Common Inductive Arguments

13 Inference to the Best Explanation

Section 1: Introduction

In the last chapter we looked at Arguments from Analogy.  In this chapter we will turn to another common non-deductive argument form: Inference to the Best Explanation.  We are constantly seeking explanations: why did my computer stop working?  Why did the housing market crash?  Why does my foot hurt so badly this morning?  This is especially common in our dealings with other people: why hasn’t she called me back? Why is the baby crying? Why does that guy always sit by himself in the cafeteria?  In general, when we reason to an explanation, we start with an observation and work backward to identify its likely source or cause.  In many cases, it is not difficult to come up with an accurate explanation, and overall we tend to do a reasonably good job at it.  Nonetheless, we do make mistakes.  In order to understand and recognize these mistakes, we will need to slow down and learn what makes Inferences to the Best Explanation distinctive, the difference between logically strong and logically weak instances of this form, and some questions to help us distinguish between the two.  We will close by considering a bias we are subject to when it comes to explaining other people’s behavior in particular.

Section 2: Inference to the Best Explanation

In most cases, to offer an explanation for some event or state of affairs is to reason to a conclusion.  To illustrate this, let us begin by looking at some examples of explanatory reasoning.

Ex. 1:

Andre: Oh no!  Somebody stole my bike!

Bobby: What do you mean?

Andre: I left it right here 10 minutes ago and now it is gone!

Andre returns to find his bike missing.  This state of affairs calls for an explanation, and Andre concludes that his bike has been stolen.  We can represent his reasoning as follows:

  1. I left my bike right here 10 minutes ago and now it is gone.
  2. So, somebody stole my bike.

Here is another example:

Ex. 2:

Jessie says: “Kayden didn’t return my text; I must not be very important to Kayden.”

Again the fact that Kayden didn’t return Jessie’s text is something that calls out for an explanation (at least to Jessie!).  Jessie reasons as follows:

  1. Kayden didn’t return my text.
  2. So, I must not be very important to Kayden.

Notice that in both cases there is a significant gap between the premise and conclusion.  In each case, the only stated premise is an observation, and the argument then jumps to the conclusion that this observed state of affairs has a specific explanation.  On what grounds should we think that these arguments are logically strong?  Let’s focus on Ex. 1—why does the fact that Andre’s bike is missing make it probable that it was stolen?  Given the Rationality Assumption, the author is assuming a connection here, so let’s go deeper and try to identify an appropriate missing premise.

Bicycle against colored wall
“RESPECT: 3-Speed Flying Pigeon Racer w. Wald Basket” by ubrayj02 CC BY 2.0

Andre expects that his bike will be right where he left it, and when he finds that it is missing he immediately wonders what has happened.  Although he quickly concludes that it has been stolen, it is important to recognize that there are many other explanations for why his bike might be missing other than that it was stolen.  After all, a friend might have borrowed it, Andre might be misremembering where he left it, his friends might be playing a joke on him and have camouflaged it in some way, and so forth.  Thus, in claiming that the bike was probably stolen, Andre is identifying one explanation for why his bike is gone as the most likely one.  Thus, the missing premise says something like this:

  1. I left my bike right here 10 minutes ago and now it is gone.
  2. The most likely explanation for why my bike is gone is that somebody stole my bike. (MP)
  3. So, somebody stole my bike.

The missing premise in this case shows us how the stated premise likely connects to the conclusion, and we can generalize on this.  After all, for any given state of affairs there are many possible explanations, and to pick one of these possible causes or explanations out is to say that it is the most likely or best explanation.  This is why such arguments are called inferences to the best explanation.  That is, an Inference to the Best Explanation argues that some specific explanation is probable given that it is the best or most likely explanation among those available for an observed state of affairs.

We should add a few brief notes about Inferences to the Best Explanation.  First, it is important to recognize that by the ‘best explanation’ we are not speaking objectively here.  We have limited knowledge and in most cases cannot realistically expect to have the actual best explanation of some event within our grasp.  What we can do is to determine the best explanation given the information we have or can access.  Second, Inferences to the Best Explanation are defeasible, and what explanation counts as the “best” can change given the acquisition of new information.  A third note is that in everyday thinking and conversation we rarely make the “best explanation” premise explicit.  That is, most cases are like Ex. 1 and Ex. 2 in that we simply jump from an observation to an explanation.

It is worth drawing special attention to a couple of contexts in which Inferences to the Best Explanation play a particularly important role.  First off, consider court cases.  Suppose Jack is charged with the murder of Xavier.  The prosecutor will make a case that the best explanation for what is known about the murder of Xavier is that Jack did it.  The defense, of course, will maintain just the opposite.  They will try to undercut the prosecutor’s evidence and to show that there is reasonable doubt that this is the best explanation (perhaps by presenting an alternative explanation for the facts surrounding Xavier’s murder).  A second prominent context is the diagnosis of illness.  You don’t feel well and go to the doctor.  The doctor will take account of your symptoms, and run some tests.  The doctor will consider various explanations for your illness in light of your symptoms and test results, and the best explanation will be the diagnosis of your illness.

Section 3: Evaluating for Logical Strength: Step One

When it comes to evaluating Inferences to the Best Explanation for logical strength, the first and most obvious step is to determine whether the proposed explanation is plausible or not.  This first step amounts to a check on the internal logical strength of the argument.  To illustrate, go back up to Ex. 2.  In this case, we need to ask ourselves whether the proposal that Jessie isn’t important to Kayden is a plausible explanation for why Kayden didn’t return Jessie’s text.  Although we don’t know all the specifics of the case, in general terms, the answer seems to be, yes, it is a plausible explanation.  After all, we often don’t return phone calls or texts promptly when the person involved is not important to us, and so this proposal would plausibly explain why Kayden didn’t return Jessie’s text.  That said, we are only part way through our evaluation, and this brings us to Step Two.

Section 4: Evaluating for Logical Strength: Step Two

The second step in evaluating Inferences to the Best Explanation is to do an external check for logical strength by looking outside of the proposed explanation for plausible alternatives.  This is a crucial step since the fact that the proposed explanation is plausible does not mean that it is the best or most likely one.  There may be other more likely explanations available, and we won’t know unless we look.  Despite its importance to evaluating Inferences to the Best Explanation, this step can be less obvious than the first one.  In fact, we often fail to consider whether there are alternative explanations, and simply take it for granted that the first explanation that we are given, or that occurs to us, is the best one.

Before saying more about Step Two, let us briefly pause to note that going with the first explanation that occurs to us is not always a bad thing.  Indeed, sometimes the first explanation that pops into our heads does so precisely because it makes the most sense in light of our knowledge and experience.  For example, suppose that class starts and you notice that your friend Ellis is absent.  There are many possible explanations for this: she dropped the class, she had to take care of a family emergency, she broke her leg on the way to class, etc.  Let’s say, however, that over the last few years Ellis has been late or missed morning classes a lot, and in every case it has been because she overslept.  Given this background information it would probably naturally and automatically occur to you that she overslept again, and this really would be the best and most likely explanation given your knowledge and experience.

Importantly, however, not all cases are like this one.  A particular explanation can occur to us for a variety of reasons—some of which have nothing to do with the explanation’s being the best or most likely one.  Return to Ex. 2: Jessie automatically infers from the fact that Kayden hasn’t returned her text that she is not important to Kayden.  Why is this the first explanation that occurs to Jessie?  It is difficult to say, of course, but it may be that she has recently been wondering about the status of her relationship with Kayden.  But the fact that she has been wondering about this, does not necessarily mean that this explanation is the most likely one.  It follows that—at least in cases where it is important—we need to take the time to consider whether there are plausible alternative explanations.  So, are there plausible alternatives for why Kayden hasn’t returned her text?  Yes: maybe Kayden hasn’t noticed the text, maybe the phone is turned off, maybe the battery is dead, and so forth.  Ultimately, then, the second step in evaluating Inferences to the Best Explanation is to ask yourself whether there are plausible alternative explanations, and in doing so prevent rash or premature explanations.  Let us call a failure with respect to this second step, a hasty explanation.  More specifically, we will say that:

A person gives a hasty explanation when they mistakenly conclude that some explanation is the most likely because they did not adequately consider whether there were plausible alternatives.

Again, a person can’t legitimately argue that some explanation is the most likely one, if they haven’t considered whether there any others that might be better.  Thus, if Jessie concludes that she isn’t very important to Kayden without having considered alternative explanations, she risks a premature conclusion, that is, a hasty explanation.

Section 5: Evaluating for Logical Strength: Step Three

Suppose that you’ve considered the plausibility of the proposed explanation, and looked outside for plausible alternatives.  Of course, if there are no plausible alternatives, your evaluation is complete.  This is uncommon, however.  In most cases, there will be plausible alternatives, and this brings us to the third step in evaluating Inferences to the Best Explanation for logical strength—identifying the best or most likely one.  This is another external check for logical strength, and the task is to compare the plausible alternatives to identify the best or most likely among them.  Returning to Ex. 2, which plausible explanation is best or most likely?  Is it that Kayden hasn’t noticed the text, that his phone is off, that his battery is dead, or that Jessie isn’t important to him?  In order to answer this question, we need to think about what else we know about Kayden.  Thus we might ask: does he constantly check his phone?  Does he regularly turn his phone off? Does he regularly let his phone’s battery go dead?  Has he been ignoring Jessie in other ways?  Answering these questions can help identify the most likely explanation all things considered.

Nevertheless, comparing explanations can be difficult.  It can be easy to get bogged down in the information that supports or detracts from the plausibility of multiple explanatory alternatives.

Fortunately, there is a short cut, or heuristic, for identifying the best or most likely explanation.  The heuristic is:

When comparing alternative explanations, look for the one that would be least surprising.

In order to see why this technique is effective, think about what it means to say that an explanation is plausible.  When we are considering the plausibility of an explanation, we are trying to figure out how likely that explanation is in light of our experience and in light of what we know about how the world works.  Moreover, when things happen in accordance with our knowledge and experience we barely notice them.  For example, when you turn on the tap and water comes out, you don’t think about it at all.  This is exactly what you’d expect given your experience.  However, if you turn on the tap and nothing comes out, you’d be surprised.  This is unexpected given your knowledge and experience of the world.  More generally, how surprising you find some situation tells you how unlikely its occurrence was, given your knowledge and experience.  Consequently, we can use surprise to compare competing explanations.

Street Magician and little girl
“Card Tricks” by canopic CC BY-NC-ND 2.0

To illustrate, imagine that you are walking down the street and come across a street magician.  He asks you to pick a card randomly out of the deck.  He turns around and you draw the 8 of diamonds.  He turns back around to face you and says he is going to identify the card you chose by reading your mind.  He closes his eyes and is silent for a moment, and then says: “The card you chose was the 8 of diamonds.”  You are surprised!  How could he have known?  You ask him how he did it, and he replies: “Just like I said…I read your mind.”  As you walk away, you consider some explanations.  He could have used a mirror, he could have gotten a sign from somebody behind you, the card is marked, all the cards in the deck could have been the 8 of diamonds, or… he could have really read your mind.  As you think about these, you note that you didn’t see any mirrors or anybody standing behind you who could have seen your card, and he showed you beforehand that the deck had different cards in it.  This seems to leave only mind-reading.  At this point, however, you might want to use the “least surprising” heuristic by asking yourself: which would be less surprising, that he really can read minds or that you just failed to see how he did it?  If you are like most people, you would be very surprised if the magician can really read minds.  We do not have any experience of mind-reading, and probably don’t know of any cases of it.  In short, it would be surprising precisely because it is unlikely given your experience and knowledge.  Thus, even though you might not know exactly how the magician identified your card, when we look for the least surprising explanation, we know that the best explanation is that the magician somehow identified your card by conventional means, not that he read your mind.

In sum, the third step in evaluating Inferences to the Best Explanation is to compare the proposed explanation with the alternatives you’ve identified in order to determine which explanation is most plausible.  As we’ve seen, one way of determining this is to ask which explanation is least surprising.  Let us call a failure to accurately compare and select from your alternatives, a poor explanation.  More specifically, we will say that:

A person gives a poor explanation when they endorse an explanation that is not the best or most likely given their information.

To say that a person has given a poor explanation is not to say that the explanation isn’t plausible, it is to say that it is not the most plausible given the information available.  As an example, a person who concludes that the magician probably did read their mind will have drawn a poor explanation since this won’t be more likely than other available explanations (e.g. mirrors, somebody behind you, marked cards).

Where does this leave us? We have identified three steps to evaluating Inferences to the Best Explanation.  First, we determine whether the proposed explanation is plausible.  Second, we consider whether there are plausible alternative explanations.  Third, we determine which of the plausible alternatives is best or most likely, and we can do this by asking which alternative would be least surprising.

Three Questions to Ask of Inferences to the Best Explanation:

  • How likely is the proposed explanation?
  • Are there other plausible explanations? (Failure: Hasty Explanation)
  • Would the truth of the proposed explanation be less surprising than the truth of any competitor? (Failure: Poor Explanation)

Section 6: Explaining People’s Behavior

Although we put a lot of effort into explaining why the world is as we find it, we tend to focus in particular on other people’s behavior.  That is, we want to know why people say the things they say, and do the things they do.  It is important to briefly emphasize this, in part, because we face distinctive challenges when it comes to other people.  Explaining other people’s behavior can be difficult.  We are complicated creatures, and we have different knowledge and experience, different personalities and values, and pursue different kinds of goals.  Thus, part of the difficulty in explaining other people’s behavior comes out of the fact that other people can be so different from us.  There is a further, and related, complication—namely our thinking about other people is often biased in a variety of ways.  We will briefly discuss one bias in particular that social psychologists call Fundamental Attribution Error.  In order to understand the bias, let’s start by identifying two broad categories of factors that influence our behavior: personal factors and situational factors.  We have goals, interests, and preferences, and of course these personal factors are often the primary factors driving our behavior.  On the other hand, sometimes our behavior is primarily a response to a situation or circumstance that we are in.  This distinction is not an exclusive one, and often our behavior is a consequence of both personal and situational factors.  The problem is that we have trouble appropriately explaining other’s behavior in terms of these categories.  More specifically,

Fundamental Attribution Error is our tendency to overestimate the influence of personal factors, and underestimate the influence of situational factors, when explaining other people’s behavior. [1]

To illustrate this, suppose that somebody is driving very fast and cuts Ben off in traffic.  Ben immediately thinks, “what a jerk!” and wonders what is wrong with that guy.  The first thing to notice about this example is that it is an Inference to the Best Explanation.  Ben has made an observation: a man is driving very fast and has cut him off, and he’s jumped to an explanation of this behavior in terms of a personal factor: namely because he is a jerk.  Of course there are other plausible explanations—many of them situational.  The driver might, for example, be having a medical emergency; alternatively, maybe he is late for his sister’s wedding.  However, Ben didn’t think of these, and this illustrates Fundamental Attribution Error in action. That is, Ben risks making a hasty explanation. Ben’s first and natural impulse was to explain the man’s behavior in terms of a personal as opposed to a situational factor.  Noting this bias is important because it tells us to double-check our explanations of other people’s behavior by taking a moment to consider whether there might be unappreciated situational factors at work.


Exercise Set 13A:


Give three examples of inferences to the best explanation from your everyday life.


Inferences to the best explanation are defeasible.  For each of the examples presented in the text (Ex. 1-3) specify one fact that you might learn that would likely change a person’s judgment of the best explanation.


You have agreed to meet up with your friend for coffee, but she cancels at the last minute.  What does Fundamental Attribution Error suggest we are prone to think, and how might you combat this?

Exercise Set 13B:

Directions: Consider each of the following situations.  First give an explanation that might immediately occur to you.  Second, identify a distinct explanation.  Third, choose between the two.  Which is more plausible and why?  The goal of this exercise is to practice identifying and comparing different explanations.


You were just singing a song in your head, and then you hear it on the radio.


Japan has the highest life expectancy of any country in the world.


You look up into the night sky and see a tiny light moving slowly across the sky.


During the economic downturn following the coronavirus pandemic women lost their jobs at a higher rate than men.


You are at a restaurant, and although the group at the adjoining table ordered after you did, they get their food before you do.


More people come down with colds and the flu in the winter than any other season.


The state of Utah has the lowest per capita rate of beer consumption in the U.S.


Briteshine is the best-selling brand of flashlight.


Students who regularly eat breakfast have a higher GPA than those who do not.


State A had more than double the number of COVID-19 cases as compared to State B.



  1. Taylor, Shelley E., Peplau, Letitia Anne, Sears, David O. (1997) Social Psychology 9th ed. Upper Saddle River, NJ: Pearson/Prentice Hall, Chapter 3.


Icon for the Creative Commons Attribution-NonCommercial 4.0 International License

Arguments in Context by Thaddeus Robinson is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, except where otherwise noted.

Share This Book