism imagined by James Clark Maxwell around 1870 that coaxes a non-random outcome from the random interaction of molecules through a simple but clever control system: a door between two containers of gas that the demon slides open only to allow fast molecules to pass in one direction slow molecules in the other. While the consensus among physicists is that Maxwell’s demon ultimately cannot defy the second law of thermodynamics (on one widely-held account, because of the unavoidable entropy cost of the information processing the demon must do), there is also a consensus that it might avoid the second law within certain limits of space and time — the piper must ultimately be paid, but later and somewhere else, not here and now.

Efficacious autoteleology, like Maxwell’s demon, seeks a local outcome that defies the second law of thermodynamics. It seeks to increase the probability of the system’s existence by means of some combination of persistence, growth and reproduction, each of which represents a maintainance or increase in local order. Moreover, this is accomplished by non-random behavior, non-random because it is driven by knowledge, whether that knowledge be engineered, evolved, acquired or computed.

Autoteleology raises many questions, including deep ontological questions. This should not be surprising, because autoteleology is among other things a form of self reference, and self reference is the source of some of the oldest and thorniest paradoxes in philosophy.

Most such paradoxes, however, are only logically paradoxical, while being physically unproblematic. There are no physical dillemas posed by a fellow who says, “Everything I say is a lie,” only logical ones. Autoteleology is different. Its implications are quite physical.

—–
Notes:

[1] The two exceptions being (1) when altruistic behavior provides significant genomic advantages (e.g. survival of genetically close relatives) and (2) when organisms capable of choosing their own goals (human beings) choose goals that are not autoteleological (such as the goal of defending one’s country). Other cases that might appear at first glance to be exceptions are not, when fully analyzed: symbiosis, for example, is a case of organisms supporting other organism’s continued existence as a means of pursuing their own continued existence.

[2] The only known exceptions being certain systems engineered by humans, such as satellites that maintain orbital stability by firing rockets when necessary.

The Anomalism of the Mental is Not an Anomaly August 9, 2010

“The anomalism of the mental” is how Donald Davidson described the fact that physical causes and effects adhere to strict laws, while mental causes and effects do not seem to do so. The cybernetic model I have proposed shows that the possibility of anomaly is inherent in cybernetic systems, and therefore in matter and physics, since cybernetic systems are physical systems, and you can make a cybernetic system out of anything.

Here is the cybernetic model, in a nutshell:

    Φ(r) & α(q) & ε(m) & ε(p) ⇒ q

where

• Φ(r) means that r is a desirable state;
• α(q) means that q is a choosable action;
• ε(m) means that q is known to cause r when p; and
• ε(p) means that the state of the world is known to be p.

I propose that this model, with a little elaboration, allows conscious behavior to be explained in physical terms in a way that doesn’t make consciousness redundant.

Consider the physics of a traffic light. The light turns green, and cars move. The light turns red, and cars stop. Clearly there is a cause and effect relationship between the color of the light and the motion of the cars. But most things in the universe don’t move when you shine a green light on them and stop when you shine a red light on them. Furthermore, there is no known simple law of physics, or complex law for that matter, that relates colored light and motion of bodies. Yet unless we admit to dualism the causal chain of events in a traffic intersection is a chain of physical events following strict physical laws. What’s the deal here?

Well, a lot of it at either end is straightforward. The physics of the light itself are pretty well understood. Light from either an LED or incandescent light source passes through a colored lens, across space and into a driver’s eye, exciting color-sensitive cone cells in the retina, which send signals via the optic nerve to the driver’s brain. Working from the other end, the change in motion is caused by changes in the flow of fuel into the engine and the application of brakes, which are caused by the pressing of the accelerator and brake pedals, which is caused by the movement of the driver’s legs, which is caused by nerve impulses originating in the driver’s brain.

In between, though, the causal chain leads into and out of the driver’s brain. Once in the brain, the simplest explanation seems to mental, not physical: the driver sees the light; if the driver perceives the light to be green, the driver decides to press the gas peddle; if the driver perceives the light to be red, the driver decides to press the brakes. Then the brain emits a signal and the chain goes physical again.

Interestingly, the brain is not actually a necessary part of the traffic light system. It would be quite possible to replace the mental parts of the causal chain with purely physical components. For example, you could attach a video camera to the hood of a car, connect it to a computer that continuously analyzes the images from the camera to see if it contains a red or green patch of the size, shape, position and context consistent with a traffic light, and connect the computer to servomechanisms which depress the accelerator or brake pedal as appropriate. Not a trivial exercise by any means, and I’d be leery of riding in such a car, but it’s certainly within the realm of possibility.

And if you find the software in such a system to be a little too brain like, you can do away with it: set up a bunch of red and green light sensors, focused in the expected direction of where a traffic light would be, and run wires through carefully adjusted amplifiers and high-pass filters to switches operating the servomechanisms. I definitely wouldn’t ride in that car, but it’s imaginable that with enough cameras and enough fine tuning it might do the right thing more often than not, which would be enough to prove the point — the point being that anomalous causal chains do not have to be mental.

But is this chain really physical? It’s certainly not natural. One could argue that it is in fact mental, because such a setup of cameras and lights would never arise naturally, and could only possibly exist by virtue of an agent desiring to set up such a system, thinking of how it should work and doing all things needed to construct it. In other words, if the causal chain itself is not mental, the cause of the causal chain is.

However, this is not true of all such causal chains. It might be true of traffic lights, but it is certainly not true of the countless biochemical signaling systems which abound in living creatures of all kinds, with and (presumably) without minds, and which behave just as anomalously as traffic lights.

It does seem to be true that persistent anomalous causal chains seem to extend only as far as living things, and things made by living things. Anomalies in the non-living world – weather anomalies, for example, such as a warm day in January where I live, or a cold day in August – do occur but are invariably ephemeral; the causal chains appear out of nowhere and disappear quickly. Perhaps being able to initiate anomalous causal chains that persist is the defining characteristic of being alive.