Gilson somewhere notes that the tendency of people in the nineteenth century to assume that Newtonian physics showed that the universe was deterministic was not based on any kind of rigorous argument, and could not be. What it actually was based on was something more like an imaginative sense of analogy based on how it was often presented. And this seems quite right. Actually moving from Newtonianism to a deterministic universe requires either ignoring or rejecting any number of possible qualifications (e.g., do we have good reason to think that Newtonian physics actually covered everything in the universe). In fact, it's at best controversial whether Newtonian physics is even fully deterministic in its own right; John Norton's dome is a famous case in which a scenario apparently consistent with all basic principles of Newtonian physics yields a result that violates determinism. But when we think of Newtonian physics we think of things like billiard balls hitting each other, and thus a billiard ball universe, etc.
Nineteenth- and early twentieth-century anthropologists often covered this kind of reasoning with the label 'magical thinking' -- or did when they found it in non-European cultures. But obviously this is not something that is unique to the 'primitives', except in the sense that everyone is primitive. And I've actually been considering the idea recently that people's evaluations of arguments are strongly affected by the kinds of associative similarity that anthropologists looked at when discussing magical thinking. In general, all their explanations of it treated the main nub of magical thinking as not making the distinction between associations between ideas and relations between things represented by ideas. It's certainly the case that people often fail to do this in evaluating reasoning, confusing associations among their own ideas of arguments and positions with actual relations among those arguments and positions. Frazer famously argued that sympathetic magic could be summarized in two laws:
Contagion: Things in contact continue to be in contact even where physical contact is broken.
Similarity: Like effects can be had by making causes alike.
Frazer, of course, was assuming that we are talking about physical objects, but one can easily find analogues of these in people's evaluation of reasoning. For instance, people have a tendency to assume that because two different philosophical positions have been found together at some point that they really have some connection, even if there is no logical requirement for this.
In fact, of course, as Levy-Bruhl pointed out, many of the instances of magical thinking were straightforwardly recognizable as logical fallacies -- post hoc ergo propter hoc was the one he focused on. But, of course, this is arguably committed as much by Europeans as by any 'primitive' tribe; we just think of it as bad magic. And this kind of thing can be found in evaluation of arguments and positions as well -- the 'genetic fallacy' is just a name for post hoc ergo propter hoc when we are talking about positions.
Likewise people have their rituals in argument evaluation. Philosophers like to set out the premises in an orderly numbered fashion, and tend to regard this as making an argument clear. Whether or not it actually does so depends; unless the argument is being made from scratch, this procedure involves rearrangement and interpretation, so whether it actually does make things more clear, in terms of increasing understanding, seems to vary considerably. But it still feels like you are bringing order and clarity to a disordered muddle, so you find people who will swear by it, even though it's not difficult to find cases where it clearly introduced a distortion. There's an argument to be made -- it would, of course, be controversial among those who engage in this kind of practice -- that such people are taking the ritual itself to be a kind of clarity, by sympathetic magic, and are taking arguments in this form to be better arguments simply because they conform to ritual expectation. It may even have good practical results, if so; a ritual might well put one in the right state of mind for a certain kind of work, and there's no reason to think that philosophical thinking doesn't sometimes need 'being in the right state of mind' as much as any difficult endeavor. And, of course, you find people who try to refute arguments by naming them -- a practice difficult to avoid, but not really all that different from shamans casting out illness by naming it.
Of course, unlike the anthropologists back then, we needn't regard 'magical thinking' as necessarily a bad thing; very likely it's just a first thing. It's rather funny reading Frazer on similarity, for instance, given that his account of the law of similarity for sympathetic magic often reads like a straightforward summary of Humean accounts of causation (which, of course, are based on associative similarities of a particular kind). It does not follow automatically from this that they are mere mumbo-jumbo. And something that may be poor reasoning in one context may be quite good in another context -- the failure to recognize this as at least a possibility is one of the most rookie mistakes when it comes to evaluating reasoning. And I'm often amused when anthropologists talk about how magical thinking often involves thinking that one's words have a causal influence on the world, since they often are sloppy enough that their formulation implies that getting someone to pass the salt by saying please is magic. Which perhaps it is; it's not as if any of us knows the precise causal connection, or even usually cares. In any case, my point here is that one need not take the label as an insult. What 'magical thinking' really is, of course, is a system of analogical inferences, and not only can we not avoid making analogical inferences, many such are quite reasonable. To that extent it's both unsurprising that people would engage in magical thinking when evaluating arguments and not necessarily in a bad way -- their 'magical thinking' might well provide a good starting point for rigorous analyses, although, of course, that might depend on any number of factors. It bears thinking, anyway.
