2004 EEG study by Kittenis et al.

I thought it might be interesting to discuss this paper from 2004, in which Kittenis, Caryl and Stevens exposed a "sender" to flashes of light while recording the EEG of a "receiver" in a different room:
https://koestlerunit.files.wordpress.com...l-2004.pdf

I saw this figure illustrating the average spatial distributions of their EEG measurements in the SPR's "Paranormal Review" last year, and was quite amazed by it. The top sequence is an average over 39 subjects viewing the flashes directly (as the "senders" did in the study) and the bottom sequence is an average over 18 receivers. Time runs from left to right, and the bar in the middle starting at 0ms marks the period of the flashes.

[Image: KittenisColourFig.jpg]

My initial reaction was that it was a very clear demonstration of telepathy, and I couldn't imagine how a sceptic could refuse to accept it. (Robert Charman, the author of the article in "Paranormal Review", was more interested in the appearance of precognition in the bottom sequence.)

Now, having thought about it more, I find it nowhere near as convincing. I was going to explain why, but I think maybe I'll just invite comments for now, and see what - if anything - people come up with.

I thought it might be interesting to discuss this paper from 2004, in which Kittenis, Caryl and Stevens exposed a "sender" to flashes of light while recording the EEG of a "receiver" in a different room:
https://koestlerunit.files.wordpress.com...l-2004.pdf

I saw this figure illustrating the average spatial distributions of their EEG measurements in the SPR's "Paranormal Review" last year, and was quite amazed by it. The top sequence is an average over 39 subjects viewing the flashes directly (as the "senders" did in the study) and the bottom sequence is an average over 18 receivers. Time runs from left to right, and the bar in the middle starting at 0ms marks the period of the flashes.

[Image: KittenisColourFig.jpg]

My initial reaction was that it was a very clear demonstration of telepathy, and I couldn't imagine how a sceptic could refuse to accept it. (Robert Charman, the author of the article in "Paranormal Review", was more interested in the appearance of precognition in the bottom sequence.)

Now, having thought about it more, I find it nowhere near as convincing. I was going to explain why, but I think maybe I'll just invite comments for now, and see what - if anything - people come up with.

I'd be interested to know if your misgivings are of a procedural or statistical/mathematical nature?

I'd be interested to know if your misgivings are of a procedural or statistical/mathematical nature?

They relate more to the analysis, and what comes out of it, than to the experimental procedure. 

As no one has bitten, I'll try to explain what I think of it, but I may not have time for a day or two.

Actually, my main misgiving was based on the fact that - although those coloured contour plots look striking, and although the results of the statistical comparisons between the EEG signals were significant - if the spatially averaged signals for the senders* and receivers are graphed as a function of time, the resemblance between them isn't particularly close. As the authors point out, the signal for receivers starts to rise at least 150ms before the stimulus (and actually it rises more gradually from at least 500ms before the stimulus). 

[* In fact, it seems they recorded only one EEG at a time, not simultaneous EEGs for senders and receivers. So strictly the "senders" signal is just for subjects observing the stimulus directly, without a receiver.]

[Image: KittenisDirectFig.jpg]

[Image: KittenisIndirectFig.jpg]

The early rise could be interpreted as evidence for precognition (which is how Robert Charman interpreted it), but if that were the case, the authors ask, why wouldn't there be a similar precognitive effect in the EEG signals of the senders, who are shown the flashes directly?

Unfortunately the base-lines of these averaged signals aren't particularly flat, and the size of the peak in receivers' signal isn't that much larger than the waviness of the base-lines. Though in theory the statistics should take care of that.

I think the choice of colours in the contour plots tends to make the correspondence between the signals appear closer than it seems from the graphs. To my eyes, the greyscale version of the contour plots doesn't look anywhere near as striking.

[Image: KittenisBWFig.jpg]

Another concern would be that there's no indication that the statistical test was fixed before the EEG signals for the receivers were seen. In fact it probably wasn't, because it was based on the measured form of the senders' signal. If so, there's a danger that the choice of comparison statistic was also influenced by what the receivers' signal looked like.

On a more positive note - 

The statistical test seems reasonable enough. It's essentially just comparing the average of the signal over the half-second after the stimulus, and the average of the signal over the half-second before the stimulus (taking logarithms first). That comparison, based on all sender-receiver pairs, gave a significant difference with a p value of 0.007. Considering the experiments that generated those graphs (including control sessions for the receivers) took only about 3.5 hours in the lab, that seems pretty efficient if it's a genuine effect.

As well as the rather crude before/after statistical comparison, it's notable that the peaks of the sender and receiver signals are only 50ms apart. The authors also comment that the spatial distribution of the signal in the receivers is consistent with what would be expected for a visually evoked response. On that basis, again if it's a genuine effect, it's plausible that a more efficient comparison statistic could be devised.

I'm not very clear about what the authors did next after this study. I can't find any more published papers by them along the same lines, but there's an abstract of Kittenis's thesis (degree awarded 2008) here:
http://ethos.bl.uk/OrderDetails.do?did=1...hos.653483
It refers to two studies which showed significant effects, but then mentions a third study in which "a variation of the experimental paradigm was adopted in order to increase the overall sample size", and says no effect was found in that study. But what the variation of the experimental paradigm was, I don't know.

Aye, gradient maps of head showing... err they don't know. Senders and receivers isolated... err no... both receiving the same auditory stream. Fig 2. If you put 'unrelated' and 'alone' together (back of fag packet stuff)... oh it's pretty much the same result as 'related'... err, we didn't find anything. 'Alone' nobody present in sender room... err... comparing apples and pears. Anyway, I thought it was so bad I couldn't motivate myself to respond in any detail.

Well, thanks for posting at all - you're the first to have made any comment on the paper.

I should be interested if you could expand one of your comments - the one about senders and receivers not being isolated, and both receiving the same auditory stream. I couldn't work out what that was referring to.

Well, thanks for posting at all - you're the first to have made any comment on the paper.

I should be interested if you could expand one of your comments - the one about senders and receivers not being isolated, and both receiving the same auditory stream. I couldn't work out what that was referring to.

Ah, sorry, you're referring to the soundtrack of drumming? I agree it doesn't seem a particularly good idea, but given the fact that the EEG signals from the senders weren't actually used in the analysis (except to inform the choice of comparison statistic), I don't think there's any way this could provide a spurious difference between stimulus and control signals from the receivers.

Thinking more about the drumming - presenting the same soundtrack to both sender and receiver simultaneously - it's the kind of thing that instinctively "feels" wrong, and which would obviously invalidate some kinds of analysis based on directly comparing the sender's EEG to the receiver's EEG. But when only the receiver's EEG is analysed in relation to the (random) time of the stimulus, I don't think there can be any objection to it. 

However, because it "feels" wrong, perhaps it hasn't been tried in other studies. One potentially good thing about it is that it would provide a kind of answer to the question "Why should the receiver pick up the sender's thought waves rather than those of anyone else in the vicinity?" (And even "why should vicinity have anything to do with it?") It would be interesting to know whether this was one of the features altered in the third, unsuccessful trial that's referred to in the thesis abstract.

In some other ways, this doesn't seem the best-designed experiment in the world. The numbers in two of the groups are clearly too small, and suggest that for some reason the planned experiment wasn't completed (another red flag, now that people are more aware of the importance of pre-planning/registration). 

Also, the gap between successive stimuli is random, but the smallest possible gap of 1 second is probably a bit small based on the graph of the sender's EEG above. The coloured contour plots do suggest that they may still be picking up some activity from the previous stimulus shortly before the current one. 

But again, because they are analysing only the receivers' EEGs, that's not a fatal flaw. In fact, one virtue of this kind of sender-receiver setup - with only the sender directly perceiving the stimulus, and only the receiver's response being analysed - is that it's immune to the kind of history-dependent averaging artefacts that can complicate presentiment experiments. There was quite a long discussion of these problems on Skeptiko, in the thread entitled "Presentiment Paper Discussion".

Someone else did a similar study... I want to say, Persinger?

Someone else did a similar study... I want to say, Persinger?

I have noted various references to similar work, though I haven't had a chance to follow them up (I can't see Persinger in my notes). Some are discussed in the "Unconscious Psi" chapter of Dean Radin's book "Entangled Minds". One of them, by Wackermann et al. (2003), is available online here:
https://www.researchgate.net/publication...n_subjects