Tuesday, December 18, 2018

The Incoherent Notions of Free Will and Determinism. Part I


The article header:The answer to its question
depends on what is meant by "Free".
I found an article in November's  Premier Christianity magazine very frustrating to read to say the least. It was on the old chestnut of "free will vs predestination". The author of the article  Justin Brierley, although a respected commentator on more general Christian affairs, seems unable to see the subject through anything other than the same old cliched lens of taken-for-granted meanings. 

Many discussions on this subject, and this one was no exception, jump in with the unacknowledged assumption that we know what we are talking about when we refer to "free will" and "predestination" and that these meanings are unquestionably non-problematical. There is therefore seldom any preliminary investigation on whether or not we are talking about two intelligible concepts and the discussions go straight in with the question as to whether we should believe in either free will or predestination.  Brierley ends the article with these words:

 "Having genuine free will really matters. It's one of the reasons why I'm neither an atheist nor a Calvinist. I'd much rather be someone who imperfectly but freely chooses  to follow Christ"

But Brierley never attempts to clarify what he means by "freely choosing" so how can we either agree or disagree with him? At this stage it is impossible to judge him as either right or wrong. We can perhaps get an inkling of the complexity of this subject from the following considerations.

If perhaps we can first define what "predestination" means we might then be able to proceed apophatically by claiming that freewill is the absence of predestination. That is:

Absence of determinism => the presence of free will.

...where "=>" reads as "logical implication". But is this relationship intuitively agreeable? A system may be mathematically non-deterministic such as we have in coin tossing and yet because it completely lacks sentience it therefore has no sense of conscious choice; presumably conscious choice is a necessary condition (although not a sufficient condition as it turns out) for "free will". What about this relationship then:

Free will => An absence of determinism ?

That is, the choices of a  "free will" agent are not deterministic. As we shall see this is, I believe, also spurious. Negating both sides of the foregoing relationship we can express it slightly differently (but still as badly) thus:

Presence of determinism => absence of free will

Clearly the issue is complex, but as long as we fail to clarify our terms we will remain in a muddle here.

***

There is some light at the end of this confusing tunnel if we turn to mathematics. For it is possible to discuss the subject of determinism with a degree of mathematical clarity. In my book on Disorder and Randomness I defined a random pattern as one that cannot be generated, predicted or specified* using small space short time algorithms; that is, algorithms which use human sized programs and/or operate in human time scales. However, it is not possible to precisely identify at what point an executing program can be considered to go beyond humanly manageable dimensions in time and space because the "cut-off" is gradual with increasing dimensions. But although limits on human computational resources may put a pattern way beyond human computation this is not to say that the pattern can't be generated by an algorithm of sufficient size and/or execution time; after all a simple counting algorithm or any other algorithm which systematically works through the possibilities, if allowed to operate for sufficient time, will eventually generate any pattern, even the random ones.

Randomness isn't a "yes or no" property but a graduated phenomenon which is measured relative to the availability of humanly manageable resources of computation. Nevertheless, we could, I suppose, define absolute randomness as a pattern which  requires  infinite computational resources to specify by algorithmic means. But most random sources we deal with have a finite manifestation in terms of what they can generate. Therefore it is conceivable in these cases that there is some underlying finite computation which generates even the most muddled of (finite) patterns but of which we know nothing, have little hope of discovering and which may in any case be beyond our computational resources. Hence, in a finite world we could be surrounded by patterns of behaviour, none of which are truly random, but which from a human perspective are to all intents and purposes random.

The situation is further complicated if we allow the introduction of expanding parallelism in our computation. In expanding parallelism we assume it is possible to introduce any number of processors to fully exploit the potential for many of the operations in a computation to be carried out in parallel. If we introduce expanding parallelism then, depending on the task in hand, this can have the effect of reducing the minimum linear time to the logarithm of the total number of operations required for the task. If resources of expanding parallelism were practically available we might have a very different view on what is deterministic and what is not. Expanding parallelism is (as far as I am aware) currently beyond human technology; the best that we can do at present is amass a large but limited number of processors in order soak up the inherent parallelisms in the task in hand; but limited parallel processing is still no match for tasks which need an exponentially growing suite of processors in order to reduce linear time to the logarithm of the number of operations the task demands. There is, needless to say, a huge motivation behind research which attempts to implement expanding parallelism via quantum computing.

***

The point of the forgoing discussion is to show that "determinism" in the mathematical sense is a question of degree and that degree is measured relative to our resources of computation. Normally we would consider a pattern generated by a relatively simple and quick rule as mathematically deterministic. But there are patterns of behaviour out there whose computational complexity ensures that they are beyond human computational resources thus rendering them practically indeterministic (although theoretically they may be deterministic). Relative to practical  human resources there is no sharp cut-off between mathematical determinism and indeterminism. Although it is possible to define absolute randomness as a pattern whose specification requires infinite computational resources, this seems rather academic and idealistic in our large but finite world.

There is also the question of "happened" and "unhappened" events. The future, if "governed" by random processes, is in mathematical sense undetermined. But when a random source has already generated its pattern is that pattern now determined or undetermined? If the pattern has already been laid down in history it looks to be determined from the perspective of those in the know. But for those not in the know indeterminism appears to reassert itself if the pattern is revealed bit by bit, because to the observer ignorant of its form this bit by bit revelation will look identical to the random source generating the pattern there and then; in fact, both cases entail a bit by bit revelation of what in one sense is already there: Viz the active source is revealing a pattern that has a kind of preexistence in "platonic space" and in case of the ignorant observer the pattern is being revealed from what already has been reified by the random source extracting it from platonic space. In fact the hidden pattern could be considered as part of the information suite inherent in the algorithm that is generating the result but with no chance that the ignorant observer is ever going to predict it even though in one sense the pattern is determined.

So are we to conclude "happened" events are determined or otherwise? In one sense it seems that every thing is determined: For either patterns are being extracted via a pattern generator from what preexists in platonic space or are being reified in the observer's consciousness as a preexisting pattern is revealed to the observer. In either case the pattern being apparently generated preexists in some sense of the word and whether it is predetermined or not  seems to depend on the level of information held by the observer. This point is important because it suggests that the question of the existence determinism is observer relative. Consequently whether the observer is omniscience or not will have a bearing on the question is what is predetermined.

***

Where do the foregoing considerations leave the vague notions of freewill and determinism? When people like Brierley talk of determinism do they mean mathematical determinism? And when they talk of "free will" do they mean freedom from mathematical determinism? Usually they do: Usually they see the mathematically deterministic billiard ball Newtonian universe as the antithesis of free-will. Hence this rather forces them into the view that mathematical indeterminism as a necessary condition for free will. (I'm going to challenge this). But as I have already pointed out even if mathematical indeterminism is a necessary condition for freewill it isn't a sufficient condition; for even if we are dealing with the next best thing to absolute randomness such as we believe to be inherent in coin tossing it hardly constitutes a form of free will; for "free will" requires the presence of a choosing sentience. But what if a choosing sentience is making choices according to some underlying deterministic physics? Does that mean that that sentience has no free will? I would answer "No" that question. In fact human choices may be entirely predicable and yet free: For example, a person making choices, may, to someone who knows them well, be entirely predicable in their patterns of behaviour. And yet it is not wrong to say that those very predicable choices are made freely, presumably because the person making those choice has the experience of wanting those choices and having the freedom to bring them about.

Deciding whether or not humans have such a thing as free choice is perhaps obscured by the sheer complexity of the human system. In contrast, for example, most computers, if not all computers, work from small space short time algorithms and they appear to be an open and shut case for mathematical determinism and therefore they may seem absent of free will. But then let us consider computers that carry out relatively complex tasks, like say controlling a manufacturing process or guiding a cruise missile. Such systems, which can be relatively complex, are making algorithmically determined choices given their perception of environmental conditions.  Is it right, then, to say that these are "free choices"?  But putting aside the fact that we are not dealing with conscious sentience here it is not unnatural to talk about even computers making free choices in a colloquial sense. Let me explain:

It is possible to imagine scenarios where the computer is no longer free to do its job. For example I might infect the programming with a computer virus, or irradiate the computer with gamma rays or even mechanically damage it in someway thus disrupting its otherwise "free choices". In each case the computer  may start doing things that are out of character with its programming and therefore it is no longer making free choices with respect to its "normal" behaviour; it is, as it were, no longer "responsible" for its outcomes and the computer is no longer true to its internal programming. Thus in a crude colloquial sense even a computer running a deterministic programming displays a kind of prototypical free will.

Usually, however, we don't think computers running short time small space algorithms, even though they may be making quite complex (deterministic) decisions, as worthy of the label "free will"; certainly not anywhere near the human sense of having freewill. The human decision making process differs radically from computers on several counts. There is of course the sheer mechanical microscopic complexity of the human system as observed from the third person perspective. Moreover, this neural system probably has non-linear feedback loops entailing the potential for mathematically chaotic behaviour making it practically unpredictable. This chaotic nature will also mean that the human system is sensitive to random fluctuations perhaps, even, fluctuations amplified up from the quantum level. Superimposed on top of all this is the fact that something about the way God has created the material world means that when matter is appropriately configured it gives rise to the "internal" first person perspective of conscious cognition. It is this human system, in all its decision making richness, which gives the hook on which we hang the idea of "free will".

But in spite of the potential for unpredictability in human behaviour much of it remains highly predictable. If circumstances allow I myself invariably have have two coffees in the morning and three teas in the afternoon and this pattern is very predicable and yet from my first person perspective it is my free choice. But it would be no longer be my free choice if I suddenly found that some awkward person had removed the means for making tea and coffee or even for that matter somehow tampered with my make up so that I'm no longer true to who I have been created to be. A complex system like a human being has free choice when it makes choices true to itself, true to its internal logic. Whether or not those choices are predicable and deterministic is I propose irrelevant. It is probably here where I differ from Brierley who, you can almost bet, will be a dualist. Using his bland Western dualist vision of a "billiard board material reality" he is unlikely to believe that "matter alone" (even if it is mathematically deterministic), when appropriately configured, is able to host "free will".

Given the richness of God's created world, and this includes complex decision making entities like human beings, there is plenty of room to identify so called "free will" as a feature of the miraculous "neural machines" we call human beings. It may be that theses neural machines are following predictable algorithms. Although I doubt that, it seems to me that there is a case, nonetheless, for claiming that the free-will vs predetermination dichotomy is bogus in sense that both categories are at once true of human beings; that is, our choices can be classified as freewill and yet predetermined at the same time. Humans beings are what they are and they have free will as long as they can act true to  the logic which makes them up.

With the foregoing frame work of ideas behind us, in the next part I will have a look at some of the contents of Brierley's article in Premier Christianity .


Foot notes
* There is a difference between generating a pattern and specifying a pattern (I gloss over this difference in the text). An elementary periodic pattern of infinite length can be specified by a simple repeating algorithm, although, of course, it would take infinite time to generate it. In contrast an absolutely random pattern requires an infinite amount of information just to specify it. In this sense random patterns are algorithmically incompressible.