FTA and Observation Selection Effects

Let’s try to get as clear as we can on what is happening in (a) Observation Selection Effect (OSE) cases, and (b) whether the Fine-Tuning Argument (FTA) is guilty of OSE. Note that figuring out just how to explain OSE is quite tricky.

We should all be familiar with the fishing case. Suppose you and I go fishing and catch a few 15 inchers and no small fish. On the basis of our evidence (the size of the fish we caught) we seem justified in concluding the following:

1. P(E/L) > P(E/S) this says that probability of E given (on the assumption of) L is greater than the probability of E given (on the assumption of) S. 

            Where: L = the lake has mostly large fish

S = the lake has mostly small fish

E = our evidence (the large fish we caught)

But, my wife, contrarian that she is, points out that the net we used is only capable of catching fish over 12 inches. Our failure to include that in our background knowledge (the stuff that we know prior to gathering our evidence). That seems like relevant information. Indeed, it seems like that fact should influence in some way the relationship between our evidence E and our hypotheses L and S. But how so? Well, it seems like that information influences the kind of evidence we are capable of gathering. So, we should include it to see what happens. 

2. P(E/L & N) > P(E/S & N)

            Where: N = our fishing net is only capable of catching fish 12” or longer

But 2 is false. Our evidence—the 15” long fish we caught—no longer supports our hypothesis L. Why? Well, we were not able to catch any small fish even if the lake had lots of them. So, E does not increase the likelihood of L at all. We would have caught large fish regardless of the size of most of the fish in the pond. So, we are only permitted to conclude:

3. P(E/L & N) = P(E/S & N)

And what is crucial to notice, I think, is that N is doing all of the explaining here. L and S are irrelevant to E. Indeed, given N, E is pretty much guaranteed. In other words:

4. P(E/L & N) = 1 and P(E/S & N) = 1

Hence, our evidence does not support our hypothesis and this is because the method we used to gather the evidence biased the kind of evidence we would gather; our method screened off or excluded any other relevant evidence. 

What have we learned? We learned from the Fish Case that if our process of gathering evidence is such that the process itself implies that we will gather only one type of evidence, then we cannot use the evidence gathered to justify certain explanations. Our process of catching fish was such that the process itself implied that we would gather only large fish. Hence, we cannot use the caught large fish to justify our hypothesis/explanation that the lake contains mostly large fish. 

Here’s the argument:

A. If our process of gathering evidence implies that we will gather only one type of evidence, then we cannot use the evidence gathered to justify certain explanations.

B. Our process of catching fish implied that we would gather only large fish. 

C. Hence, we cannot use the caught large fish to justify our hypothesis/explanation that the lake contains mostly large fish.

Consider now the Firing Squad Case. Mary has 10 machine guns aimed at her head. At the same moment all fire and miss. Mary is unscathed. Mary concludes that the shooters intentionally missed her. Mary, it seems, is justified in believing:

5. P(E/I) > P(E/A)

            Where: I = intentionally missed

                        A = accidentally missed

                        E = observes herself being alive after the shooting

Ah, but it seems that we are ignoring a relevant piece of information here as well, namely, that Mary could not have gathered E without being alive. Mary’s process of gathering her evidence may be defective since it guarantees the result. So, we need to know if the following is true:

6. P(E/I & L) > P(E/A & L)

            Where: L = I am living

Notice that just as in the Fish Case, 6 above seems to show that I and A are irrelevant. Why? Because L alone implies E. So, parallels 2 and since 2 is false, 6 is as well. Hence, if the Firing Squad Case is also an instance of OSE, then Mary is not justified in concluding 5 or 6. She is only justified in saying:

7. P(E/I & L) = P(E/A & L)

And this is because: 

8. P(E/I & L) = and P(E/A & L) = 1. 

In other words, we add the fact that Mary is alive to our background knowledge (the stuff we know prior to gathering the evidence) our evidence is guaranteed and cannot be used to support our hypothesis. 

YIKES! That is a crazy result for the Firing Squad Case. Of course, Mary is justified in believing that the shooters intentionally missed, and not justified in believing that the shooters accidentally missed. What went wrong? Well, we should not have included L in our givens. Mary’s being alive after the shooting surely is evidence of something. But by including that fact in our background knowledge we made it evidentially silent. How do we fix this? 

In class, I said to move L over to the E side of slash. When we do that we get the right result. So, we have:

9. P(E & L/I) > P(E & L/A)

Is this cheating? Not at all. It is intuitively obvious that Mary is justified in believing that her shooters intentionally missed. When we have L as part of our background knowledge we failed to respect that intuition and wound up with something obviously false. It is false that Mary’s being alive after the shooting is not any evidence that the shooters intentionally missed. 9 saves us from saying something clearly false. 

Crucial Question:  Is the FTA more like the Fish Case or the Firing Squad Case? 

A lot of people have thought that it was more like the Fish Case (and not for the reasons Plantinga gives in the book). The worry is that in the Fine-Tuning case, my being alive guarantees that the universe is Fine-Tuned for life. So, my being alive entail Fine-Tuning. So, the D hypothesis is not doing any real work. Just as having a net with 12-inch holes guarantees catching only large fish, my being alive guarantees Fine-Tuning. Hence, FTA parallels the fishing case more strongly than the firing squad case. Hence, FTA is fallacious. 

Is the above argument a good one? Let’s see. According to FTA we are justified in believing:

10: P(FT/D) > P(FT/C)

            Where: FT = the universe is fine-tuned for life

                        D = the universe is intentionally designed for life

                        C = the universe produced life by chance alone

In order for the OSE objection to FTA to be persuasive we need to be able to show that 10 somehow commits the same type of mistake that we found in the Fish Case. So, we need to show that instead of 10 we should have written something like:

11. P(FT/D & L) > P(FT/C & L)

            Where: L = I am alive

But, 11 looks false. It is no longer the case that FT supports D over C because once we include L in our background knowledge (the stuff we know prior to our gathering the evidence FT), FT is guaranteed. Why? Because, we already know that the universe must be life-sustaining prior to our investigation of it. So, L needs to be one of our givens. L entails FT. So,

12. P(FT/D & L) = P(FT/C & L)

And 12 is true because 

13. P(FT/D & L) = 1 and P(FT/C & L) = 1

Both of the above equal 1, according to this objection, because L entails FT. That is, our being alive guarantees that we live in a universe that is fine-tuned for life. 

What should we make of all of this? In particular, what should we make of 12?

It is true that our being alive is a background condition. In order for us to gather the evidence of FT we have to be alive to do so. But I think there are a few good responses to this. 

Responses

The first is that, like the Firing Squad Case we should move L to the other side of the slash. So, now we have:

14. P(FT & L/D) > P(FT & L/C)

One reason to think that this move is justified is that what we are trying to explain is the fact that there is life and that life requires FT. Our process of gathering the evidence for FT does require that we are alive, but there is no reason to suppose that our being alive somehow biased the evidence gathering process (remember that we assuming FT; we are not doubting FT while considering the OSE objection to the FTA). So, the L really should be on the left side of the slash. We can think of it like this: while it is true that oxygen must exist for us to gather evidence for the origin and conditions necessary for the production of oxygen, we should not put the existence of oxygen in our background knowledge when attempting the explain the existence of oxygen. If we did that, then it would be impossible to explain the existence of just about everything (the origin of the universe, carbon, plants, etc) and then science itself is useless. 

Here’s a different take on all of this. We can’t forget that we discovered FT. Prior to our discovery of FT we knew that the universe was life-conducive. So, we knew that the conditions sufficient for life were met in our universe. But we did not yet know what those conditions were. After hundreds and hundreds of years, we have discovered that the conditions needed for life in our universe are extremely precise. That was shocking. So, we have to represent that in our formulations. Since, 12 clearly does not represent just how surprising those discoveries we, 14 is the better way to go. Hence, the FTA seems to avoid the OSE objection.