Life, Mind and Engineering April 29, 2011
Posted by Michael in Cybernetics, Epistemology, Natural Philosophy.add a comment
The general view among materialists, physicalists and scientific realists has been that, while physics does ultimately explain everything, the physical explanations get exceedingly messy, sloppy and effectively incomprehensible once you get to life sciences, mind sciences and beyond. Life and mind may reduce to the dynamics of molecules, but that is not at all the same as reducing to molecules. A creature who has just died still has, for a while at least, all the same molecules as when it was alive, but it is not at all the same. Being alive consists of many high-level behaviors involving the coordinated dynamics of large aggregations of molecules, and when those behaviors cease, the molecules may remain but the life is gone.
Yet the notion that the dynamics of the inanimate world are simple and predictable while the dynamics of life are messy, complex and chaotic is simply false. For one thing, the inanimate world is also messy, complex and chaotic — consider the weather. For another, the fact that simple principles are indeed useful in understanding and predicting many inanimate behaviors, such as using Newton’s equations to calculate the trajectory of a projectile, does not rule out the possibility that life itself follows its own simple principles. My current proposal is exactly this.
The simple principle I have in mind is one which describes the behavior of cybernetic systems, which I formulate as follows:
Φ(r) & α(q) & ε(p) & ε(m) ⇒ q
where
- Φ(r) means that r is a goal, i.e., a desired state of the world;
- α(q) means that q is a choosable action;
- ε(p) means that the state of the world is known to be p; and
- ε(m) is short for ε(p & q ⇒ r), which means that it is known that when p, doing q leads to r.
In other words, if a system has a goal, and it knows what action will achieve its goal, and it has both the opportunity and the capacity to take the action, then the system will take the action.
To be sure, this is a simplification. For example, it doesn’t describe how a system might balance competing goals that call for contradictory actions. Nor does it discriminate between broad, long term, high level goals/actions and all the smaller, more immediate goals and actions they decompose into. But it’s straightforward to accommodate these and other complications by extending and elaborating on the formula without weakening the underlying principle.
The four terms in the formula represent the components that a system must have to be cybernetically successful, that is, to respond advantageously to opportunities in the environment and thereby achieve its goals.
The first term, Φ(r), is teleological; it defines the system’s purpose or goal. Generally this is a given, though some cybernetic systems have the ability to choose some goals for themselves.
The second and third terms, ε(p) and α(q), describe the system’s interaction with the world. They are functions of what the system is physically capable of sensing and doing. As such, they may vary widely from system to system, but the principle is always the same: to achieve its purpose, a cybernetic system must have the ability to obtain information about the state of the world, and the ability to perform useful actions.
The final term, ε(m), spells out the critical knowledge that is required for the system to achieve its goal. How this knowledge is acquired varies greatly; in my analysis, the nature of this knowledge acquisition is the primary characteristic that sets apart the main classes of cybernetic systems.
Let’s see how this applies in the case of a very simple cybernetic system: a furnace connected to a thermostat. We can describe the behavior of this system in two simple sentences:
- If the temperature is below the target level then if the furnace is off, turn the furnace on.
- If the temperature is above the target level then if the furnace is on, turn the furnace off.
Three of the four terms of the cybernetic formula are readily identifiable in these sentences:
- Φ(r), which means that r is the desired state, where r = “temperature is at the target level”
- ε(p), which means that p is known, where p has a different meaning in each of the two sentences:
- p1 = “the temperature is below the target level”
- p2 = “the temperature is above the target level”
- α(q) means that q is a choosable action, where q also has a different meaning in each sentence:
- q1 = “turn the furnace on”; α(q1) is true if the furnace is off.
- q2 = “turn the furnace off”; α(q2) is true if the furnace is on.
But what of the remaining term, ε(m)? This term refers to knowledge of the causal relationship between the action and the goal. In an engineered system such as a heating system, this knowledge is first and foremost in the mind of the engineer. From there it translates into interconnections between sensors and effectuators — ε(p) and (q if α(q)) — that the engineer builds into the system. In our heating system, the knowledge is embodied in the wiring between the thermostat and the furnace.
There exists another class of cybernetic systems where ε(m) comes to be embodied in the system without the assistance of an engineer. This is the class of systems whose designs are the product of Darwinian evolution, a category which includes all living organisms. Darwinian evolution emerges from the interplay of three processes: reproduction, mutation and selection. Selection operates on individual organisms, but reproduction and mutation are different: they depend on the existence of encoded information that persists beyond the individual and in fact accumulates and improves over generations.
In living organisms, this hard-won knowledge is encoded in nucleotide chains, that is, DNA. ε(m) is not explicitly encoded in the organism’s DNA. Rather, just as ε(m) in a heating system is embodied in the particular way the engineer connects the wires between the thermostat and the furnace, in a living creature ε(m) is embodied in the (primarily) biochemical interplay among the mechanisms that comprise it. The way these interactions are distributed and regulated is encoded in the organism’s DNA.
A third class of cybernetic systems is the class of systems capable of learning, which includes most animals to one degree or another. In such systems, a mechanism exists allowing the system to acquire and embody new knowledge, in addition to the knowledge it inherits. Such a system can potentially be much more adaptable than one whose behavior is hardcoded, whether by evolution or engineering.
Learning can take many forms. Essentially the task is to find actions (q) that succeed in a given circumstance (p). One way is to randomly generate candidates for q (perhaps by stringing together smaller, hardcoded actions), test them, and keep the ones that work. Another way is to start with q, make incremental changes, and reinforce or reject each change according to the result. Yet another way is to try a single behavior in different situations to find a beneficial combination — in other words, keep q fixed and vary p. In none of these cases does ε(m) need to be fully and explicitly represented, but at least some part of it (p or q) must be exposed in some fashion that translates into the ability to modify ε(m).
The fourth and final class of cybernetic systems are those capable of creating new ε(m) through an internal logical process of some kind, for example a mind. This class includes humans, possibly a few other species, and arguably some experimental computer systems. Because ε(m) in such a system is explicitly arrived at, it is explicitly represented. (This presumes a broad reading of the term “explicit”. For example, it includes algorithmic representations.)
The ability to synthesize ε(m) through internal processes is an enormous advantage. In contrast to acquiring information across generations, or over a lifetime of trial and error, a system with the ability to synthesize knowledge can generate new ε(m) on demand. In the case of humans, inference is so woven into our perception that we seldom acknowledge it. I walk into a home I’ve never been in before, and in the distance through a doorway I see just a sliver of a tall white object with two handles — a refrigerator. From a smidgen of evidence combined with knowledge I already possess I can now infer huge amounts of new knowledge. For example, if I get thirsty later in the evening and the host invites me to help myself to a beer, I instantly infer exactly where to look for one.
To sum up the different ways cybernetic systems get their ε(m), that is, usable knowledge of causal connections:
- burned — ε(m) is engineered into the system
- earned — ε(m) is acquired through evolution by natural selection
- learned — ε(m) is acquired through trial and error learning
- discerned — ε(m) is synthesized by an internal logical process
My contention is that the above processes, which are very different in the details of their operation, are all the same in terms of the principle that drives them and the nature of what they produce. For a single principle to underlie such varied phenomena — life, mind and engineering — it must be fundamental. And so it is: the physics of knowledge.
Knowing the Future February 27, 2010
Posted by Michael in Acausal efficacy, Cybernetics, Epistemology.add a comment
For the last several weeks I’ve been working on something that started as a post, grew steadily into an essay and is now threatening to approach a monograph level of voluminosity. The paper describes a cybernetic model of mental causation which elides the whole problem by shifting it to pure physics and the multiverse — a scope that contains all possible worlds including the one we inhabit. I sketch out in the paper a description which is causal at the level of the multiverse but acausal from the point of view of the realized universe. Not just acausal — acausally efficacious. I won’t repeat here the argument I make for acausal efficacy. But I do want to note where this mulitversally causal/universally acausal process gets its power: knowledge of the future.
By “knowledge of the future” I don’t mean by powers of prognostication or parapsychology; I mean knowledge that is earned (via evolution) or learned. Nor do I mean knowledge of the exact future that will happen; I mean knowledge of various futures that may happen, depending on one’s choices. Nor, finally, need this knowledge be perfect; just somewhat better than random. The point is that if one can make choices, and can make better-than-random predictions about the possible outcomes of the choices, then one is in a position to extract value from something that in conventional terms doesn’t physically exist — the difference between what is and what might have been.
A Cybernetic Theory of Representation December 17, 2009
Posted by Michael in Consciousness, Cybernetics.add a comment
When it comes to mental representation, I think the teleosemanticists are largely right, except that they are overfocused on biology and evolution rather than the more general underlying principles that I believe are in play.
Teleosemantics is a theory of representation that says that mental representations reference the external world via their purpose, not their cause — that is, a mental representation is the way it is not because the thing it represents caused it to be that way, but because that’s the best way for it to achieve the function it was evolved to do. Teleosemantics was formulated as an alternative to causal theories of representation, which hold that a causal chain of some sort connects things in the real world to their mental representations. Imagine you see a snake and you think “snake”: a causal theory asserts that “snake” (i.e., the representation of a snake in your mind) means snake because it’s caused by an actual snake. A teleosemantic theory, however, asserts that “snake” means snake because you have evolved to react to snakes, and a mental representation of a snake such as the one in your head when you think “snake” is useful as a part of the reaction mechanism.
The notable advantage of teleosemantics over a causal theory is that it explains how you can see a rubber snake and think “snake”, without “snake” having to mean rubber snakes as well as real snakes. After all, if the meaning of any mental representation is the thing that caused it, then pretty much any mental representation means just about anything, because humans are very good at mistaking things for other things. Teleosemantics doesn’t have this problem, because if meaning derives from purpose, and the purpose of a mental image of a snake is to avoid real snakes, then it’s still about real snakes even when prompted by the observation of a rubber snake.
But teleosemantics has its own problems. For one thing, it seems too narrow: how could such a theory account for all the things we can think about that don’t have an apparent purpose? For another thing, it seems too tied to biology and evolution: teleosemantics says that mental representations are determined by biological needs and evolutionary history, and are only indirectly connected to the things represented, whereas intuitively it seems that the connection must be much more direct.
I agree with the critics that there must be a deeper connection between mental representations and their referents, something mediated perhaps by biological mechanisms but not fundamentally defined by them. But I agree with the teleosemanticists that the connection is by way of function. My solution is essentially to recast the teleosemantic explanation in cybernetic terms and transport it to the level of basic physics and logic, thereby becoming fundamental and universal (to a physicalist anyway).
To see how this works, let’s revisit the snake example. What is in your head when you see a snake, and what connection does that thing in your head have with the actual snake?
The teleosemanticists say, for starters, that what happens in your head is an evolved capability. I fully agree. Without doubt we are what we are because of evolution, including our mental faculties. If you see a snake and move carefully away, it makes perfect sense to say you are exhibiting behavior that was favored by evolution, and that what happens in your head has something to do with that evolved capability. But the fact that capabilities are evolved doesn’t tell us what they are and how they work. Evolution can tell you why birds have wings, but to know how those wings work you need to know aerodynamics. Likewise, it’s certainly true that you are able to evade the snake because of the abilities afforded you by evolution. That’s why you can evade the snake. What I would like to consider is how you can evade the snake. How is this behavior implemented? This is not a question evolution can answer.
Avoiding snakes is a directed cybernetic process. A cybernetic process is one that operates on feedback: the state of the system at time t + 1 is a function of the state of the system at time t and the state of the world, as sensed by the system. A directed cybernetic process is one that functions towards a goal, defined as a particular state of the world at time t + 1. In our example, the goal is the state of not being imminently threatened by snakes.
But we are not talking about hard-wired snake avoidance, we are talking about seeing something, thinking it might be a snake, and deciding to avoid it — in other words, not just avoiding snakes, but consciously avoiding snakes. Consciously avoiding snakes is a self-directed cybernetic process.
This is not to say that self-directed cybernetic processes are always conscious — but they always involve something similar to mental states: information about the state of the world, for one thing, and goals, for another. They are functionally similar, to be precise, which does not require being phenomenologically similar.
We can describe a directed cybernetic process functionally. The physical embodiment of the cybernetic process is a cybernetic system, conventionally referred to as an agent. Let p be a particular state of the world. Let q be a particular action by the agent. Let r be the state of the world that results. So, we can say: if p is the case, and an agent does q, the result will be r. We can also express this as a logical proposition:
p & q ⇒ r
Let m stand for the above proposition:
m = (p & q ⇒ r)
A directed cybernetic system works toward a state of the world. Let the function Φ(r) mean “works toward r”, or, in other words, the agent has chosen r to be a goal. Let the funtion α(q) mean that doing q is an available choice to the agent, that is, q is actionable. Let ε(p) mean that the agent knows p to be the state of the world., or one might say, the agent knows p. Let ε(m) mean that the agent knows m, that is, knows that if p is the state of the world, then doing q will result in r. Then the behavior of a directed cybernetic system may be described as follows:
Φ(r) & ε(m) & ε(p) & α(q) ⇒ q
Or, restated in plain English: if r is a desirable goal, and an agent knows that doing q when the state of the world is p results in r, and the agent believes the state of the world to be p, and q is an option available to the agent, then the agent will do q.
To be sure, this is a simplification. For one thing, it’s just a snapshot; in reality, the state of the world, the available actions and even the goals change over time. For another thing, the state of the world is too complex to consider in its entirety; in practical terms, only a relatively small number of properties of the world can be considered at once. Finally, there are many goals and actions in play at any given moment, and they are not guaranteed to be consistent or compatible.
But such complications don’t negate the fundamental principles expressed in this simplified formulation. One such principle is that what the system needs to know is not the same as what the system’s goal is. Or, in terms of the formula, to achieve r, the system needs to know p and m. This explains the apparent disconnection between a functional explanation and the intuition that a representation is related to what it represents: in a cybernetic system, while the function of a representation does determine content, the content is not of the function, but of the state of the world that causes or enables the function. I call this a reverse-causal theory of content: the mental chain of causation, from function to representation, is the reverse of the physical chain of causation, from content of representation to function.
So the cybernetic model explains why the cause of our mental representations can be so different from their content. As stated so far, however, it doesn’t explain why there are mental representations in the first place. After all, it holds for thermostats as well as human minds, and themostats do not need mental representations.
Still, the cybernetic model is the starting point for such an explanation, once we examine the complexities and opportunities the human cybernetic system has evolved to handle.
The first complication is the fact that our knowledge of the world comes to us in bits and pieces over time. This means that ε(p) — our knowledge that the state of the world is p — will in general require memory and integration of sense data over time. A mental representation of some sort seems a natural model for storing the result of such a process.
A stronger rationale for mental representation comes from a consideration of the practical limitations of the cybernetic model as described so far. When we talk about the state of the world, of course we don’t mean every single detail, or even every single detail it is possible for us to know; there are simply too many. Luckily in practice most of the details we might consider will be irrelevant for most purposes. In the snake example, what counts is a fairly small subset of details: those that determine that a snake is within striking distance of us. Even so, the possible combinations of details this encompasses is astronomically large: think of all the ways a snake might look, all the sizes, shapes, colors and patterns in all the possible orientations and contexts. And this is just for a single snake: of course we want the same behavior to hold if we see two or more snakes, or, for that matter, if what we see is not a snake but another animal that poses a similar risk.
The only practical way that I can think of to handle such complexity is through some kind of generalization. Let’s define P as a set of world states that includes p, Q as a set of specific actions that includes q, and R as a set of world states that includes r. Let’s stipulate that the relation p & q ⇒ r holds for any p ∈ P, q ∈ Q and r ∈ R; we will call this proposition M.
We can now write a generalized version of the cybernetic formula:
Φ(R) & ε(M) & ε(p) & ε(p ∈ P) & ε(q ∈ Q) & α(q) ⇒ q
Or: if R is a desirable type of goal, and an agent knows that doing an action of type Q when the state of the world is of type P results in a R-type state, and the agent believes the state of the world to be p, and knows that p is a P-type state, and knows that q is a Q-type action, and knows that q is an option available to the agent, then the agent will do q.
This may be more complicated to describe, but is much easier to implement on a practical basis. Without generalization of this kind, an agent would have to essentially relearn a lesson for every minor variation of the relevant factors. But it does impose an additional requirement on agents: they must have an understanding of types. A type is by definition abstract: it encompasses many instances, including the possibility of instances that have not yet been encountered.
Whereas a system that deals only with concrete instances could conceivably work entirely with sense data rather than representations, this seems less feasible in the case of a system that also deals with abstractions. On the other hand, if a system uses mental representations to deal with concrete instances, handling abstract types may be a small step: all the system needs to do is to reuse representations. If “snake” is a representation, and this representation is applied whenever the sense data fit certain criteria, and a number of different concrete instances can evoke the suitable sense data, then “snake” is a type, and the system is capable of generalization.
I will look at one more factor that argues for the existence of mental representations. The value of a cybernetic principle expands enormously if it can be created proactively rather than in hindsight: i.e., if the agent can come up with new formulas based on existing formulas, existing knowledge and mental processes (logic, intuition, etc.). In fact, we do this all the time, by using our imagination. We imagine possible future states of the world; we imagine possible actions we might take; and we imagine the result that might occur from an action. What could imagined states and actions and results possibly be, other than mental representations?
Note that none of the foregoing requires biology or evolution; rather, biology and evolution take advantage of possibilities inherent in the way the world operates.
In short: the laws of physics predict the power of imagination.
The Cybernetic Explanation October 12, 2009
Posted by Michael in Cybernetics, Observations, Ontology.1 comment so far
One thing materialists have come to grips with is that there may be multiple ways of describing nature that are fundamental and true. Most famously there is wave-particle duality, which gives us two equally true, equally fundamental descriptions of all matter and energy. Then there are the cases where two descriptions are each true and fundamental, but in different ways. Take the case of the temperature of a body: we can view it as the kinetic energy of individual molecules, or we can view it as a property of the body as a whole. The former is ontologically fundamental, being the most precise and complete description possible, but is computationally intractable, therefore unknowable. So we choose the epistemologically fundamental description — temperature of the body as a whole. Yes, such a description brings with it imprecision and incompleteness, but these uncertainties are strictly governed by the mathematics of large numbers, to the degree that it is possible to formulate laws such as Boyle’s law relating pressure and volume in gases that are statistical in nature yet ironclad and true.
And then there is entropy and the arrow of time.
The second law of thermodynamics predicts a steady increase in disorder (entropy) in a closed system over time. Like other laws that operate at large scales, the second law can be derived from the statistical behavior of large numbers of molecules. It has also been abundantly confirmed in empirical observations.
But there is a fundamental truth introduced by the second law which is oddly inexplicable at the micro level: the arrow of time. The second law of thermodynamics says that balloons don’t unpop, rust doesn’t turn to iron, and eggs don’t unscramble. The funny thing is that no other physical laws prohibit these things from happening. Is it possible that something we experience so immanently and intimately — the one-way flow of time — is just a statistical artifact? Many physicists and philosophers are unconvinced.
I won’t weigh in on that particular question. But I will say that there is a loophole in the second law: it only holds for a closed system as a whole, and doesn’t rule out local increases in order. In fact, it happens all the time: it’s called life. Every living creature is an island of reverse entropy, every living cell a little machine that absorbs disordered matter, burns some of it as fuel, and uses the rest to build things — such as new little machines.
Is life a special case? No. Every air conditioned home or car is also an island of reverse entropy. Perhaps you may consider air conditioners to be extensions of life, “things-engineered-by-life”, along with spiders’ webs, termite hives, hermit crab shells, water wheels and semiconductors. Even in the realm of life itself, discernible order occurs at many levels: proteins, organelles, cells, organisms, societies. If life is a necessary cause, it’s one with a multiple personality disorder.
From a reductionist physicalist point of view, it’s hard to get much of an explanation for this order — it seems to be random, a statistical artifact like entropy itself. In essence, from the reductionist point of view, there is nothing to explain: life follows the laws of physics like everything else, and in the closed system of which it is a part, the second law is honored.
But this is willful blindness. There is clearly something to be explained, and if science cannot explain it then magical and supernatural explanations will fill the void.
Not to fear, though. Indeed there is a logical explanation. It goes like this:
- Systematic local reversal of entropy is caused by systematic feedback.
- Systematic feedback is caused by directed behavior.
- Directed behavior is caused by a Turing machine or the equivalent.
- There exist Turing machines.
- Therefore there may exist systematic local reversals of entropy.
Shorter version: life is a computation.
But the explanation is actually broader than life. It begins with a fundamental relationship between feedback and entropy. And this makes sense, because feedback can be seen as a way to reverse the arrow of time in a limited but real sense.
To understand this, consider the iconic case of the rudder of a boat and the steersman who controls it. Moving the rudder causes the boat to turn. The arrow of time leads from cause to effect, from changing rudder position to changing boat direction. But in the mind of the steersman, changing the direction of the boat comes first, changing the position of the rudder second. Boat direction is the independent variable; rudder position the dependent variable. This is easy to see by considering the possibilities: the steersman could just as easily decide to turn right as turn left, but once the boat direction is decided the rudder position is determined and cannot be otherwise.
So, in this sense, in a cybernetic system (a system based on feedback) an effect may lead to a cause. This, I would argue, is the root principle underlying the reversal of entropy. It’s the first and most important proposition in the five-point explanation above.
The second point states that the whole arrangement of rudder and steersman is not random; the rudder exists in the first place explicitly to enable the cause, in order to enable the achievement of the effect. So not only does effect precede the cause, desire for the effect precedes the possibility of the cause. The feedback system exists to fulfill a purpose.
The third point states that for the purposeful behavior to work, it must entail knowledge, in this case of the motion of boats and the effect of rudders. This could be conscious knowledge achieved through observation and reason. Or it could be genetic knowledge acheived through evolution. Either way, it is the result of computation, and computation requires a Turing machine.
The fourth point posits the existence of one or more Turing machines. So where do Turing machines come from in the first place? This is, of course, the perfect spot to throw in a supernatural explanation, an outside intelligence to act as the first cause. But a naturalistic account is surely more plausible: given that local systems of reverse entropy (life) have arisen naturally in the material universe, Turing machines must be endemic to nature.
As a materialist, I find the natural explanation better than the supernatural one, even if it requires going further than the conventional physicalist argument generally runs. In essence, the cybernetic view requires accepting a new duality: the duality of information and matter. Duality is not dualism; it is dual identity, two equally valid ways of seeing the same thing. The materialists says (correctly in my view) that there can be no information without matter. The cybernetic materialist says that, in addition, there can be no matter without information.
This is not as mystical as it might sound. It takes nothing away from physicalism, and offers nothing to the dualist. On the contrary, it seeks to expand the physicalist explanation to concepts, abstractions and logic, in other words, the building blocks of thought: they are fundamental attributes of matter, like charge and mass, that are inseparable from matter yet may be described on their own terms. Do charge and mass exist the same way matter exists? I don’t know the answer, but whatever it is I believe it also holds for logic and arithmetic. Perhaps they don’t extend quite all the way to fundamental particles; perhaps, like entropy, they emerge at larger scales. Either way, they inhere in physical existence.
Finally, we get to point five, the conclusion: if you are a cybernetic materialist, naturally there is life, and life reflecting on life. That’s the cybernetic explanation.
