In the comments to a recent post a sceptic tried, in the usual way, to convince us to abandon our delusions. He cited various references, including a fMRI study published in the Journal of Cognitive Neuroscience in 2008, which claims to have demonstrated that psi does not exist.
Again, as is often the case, the study says more about the urgent disbelief of debunking psychologists than it says about psi. It's shortcomings were widely noted at the time, but since sceptics take it so completely at face value I thought I'd give it another look.
The experiment was carried out by Samuel Moulton, a Harvard psychology graduate, and his mentor Stephen Kosslyn, who is well-regarded for his brain studies on mental imagery and visual perception. The authors consider that the evidence for psi so far is highly dubious. Anecdotal experiences can be attributed to cognitive illusions, while the outcome of ESP experiments is ambiguous. By contrast, they think that brain scanning technology is sufficiently reliable to settle the matter. If psi exists then, ergo, it ought to show up in fMRI scans.
Accordingly they recruited 19 bonded pairs of people - twins, couples, friends - which they whittled down to 16, one acting as sender and the other placed in the scanner as receiver. The receiver was shown two pictures, one being 'sent' by the sender and the other a control, and asked to guess which by pressing a button. The results came out at 50%, as expected if chance alone was operating, nor did anything unusual show up on the scans.
Analyses of group data revealed no evidence whatsoever of psi. Psi and non-psi stimuli evoked widespread but indistinguishable neuronal responses . . . The results support the null hypothesis that psi does not exist.
The study shows an impressive attention to detail, and in its technical aspects looks extremely solid, eg:
Functional series were analyzed using FMRIB's Improved Linear Model (FILM; Woolrich, Ripley, Brady, & Smith, 2001), which removes nonparametrically estimated temporal autocorrelation in each voxel's time series before applying the general linear model (GLM). For each series we modeled neural responses to the psi stimulus, non-psi stimulus, and feedback stimulus, as well as their temporal derivatives by convolving the basic waveforms (based on onset time and duration) of these variables with a double-gamma canonical hemodynamic response function (Glover, 1999). The same temporal filtering that was applied to the data was also applied to the model. For every functional volume, the following linear contrasts were employed to create statistical parametric maps (SPMs): non-psi > psi, and psi > non-psi.
So these guys know their stuff. Or do they?
Moulton and Kosslyn talk as if they're pioneering a new method which will settle the psi controversy once and for all. But they make no mention at all of any of the successful previous psi research using scanners, which even by this time was quite significant. Critics gave them such a hard time about this, they eventually went to Dean Radin to ask him what they'd missed (see a description here) - something they might have usefully done at the outset.
Then there's their method of producing ESP, which I did not at all recognise. If it wasn't tried and tested, it's not surprising that it was unsuccessful here. The more so, considering that the receiver is in a noisy machine and being asked to perform tasks in quick succession - hardly conducive to the relaxed state of semi-disassociation that is known to be ideal for generating psi.
And interestingly, despite their insistence that their methodology beats 'behavioural' research, it still centred on a cognitive process, of conscious knowing. At least one of the earlier fMRI studies bypassed this by looking to see if a light flash stimulus on the sender provoked the same response in the receiver's brain as in the sender's - it did.
In the earlier studies, moreover, efforts had been made to find participants that really could have produced psi. These authors knew enough to understand that bonded pairs are more likely to produce psi interactions than those that are not. But they seemed to think that alone would guarantee it. Earlier successful fMRI experiments selected people who had demonstrated psi abilities, for instance having worked as healers or taken part in a successful previous study.
There are other possible explanations for the negative results. Perhaps the ESP activity was there, but masked by the activity relating to visual perception. Or the signal might have simply been too weak to pick out from the noise. It's not clear why the experimenters were impressed by the absence of activity in the brain relating to psi if there was no psi in the first place - unless they saw the sender's intention as a material entity that was bound to register visibly among the receiver's neurons.
What really catches the eye is that one of the receivers actually did show potential evidence of psi - in the form of a relative absence of brain activity that correlated positively with the psi stimulus. The authors concede this would count as a positive result, but devote the best part of two pages trying to explain it away as an 'uninteresting artefact'. Having discussed and rejected two hypotheses, they eventually decide that the anomaly reflects the participant's 'idiosyncratic reactions to perceptual, conceptual, or affective differences between the psi and non-psi stimuli'.
Their reasoning is highly technical and difficult to appraise. But it begs a number of questions. If a single positive result can be neutralised in this way, could not similar arguments be applied in the event that other participants showed positive effects? Is there any guarantee that any of the three people whose data was rejected did not also show positive results? And what is the point of looking for evidence in the first place if it can be so easily dismissed? Surely this weakens the authors' insistence on the power of fMRI scanning to settle matters once and for all.
This is one of the more curious features of debunking studies. The sceptics embark on a process in the expectation of getting null results, are embarrassed to get positive results after all, but save the day with post hoc reasoning such as, 'yes but it wouldn't have happened if we'd done the randomisation differently'. (Radin references a wonderful study published in the The Humanistic Psychologist in 2006 by two psychologists, who carried out eight ganzfeld experiments and were shocked to get the usual 32% significance rate. To deal with this they redesigned the experiment to their own specification, and got a null result, which they then used to explain away all the rest.)
Someone who is not an out-and-out sceptic might spot the authors' hostility towards the idea of psi and wonder what effect this might have on their reasoning. To those of us who know about psi research, it stands out a mile. Their preamble makes it abundantly clear. They go to the trouble of quoting in full a rather good example of a crisis ESP episode - a woman waking one night experiencing powerful symptoms of choking and blood falling down her head, and learning two days later that her son was shot in the head at this time - but wave such testimony away on the grounds that it's probably all caused by confirmation bias, clustering illusions, etc. Even if ESP was proved, they say, it would just be an unexplained anomaly. They admit that claiming a single null result to be proof of non-existence is problematic, but consider that their results are so clear-cut it's fair to make an exception.
As an exploratory experiment it has value, and arguably forms an interesting addition to the parapsychological database. But their ambition is manifestly not to add to the database but to discredit it, backed by the big guns of contemporary neuroscience. The real story here is one of individuals setting out to rid themselves of their uncomfortable cognitive dissonance. Hence the triumphant press release that one can guess generated headlines around the world: 'ESP does not exist, Harvard scientists prove.'
The result is a curious illusion - an apparently careful, well-designed and thoroughly scientific study which is in fact only superficially so, and whose real effect is to reinforce materialist prejudices. Science as propaganda, one might say.