Does this philosophical argument about accidents and universal origin contain valid logical structure?

Question

I'm not enough of a philosopher to properly analyze this particular argument, but I'm curious about it and I'd like to see it dissected and explained.

"If the solar system was brought about by an accidental collision, then the appearance of organic life on this planet was also an accident, and the whole evolution of Man was an accident too. If so, then all our present thoughts are mere accidents – the accidental by-product of the movement of atoms. And this holds for the thoughts of the materialists and astronomers as well as for anyone else’s. But if their thoughts – i.e., Materialism and Astronomy – are mere accidental by-products, why should we believe them to be true? I see no reason for believing that one accident should be able to give me a correct account of all the other accidents. It’s like expecting the accidental shape taken by the splash when you upset a milk-jug should give you a correct account of how the jug was made and why it was upset." - C.S. Lewis, The Business of Heaven p. 97

Maybe this can't and shouldn't be altogether answered in a "yes or no" format without pulling from subjective opinion, but perhaps there is or is not invalid logic or circular reasoning or perhaps it's a perfectly acceptable logical argument, so far as structure is concerned.

Answer 1

I'm not going to pass final judgment on this, but for an argument with virtually the same structure, see Plantinga's evolutionary debunking argument against naturalism. He credits this argument by Lewis for the idea, and the main idea of the argument is that if natural selection selects for survival we have no reason to think it selects for believing truths. "Thus, Plantinga argued, the probability that our minds are reliable under a conjunction of philosophical naturalism and naturalistic evolution is low or inscrutable. "

As a final note, this way of construing the argument is invalid. But that's not because it is necessarily a bad argument, but because the most plausible version of it is an inductive argument. That's why Plantinga's conclusion is in terms of probability. If you tried to rigidify Lewis's argument it would also be more cheritable to construe it as an inductive argument, not a valid deductive argument.

There is a surprisingly high quality wikipedia article on the subject and you're free to check it out for some rebuttals and further discussion on the subject. I normally wouldn't trust wikipedia on philosophy but some really high quality editor must have gotten her hands on this one: evolutionary argument against naturalism

Answer 2

Lewis seems to be arguing against materialism and astronomy. Let's call this combination science for short.

Assumptions

• (A1) Whatever has an accidental cause is also accidental.
• (A2) There is no reason to believe accidental thoughts.
• (A3) Science implies that the emergence of the solar system was accidental.
• (A4) Science implies the evolution of life on earth was caused by the emergence of the solar system.
• (A5) Science implies that our thoughts are caused by the evolution of life on earth.
• (A6) Science is just our thoughts.

Derivation

• (1) Assume that science is correct.
  • (1.1) The evolution of life on earth is accidental. (A1, A3, A4)
  • (1.2) Our thoughts are accidental. (1.1, A1, A5)
  • (1.3) There's no reason to believe science (1.2, A2, A6)

Conclusion

• (C1) If science is correct, then there's no reason to believe it. (1-1.3)

We can take one step further and derive a conclusion that is not explicitly stated:

• (C2) Either science is wrong, or there's no reason to believe science. (C1)

If my analysis is correct, then this conclusion does follow from the above assumptions, and the argument is valid.

However, a few things to note:

• The validity of an argument doesn't mean a lot. It is of course a necessary condition for it to be taken seriously, but it is not at all sufficient.
• Debates over arguments in philosophy usually center on the assumptions (however implicit they may be) these arguments make, rather than on their validity (making valid arguments is not very difficult).
• To make a more convincing case, Lewis must also explain how his conclusion is compatible with the apparent success of the scientific enterprise.
• A2 seems to me very questionable. In particular, the relevance or importance of something's being an accident might be questioned. It may even be a genetic fallacy. A6 might be questioned as well.
• This strikes me as a particularly bad argument against science as a whole. But this is my opinion.

Answer 3

The answer is no, this argument is invalid even as informal one, and for several independent reasons. First, it conflates causes with reasons. Our beliefs being accidental has no relation whatsoever to said beliefs being rationally justified or not. A necessary event is no more or less prone to produce a correct belief than an accidental one. Second, near the end Lewis confuses "accidental" with "randomized" in his analogy. The shape of the splash contains little information about the milk jug not because it is accidental, but because the process of splashing randomizes the outcome. In contrast, the shape of a drum, for example, can be largely recovered from the sounds it makes under "accidental" drumming, see Can One Hear the Shape of a Drum? It is unclear why formation of the solar system, etc., is more analogous to the splash than to the drumming.

Plantinga attempted to "save" Lewis's argument by linking reliability of reasons to their causal origins through evolution theory, in other words he argued that evolution, as described by biology and materialism, is "unlikely" to produce reliable knowledge:"what is the likelihood, given materialism, that the content that thus arises is in fact true?" This modified argument suffers from a different problem, the same one as Pascal's wager, it injects probability where it doesn't belong. More specifically, Plantinga commits an instance of base rate fallacy, by assigning "prior probabilities" to "events" like "materialism is true", for which probability makes no sense.

It is interesting that essentially the same type of base rate fallacy is committed in the "no miracles" argument, which is used to support scientific realism ("it is unlikely that what science says about reality is wrong given its success"). I guess fallacy, like money, does not stink.

Answer 4

Not only is C.S. Lewis's argument valid, but it is also finding its quantification and articulation in computer science.

Unsupervised Learning

The closest approximation that we can come to Lewis' idea of an accident giving "a correct account of all the other accidents" is in the field of machine learning. Within that field, there is a distinction made between supervised and unsupervised learning based on whether or not training data is utilized to prepare the computer for the actual test data. In the case of unsupervised learning, the computer is presented with the test data with no preparation. Although this may appear to be learning from a Lockean tabula rasa, the fact is that even in unsupervised learning, certain assumptions are always incorporated to make it possible for the computer to process the data:

"In the classical literature, most unsupervised learning algorithms are essentially algorithms for performing on-line clustering of the data. They are based on the assumption that clusters are likely to correspond to categories — an instance of the general epistemological assumption that 'nature is not a cryptographer.' This important assumption, which has extensive empirical support even if its philosophical status is not entirely clear (Wigner 1960), provides justification for the study of unsupervised learning algorithms. The need for such assumptions however is not restricted to the unsupervised learning paradigm - assumptions that nature is 'simple' are also necessary for the theory of supervised learning. The major problem in supervised learning is that of interpolating from the categories of the data in the training set to the categories of novel inputs. Such interpolation can only be based on assumptions about the simplicity of the natural process that generates the data." (Michael I. Jordan and Robert A. Jacobs, "Modularity, Unsupervised Learning, and Supervised Learning")

Peter Dayan of MIT also comments on the need for a priori information as one of the prerequisites for unsupervised learning:

"The only things that unsupervised learning methods have to work with are the observed input patterns xi, which are often assumed to be independent samples from an underlying unknown probability distribution PI [x], and some explicit or implicit a priori information as to what is important." (Peter Dayan, "Unsupervised Learning")

Any type of algorithm that is used for machine learning starts with assumptions with respect to the existence and value of data. These assumptions are programmed into the algorithm, and the computer is given all the tools that are necessary to collected data and perform various statistical operations on it. Far from being accidental, unsupervised machine learning involves sophisticated statistical operations and computation.

No Free Lunch

In spite of the fact that machine learning has a priori knowledge as its starting point, not all algorithms are created equal and no given algorithm is successful for all types of data. The study of this particular problem resulted what is known as the "No Free Lunch" (NFL) theorem, published in 1997 by David H. Wolpert and William G. Macready.

"Roughly speaking, we show that for both static and timedependent optimization problems, the average performance of any pair of algorithms across all possible problems is identical." (Wolpert and Macready, "No Free Lunch Theorems for Optimization")

Even though the average performance of algorithms is identical when tested with all types of data, a given algorithm may be especially suited for a particular type of data. However, the NFL theorem also implies that its performance will suffer when tested with data for which is not suited.

"[T]he main message of the NFL theorems may be summarized as follows: If there is no restriction on how the past (already visited points) can be related to the future (not yet explored search points), efficient search and optimization is impossible." (Christian Igels, "No Free Lunch Theorems: Limitations and Perspectives of Metaheuristics")

Succinctly expressed, the NFL theorem holds that not only are assumptions made when designing algorithms but their success is fully dependent on those assumptions:

"[I]f the practitioner has knowledge of problem characteristics but does not incorporate them [...], the NFL theorems establish that there are no formal assurances that the algorithm chosen will be at all effective." (Wolpert and Macready, "No Free Lunch Theorems for Optimization")

"[Y]ou can't make a clustering algorithm without making some assumptions about the nature of those clusters." (David Robinson)

Even if it were possible for a genetic mutation to accidentally equip an organism with the ability to perform the biological equivalent of statistical calculations on its own physiological states, the practical utilization of those calculations has been demonstrated to be useless without some sort of a priori guidance. For this reason, the idea of accidental knowledge is and always will remain theoretically implausible.