Lots of fantasy and science-fiction writers got ideas about time travel from a book called An Experiment with Time by J.W. Dunne. This book is Dunne’s effort to establish that dreams are prophetic because they exist outside of our normal perception of time. In keeping with the tradition of metaphysical treatises written by engineers, it is hogwash.

The design of his “experiment” begins with collecting data. That is a commendable place to start, if you do it right. But he collects only data that support the hypothesis, ignoring everything else. After that, it is unsurprising that he judges his hypothesis has been supported.

His method works like this: Dunne recommends to his readers a method for writing down dreams in a log, which is good. But then he recommends going back at the end of the day, or a few days later, re-reading the dream log, and thinking of events since the dream that might have a relationship to it. This is wrong – it sets up the researcher for confirmation bias. Better would be to keep two notebooks. In the other one, write down all the extraordinary events of the day each evening, and then match them up with dreams some time later, counting how many are unmatched and how many seem to have been foretold. A method like this permits negative information – how many events didn’t come with dreams and how many dreams didn’t relate to events – to be considered alongside positive information. It would also allow study of the time-lag between dreams and events, which Dunne sweeps under the rug. (If prophetic dreams existed, this last bit would be the most important part.)

As presented, this method of data collection is exactly the same used to prove Pennsylvanians are the worst drivers. After we develop that hypothesis, we drive around on our daily business and whenever some driver makes an obnoxious move we note their license plate. After we see a bunch of Pennsylvania plates, we announce that the hypothesis is confirmed. The method is logically wrong and you can’t base conclusions on it.

This is bad enough, but it gets worse. As I’ve noted before, I suspect the reason science progresses rapidly, while the humanities move at a snail’s pace, is that science can be proven wrong. Once something is proven wrong, future scientists can ignore it. Humanities scholars don’t have that option; they have to read everything, whereas scientists don’t have to familiarize themselves with a lot of material that won’t help their work. In this case, I have reverted to the scientific tradition. I ignored the second half of the book, where Dunne creates a (cumbersome) theory to explain his (bogus) observations.

And “cumbersome” is an understatement. Kristine Larsen wrote a précis of the theory1:

Dunne’s new concept of time, called serialism, is based on the idea that what we call the present is a window through which we view the entirety of reality, like a train rider looking at a landscape through a small window. Like the train, our present moves along a track, in the dimension we call time T1. Dunne argued that like the train track, T1 is actually a spatial dimension, and that there must exist a higher dimension, a more ultimate time T2, in which we measure the rate of motion of our journey along time’s track. There exists a higher-dimensional observer in this T2 who has the freedom to observe all events in T1 (all events in our perceived time) in whatever order he or she wishes. He argued that there exists an infinite regression of these higher times and observers, leading to some Ultimate Time and an Ultimate Observer, which Dunne calls the “Super-Mind”.

Creating an infinite number of time dimensions to go along with the three spatial ones we can see might be the worst violation of Occam’s Razor ever achieved. It is particularly bad because it breaks the way physics works. When a scientist builds a model of some phenomenon, the idea is to start with a simple expression that gets things almost right. The universe is complicated, though, so correction terms will have to be added to make the model precise enough to use. This is expected. But what’s most important is that the correction terms should be small compared to the base model. If they aren’t, the model is useless for computing things. Dunne’s model says all the extra dimensions are equally important, which means you can’t add up just a few of the different contributions to get an approximate answer.2 This infinite number of time dimensions doesn’t even add anything to the model. I suspect that the only reason Dunne included them is because an Ultimate Observer was an appealing notion. The first premise of Relativity, by the way, is that there is no preferred observer. Dunne makes a big deal about deriving his ideas from Einstein in his book, but seems to have missed that.

Digression

This reminds me of my old professor. I studied Special and General Relativity with Edward Desloge, S.J.. One day in class he said a fascinating aside:

St. Augustine almost discovered Relativity. He asked, ‘is there just one time in the cosmos, or can everyone have their own time?’ Which was exactly the question Einstein asked, except to Augustine the answer was obviously that God is a preferred observer, and the time perceived by God is the universal truth.

Once you establish a preferred observer, Galilean relativity and Newtonian mechanics eventually follow, so it’s hard to say that’s a bad thing. It could be a necessary step toward the truth. Still, I’d love to know what Augustine might have come up with, had he taken the other choice at that moment. I guess that will have to wait for a really good alternative-history writer.

And we’re back on topic

To Dunne’s credit, he does include criticism in an appendix to his Third Edition, ranging from a pie-fight with H.G. Wells to a diplomatically-phrased excerpt of a letter from Arthur Eddington. The excerpt is so short that I suspect it was the only vaguely-positive thing in the letter. Eddington, by the way, is as close to a nutcase as you can be while still being a brilliant scientist. But that excerpt does get the name “Minkowski” into the book. If Dunne had understood the minus sign in Minkowski’s metric, maybe all of this could have been avoided.

Conclusion

I’ve been pretty harsh, here. Fortunately, Verlyn Flieger can give us a more graceful conclusion. As she puts it, “It is important to remember here that the validity of Dunne’s theory … is not the point.” 3 The point is the inspiration Dunne provided to various authors around England in the 1930s and 1940s.

To oversimplify a bit, authors stripped down Dunne’s theory to the idea that people have two ways of perceiving time, of which the one closer to our bodies looks out through a window to see only what we call “the present”. That’s our everyday mind. The further mode of consciousness sees time differently. It can can range across the past and future. If a character has prophetic dreams, they come from communications between events perceived in that further “time field” (Dunne’s word, whatever that means) and those in our usual one. But that will be enough to go on with.

 

Notes

  1. Larsen, Kristine. “From Dunne to Desmond: disembodied time travel in Tolkien, Stapledon and Lost.” Mallorn: The Journal of the Tolkien Society 53 (2012): 26-30.
  2. As it turns out, nobody who uses his ideas keeps all those extra dimensions, so the harm done is minimal.
  3. Flieger, Verlyn. A Question of Time: JRR Tolkien’s Road to Faërie. Kent State University Press, 1997. p.46