Author Michael Prescott has recently provided some critical comments on my recent blog post on Chris Carter’s defense of empirical arguments for postmortem survival. Prescott’s widely read blog often addresses the topic of life after death. However, unlike many other bloggers, Prescott brings some quality insights to the topic, so naturally I’m happy to see a discussion of my arguments on his blog, and I’m happy to respond to his comments.
In his most recent blog, Prescott focused on what I call the “problem of auxiliary hypotheses.” On my view, empirical arguments for survival depend on auxiliary hypotheses that are not independently testable. In this way survival arguments are very much unlike empirical arguments we encounter in other domains of inquiry (e.g., detective work, jury deliberations, and the sciences), where the predictive consequences of hypotheses are derived with the assistance of added assumptions that can be independently tested and for which there is independent evidence. As a result of reliance on auxiliary hypotheses that lack independent support, survival arguments carry significantly less force than their proponents claim.
Below is my response to Prescott’s criticism of my argument concerning auxiliary hypotheses. The specific context here is my application of the problem of auxiliary hypotheses to one of Chris Carter’s defenses of survival arguments against appeals to living-agent psychic functioning as a viable counter-explanation of the relevant empirical data. As explained in “Chris Carter’s Challenge: Survival vs. Super-Psi,” Carter rejects living-agent psi explanations of the data because they can only account for the data by being expanded into a fairly robust version of living-agent psi called super-psi, but – so Carter contends – super-psi lacks independent support.
My response to Carter is a simple parity argument. Survival can only account for the relevant data by being expanded into a fairly robust version of survival, one that, like super-psi, involves a large number of auxiliary hypotheses for which there is no independent support. Hence, if “lack of independent support” is a reason to reject non-survival explanations of the data, it is equally a reason to reject survival explanations of the data. As I see it, Carter and other survivalists who reject non-survival counter-explanations of the data on the grounds that these alternative explanations lack independent support are ignoring the extent to which the survival hypothesis fails in precisely the same way. The integrity of survival arguments is undermined by reliance on what amounts to an epistemic double standard.
Prescott attempts to rescue Carter from my critique by arguing that, while survivalist auxiliary hypotheses are not independently testable, they do fit with our background knowledge. While the strategy of generating a salient difference between the auxiliary hypotheses of competing explanations is in principle sound, Prescott’s particular argument does not work. The main problem in Prescott’s argument is that there are many different auxiliary hypotheses that (i) are consistent with the survival hypothesis, (ii) fit with our background knowledge, but (iii) generate very different predictive consequences, many of which would disconfirm the survival hypothesis. Otherwise stated, there are many different survival hypotheses. Only a very narrow range of these survival hypotheses would lead us to expect the data adduced as evidence for survival. We simply don’t know how the world should look if survival is true, which is why empirical survivalists can’t tell us how the world should look if survival is false.
The upshot: in the absence of independent testability/support, we have no way of selecting auxiliaries in a way that does not appear to be a case of explanatory retrofitting. Particular facts are judged salient and consequently selected because they fit the auxiliaries one’s favored hypothesis needs to generate successful predictions. This is not a truth-conducive policy in explanatory reasoning in any other domain in which we aim to weigh empirical evidence. The burden is on the survivalist to show that the survival hypothesis is an exception to this rule.
– M.S.
=========================================================================
Response to Michael Prescott
Michael:
Thanks for your continued discussion of my critique of Carter. I’ve provided some comments here on your blog “More on Super-Psi.”
You have my permission to post this response on your blog. I’ll probably post it on my own blog this coming week.
Before digging into your main argument, let me offer an initial clarification. In your blog, you wrote:
“I don’t think that a rigorously logical proof, along the lines of proving a mathematical theorem, is possible when dealing with empirical evidence, especially when the evidence involves something as in inherently ambiguous and subjective as states of consciousness (incarnate or discarnate). Instead, I think what is needed is something more like the reasoning we hope to find in a jury’s deliberations.”
I agree that a “rigorously logical proof, along the lines of proving a mathematical theorem,” is not possible when dealing with empirical evidence. Of course I’m not asking survivalists to produce such an argument. This is not the problem with empirical survival arguments. My criticisms are directed towards what survivalists claim on behalf of their arguments. Survivalists make claims (of varying sorts) about the force of evidence for the hypothesis of survival, often claiming that the evidence confers some favorable probability on this hypothesis. I’m subjecting these claims and their supporting arguments to critical evaluation, and I’m relying on principles that are broadly applicable to evidence assessment across different domains of inquiry. I really don’t see how we can do justice to survival arguments and avoid technical issues in confirmation theory and general epistemology.
With reference to my critique, you’ve mainly focused on one of my several criticisms of Carter’s arguments. Carter objects to the super-psi hypothesis on the grounds that it lacks independent support. I had argued that the empirical survivalist is in exactly the same position. The only kind of survival hypothesis that generates anything in the way of even general predictive consequences depends on a range of auxiliary hypotheses for which there is also no independent support. My argument is a straightforward parity argument: Carter demands “x” of super-psi arguments, but survival arguments don’t satisfy “x.” It’s also an application of one of my more general criticisms of survival arguments, namely that the lack of independent testability/support for auxiliary hypotheses significantly deflates the force of empirical survival arguments in their classical formulations.
Now it appears that we agree on at least two issues. You agree that empirical survival arguments depend on auxiliary hypotheses. You also appear to agree that these auxiliary hypotheses are not independently testable and lack independent support. The point of disagreement concerns whether this fact undermines Carter’s particular criticism of the super-psi hypothesis. You seem to think not. And here you make the observation that survivalist auxiliaries are consistent with our background knowledge. It’s not entirely clear how this observation, which is surely correct, deflates the force of my criticism of Carter, but I suspect you intend something of the following sort: while it may be true that survivalist and living-agent psi explanations of the relevant data each depend on auxiliary hypotheses that are not independently testable, the survivalist assumptions at least fit with our background knowledge, whereas living-agent psi auxiliaries do not, or at least the former fit better with our background knowledge than the latter. So it seems that your answer to my parity argument is a “disparity” counter-argument.
Well, this is an interesting approach. In principle it’s the right kind of move to make. To deflate a parity argument you’d need to show an overriding salient disparity, a significant difference between the survival hypothesis and its explanatory competitors that favors the survival hypothesis, even if my parity thesis is true. However, I don’t think you’re going to get the necessary mileage out of this particular argument. In fact, I’m inclined to think that it actually highlights precisely what I think is wrong with survival arguments. So it’s worth looking at this.
First, there’s something of a challenge here in determining the precise role of fit with background knowledge in the larger framework of the survival arguments we’re critically engaging, and this includes Carter’s defense of survival arguments. “Fit with background knowledge” is plausibly a virtue of some sort for hypotheses. But what sort? And what sort of weight do we give it in the larger context of other criteria we’re invoking in evidence assessment. The same is true with respect to “simplicity,” which participants on both sides of the debate tend to wield in an incautious manner. Until this is explored, it’s hard to see the net impact of your observation on either survival arguments or my critique of Carter’s objections to super-psi.
Now one way “fit with background knowledge” often enters the structure of empirical arguments is as a determinant of the prior probability of a hypothesis. By “prior probability” I mean the credibility of a hypothesis independent of the evidence it’s adduced to explain. So we might treat your observation as offering “credit” to the prior probability of the survival hypothesis. However, assigning a prior probability to the survival hypothesis is a notoriously difficult matter. I’m highly skeptical that such assignments do anything more than express the arguer’s subjective degree of certainty in the hypothesis of survival. And this is one of the several problems that infect (Bayesian) formulations of survival arguments that incorporate claims about prior probabilities, which is why Likelihood formulations are better suited to survival arguments, even if the conclusion must be a bit more modest. (See my blog series “Getting Sober about Survival” for a discussion on Likelihoodism).
Second, with respect to the alleged advantage of survival-friendly auxiliaries over living-agent psi-friendly auxiliaries, it’s important to note that there are lots of different auxiliary hypotheses that may be used to generate robust living-agent psi hypotheses that account for the data in a way consistent with our background knowledge. Here I’ll refer only to Braude’s well-developed living-agent psi hypothesis. There simply isn’t a single super-psi hypothesis. Both Hodgson and Carter adopt a fairly narrow set of auxiliary assumptions about super-psi (as well as auxiliaries about the relevant psychodynamics that might play a role in its operation), and from this position they try to show that super-psi generates predictive consequences that are contrary to our observational data. However, in the absence of independent support for these auxiliaries, we really don’t know whether the alleged observational data count against the super-psi hypothesis or count against the auxiliaries Hodgson and Carter have adopted. I’ve discussed this problem in Sudduth 2014b.
Third, and this is really the crucial issue, while there’s no doubt that the survival-friendly auxiliaries you cite fit with our background knowledge in precisely the ways you’ve suggested, there are dozens of other auxiliary hypotheses that are (i) consistent with the survival hypothesis, (ii) fit with our background knowledge, but (iii) generate very different predictive consequences of varying degrees of specificity. Broad, Price, and Ducasse each outlined a range of different robust survival hypotheses, each of which has analogues with our current experience (e.g., dream consciousness, dementia, dissociative identity disorder, psychogenic amnesia). But these alternate robust survival hypotheses do not lead us to expect the data adduced as evidence for survival. For example, they would not lead us to expect the persistence of ante-mortem autobiographical memories, intentions and purposes, skills, or the personality traits/profiles of relatively unified selves.
The root of the problem is apparent if we look more carefully at how you make use of fit with background knowledge. What you’ve done is show how the auxiliaries needed for predictive success in survival arguments fit with a selected subset of our total relevant background knowledge. But note – you could have selected very different auxiliary hypotheses about the nature of postmortem consciousness, and these would have been equally compatible with our background knowledge. Had you selected a different subset, the survival hypothesis would not have had predictive success because, relative to these alternate hypotheses, we would not expect survivors to have strong psychological continuity with their ante-mortem lives. Why privilege your subset then? That’s the crucial question. Since you acknowledge you have no independent support for your auxiliary hypotheses, your selection procedure is open to the charge of being more accurately a retrofitting procedure. Particular facts are judged salient and consequently selected because they fit the auxiliaries one’s favored hypothesis needs to generate successful predictions.
If defenders of living-agent psi hypotheses appear to adopt ad hoc auxiliaries, survivalists are guilty of adopting auxiliary hypotheses that beg the question. In the absence of independent support for auxiliary hypotheses, we have no way to sensibly navigate the vast array of options in the logical space occupied by auxiliaries that are consistent with our background knowledge but that produce very different predictive consequences. This is why survivalists have a hard time stating what observation would be evidence against survival. It’s simply not clear what the world should look like if survival is true, which is rather unfortunate for empirical arguments for survival. Similarly, and contrary to Hodgson and Carter, it’s also not clear what the world should look like if “super-psi” is true. It follows that we don’t know whether the relevant data are more to be expected given survival or super-psi. What then of arguments that purport to show that the evidence favors survival over living-agent psi? Perhaps this explains why the jury is still out on this one.
Michael
Follow