Last time, I thought a bit about mediumship, and how the observational evidence doesn’t lead us to a supernatural spirit-world medium hypothesis rather than a “mundane” supernatural psychic mind reading that does not involve undead spirits communicating with the medium.

This time, I’d like to think a little about how Sudduth shows the natural explanation is preferable to the supernatural one. Sudduth frames the survivalist position in this way:

• Parapsychologists and survivalists have presented arguments to show that, with respect to mediumship, the fraud and chance hypotheses are improbable. Unfortunately, the form of argument on which they have relied to show this is bogus … One salient point raised by Broad, Royall, and Sober is that hypotheses must be tested against an alternative … In the case of mediumship, the survivalist needs to show that the survival hypothesis confers a greater probability on the observational evidence than does the fraud hypothesis. The observations will then favor survival over fraud.

Noting the difficulty here for survivalists, Sudduth points out:

• If one does not find the fallacious nature of the inference in question intuitively obvious, it is very easy to find examples of hypotheses that confer hugely low probabilities on an observation without the hypotheses themselves plausibly being regarded as having (hugely) low probabilities. Twenty-six consecutive black numbers came up on the roulette wheel at Monte Carlo in 1913, with odds of about 1 in 137 million (Hand, 2014, p. 83). This outcome was hugely improbable given that the roulette game was fair, but it is clearly implausible to infer that a fair roulette game was improbable merely because that hypothesis confers a hugely low probability on the outcome. Evelyn Marie Adams won the New Jersey lottery twice in four months in the 1980s, with odds of one in a trillion (Ibid, p. 86). This outcome was also hugely improbable given that it was a fair lottery. It is more impressive than Mrs. Warren Elliott’s mediumship, the results of which were one in a billion by comparison. So, if Saltmarsh’s reasoning was correct in the case of Elliott, a fortiori the chance hypothesis should be excluded in the case of Evelyn Marie Adams. But this is absurd. We rightly do not conclude that the hypothesis of a fair lottery was improbable, and so must be rejected. We also should not regard the observation as evidence against the hypothesis… When a hypothesis confers a low probability on an observation, it is not reasonable to conclude that the hypothesis is improbable, that the observation is evidence against the hypothesis, or that we should reject the hypothesis… As Fisher correctly pointed out, if a hypothesis H says an observation O is improbable and O occurs, then one of two things is true, either H is false or something very improbable has happened (Sober, 2008, p. 56). Survivalists have not offered any good reason to prefer the former possibility to the latter.

In fact, survivalists overestimate their hands because they don’t factor in recalcitrant evidence when touting the success of the medium’s reading:

• Augustine’s criticism is that Nahm and DRW ignore or fail to properly weigh salient counter considerations – (i) the significant number of ostensible spirits being fictitious constructions of the medium’s own mind, (ii) the mixture of accurate statements and twaddle, and (iii) Mrs. Piper’s access to gossip as an ordinary source of information. (i) and (ii) are relevant to net plausibility assessments. They more than offset any gain the “never caught cheating card” provides, which cannot be conclusive on account of (iii). These considerations require that Nahm and DRW downgrade their highly favorable assessment of the evidence from mediumship or explain why such data make no difference to their favorable assessment… Augustine’s suggestion is that the data on fictitious controls and twaddle count against or lower the probability of the survival hypothesis in a way not acknowledged or anticipated by Nahm and DRW… [Fictitious controls] may be unsurprising given [survival hypothesis] either because the survival hypothesis can be forced to fit any observational data – fictitious controls, fraud, alien abduction experiences – or because it has been bulked up enough to make predictions… By contrast, given the well-understood human motivations that underlie fraud (mediumistic and otherwise) and the varied phenomena of abnormal psychology – for example, dissociative phenomena and savant syndrome – neither the existence nor the nature of fictitious controls is surprising if the survival hypothesis is false. And, unlike the survival hypothesis, no extravagant assumptions are required. To that extent, there will be precious little in the way of observational evidence to support the survival hypothesis as well. Predictively impotent hypotheses may be shielded from empirical disconfirmation, but this comes with a steep cost: the loss of empirical confirmation. So, this move offers no help to the empirical survivalist

,