Header 1

Our future, our universe, and other weighty topics


Sunday, June 16, 2019

Study Hints Your Brain Isn't Making Your Dreams

There are many problems with neuroscience studies that greatly affect their reliability, and should cause us to believe that a large fraction of them are false alarms. One of the biggest problems is insufficient samples sizes, a problem so bad that it led one neuroscientist to conclude that most published neuroscience studies are false. Another huge problem is the low use of what are called blinding protocols.

Imagine a study involving two different groups of subjects, either humans or animals, who differ in some way. Perhaps one group received some stimulus, and the other did not; or perhaps one group reported some tendency or experience, and the other did not. When a blinding protocol is used, the scientists analyzing these subjects will not know which group the subjects belonged to. So, for example, if 10 test subjects are given a pill, and 10 control subjects are not given the pill, then when scientists are studying data from the subjects they will not know whether the subjects got the pill or did not get the pill.

A blinding protocol such as this can be important for reducing experimental bias. For example, if such a protocol were not used, and you were a scientist asked to compare 10 subjects who you knew were given a pill and 10 subjects who you knew were not given the pill, it would be all too easy for you to show some bias in your analysis caused by your knowledge of whether or not the subjects had been given the pill.

Proper blinding protocols are not often used in neuroscience studies. But recently we had an example of an interesting study that used such a protocol. The study (called the Dream Catcher study) was about whether or not scientists could detect a brain signature of dreaming. Nine subjects went to sleep in a laboratory, each with an EEG reader attached to his or her head. The brain waves of the subjects were recorded, and at random intervals subjects were woken up. The subjects were then asked to recall any dreams they were having when woken up. From such cases a Data Team accumulated 27 cases of dreamless sleep, and 27 cases of dreaming sleep, along with the corresponding EEG readings from the brain.

The EEG readings were then given to some other people in an Analysis Team, consisting of people who did not know whether any particular case they were analyzing was a case of dreaming sleep or a case of dreamless sleep. These “blinded” analysts were asked to predict from the EEG readings whether particular cases were examples of dreaming sleep or dreamless sleep.

The result was a null result. The predictions of the analysts (using the EEG data) were not better than what would be expected by chance. The experiment is consistent with the hypothesis that your brain is not actually the source of your dreams.



A previous study by Tononi and others claimed to find some neural correlate of dreaming. In a science news article, Tononi tried to suggest that the difference was due to “trouble” in the methodology of the “Dream Catcher” study finding no evidence of a neural correlate of dreaming. But such an insinuation does not seem fair. Although it involved only 9 subjects, the “Dream Catcher” study involved 54 different cases, and a sample size of 54 is generally regarded as adequate. The “Dream Catcher” study actually involved a protocol much better than that in the Tononi study, which failed to use blinding.

The Tononi study has one claim of predictive success, but only a very dubious one. In one experiment, 84 times sleeping people (connected to EEG brain wave readers) were awoken based on some criteria in their brain waves that might predict that they were dreaming. The paper tells us that the vast majority of these observational cases were thrown away, leaving only 36 cases that were used to judge predictive success; and in that 36 the prediction was pretty good. But this “picking 36 out of 84” smells like cherry-picking to get the desired predictive success, so it is very unimpressive. Among the reasons for discarding observational cases mentioned in the Tononi study (in the Methods section for Experiment 3) is when "sleep stage could not be confirmed," but that is a most dubious procedure, since the whole idea is to show whether we can tell whether people are dreaming from their brain waves; and all the dreamers had their brain waves continuously monitored.   If there was actually a brain wave signal showing dreaming, there should be no reason to throw out most of the observational cases on the basis that "sleep stage could not be confirmed," since the brain waves in such a case would let you know what the sleep stage was. 

What I would like to see is many more neuroscience experiments using proper blinding protocols. Here is an experiment that neuroscientists have not done (to the best of my knowledge), but should be doing:

  1. Do a brain scan on 20 subjects (called Group A). Tell the subjects to think of absolutely nothing during the brain scan other than the blackness of outer space.
  2. Do a brain scan on 20 other subjects (called Group B). Tell the subjects to do some mental task, such as creating a summary total of the first 20 integers. For example, 1+2=3, 1+2+3=6,1+2+3+4=10, 1+2+3+4+5=15, and so on and so forth until a total for the first 20 integers is reached.
  3. Shuffle the brain scans, and submit them to some other “blinded” scientists who do not know whether the subjects were in Group A or Group B. Ask the scientists to predict whether the people were actively engaging in calculation, or simply thinking of the blackness of space.

I predict that the predictive success would not actually be better than chance. The likely reason is that the human brain is not actually the cause of human thought. No one has a coherent idea as to how neurons could produce thinking or ideas. There are strong reasons for believing that fast accurate complex thought should be impossible for a brain, because of the very high noise levels in a human brain (as discussed here), and because signal transmission should actually be very slow in a brain (for reasons discussed here).

Here is another experiment that neuroscientists have not done (to the best of my knowledge), but should be doing:
  1. Do a brain scan on 20 subjects (called Group A). Tell the subjects to think of absolutely nothing during the brain scan other than the blackness of outer space.
  2. Do a brain scan on 20 other subjects (called Group B). Tell the subjects to do some task involving memory recall, such as remembering all the vacations they have ever had (or trying to recall everyone they can remember with a name beginning with the letter “A,” everyone they can remember with a name beginning with the letter “B,” and so forth).
  3. Shuffle the brain scans, and submit them to some other “blinded” scientists who do not know whether the subjects were in Group A or Group B. Ask the scientists to predict whether the people were actively engaging in memory recall, or simply thinking of the blackness of space.

I predict that the predictive success would not actually be better than chance. The likely reason is that the human brain is not actually the cause of human recall. Given the short lifetime of synapse proteins and other forms of instability in the brain, no one has a coherent idea as to how a brain could store memories lasting for decades, or how a brain could instantly recall memories without any addressing system that might allow such a thing. There are strong reasons for believing that the brain is not the storage place of human memory. 

Studies with the protocol above have not been done (to the best of my knowledge). But scientists have done studies in which people have their brains scanned while the people are thinking or recalling. Such studies show no real evidence of neural correlates of thinking or neural correlates of recall. Typically the change in signal strength from one brain region to another (which is the most important thing to consider) is no greater than 1%, about what we would expect from random variations. Such results (discussed here) are consistent with what we would expect if the brain is not a storage place for memories, and if the brain is not the source of our thoughts.  Memory and thought are very likely aspects of a spiritual aspect of man, something quite distinct from the brain. 

An interesting aspect of dreaming is how we can recall names, locations and even intellectual principles during dreaming, even though we may have never thought of such things in years.  Recently I had a dream in which I recalled the principle that you can compute the price of a bond from its yield, a principle I haven't used, read about or thought about in many years.  Nobody has a coherent detailed explanation as to how such abstract principles could ever be stored as neural states or synapse states, and it is all the more impossible to explain how a sleeping person's brain could recall such a principle. 

No comments:

Post a Comment