Wednesday, October 18, 2006

Occam's Razor

[By Alex Gregory]

I'm running with a kind of "philosophical methodology" theme on here so far, so I'll run with that for this post.

---

"entities should not be multiplied beyond necessity."

"the fewer assumptions an explanation of a phenomenon depends on, the better it is."

"preference for the least complex explanation for an observation"

There are various ways in which this ideally needs qualifying. First, we must balance up the theoretical virtue of simplicity against other theoretical virtues, such as predictive power. Are other virtues lexically prior to simplicity, or is simplicity more weighty than that? Second, we need some idea of what counts as simplicity. Often fewer entities entails more complex processes about how those remaining entities provide an explanation. How do we weigh the various facets of simplicity against one-another? Still, despite these problems, Occam's Razor is a popular tool, both in and out of philosophical discourse.

An incredibly dominant atheist suggestion is that it is up to theists to prove that God exists, in exactly the same way that it would be up to someone to prove that Santa Claus, or Fairies, existed, if they were to hold that belief. That is, the burden of proof is assumed to be on the theist. This seems to be the result of Occam's razor type considerations.

But I think this picture is misleading. The theist does not assert everthing the atheist does and more. They assert a whole picture that looks very different. The atheist (I simplify) is often thought to merely assert that natural facts exist, nothing more. In contrast to this, some take the theist to be asserting this picture, plus a God overseeing it all. But this is false, at least as far as my understanding of religion goes. Theists tend to believe that the natural facts are caused by, maintained by, and/or even constitute, God. In other words, I wonder if the natural facts are not something in addition to God for the theist. Both therefore assert the existence of one entity - for the atheist: the natural facts, and everything associated with them; for the theist: God, and everything associated with him.

If this picture is correct, there is no obvious sense in which Occam's razor supports atheism over theism. There is no burden of proof particuarly on the theist. Atheists must provide reasons to think that God doesn't exist just as much as theists must show that she does.

Of course, I think that there are other, very obvious, considerations which do such provide positive reason to think theism false. The problem of evil is an obvious choice, but also other things such as that most major religious texts contain statements that seem flat out false.

There's two reasons why I think this stuff is interesting. First, given the huge numbers of believers in the world, I find it hard to believe that Occam's razor provides such a knockdown refutation of theism as some people like to believe (maybe I'm relying on something related to Richard's epistemic principle here?). Second, I've often thought that the sceptic elsewhere (epistemology, ethics) is wrong to assert that the burden of proof is on the objectivist. I've recently realised that this is at odds with the view that the burden of proof is on the objectivist in the religious case.

9 comments:

  1. Alex,

    By what definition of an explanation is God explanatory?

    God is almost infinitely fine-tuned to predict exactly what we have seen, yet never one thing more. That's not explanation, it's restating our observations and naming their undefined cause.

    I think the best definition of a materialist is a person who rejects non-predictive explanations. For the materialist, an explanation is a set of facts and rules that preferentially predicts our observations. Supernatural entities don't do this. If they did, they would be natural entities. To the materialist, the supernatural explanation is an oxymoron. The materialist acknowledges that there are unexplained events, but accepts that either they are explicable by natural laws, or else they are simply inexplicable.

    So, I would suggest that Occam's Razor doesn't apply to theism, not because of the complexity of theistic arguments, but because theism is a non-explanation.

    I know that most people feel as though theism is explanatory, even when it isn't. Perhaps this is because most religions prescribe behaviors (e.g., "don't blaspheme"), much the way one would prescribe behaviors based on naturalistic explanations (e.g., "don't stand under that tree in a lightning storm").

    ReplyDelete
  2. Dr. Logic,

    Your characterization of theism as a non-explanation because it assigns more than a material cause is quite odd. If the theism you are thinking of is some form of medieval peasant characterization of God as the one who makes thunder happen because he is angry, then you might be right.

    Under your definition of supernatural, any exhibition of human freewill is supernatural. The theist does not reject natural explanations, but simply posits freewill of beings as an additional cause. Does the fact that a cause or effect cannot be explained by facts or rules, mean that it is inexplicable? This would only seem to be true if you set up the definition of an explanation to be just that, and the reason this definition is assumed is because materialism won't allow any other way of explaining. You have to assume your conclusion to make this argument.

    ReplyDelete
  3. Belief,faith,make believe are all the same. Fortunately they have no bearing on the truth whatsoever.

    ReplyDelete
  4. Bryan,

    Under your definition of supernatural, any exhibition of human freewill is supernatural.

    No, not at all. I think our intuitive sense of free will specifically neglects the causes of our preferences, of who we are.

    For example, suppose I am choosing which breakfast cereal to eat in the morning, and that Corn Flakes, Special K and Count Chocula are available in the cupboard. Corn Flakes is my favorite, so I freely choose it. I don't intuitively regard my life history and genetic makeup as coercing my choice to eat Corn Flakes, even though they strongly (if not totally) determined why Corn Flakes is my favorite. Similarly, if I have eaten Corn Flakes every day for the last week, I don't intuitively regard that fact as coercing my choice to eat Special K for a change.

    So, our intuition of free will is generally relative to our preferences, and we generally exclude factors that gave us those preferences in the first place. Free will is a psychological phenomenon, and it is largely independent of whether, say, the universe it totally deterministic.

    I think this is a better way to approach the question of free will than to beg the question by inherently linking it to determinism. We definitely intuit something when we speak of free will, but, a priori, determinism is a different animal.

    The theist does not reject natural explanations, but simply posits freewill of beings as an additional cause.

    If an event (e.g., a choice) is unconstrained by past history, then it is random. Surely, that's acausality, not a form of causality?

    Even when "free will" isn't acausal, I don't see how free will is broadly explanatory of low probability events.

    Suppose a man jumps from a bridge, and nothing in his known life history suggested he would take such an action. Should we feel satisfied that the action is explained when we say he had the free will to jump? Or is the event still unexplained?

    It seems to me that our stating that the man had the free will to jump merely states that it was physically possible for him to decide to do so, as demonstrated by the fact that he actually jumped. Well, does anyone really doubt that we observed what we observed? We don't need explanations for trusting experiences, we need explanations to link related experiences, and we can only do that if the explanations are predictive.

    (BTW, I also fail to see why human free will is any more explanatory than chance is at explaining, say, why a boulder falls off a mountaintop.)

    Does the fact that a cause or effect cannot be explained by facts or rules, mean that it is inexplicable?

    IMO, yes. As you suggest, definitions are conventional, so you are welcome to devise your own definition of an explanation. How would you prefer to precisely define an explanation?

    ReplyDelete
  5. I don't intuitively regard my life history and genetic makeup as coercing my choice to eat Corn Flakes, even though they strongly (if not totally) determined why Corn Flakes is my favorite.

    To explain freewill in this way seems to me to be engaging in nothing more than a game of semantics. By freewill I mean that there is the possibility of self-caused action (along with possibilities of action caused by another and uncaused action). The sort of freewill you have described really has no bearing on the term "will" other than the fact that it is falsely assumed that we have that ability. So in your example, regardless of the fact that you don't intuitively regard the deterministic effects as making you choose Corn Flakes, there is no positive freewill by the definition of being a self-caused action (it is still caused by another influence).

    So you aren't really approaching the idea of freewill but rather the fantasy there of. This is strange in the fact that, for us to a decision on whether a statement is logical or truthful, a self-caused action seems to be necessary (otherwise the actual truthfulness of the statement is irrelevant because my response is already determined).

    If an event (e.g., a choice) is unconstrained by past history, then it is random. Surely, that's acausality, not a form of causality?

    As I have already suggested, a self-caused event is not acausality but rather a break in determined causality that allows a being (whether necessary or contingent) to make its own cause. In the case of God every action is self-caused. A self-caused event is not necessarily devoid of outside influence, but is often a response to it.

    Suppose a man jumps from a bridge, and nothing in his known life history suggested he would take such an action. Should we feel satisfied that the action is explained when we say he had the free will to jump? Or is the event still unexplained?

    I am not implying that the simple response of "freewill" is adequate to give an explanation acceptable in all circumstances. The possibility of self-caused events does not eliminate outside influence nor the fact that events can be caused by other events. I would not feel satisfied in the case of the man jumping by simply responding with "freewill."

    It seems to me that our stating that the man had the free will to jump merely states that it was physically possible for him to decide to do so, as demonstrated by the fact that he actually jumped.

    This is by far more true under your definition of freewill as a psychological fantasy.

    And on your final point of defining an explanation:

    I honestly don't have a good solid definition for this word, but it seems that there are some things that can be sorted out. Explanation can involve three different sorts of events: uncaused, caused by someone or something else, or self-caused (not a self-caused being, but action caused by myself). So maybe it is correct that facts and rules are the explanation, but these facts and rules are more broad than defined by determinism.

    Also, on the concept of cause and effect I would suggest referencing Hume's work in An Enquiry Concerning Human Understanding. Particularly Section IV on the Sceptical Doubts Concerning the Operations of the Understanding. I have addressed some his discussion on Causality, the Relations of Ideas, and Matters of Fact in relationship to Economics.

    ReplyDelete
  6. Bryan,

    By freewill I mean that there is the possibility of self-caused action...

    I don't think your definitions are sufficiently precise.

    First, I don't see any conflict in a "self" causing events materialistically. That is, the self is a materially identifiable thing, there are material rules that govern aspects of the self, and we can thus identify materially self-caused events.

    For example, we are comfortable saying that the Pacific Ocean caused the erosion of Monterey Bay. We generally accept this claim, and ignore the fact that the chain of causation could be extended back to the comets that deposited the oceans on the Earth's surface billions of years ago, the Moon's tides, the wind's waves, the Sun's energy, etc. How are Pacific-caused events not analogous to self-caused events (i.e., implicitly neglecting causes of the ocean/self)?

    Second, the determinism/randomness distinction applies to the non-physical as well as the physical.

    An event is deterministic to the extent that it is determined by past events. The logical complement of the deterministic extent is an acausal and random extent. This is true whether or not one regards the event in question as physical. As long as we're using binary logic, there is no third, "free" category of causation (~deterministic=random).

    Relative to thought or "will," either our decisions have reasons as causes or they don't. If they don't, then they are truly random.

    This is strange in the fact that, for us to a decision on whether a statement is logical or truthful, a self-caused action seems to be necessary (otherwise the actual truthfulness of the statement is irrelevant because my response is already determined).

    I have never understood this criticism.

    Concluding that "1+1=2" really is a matter of making a determination that "1+1=2".

    It is the truth of the statement that causes us to reach our conclusion. So, it would seem to me that the reverse is true, that reliability of rationality is proportional to its determinism.

    This is by far more true under your definition of freewill as a psychological fantasy.

    Under my definition, free will is no more a fantasy than is hunger. Free will is a real, observable mental phenomenon. What is a fantasy is this supposed connection between the phenomenon of free will and acausality.

    I grant that this idea is counterintuitive, but it is rigorous.

    Explanation can involve three different sorts of events: uncaused, caused by someone or something else, or self-caused (not a self-caused being, but action caused by myself).

    How can an uncaused event be explained? At best it can be described. There is no cause to point to and say, "Ah, there's the explanation!"

    I read your blog entry, but haven't read Enquiry in a while. How do you see its particular relevance to our discussion?

    ReplyDelete
  7. I thinking you are missing my point on a self-caused event. The effect (that in an event "caused by another" becomes the cause) in a self-caused is the individual. This is only possible for an animated object with its on will.

    For example, we are comfortable saying that the Pacific Ocean caused the erosion of Monterey Bay. We generally accept this claim, and ignore the fact that the chain of causation could be extended back to the comets that deposited the oceans on the Earth's surface billions of years ago, the Moon's tides, the wind's waves, the Sun's energy, etc. How are Pacific-caused events not analogous to self-caused events (i.e., implicitly neglecting causes of the ocean/self)?

    Your example is not of a self-caused event, it is still an example of an event "caused by another." That is to say the movement of the Pacific Ocean does not originate within itself, but by the effect of these previous events. The distinction is that a self-caused event is a break in the causal chain. A self-caused event (not a self-caused being, which is impossible) originates within the individual, it is not just a recognition of an individual within a causal chain. The Pacific Ocean is an object, a "self", but the event of its motion is not self-caused.

    The debate is not whether there is anything but caused (determined) or random (undetermined), but rather the beginning of the causal sequence. I agree there is only caused or uncaused events, but there are variations within caused events (caused by me or not caused by me).

    I have never understood this criticism.

    Concluding that "1+1=2" really is a matter of making a determination that "1+1=2".

    It is the truth of the statement that causes us to reach our conclusion. So, it would seem to me that the reverse is true, that reliability of rationality is proportional to its determinism.


    You are using determination in a different sense here. Determination here means to reason out and decide compared to "deterministic" which means the effect is already a sure thing based on the cause. It's quite a simple criticism in that, what if you are determined (because of a previous causal sequence) to believe that 1+1=3. Your belief on the validity of 1+1=3 or 2 or whatever is determined by causation (causes created by events in your past, i.e. everyone telling 1+1=2, but what if it is not so?). So it’s not "determination" but rather it is determined. The same goes for your belief in materialism and mine in theism, they are "determined."

    Under my definition, free will is no more a fantasy than is hunger. Free will is a real, observable mental phenomenon. What is a fantasy is this supposed connection between the phenomenon of free will and acausality.

    Hunger is a bodily reaction telling us our status. I see no connection between these two events. It's observable that someone thinks they have freewill, but if you’re right, they don't. I still insist that it is a game of semantics to call that freewill.

    I agree that an uncaused event can't be explained by rules. The explanation itself is to say "indeterminism." "It just as well could have been otherwise."

    I see Hume relating in a tertiary sort of way when he tells us that we have no rational basis for expecting the future to be like the past. He says science is based on nothing other than convention or “habit of thought.” The simple point is that cause and effect is not discoverable a priori. It is an a posteriori conclusion that an event (cause) is followed by another event (effect). The inference of the temporal uniformity of nature is read from the past but there is no way to force this on the future (you can't even use the example of past futures because they are past). So the only way to come up with the principle of the temporal uniformity of nature is through induction, but induction doesn't give an a priori logical connection. Hume also criticizes the supposed connection of cause and effect by saying we don't know the "secret power of nature" that connects the two events.

    I would say read Hume for yourself, because that was probably a terrible summation of his argument. Anyway, that how I saw it relating to some extent.

    ReplyDelete
  8. Bryan,

    I think this is the crux of the debate:

    The debate is not whether there is anything but caused (determined) or random (undetermined), but rather the beginning of the causal sequence. I agree there is only caused or uncaused events, but there are variations within caused events (caused by me or not caused by me).

    A caused event is a determined one, and an uncaused event is acausal and random. These are not separate, independent attributes.

    An event that is a break in the causal chain is an acausal and random event. (Technically, an individual event may be partially determined by the past, but in that case, the indeterminate part is random and acausal.)

    Suppose the self were capable of causing its own events. To the degree that the self's actions are not caused by its past self or its past environment, they are acausal and random actions. I see no escape.

    It's quite a simple criticism in that, what if you are determined (because of a previous causal sequence) to believe that 1+1=3. Your belief on the validity of 1+1=3 or 2 or whatever is determined by causation (causes created by events in your past, i.e. everyone telling 1+1=2, but what if it is not so?).

    Well, let's use a more realistic example. "There is a lion on this plain." Those humans whose genes created nervous systems that incorrectly reported this kind of fact are no longer with us.

    Another example would be computer software. Computer software is highly deterministic, and it successfully reasons about the world, even learns from the world. Contemporary computers are not designed to capture knowledge like we do (to know is to know that you know), but the principle that a system can be made to know about its environment deterministically is well-established.

    One more example. Ants. They do not reason, as far as we know. Most theists would not say they have free will. And yet, they establish accurate models of their environment mechanistically.

    Hume's criticism of cause and effect is a valid one. However, the assumption that there are some knowable cause and effect relationships is an axiom of rationality. We have to make this assumption to reason about anything. The rules of logic, the associations between past thoughts and our memories of them must be assumed without proof. Without these assumptions, we could not formulate an argument.

    This isn't dedictive proof that some cause and effect relationships can be found, but it does show that, by reasoning, we have already accepted the assumption.

    ReplyDelete
  9. I have the beginnings of a continuation of this discussion posted here.

    ReplyDelete

Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)

Note: only a member of this blog may post a comment.