A Method for Evaluating the Strength of Causal Claims

An elementary school student might think to himself something like the following,

"When scientists figure out the cause of some effect, they start saying that such a cause is the reason why the effect occurs. If they don't know what causes the effect, they just say they don't understand what causes it."

Such a generalization is naive. A more accurate and realistic description is the one below:

Scientists often determine the cause of an effect, and begin to state that cause as the reason why that effect occurs. In many other cases, humans are unable to figure out what the cause of an effect is. But that doesn't stop scientists from making claims about what causes the effect. In many cases such claims are weakly supported by evidence or logic, but scientists continue to make such claims because the assertion of such claims has become a speech custom of scientists, or because the claims help the scientists avoid or sweep under the rug some possibility they prefer not to believe in.

Given such a situation, it is helpful to have some method for evaluating the strength of causal claims. Using such a system, we might be able to distinguish between strong causal claims and weak causal claims. I can think of such a method, which I list below.

In the method I propose, we ask many different questions, and either add or subtract points based on the answers. Here are the questions:

Y= Effect, X=proposed cause of the effect
1. Has Y always been observed occurring whenever X occurs?	+1
2. Has Y usually been observed occurring when X occurs?	+1
3. Has Y sometimes been observed occurring when X occurs?	+1
4. Has Y never been observed occurring when X occurs?	-1
5. Does X usually occur without producing Y?	-1
6. Does X sometimes occur without producing Y?	-1
7. Does Y always occur very close in time and space to X?	+1
8. Can we think of some reason why X would always produce Y?	+1
9. Can we think of some reason why X would usually produce Y?	+1
10. Can we think of some reason why X would occasionally produce Y?	+1
11. Can we think of some reason why X would never produce Y?	-1
12. Can we think of some reason why X would usually not produce Y?	-1
13. Are there no other reasonable theories of how Y could be produced other than assuming X produced it?	+1
14. Can we think of a fact or facts that are true, but which are surprising or unexpected if X causes Y?	-1
15. Can we think of some fact or facts that are true, but hard to explain unless X causes Y?	+1
16.  If X an undisputed reality that has been instrumentally measured or photographed?	+1

Using this system, we should find that strong causal claims end up with a positive score, and weak causal claims end up with a zero or negative score. The stronger the causal claim, the higher the should be, and the weaker the causal claim, the lower the score should be.

Let's try this system, using a specific example. An example of a causal claim is the claim that water freezing is caused by temperatures below 0 degrees Celsius. Below is how we can compute the score of this causal claim. (For the purpose of simplicity, the only places being considered are the surface of Earth and the surface of the moon.) 

Y= Water freezing, X=Temperatures below 0 degrees Celsius
1. Has Y always been observed occurring whenever X occurs?	0	An arid landscape (such as on the moon) can show no signs of water freezing even when temperatures are below 0 degrees Celsius.
2. Has Y usually been observed occurring when X occurs?	+1	We usually observe freezing at temperatures below 0 degrees Celsius.
3. Has Y sometimes been observed occurring when X occurs?	+1	We usually observe freezing at temperatures below 0 degrees Celsius.
4. Has Y never been observed occurring when X occurs?	0	Water has been observed freezing at emperatures below 0 degrees Celsius.
5. Does X usually occur without producing Y?	0	Temperatures below 0 degrees Celsius usually do not occur without producing water freezing
6. Does X sometimes occur without producing Y?	-1	Temperatures below 0 degrees Celsius sometimes occur without producing water freezing, in arid places like the moon.
7. Does Y always occur very close in time and space to X?	+1	Water freezing always occurs at the same time and place as temperatures below 0 degrees Celsius.
8. Can we think of some reason why X would always produce Y?	0	We can think of no reason why temperatures below 0 degrees Celsius would always cause water freezing, because on some landscapes like the moon there is no water.
9. Can we think of some reason why X would usually produce Y?	+1	We can think of a reason why temperatures below 0 degrees Celsius would usually cause water freezing, given that most places have water that might freeze.
10. Can we think of some reason why X would occasionally produce Y?	+1	We can think of a reason why temperatures below 0 degrees Celsius would sometimes cause water freezing, because most places have water that might freeze.
11. Can we think of some reason why X would never produce Y?	0	There is no reason why temperatures below 0 degrees Celsius should never produce water freezing.
12. Can we think of some reason why X would usually not produce Y?	0	Given that most places have water, there is no reason why temperatures below 0 degrees Celsius should usually not produce water freezing.
13. Are there no other reasonable theories of how Y could be produced other than assuming X produced it?	+1	No one has advanced an alternate theory of how water freezing could occur other than temperatures below 0 degrees Celsius.
14. Can we think of a fact or facts that are true, but which are surprising or unexpected if X causes Y?	0	No, we can't.
15. Can we think of some fact or facts that are true, but hard to explain unless X causes Y?	+1	Frozen water (ice) is very cold, and we would not expect it to be so cold unless temperatures below 0 degrees Celsius had caused it get so cold.
16.  If X an undisputed reality that has been instrumentally measured or photographed?	+1	No one disputes that there are temperatures below 0, and such temperatures have been instrumentally  measured with a thermometer.

So using this scoring system and adding up all the numbers, it turns out that the claim that water freezing is caused by temperatures below 0 degrees Celsius ends up with a score of +7. In this case, the system seems to work well. We have a very strong cause and effect relation in this case, and the causal explanation has scored highly using our system.

Let's try this system, using another specific example. An example of a causal claim is the claim that thunder is caused by lightning. Below is how we can compute the score of this causal claim.

Y= Thunder, X=Lightning
1. Has Y always been observed occurring whenever X occurs?	0	While thunder is believed to occur whenever lightning occurs, we sometimes hear no thunder when we see lightning.
2. Has Y usually been observed occurring when X occurs?	+1	We usually do hear thunder when lightning occurs.
3. Has Y sometimes been observed occurring when X occurs?	+1	We usually do hear thunder when lightning occurs.
4. Has Y never been observed occurring when X occurs?	0	Thunder does occur when lightning occurs.
5. Does X usually occur without producing Y?	0	Visible lighting usually produces audible thunder.
6. Does X sometimes occur without producing Y?	0	It is believed that lightning always produces thunder, even when we fail to notice the thunder.
7. Does Y always occur very close in time and space to X?	+1	Thunder is thought to always occur at the same time and place as lightning.
8. Can we think of some reason why X would always produce Y?	+1	There are scientific reasons why lightning should always produce thunder.
9. Can we think of some reason why X would usually produce Y?	+1	There are scientific reasons why lightning should always produce thunder.
10. Can we think of some reason why X would occasionally produce Y?	+1	There are scientific reasons why lightning should always produce thunder.
11. Can we think of some reason why X would never produce Y?	0	No.
12. Can we think of some reason why X would usually not produce Y?	0	No.
13. Are there no other reasonable theories of how Y could be produced other than assuming X produced it?	0	There is the theory that thunder is a roar of angry gods or spirits.
14. Can we think of a fact or facts that are true, but which are surprising or unexpected if X causes Y?	0	No, we can't.
15. Can we think of some fact or facts that are true, but hard to explain unless X causes Y?	0	There may well be, but I can't think of any.
16.  If X an undisputed reality that has been instrumentally  measured or photographed?	+1	No one disputes that lightning exists, and it has been photographed.

So using this scoring system and adding up all the numbers, it turns out that the claim that thunder is caused by lightning ends up with a score of +7. In this case, the system seems to work well. We have a very clear cause and effect relation in this case, and the causal explanation has scored highly using the method I have proposed for judging the strength of a causal effect.

Thunder is actually produced by lightning

Now let us consider the claim that complex visible biological innovations are caused by random mutations and what biologists call "natural selection" (a term that is not literally accurate because blind unconscious nature does not literally select things).  By "complex visible biological innovations" I refer only to things appearing in nature, not anything produced artificially by humans, and I refer only to things involving a high level of complexity (not mere superficial changes such as darkening).  Below is how we can compute the score of the causal claim that complex visible biological innovations are caused by random mutations and natural selection.

Y= The appearance of complex visible biological innovations, X = random mutations and natural selection
1. Has Y always been observed occurring whenever X occurs?	0	Humans have not actually observed any complex visible biological innovations appearing. All such innovations appeared before 3000 B.C.
2. Has Y usually been observed occurring when X occurs?	0	Humans have not actually observed any complex visible biological innovations appearing.
3. Has Y sometimes been observed occurring when X occurs?	0	Humans have not actually observed any complex visible biological innovations appearing.
4. Has Y never been observed occurring when X occurs?	-1	Humans have never observed any complex visible biological innovations appearing.
5. Does X usually occur without producing Y?	-1	Scientists say that natural selection is constantly occurring, and that random mutations are constantly occurring. But still humans have never observed any complex visible biological innovations occurring.
6. Does X sometimes occur without producing Y?	-1	Scientists say that natural selection is constantly occurring, and that random mutations are constantly occurring. But still humans have never observed any complex visible biological innovations occurring.
7. Does Y always occur very close in time and space to X?	+1	Although we have not observed complex visible biological innovations occurring, it is claimed that random mutations and natural selection have always been occurring, so we can generously grant that when previous complex visible biological innovations occurred, they were close in time in space to random mutation and some type of survival of the fittest that might be metaphorically called "natural selection."
8. Can we think of some reason why X would always produce Y?	-1	We can think of no reason why random mutations and natural selection would always produce complex visible biological innovations, and scientists think that in most years they do not produce such a thing.
9. Can we think of some reason why X would usually produce Y?	-1	We can think of no reason why random mutations and natural selection would usually produce complex visible biological innovations, and scientists think that in most years they do not produce such a thing.
10. Can we think of some reason why X would occasionally produce Y?	-1	We can think of no reason why random mutations and natural selection would occasionally produce complex visible biological innovations, just as we can think of no reason why tornadoes passing through a city (which produce random effects and a kind of architectural "survival of the fittest") would occasionally produce complex new architectural innovations.
11. Can we think of some reason why X would never produce Y?	-1	We can think of several  reasons why random mutations and natural selection would never produce complex visible biological innovations: (1) the fact that such complex visible biological  innovations require such extremely high states of organization  and fine-tuning that it seems impossible that they would ever be produced accidentally, just as houses of cards should never be formed by throwing a deck of cards into the air, and just as log cabins should never form from falling trees; (2) the fact that individual random mutations (point mutations) are merely tiny fragments of what is needed for a complex visible biological innovation, typically less than 1%; (3) the fact that natural selection (or a superior reproduction rate) never works in regard to some complex biological innovation until such an innovation has already appeared.
12. Can we think of some reason why X would usually not produce Y?	-1	Same as above.
13. Are there no other reasonable theories of how Y could be produced other than assuming X produced it?	-1	There are three other reasonable theories of how complex visible biological innovations might have happened (1) they might have been introduced by a divine creator; (2) they might have been introduced by extraterrestrial visitors wanting for Earth to eventually produce intelligent life; (3) they might have been produced by some cosmic life-force.
14. Can we think of a fact or facts that are true, but which are surprising or unexpected if X causes Y?	-1	The Cambrian Explosion in which most animal phyla originated rather suddenly is unexpected under the theory that complex visible biological innovations are produced by random mutations and natural selection, which have always been described as very slow and gradual effects.
15. Can we think of some fact or facts that are true, but hard to explain unless X causes Y?	+1	There probably are no such facts, given a wide variety of alternative possibilities, but let's grant this +1 point just to be charitable.
16.  If X an undisputed reality that has been instrumentally  measured or photographed?	0	"Natural selection" is disputed as a literally accurate term, on the grounds that only conscious agents select things.  The main idea behind natural selection (that fit things reproduce more) has been disputed, on the grounds that it is unproven (for example, not-very-fit bacteria seem to reproduce at a much higher rate than mammals).  Natural selection has never been instrumentally measured or photographed.

Shockingly, the theory that random mutations and natural selection produce complex visible biological innovations produces a score of only -8, which is 15 points lower than the score produced by the claim that lightning produces thunder and the claim that temperatures of 0 degrees Celsius or lower produce freezing.  This very low score suggests that the theory that random mutations and natural selection produce complex visible biological innovations is far from being a strong causal claim. 

Could it be that something is wrong with my method for evaluating the strength of causal claims? Let's try it again, testing it with the claim that accidental falls from high altitudes produce fatal injuries. 

Y= Fatal injuries, X= accidental falls from high altitudes
1. Has Y always been observed occurring whenever X occurs?	0	Occasionally someone will survive an accidental fall from a high altitude.
2. Has Y usually been observed occurring when X occurs?	+1	Most people who accidentally fall from high altitudes have fatal injuries.
3. Has Y sometimes been observed occurring when X occurs?	+1	Most people who accidentally fall from high altitudes have fatal injuries.
4. Has Y never been observed occurring when X occurs?	0	Most people who accidentally fall from high altitudes have fatal injuries.
5. Does X usually occur without producing Y?	0	Most people who accidentally fall from high altitudes have fatal injuries.
6. Does X sometimes occur without producing Y?	-1	Sometimes people accidentally fall from high altitudes without having fatal injuries.
7. Does Y always occur very close in time and space to X?	-1	Fatal injuries can occur without falling, such as in auto collisions.
8. Can we think of some reason why X would always produce Y?	0	A person accidentally falling from a high altitude may land in a soft surface like a haystack.
9. Can we think of some reason why X would usually produce Y?	+1	Given the acceleration produced by accidental falls from high altitudes, we can think of a reason why they would usually produce fatal injuries.
10. Can we think of some reason why X would occasionally produce Y?	+1	Given the acceleration produced by accidental falls from high altitudes, we can think of a reason why they would usually produce fatal injuries.
11. Can we think of some reason why X would never produce Y?	0	No.
12. Can we think of some reason why X would usually not produce Y?	0	No.
13. Are there no other reasonable theories of how Y could be produced other than assuming X produced it?	0	Fatal injuries can be produced by car crashes that do not involve falls from high altitudes.
14. Can we think of a fact or facts that are true, but which are surprising or unexpected if X causes Y?	0	No, we can't.
15. Can we think of some fact or facts that are true, but hard to explain unless X causes Y?	0	No, we can't.
16.  If X an undisputed reality that has been instrumentally measured or photographed?	1	No one disputes there are accidental falls from high altitudes, and they  have been photographed.

So according to this system, the overall strength of the claim that accidental falls from high altitudes produce fatal injuries is +3. While not as strong as the +7 scores reached by the theory that lightning causes thunder and the theory that temperatures below 0 degrees Celsius produce freezing, at least the scoring system has given a positive score to a strong causal claim.  So it seems that the method I have suggested here works well in distinguishing between causal claims that are weak and those that are strong. 

So why is it that we continue to have so many assertions of the claim that complex visible biological innovations are caused by random mutations and natural selection, despite the weakness of such a causal claim? It would seem the answer is: largely because this is a speech custom that is imposed by institutional powers that compel or prod people to say a particular thing.

We might also ask: why do millions of children in school state every day the false claim that there is "liberty and justice for all" in America? It is a fine thing for school children to "pledge allegiance to the flag, and to the republic for which it stands." But a BBC article tells us, "Prison rates in the US are the world's highest, at 724 people per 100,000." Very many of these people in prison should not be there, and are there because the people could not afford good lawyers, were imprisoned because of over-zealous drug laws, or were the victims of racial prejudice. Meanwhile many people who commited the worst kind of crimes between 2003 and 2009 suffered no legal penalties at all, because they were rich enough to afford skillful lawyers or politically protected.  So it seems that while in the United States there is liberty and justice for most, there is neither "liberty for all" nor "justice for all."  So why is it that millions of children every school day assert the false claim that in America there is "liberty and justice for all"? The answer is:  because this is a speech custom that is imposed by institutional powers that compel or prod people to say a particular thing.

Psi-Denying Professors Are Notorious Cherry Pickers

In the recent book Knock on Wood: Luck, Chance and the Meaning of Everything by statistics professor Jeffrey S. Rosenthal, we get no real insight about the meaning of everything. We also very much fail to get something vastly easier to produce: a reasonably fair treatment of the topic of evidence for human paranormal abilities, which are collectively referred to as psi. Rosenthal's treatment of the topic gives us quite a few examples of atrocious cherry picking, which is when someone cites only a few isolated things supporting his position, while failing to mention very strong cases, examples or reasons that argue against his position.

Chapter 21 of Rosenthal's book is entitled “Mind Over Matter?” His first form of cherry-picking is to bring up the topic of extrasensory perception (ESP), but fail to inform his reader about any of the massive laboratory evidence that exists for telepathy. Very inaccurately, he suggests that there is no evidence for it. You would never know from reading Rosenthal's book that university scientists spent many years and countless hours testing people for ESP. Excluding his discussion of remote viewing and precognition tests, his sole discussion of such laboratory tests is to tell us about a single unsuccessful ESP experiment that was done on a pair of twins. You might think from his discussion that this was the only time professors ever tested people for ESP. But why do we not hear about the countless hours of university tests done by researchers such as professor Joseph Rhine, who Rosenthal fails to mention? I suspect the reason is that such tests produced very dramatic successes proving the reality of ESP (as discussed here and here).  Similarly, Rosenthal tells us not a word about the more modern ganzfeld ESP tests under which countless experiments very consistently demonstrated results far above that which we would expect by chance.

Rosenthal discusses remote viewing. He states on page 247, “There are indeed some experiments that have shown remote viewing accuracy slightly higher than what would be expected by chance alone.” It is extremely misleading to characterize remote viewing tests as something that merely showed “accuracy slightly higher than what would be expected by chance alone.” The reason why the US government kept funding the Stargate Project from 1978 to 1995 is that there were many spectacular successes by people attempting remote viewing.

In the current version of the Journal for Scientific Exploration (Volume 33, Issue 4), which you can download at this link, we have on page 33 a review article by Russel Targ entitled, “What Do We Know about Psi? The First Decade of Remote-Viewing Research and Operations at Stanford Research Institute.” Targ states, “Many of our experimental series were statistically significant at four standard deviations from chance expectation, with effect sizes greater than 0.6.” Results of four standard deviations above chance (the same as a four-sigma evernt) are vastly greater than results “slightly higher than what would be expected by chance alone.” In the statistical paper here, we are told, "a 4-sigma event is to be expected about every 31,560 days or about 1 trading day in 126 years." Getting many such 4-sigma events is a result enormously higher than what would be expected by chance alone, not merely "slightly higher than what would be expected by chance alone," as Rosenthal asserts. Targ gives many specific examples of astounding successes. 

Rosenthal mentions "replication failures" as an excuse for rejecting evidence for paranormal abilities.   This doesn't make sense. If some tests are done showing results that no human being should ever be able to have achieved by chance (if paranormal abilities do not exist), you do not debunk such evidence for the paranormal by doing some other experiment (with the same subjects or other subjects) that produces only chance results.  Similarly, if I do tests showing that some people have extraordinarily great memory recall, you do not debunk such tests by testing some other people who show only average memory recall. And if I do tests showing that on some days certain baseball hitters can hit baseballs into the bleachers, I do not discredit such evidence if I do other tests in which the same baseball hitters fail to produce such home runs. 

Rosenthal's idea of debunking is to do something like cite on page 248 an analysis that claimed that "of the 332 experiments performed by PEAR between 1969 and 1987, only 71 gave results consistents with telekinesis, while 261 did not."  That would actually seem to be massive evidence that mind-over-matter or telekinesis can actually occur, since you never debunk a positive experiment for an irregular effect by doing some other experiment in which the irregular effect does not occur. 

Rosenthal mentions the experimental evidence for precognition produced by professor Daryl Bem. As we read in Bem's original paper, one of his original experiments showed a success rate of 53.5% against an expected result of only 50%.  This was one of the less impressive positive results in parapsychology, with many other experiments showing success rates vastly higher than what should be expected by chance.  But still, given the number of participants (100) and the number of trials per subject (32), such a result is extremely unlikely to occur by chance. According to the binomial probability calculation below using the Stat Trek calculator, the chance of getting a result as good as this by chance is only about .00003.

Rosenthal attempts to discredit this evidence by mentioning three unsuccessful attempts to replicate the experiment.  This is quite the outrageous case of cherry-picking. What he fails to tell us is that Bem's original experiment has been successfully replicated by far more attempts than those three replication attempts.  The meta-analysis here tallies all attempts to replicate Bem's original results.  Table 1 shows that when you tally the results from all of the replication attempts, you are left with a strong effect size, and a statistical significance that easily qualifies as good solid empirical science and good solid evidence for a very real effect. 

On page 244 Rosenthal discusses the possibility of twin telepathy. The only evidence he mentions is some negative experiment done by a psychologist. He fails to mention any of the evidence on the topic reported by the Psi Encyclopedia.  In a large survey of twins in 1961, a full third reported that they could communicate by telepathy, particularly in times of danger. We also read this:

"Two ophthalmologists from Philadelphia published a brief paper in a leading science journal in which they claimed to have shown that telepathic influences could actually be instrumentally recorded. By artificially inducing alpha rhythm in one of a pair of twins, they showed that the other’s brain went into alpha at exactly the same time through what they controversially called ‘extrasensory induction.’ "

We also read this about 1967 results from a team at Rockland State Hospital:

"They used just one pair of twins, but reported confidently that ‘in a physically isolated subject, we have observed physiological reactions at the precise moment at which another person was actively stimulated’. They even printed the entire chart record ‘to show how obvious the [reactions] are.' "

The same Psi Encyclopedia tells us of a televised experiment and its aftermath:

"A popular television programme included an experiment in front of a live audience: here twin A, seated in front of a large pyramid, was put into a light trance state, while her sister B was in a distant and soundproof room wired to a polygraph that recorded her respiration, heartbeat, and galvanic skin response. After a period of relaxing meditation, the pyramid was made to explode with a loud bang, giving twin A a considerable shock. At the same time all channels monitoring B peaked sharply. This was a useful pilot project for more tightly controlled experiments to follow. It showed that when twins are carefully selected, as these were, and submitted to a surprise stimulus, they are far more likely to react than they might succeed in the type of telepathy experiment that involves guessing symbols on cards. Following this programme – along with a popular book and a journal article on twin telepathy – four similar experiments using the polygraph were shown on different channels. This gave some support to the view that telepathy can be shown not only to cause a physical response, but also to be instrumentally recorded as it does so."

According to the link here, twins tested under such separated and shielded conditions had unexplained correlations in reactions, with a statistical significance of .0001, which is much better than the .05 statistical significance needed to get a result published in a scientific journal. 

These results are little known, so we can forgive Rosenthal for failing to mention them. But there is no excuse for failing to mention the very well-known research of professor Joseph Rhine and his colleagues in Rosenthal's 14-page chapter on the evidence for paranormal abilities.  Rhine produced enormously convincing evidence for telepathy, such as I discussed in my post "When Rhine and Pearce Got 'Smoking Gun' Evidence for ESP," and in my other post reviewing results such as Rhine reported in his monumental work Extra-sensory Perception After Sixty Years. Even greater evidence was produced by professor Riess, who (as I discuss here) did a remote-location test (involving 1850 card guesses) with a woman who guessed an average of 18.24 cards correctly per 25 cards, achieving a phenomenal 73% accuracy rate (instead of the expected accuracy rate of 20%).  This link estimates the chance likelihood of such a result as being about 1 in 10 to the 700th power. 

Of course, we read about no such results from psi-denying professors such as Rosenthal, who are very careful to only mention weak or negative results in their books.  Similar cherry-picking might be involved if someone claimed that New England quarterbacks haven't been very good, and mentioned only the career of forgettable New England quarterbacks, failing to mention the case of six-time Super Bowl champion Tom Brady.  

A Brain Would Never Know Where to Read or Write a Memory

There was never any observation that forced scientists to start claiming that human memories are stored in brains. Scientists simply started gradually claiming such a thing, and the idea spread to the masses through a process of social contagion. 

There is no robust evidence that memories are stored in brains, but every now-and-then the press claims that some scientists have done something they could only do if memories were stored in brains. Maybe a claim is made that memories were erased from a brain, or maybe a claim is made that memories were implanted in a brain, or maybe a claim is made that memories were transferred from one brain to another.  It will inevitably be true that if you examine the research in detail, you will find that no robust evidence has been produced for any such thing. Such research suffers from the types of flaws discussed here. 

Given that there are many thousands of neuroscientists funded with so many hundreds of millions of dollars of research money, we should expect that exactly such reports would occasionally appear, even if memories are not stored in brains. In considering matters such as these, I like to remember a particular rule:

The rule of well-funded and highly motivated research communities: almost any large well-funded research community eagerly desiring to prove some particular claim can be expected to  occasionally produce superficially persuasive evidence in support of such a claim, even if the claim is untrue. 

For example, if there were a group of 40,000 researchers who were believers in Bigfoot creatures, and such a group were to each year receive hundreds of millions of dollars in funding for their research, we should then expect to occasionally get superficially persuasive evidence in support of the existence of Bigfoot creatures, even if they don't exist. 

Sometimes an idea may seem fairly believable when it is painted in broad brushstrokes, but we may recognize the idea as being untenable once we start to examine the idea in detail. When a small child loses a tooth, her mother may tell her the story of the Tooth Fairy. “Just place the tooth under your pillow,” says the mother, assuring the child that when she wakes up there will be some cash under her pillow and that the tooth will be gone, because of the action of the Tooth Fairy. The next morning the child will probably be woken up by her mother, who will report finding some cash under the pillow. The story of the Tooth Fairy doesn't seem too unbelievable, unless the child subjects the story to detailed scrutiny. For example, she might ask: how could the Tooth Fairy ever have known that I had lost a tooth? Or she might ask: how could the Tooth Fairy ever have known where I lived, or removed the tooth and placed the money, without waking me up? After such scrutiny the child may realize that the mother probably just removed the tooth and put the money under the pillow when she woke up the child .

Like the story of the Tooth Fairy, the story that memories are stored in brains does not sound very unreasonable if we hear the claim painted in broad brushstrokes. It is when we start to subject this claim to detailed scrutiny that all types of credibility problems arise.

Let us consider two of these credibility problems: that a brain would never exactly where to read a memory, and that a brain would never know exactly where to write a memory.

Problem #1: A Brain Would Never Know Exactly Where to Read a Memory

One of the worst problems associated with the idea that a brain stores memories is what I call the navigation problem. The navigation problem is the problem that if a memory or some particular piece of knowledge was stored in your brain, your brain would never know where exactly to read that exact memory.

Let's consider an example. You are in school, and you come to a test question asking about some particular human, perhaps a scientist or a general or the leader of a country. You then have to recall what you know about that person. Under the theory that our brains store memories, such information would presumably be stored in some exact tiny spot in your brain. But for you to answer the question using the brain to retrieve a memory, your brain would presumably have to know the exact tiny spot where to read that information. How could the brain possibly know where that exact spot was, so that you instantly recall the memory?

I can imagine an extraterrestrial organism for which such a thing would be easy. It might work like this. The organism might have memory addresses, a position location system with numerical identifiers for each of the little storage locations (comparable to post office boxes in a gigantic post office). So when the organism formed a new memory, it might put that memory into some numbered storage location (for example, memory slot #822,235). Also, when the organism formed a new memory, it would always be associating these memory addresses with particular names, facts, and faces. So, for example, if the organism formed a new memory that Jokonto was the ruler of Zunando, then it would always remember a particular number (such as #532,233) that it would associate with the name Jokonto. Then when the organism heard the name Jokonto, it would remember that memory address and read the information from exactly that tiny little spot in its brain (memory slot #532,233 in this example).

But nothing like this can be occurring in the human brain. Particular neurons in the brain are not addressable. There is no position location system that could possibly be used by the brain to identify the exact tiny location of a stored memory. Also, humans do not at all remember any numbers associated with a storage location in the brain. I may learn the fact that George Patton was a skilled general during World War II, but I do not at all learn any “brain location coordinate” or location address that I associate with the name of George Patton, some neural location number that I could use to instantly retrieve the exact location of the information that I had learned about George Patton.

So if a memory or learned information is stored in some exact spot of the brain, how could your brain ever instantly find that exact spot? It seems that it could never do this. The brain would never know the exact spot to read a particular memory.

You do not at all get around this difficulty by suggesting the idea that a memory or a piece of learned information is scattered in multiple locations across the brain. The difficulty is explaining instantaneous recall. If a brain has to search scattered storage locations in the brain, that would not be any easier than finding a single storage location. We would then have the same problem: how is it that those exact locations can instantly be found? Similarly, if  a family is somewhere in New York City, and you don't know their address, you won't be able to find the family very quickly; and it's not going to be any easier if the family is scattered across three different apartments.

You also do not at all get around this difficulty by suggesting the idea that the brain reads all of its information each time a memory is retrieved. For one thing, such an idea does not correspond to human experience. If I hear a name, I recall only what I have learned about that name, and do not at all have some experience of recalling or reading some huge amount of information. Moreover, the idea of the brain reading all of its stored information (rather than one tiny spot) just worsens the problem of explaining how instantaneous recall can occur. Instantaneous recall could never happen if a brain was reading all of a large amount of information stored in it, or even a tenth of such information. 

Problem #2: A Brain Would Never Know Where to Write a Memory

Now let us consider a separate problem regarding the idea that brains store memories: the problem that a brain could apparently never derive an exact suitable specific location at which a new memory should be written.

With certain types of systems, there is no problem about where to write a new piece of information. Consider a diary. A diary typically consists of dated pages. So it's always obvious where you should be writing when you make a diary entry: in the page that has the name of today's date. It is also rather obvious in a student's notebook where new information should be written: at the end of the last place where something was written.

But a brain is not at all like a diary or a notebook. There are no date-marked places to put information acquired on some particular date. And given the organization of the brain, there is nothing like some place corresponding to the first blank page of a notebook. So, if a brain were to be writing some new information acquired on a particular day, how could the brain figure out or derive some appropriate position to write at?

We will not get any insight into this question by considering how a computer stores data. Imagine I open an application, and write some text. I then try to save my information as a particular file. How does the computer figure out where on the computer to store this information? What seems to happen is that the software application uses the computer's operating system to save the file. The operating system is a core set of software routines used for common tasks such as saving files. If we were to delve into the software code of the operating system, we would probably find that the operating system code searches for a random block of free data on the hard drive, a block with enough space to store the data.

For example, suppose I open the “Notepad” application, and type 100 words. When I choose File/Save from the menu, the application calls some software (probably operating system software) that looks for a random position on the disk with enough space to write 100 words. Random selection is very easy for a computer (for example, it's easy for some sofware routine to pick a random number between 1 and 100).

Here is the type of algorithm that an operating system might use to save data at a random free location:

1. Pick a random position on the disk.
   
2. Scan ahead X bytes to make sure that the next X bytes are free space on which nothing is written (where X is the amount of data to write).
   
3. If the next X bytes are free disk space, write the data to be written at the random position.
   
4. If any of the next X bytes are not free disk space, go back to step (1).
   

Can we imagine that something like this goes on in the brain, even though we have no mental experience of any such logic or calculation going on when we form a new memory? It does not seem that something like this could be occurring in the brain. The brain has nothing like the operating system in a computer. A brain cannot secretly be doing logic like the logic above to determine where to write a memory.

We may consider the simple matter of picking a random position in the brain to write a memory. For a conscious agent, it is very easy to pick a random position. But it would seem to be impossible for a brain to pick a random position in itself without any conscious choice being made by a mind.

There are certain physical arrangements that can kind of guarantee a random positioning effect. For example, if I throw a small ball on a large grate covering a hole, such an arrangement will make it likely that the ball will fall through one of the holes in the grate, with the ball going through a random hole. But there seems to be no physical arrangement in the brain by which some random position in it could be selected as the place to write a memory.

One possibility is the possibility of a cursor. In a word processing document, a cursor is blinking position marker indicating the current writing position. We can imagine something similar in a brain. There could be something like a “moveable write unit.” When a memory is stored using the write unit, the write unit could write wherever in the brain it was located. The write unit could move along as it wrote. Under such a system, there would be no need for a brain to be selecting a random position to write at. The brain would simply write at wherever brain position the write cursor was located at.

However, there is no sign of such a cursor or movable write unit in the brain. Other than electricity and chemicals and blood moving around in the brain, there are no moving parts in the brain. Electricity and chemicals are evenly distributed in the brain, and there is no concentration of electricity or chemicals that could be anything like a memory cursor or a moving write unit. The human brain bears no resemblance to a system for storing a particular memory in one specific spot. Similarly, the human body bears no resemblance to a system for storing the nutrients from a meal in only one specific spot of the body. 

A scientist may claim that when some new memory is acquired, that memory is stored in some exact tiny spot of the brain. But such a person could never give a credible answer to the question: why would such a memory have been stored in that exact tiny spot rather than any of a million other tiny locations in the brain?

Faced with such difficulties, someone may throw up his hands and say, "There must be some way by which some particular spot in the brain becomes the current spot where you store what you are learning right now." To this I say: no, there isn't any such thing. The difficulties I mention are only two of a host of prohibitive difficulties involved in the idea of a brain storing a memory. Others discussed at great length here include the lack of any known information reading mechanism in a brain, the lack of any known information writing mechanism in a brain, the complete lack of any credible theory by which conceptual and episodic information could be translated into neural states or synapse states, and the lack of any credible theory of how memories could be stored in a brain for decades given high molecular turnover such as the very rapid protein turnoever in synapses.  The way to overcome such difficulties is to abandon the never-proven claim that memories are stored in brains, and to move to the idea that memory must be a spiritual facility.