Friday, March 26, 2010

Inference to the best explanation

       Aleksandar Jokic, a philosopher who teaches at Portland State, sent me a friendly amendment about my use of a fundamental philosophical term inductive inference to the best explanation (Gilbert Harman) which comes from C.S. Pierce (abduction, as Aleksandar rightly calls it in the note). This is perhaps the leading idea in the study of non-pure mathematics explanations.  So it would be valuable, I think, to many readers, to see what is at issue in this argument, as opposed to common but distorted views, for instance, empiricism (more accurately, deteriorated logical positivism via IQ testing in psychology) which dominates political and social “science.”   Understanding inference to the best explanation will also prove valuable in a discussion of moral objectivity or moral realism (with Hilary Putnam and Steve Wagner) which I will pursue in the near future. For a longer consideration of these matters, see Democratic Individuality, chs. 1-4.

     In many cases, from ordinary reasoning to detective work to physics, what we do is to compare significant evidence* with reasonable inferences, and decide which is an inference to be best explanation.  This may take a long time, we may  get it wrong for a time, or, as Popper suggests, the best explanation may be surprising or counterintuitive.  For instance, many of us initially have intuitions about American foreign policy which are, given evidence, mistaken but hard to shake; a counterinference that these policies arise from a destructive war complex or empire may be hard, at first, to reach, harder still to unravel the implications of fully.  Or Greeks discovered the novel human possibilities of the polis – of democratic debate on great issues like war and peace as opposed to the religiously-“inspired” commands of a despot, the initial and then uncontroversial form of rule (cf. Leo Strauss today) – as Aristotle suggests in book 1 of the Politics.  Or we have learned that Aristotle offers an uncontroversial notion of just war – one of self-defense by a free regime, one which has  at least a limited common good – and a mistaken view that slave-hunting is a form of just war.  After two millennia of struggle, Montesquieu and Hegel rightly rejected this view of bondage and “justice” and founded modern liberalism.  The last two are striking examples of what I call moral progress in the first chapter of Democratic Individuality (Hilary makes a related point about moral learning in a recent lecture), and the first is a candidate if we ever manage to get beyond it (even Obama, despite his many wars, gives some halting signs of it on torture or Iran or Israel or agreements to reduce the danger of nuclear war with Russia…). 

        The course of reasoning is not deductive (from premises, as in mathematics), but from relevant facts. It is thus like induction or as Harman also says, an inductive inference to the best explanation.  But it can also employ highly developed theories (as in physics), and arrive at deductions from them.  The reasonableness of those theories is dependent on argument, given striking issues or anomalies, with leading, competing theories; when a leading theory goes awry about such examples, scientific discovery involves, as physics strikingly reveals, remarkable and before unimagined inferences to the best explanation (quantum mechanics, relativity).  One scientific realist amendment to Aleksandar’s account below – the second step in abduction, that my hypothesis explains the surprising fact – is only in the context of a web of scientific (or practical) reasoning which includes leading alternate hypotheses which make the phenomenon significant, the hypothesis plausible (i.e. the Michelson-Morley experiment and Einstein’s rejection of aether in his 1905 paper on special relativity).  One needs to explain things which the prevailing theory or explanation does while also and unexpectedly illuminating anomalies for it.

       Aleksandar is also responding to what one might call genocide-shock. In Europe and in the United States, political murder, associated with fascism, is all around us.  If one takes on what torture is (read Andrew Sullivan or Scott Horton over the last years; Andrew pinpoints the moral issue with photographs from Abu Ghraib or the murders in American custody), one will be repulsed not only by the Cheneys, Condi Rice or Rumsfeld but by Democrats, even  Obama, who have tried to hide and protect them.  The US often indicts others for crimes (sometimes exaggerations of crimes) which its own record of murder (or even collaboration in, as with Saddam) sharply outweighs.  In the late 1990s, for example,  the assistant General Secretary of the UN Denis Halliday and 2 other UN humanitarian leaders resigned because of the genocide – some hundreds of thousands of children unnecessarily dead - caused by the UN boycott of Iraq, pushed by the United States and Great Britain (see John Pilger’s “Paying the Price” made for BBC and never shown on American television).

     But we need to heal up the world.  This means stopping the crimes, holding investigations and hearings to restore the rule of law, and exacting punishment.  I prefer Truth and Reconciliation Commissions.  This is, however, the nonviolence of the strong (Tutu, Gandhi)  In America, it would be hard to achieve more than punishment (leaving Cheney disgraced and isolated on some small island, say a Guantanamo without torture, with clothes, food and Fox news – where he can do no more harm and live out his madness and maybe even heal some – say, turn off Fox News - strikes me as sufficient).  But this is a long way up from here.

      In this murderous climate, Aleksandr speaks of revulsion for Nietzsche and Heidegger.  With Hilary Putnam and Tracy Strong, I read Nietzsche as having many ideas, ones which do not simply or even mainly point in a fascist direction (though there is a nasty current that does). And of course I am struck also by what Leo Strauss calls Nietzsche’s “inexhaustible power of passionate and fascinating speech.” Heidegger’s writings on poetry, such as his 1946 “What are Poets for?” (Wozu Dichter?), though stained by German reactionary nationalism of a sort alien to Hoederlin (he prefers Hoelderlin to Rilke largely on nationalist grounds) are often quite wonderful.  His critique of technology and arguments about mortality also open or develop paths through forests.  But I know exactly what Aleksandar is speaking of.  As I said in the first post on Enmity and Tyranny here, I looked into the darkness of Schmitt’s medieval anti-semitism and it to some extent possessed me (I felt ill and had to take steps to dispel its effects).  Nietzsche is a brilliant psychologist: when you look into the abyss, the abyss looks into you.  As Aleksandar suggests, some things are so awful that they can burn right through you.  To walk a path, one must be careful about how one looks.

“Dear Alan, I read your blog post today and I am writing to urge a friendly amendment. Let me just first mention that I was quite happy to discover the blog almost simultaneously with but independently of Tiphaine around the time she was working on her Strauss paper. You deal with some issues that are of interest to me in relation to the phenomenon I call genocidalism that I have been exploring for some time. However, those things you present in a way that is highly readable, with clear argumentation, when explored in the original are too painful for me to engage in. By this I mean to tell you that I could no longer summon the force to read Nietzsche or Heidegger and their cronies. In other words, continental philosophy  is something I find almost impossible to read. Yet, you are able to render people like Strauss or Schmitt understandable, you reveal with great clarity what their views were, their arguments (or lack of), and influence that spare me the necessity to go after such interpretations myself to see their covert and open support for fascism and the rest. My friendly amendment concerns the phrase "inductive inferences to the best explanation". Now the phrase suggests a single type of reasoning where there are two. And my reading of Harman and others is that they are saying as much about the two different kinds of reasoning processes. To make the point that there is no such thing as inductive reasoning that results in the best explanation let me start with a few definitions: An *argument* is a group of statements divided in two parts: (i) the claim, and (ii) evidence offered in support of the claim. *Reasoning* is the process of connecting evidence and the claim. There are exactly three kinds of reasoning processes: 1. Deductive (D)
2. Inductive (I) 3. Abductive, also called Retroductive, also called Inference to the Best Explanation (IBE) Now, the phrase from your blog suggests that IBE is a kind of I (inductive reasoning) when in fact these are two very different kinds of reasoning processes. What they have in common is that they both, unlike D, allow for the possibility that even the best I and IBE arguments in fact have true premises and false conclusions--that is because the intention is not, like in deductive reasoning--to provide an absolute, 100% guarantee that if the premises are true then the conclusion also must be true. Instead, the intended connection between premises and the conclusion is something less than absolute guarantee, that is good likelihood or probability that if the premises are true so is the conclusion. But the difference between I and IBE is that IBE does not proceed from a sample and does not extrapolate that a population that the conclusion talks about has a certain property based on the fact that the population that the premises talk about has a given property. Instead, IBE is based on the explanatory power of the hypothesis asserted in the conclusion (and offered in the second premise). In other words, an Abductive argument has roughly the following form: 1. A surprising phenomenon, P, has been observed. 2. If my hypothesis, H, were true, then P would be explained. Therefore: 3. H is true. Now, what Harman is arguing in the paper you are citing is not that IBE is the same sort of reasoning as I. He is arguing that a certain type of argument called enumerative induction is better understood as an abductive rather than inductive argument. It remains true that I and IBE are different kinds of reasoning, producing different kinds of arguments that would have to be evaluated differently. That is, if we are considering an I argument the conclusion of which we don't like we would have to go into the questions about the sample: is it representative, large enough, etc. And if we are evaluating an IBE whose conclusion we do not like the challenge is very different: we must offer a better explanation, a different hypothesis that does a better job. In conclusion, I am presenting you with reasons to reformulate your point in such a way that does not suggest that IBE is a kind of I, but a different kind of reasoning altogether. Best regards, Aleksandar”

*significant - determined by leading theories in a domain of investigation at the time.  All scientific investigation of facts is, as has been said by Norbert Hanson, "theory-loaded" or theory-governed. The common forms of empiricism or behaviorism, as philosophical views purporting to account for the successes of science, don't have a chance.


Amentahru said...

great post. Of course, I already think Dr. Jokic is a stud. I am curious to what extent the above cited authors differ from Rawls in any signficant way. For example, Rawls writes in "Political Liberalism" that reason is thought, inference, and judgment (I don't actually have the text in front of me but that's what i remember without going directly to it).

Finally, given that we are not dealing specifically with numbers, how do we "decide" on the better explanation (in the case of IBE)? Ideology seems to routinely come in to play. To wit: over 30% of Republicans think that Obama is doing what Hitler did (see the recent SPLC Intelligence report).

Alan Gilbert said...


Last point - ignorance is not inference to explanation. Or this would not be inference to the worst explanation.

Most inference to the best explanation does not involve numbers (viz. how to find one's way to a nearby good restaurant). And mere numbers as in David Lake's "Powerful pacifists" are no evidence of common sense, let alone some scientific ibe as I indicated in a post of that title some months back.

Rawls know of ibe and his reasoning often brilliantly illustrates it.

Post a Comment