I’ve talked about deductive reasoning and inductive reasoning before. Both of these are important mechanisms we use for arriving at conclusions. The first makes use of the rules of logic – which we can essentially view as limiting ourselves to the definitions of words like true and false and not allowing ourselves any other resources. The second we can view as showing that something is probably true be accumulating lots of examples of that truth occurring in various different situations. Both of these are important and useful, but there is yet another way of thinking that is also essential – which is often called abduction or abductive reasoning.
What is Abduction?
The deductive method emphasizes the rules of logic and doesn’t allow much else into the picture. The inductive method emphasizes accumulating many examples to point towards a specific explanation of those examples. Abduction is very different from both of these. The idea with abduction is that we want the best explanation of the information available to us. Let’s give an example before we try to get too specific.
Suppose that you walk outside in the morning and all of the ground around you is wet, but it is not currently raining. You probably will immediately think that it probably had finished raining not too long ago. But, for the sake of discussion, how is it that you reject the idea that someone can in the middle of the night and dumped buckets of water everywhere you can see? This idea clearly explains the wetness of the ground, and since there isn’t anything we can “repeat” in an experimental way, we can’t use induction. And yet, I doubt any of you reading this paragraph would doubt your belief that it rained not too long ago based on my alternative explanation.
What you are doing subconsciously here is using a kind of abductive argument. You are thinking to yourself something like “Well, yes, your alternative does explain why the ground is wet. But the explanation I originally came up with is better, so I would need you to show me some proof before I give up my belief that it rained yesterday.” You think this way because the hypothesis of rain is better at explaining what you know than the hypothesis of the buckets of water.
Now, suppose that someone showed you a time-stamped video of someone dumping buckets and buckets of water all over the area you are standing last night. In that situation, you may well shift what you believe. If you walk a ways away and the ground if entirely dry, this will probably convince you. This is because you’ve received new information that doesn’t make sense if it actually had rained earlier, but does make sense if someone had come by while you slept and dumped water everywhere near you.
This is exactly the idea that goes behind abduction. Abductive arguments collect all the information we know about a certain situation, formulate the various ways that we might explain that information, and then try to find which of those does the “best job” of explaining what you see. When all you knew was that the ground was wet, rain made the most sense. But when you found dry ground nearby and saw the video, you gained more information and that new information caused your belief to shift. This is the heart of the abductive method.
But, we must ask, how do we know which option is the best one? What do we mean by best anyways?
What Criteria Do We Use?
When using abductive arguments, we need to be clear about how we are judging which explanation is the best. For example, it wouldn’t be acceptable to use your own emotional preferences as part of what you mean by “best” since that makes the entire discussion depend on your personal opinion, which is exactly what we are trying to avoid. But, what criteria can we use then? There are a lot, which may apply to greater or lesser extents depending on the circumstances. The best way to approach this situation (notice I used the word best… what do I mean by best here?) is probably to make a list of various criteria that are often used in abductive arguments. These are often called explanatory virtues, because… well… it is a “good” thing for an explanation to have them. Fairly intuitive name for them. But what are these explanatory virtues? There are many. Here are some examples of qualities that might count as explanatory virtues.
- Explanatory Power: How well does the theory explain what we know?
- Example: Einstein’s theory of gravity predicts orbits more accurately than Newton’s theory of gravity does.
- Explanatory Scope: How many different things that we know does the theory explain?
- Example: Einstein’s theory of gravity does everything old theories of gravity do, and it does even more.
- Simplicity: How many new things do we have to accept to adopt this theory?
- Example: If someone hid a practical joke in your room, it was probably one or two people, not one hundred. And it was probably your brother, not aliens.
- Coherence: This theory meshes well with things I already know about the world.
- Example: If you know I like math, then you learn that I have a lot of math and science books in my house, that fits well with what you already know about me. If you know I hate math but learn that I own dozens of math books, that does not fit well.
- Not Ad-Hoc: Does this theory seem arbitrary or contrived? Does it feel like someone ‘pulled it out of thin air’?
- Example: Explaining Christmas presents by appealing to Santa Claus, a person whose sole purpose of existence is to make the presents appear under the Christmas tree.
- High Prior Probability: Is there something intuitively (or logically) plausible about this theory? Does it ‘make a lot of sense’?
- Example: Even before you count up ballots, you pretty much know who California voted for. Same with Mississippi.
- Multiply Evidenced: Is there evidence for this theory from a variety of different sources and different types of sources?
- Example: If your theory has evidence from cosmology, biology, history, and archaeology, you’re looking pretty good.
Why Does Abduction Work?
It is actually very strange that abduction works at all. This is because, on the level of deductive logic, abduction is a fallacy. When we use abduction, we are considering some reality B and try discover the cause of B (call this cause A). In terms of logic, we are saying something like “since we know B, and A best explains B, therefore B probably implies A.” This is extremely similar to the fallacy of affirming the consequent – the fallacy of saying that if we know that B and that A implies B, then we know A. This reasoning is false, and yet in the case of abduction it tends to work pretty well (although it is not perfect). How is this so?
If I’m honest, I don’t think we can ever be absolutely certain about this. Especially because, sometimes, it doesn’t work. Sometimes life is more complicated than that, and sometimes we don’t have all the evidence. And even if we do – why should the concepts that we humans happen to think are explanatory virtues have anything to do with whether the theory behind the virtues is really real?
This reminds me of Einstein’s keen observation about science, “The most incomprehensible thing about the world is that it is comprehensible.” It is a daunting question, in my opinion. I think that something like theism would help, as in that case the universe may well have been created by a being who made it orderly and who made human minds in such a way as to discover that order. But, nonetheless, I believe that we are all should continue believing that abduction will work, even if we don’t always fully understand why. It is a great blessing that the way we think about goodness and rationality so often lines up with the world in which we live. And using abduction well and thoughtfully is a way we can express our gratitude for that gift.