Exhibit B Suggesting Scientists Don't Know How a Brain Could Retrieve a Memory

In a 2019 post “Exhibit A Suggesting Scientists Don't Know How a Brain Could Retrieve a Memory,” I took a close look at 68 “expert answers” given on one page of an “expert answers” site, a page with the topic of "how are memories retrieved in the brain?" I argued  that none of the experts had a coherent and convincing answer to the question “how are memories retrieved in the brain?” I maintain that answering such a question convincingly will always be impossible, because human memories are not stored in brains, and nothing in the human brain bears any resemblance to either  a device for retrieving factual information learned during human experience or a device for storing memories for years. In particular, there is not any thing in the human brain that can explain how a human brain can instantly retrieve detailed information learned long ago about about some obscure person, place or event. Since the brain lacks any addressing system, any indexing system, and any position notation system, it should be absolutely impossible for a brain to instantly recall obscure information, such as we see happening on the long-running television quiz show Jeopardy. For example, if someone asks you (for the first time ever in your life) to name three Russian composers, and you instantly answer “Tchaikovsky, Borodin, and Rimsky-Korsakov,” you are doing something absolutely inexplicable in terms of brain activity.

Now I will give a kind of “Exhibit B” suggesting that scientists don't know how a brain could retrieve a memory: a 2019 paper entitled “The neurobiological foundation of memory retrieval.” When we get beyond the hype and unwarranted braggadocio of this paper, we find that it fails to convincingly portray any such foundation at all.

A great deal of the paper is involved with trying to persuade us that experimental studies have made great progress in identifying memory storage sites (called engrams). The authors state, “In the last decade, enormous progress has been made in identifying and manipulating engrams in rodents.” This statement is not at all correct. A few scattered studies have claimed to identify and manipulate such alleged engrams, but such studies have failed to provide any convincing evidence that such engrams really exist. The studies typically suffer from several of the following methodological sins:

Sin #1: assuming or acting as if a memory is stored in some exact speck-sized spot of a brain without any adequate basis for such a “shot in the dark” assumption.

Sin #2: either a lack of a blinding protocol, or no detailed discussion of how an effective technique for blinding was achieved.

Sin #3: inadequate sample sizes, and a failure to do a sample size calculation to determine how large a sample size to test with.

Sin #4: a high occurrence of low statistical significance near the minimum of .05, along with a frequent hiding of such unimpressive results, burying them outside of the main text of a paper rather than placing them in the abstract of the paper.

Sin #5: using presumptuous or loaded language in the paper, such as referring in the paper to the non-movement of an animal as “freezing” and referring to some supposedly "preferentially activated" cell as an "engram cell."

Sin #6: failing to mention or test alternate explanations for the non-movement of an animal (called “freezing”), explanations that have nothing to do with memory recall.

Sin #7: a dependency on arbitrarily analyzed brain scans or an uncorroborated judgment of "freezing behavior" which is not a reliable way of measuring fear.

I fully discuss all of these methodological problems in my post “The Seven Sins of Memory Engram Experiments,” and I give very many examples of how the papers cited as evidence for engrams in rodents are guilty of such procedural sins. So when the authors of the paper “The neurobiological foundation of memory retrieval” assert that "enormous progress has been made in identifying and manipulating engrams in rodents,” they do not speak correctly at all. There still exists no robust well-replicated evidence that any such thing as an engram (a neural site of stored learned information) exists in any animal. 

The authors present a lengthy, credulous and uncritical review of weak neuroscience studies that have attempted to find evidence for memory engrams (neural storage sites for memories). Their review repeatedly fails to subject such studies to an appropriate level of scrutiny. The authors  trumpet weak and poorly replicated studies as evidence for the memory engrams that they  want to believe in. We hear no mention of the very many problems in such studies, such as the fact that they typically use unreliable bias-prone techniques for judging the degree of fear in rodents (subjective judgments about "freezing behavior") rather than reliable objective techniques such as heart-rate measurement (the heart rate of a rat dramatically surges when the rat is afraid). 

In the section entitled “Retrieval as neuronal reinstatement,” we have the main part of the authors' ideas about how memory retrieval might work in a brain. Get beyond the dense layers of jargon, digressions and circumlocutions, and we find very little of substance. Their basic idea is that “natural retrieval cues reactivate neural ensembles active at encoding.” “Encoding” is a jargon term used by neuroscientists to describe some process that allegedly occurs when learned information is translated into neural states or synapse states. Despite the fact that the term “encoding” has been constantly used in scientific papers, we have neither any good evidence that such encoding occurs (in the sense of knowledge being translated into neural or synapse states), nor any coherent theory as to how it possibly could occur (there being an ocean of difficulties in the idea that human experience or conceptual knowledge could ever be translated into neural states). We merely have evidence that human beings remember things.

Neuroscientists so often use the term “encoding” that one way to interpret the word is to simply use it as a synonym for learning or memory acquisition. So using that interpretation, we can regard “natural retrieval cues reactivate neural ensembles active at encoding” as simply meaning “when you recall something, your brain reactivates some part of the brain that you used in learning the thing or experiencing the thing recalled.”

When we consider how a brain works, and the fact that all parts of it are constantly active, we can realize that such an explanation for memory retrieval is vacuous or untenable. All neurons in the human brain are constantly firing. Each neuron fires multiple times per minute. So we cannot at all explain a memory recollection as being a case where some tiny part of the brain was “activated,” as if that tiny part was the only part active. All neurons are constantly active.

Brain scanning studies contradict the claim that some little part of the brain (where some memory might be stored) is activated to a higher degree during memory recall.  Excluding the visual cortex that may be to used to kind of visually enhance some memory that was retrieved, such studies show that when humans recall things, there is no brain area that has even a 1% greater activation than any other brain area. Here are some specific numbers from particular studies:

• This brain scan study was entitled “Working Memory Retrieval: Contributions of the Left Prefrontal Cortex, the Left Posterior Parietal Cortex, and the Hippocampus.” Figure 4 and Figure 5 of the study shows that none of the memory retrievals produced more than a .3 percent signal change, so they all involved signal changes of less than 1 part in 333.
• In this study, brain scans were done during recognition activities, looking for signs of increased brain activity in the hippocampus, a region of the brain often described as some center of brain memory involvement. But the percent signal change is never more than .2 percent, that is, never more than 1 part in 500.
• The paper here is entitled, “Functional-anatomic correlates of remembering and knowing.” It shows a graph showing a percent signal change in the brain during memory retrieval that is no greater than .3 percent, less than 1 part in 300.
• The paper here is entitled “The neural correlates of specific versus general autobiographical memory construction and elaboration.” It shows various graphs showing a percent signal change in the brain during memory retrieval that is no greater than .07 percent, less than 1 part in 1000.
• The paper here is entitled “Neural correlates of true memory, false memory, and deception." It shows various graphs showing a percent signal change during memory retrieval that is no greater than .4 percent, 1 part in 250.
• This paper did a review of 12 other brain scanning studies pertaining to the neural correlates of recollection. Figure 3 of the paper shows an average signal change for different parts of the brain of only about .4 percent, 1 part in 250.
• This paper was entitled “Neural correlates of emotional memories: a review of evidence from brain imaging studies.” We learn from Figure 2 that none of the percent signal changes were greater than .4 percent,  1 part in 250.
• This study was entitled “Sex Differences in the Neural Correlates of Specific and General Autobiographical Memory.” Figure 2 shows that none of the differences in brain activity (for men or women) involved a percent signal change of more than .3 percent or 1 part in 333.

So it simply is not true that when you recall something, there is some substantially greater activation of some region of your brain where the memory is stored.  The claim that "natural retrieval cues reactivate neural ensembles active at encoding" basically means merely "your brain uses the information that it stored somewhere," but such an idea doesn't explain how a human brain supposedly storing very many thousands or millions of learned items of information could ever instantly find just the right neurons to use to cause you to instantly recall just the right piece of information when you are asked a specific question such as "What jobs did Ulysses Grant have?" 

There are many seemingly insurmountable problems that would have to be tackled by any theory of neural memory retrieval. The first is what I call the navigation problem. This is the problem that if a memory were to be stored on some exact tiny spot on the brain, it would seem that there would be no way for a brain to instantly find just that little spot. For that to occur would be like someone instantly finding a needle in a mountain-sized haystack, or like someone instantly finding just the right book in a vast library in which books were shelved in random positions. Neurons are not addressable, and have no neuron numbers or neuron addresses. So, for example, we cannot imagine that the brain instantly finds your memory image of Marilyn Monroe (when you hear her name) because the brain knows that such information is stored at neural location #239355235.  There are no such "neural addresses" in the brain. 

Then there is also the fact that the brain seems to have nothing like a read mechanism by which some small group of neurons are given special attention. The hard disk of a computer has a read/write head, but there's nothing like that in the brain. 

Then there is the fact that if memory information were encoded into neural states, the brain would have to decode that encoded information; but such a decoding would seem to require time that would prevent instantaneous recall. When cells do vastly simpler decoding involved in decoding DNA information, it takes cells many seconds or minutes. We would expect that any decoding of encoded information stored in a brain would take many seconds or minutes, preventing any such thing as instantaneous recall of rarely-remembered data items. In addition, we have not the slightest idea of how human learned information (with so many diverse forms) could either be translated or encoded into neural states, or decoded back into thoughts once such translated or encoded knowledge was decoded.  There exist hundreds of genes for the relatively simple job of decoding the genetic information in DNA. If human learned information and experiences (with so many diverse forms) were to be translated into neural or synapse states, so that learned information could be stored in a brain, there would need to be many hundreds or thousands of genes and proteins devoted to so complex a task. But no such genes and proteins seem to exist, and no one has proven that any gene or protein is dedicated to the task of memory encoding or decoding. 

None of these problems are addressed by the paper "The neurobiological foundation of memory retrieval."  The authors simply ignore the whole speed problem of explaining instant memory recall.  Their paper makes no mention of such a thing, and doesn't use words such as "speed" or "quick" or "fast" or "instant" or "instantaneous."  The authors also ignore the issue of how a brain could decode (during memory retrieval) encoded information stored in a brain. Their paper does not use the words "decode," "decoding" or "translate."  The paper merely refers in passing to some research they claim has "potentially interesting translational implications," but give no details to clarify such a claim.  Nor does the paper have any discussion of some theory of a read mechanism that could be used to read memories from brains. Searching for the word "read" in the paper produces no relevant sentences. 

Any real theory of a neural retrieval of memories would have to also be a theory of the storage and encoding of such memories. There can be no understanding of how some memories could be read from neurons or synapses or decoded unless you had an understanding of how such memories were stored and encoded in neurons or synapses.  But the paper "The neurobiological foundation of memory retrieval" gives no theory of how a brain could store learned information. The paper does make quite a few uses of the word "encoding," but simply uses that as a synonym for "learning" or "memory acquisition" without doing anything to explain how learned information could be translated into neural states. 

So the paper claiming to elucidate a "neurobiological foundation of memory retrieval" fails to discuss in any substantive way any of the main things that would need to be explained by an actual theory or understanding of how a brain could retrieve a memory: (1) how a brain could instantly find just the right tiny engram where a memory was stored in it; (2) how a brain could read information stored in it; (3) how a brain could perform the miracle of instantly decoding such learned information that had been encoded in neural states or synapse states, acting 1000 times faster than cells do when they decode DNA information; (4) what miracle of translation would have allowed information so diverse to ever have been encoded as neural states or synapse states in the first place. The paper is additional evidence that our scientists have no actual understanding of how a brain could instantly retrieve a memory. There does not exist any such thing as a "neurobiological foundation of memory retrieval." Humans and animals remember things, but neither scanning their brains during memory activity nor rat experiments provide any insight as to how instantaneous recall of specific learned items (or any recall at all of such items) can occur. 

The lack of any real understanding on this matter is almost admitted by the paper in question, which states at its end, "Our understanding of the neurobiological underpinnings of retrieval remains rudimentary." That is not how it would be in the year 2020 (70 years after the discovery of DNA) if human brains actually performed memory retrieval.  In a brain that stored and retrieved memories, there would have been signs of its memory storage and retrieval mechanism discoverable around 1950; and around the same time we discovered the readable microscopic encoded information in DNA, around 1950, we would have discovered readable encoded memory information in brains (something which still has not been found).  Instead of finding any evidence for proteins dedicated to encoding memories,  which would have to exist in massive numbers if a brain stored memories, what was found was that the proteins in synapses (the alleged storage place of memories) have lifetimes 1000 times shorter than the maximum age of human memories.