Why 2 New Studies Fail to Show Memories Are Stored in Brains

There is no real evidence that memories are stored in brains, and there are some very good reasons for thinking that long-term memories cannot be stored in brains. Among such reasons are the following:

1. the fact that no one has ever found any sign of encoded information in a brain (other than the genetic information in DNA);
   
2. the fact that the best candidates for a storage place for memory (synapses and dendriditic spines) are unstable structures that don't last for years, and the fact that synapses and dendritic spines are made of proteins with short average lifetimes of only a few weeks;
   
3. the fact that the brain has no addressing system that might allow the exact tiny location of a memory to be instantly found, and make possible the instantaneous recall that humans display;
   
4. the fact that brains and synapses are very noisy and slow, making them unsuitable as an explanation for memory recall that can occur instantly and with 100% reliability.
   

But scientists occasionally produce research papers trying to persuade us that memories can be stored in a brain. In this post, I will discuss two such papers that appeared in the news this week, and explain why they are not good evidence that memories are stored in brains.

Study #1: Another Optogenetic Study on Memory

This week a paper was published with the grandiose title “Artificially Enhancing and Suppressing Hippocampus-Mediated Memories.” The paper was authored by Ramirez and others. There were two experiments. In one experiment, mice were exposed to training to give them a memory regarding a stimulus, and an attempt was made to reactivate that memory by exposing the mice to the same stimulus. The brains of the mice were then examined, and the scientists claimed to found evidence that some part of the brain “showed significant increases in re-activated cells.” This type of finding is not good evidence for anything, because judgment of whether a cell has been "reactivated" is arbitrary, the very notion of "cell reactivation" is not a precise one, and any one looking for an evidence of a significant increase in something (and freedom to check many different areas) will always be able to find something. For example, if I have freedom to check 100 different parameters related to your body and chemistry after a physical and a blood test is performed on you, I will always be able to find some things that made a “significant increase.”

In another experiment, the scientists trained some mice to fear a stimulus, and then sent an optogenetic stimulation to a part of the brain that the scientists guessed was the storage place of the memory. The optogenetic stimulation can roughly be described as a brain zap. The scientists claimed that this resulted in increased “freezing behavior” suggesting that a fear memory had been artificially activated. 

There are two major reasons why the methodology is flawed. The first reason is that when part of an animal's brain is zapped, we have no idea whether a fear memory is being recalled after such a brain zapping. The alleged “increased freezing behavior” could be caused by the brain stimulation itself rather than a recall of a fear memory. A science paper says that it is possible to induce freezing in rodents by stimulating a wide variety of regions. It says, "It is possible to induce freezing by activating a variety of brain areas and projections, including the hippocampus (Liu et al., 2012), lateral, basal and central amygdala (Ciocchi et al., 2010); Johansen et al., 2010;  Gore et al., 2015a), periaqueductal gray (Tovote et al., 2016), motor and primary sensory cortices (Kass et al., 2013), prefrontal projections (Rajasethupathy et al., 2015) and retrosplenial cortex (Cowansage et al., 2014).”

The second reason such a methodology is dubious is that it is hard to reliably measure freezing behavior in rodents in an objective, standardized way without arbitrary judgment factors being involved. Consider a scientist doing an experiment that hinges upon an analysis of freezing behavior in rodents. There would be lots of arbitrary choices to be made, such as the following:

1. What technique should be used to measure the degree of freezing in the rodent's movement? If automated software is used, what startup parameters or program inputs should be used? Such parameters and inputs can have a huge effect on the output of the program.
   
2. What exactly should be counted as an indication that an animal's movement has frozen – complete lack of motion, no more than a slight movement, or no major movement?
   
3. If the experiment is testing avoidance of some area where the animal previously experienced a fear stimulus, how far away should the animal be from that area for it count as an example of freezing?
   
4. What time frame should be used in measuring the freezing? Should freezing only be considered exactly at the instant some brain stimulation is produced, or should the measurement time span be some length of time following that? If the latter, how long a time frame should be used?

With all these parameters for an experimenter to modify or vary as he wishes, it would not be too hard for anyone wanting to show “increased freezing behavior” to do so, just by modifying the parameters or how the data is accumulated, until the desired result is achieved. So in general, “increased freezing behavior” is not a very persuasive scientific result. It's too easy to get such a result just by fiddling with the measurement technique or data analysis details until an experimenter gets what he wants.

As discussed in this paper, randomization and blinding techniques are a very important scientific technique for avoiding experimenter bias. For example, what is called the “gold standard” in experimental drug studies is a type of study called a double-blind, randomized experiment. In such a study, both the doctors or scientific staff handing out pills and the subjects taking the pills do not know whether the pills are the medicine being tested or a placebo with no effect.

A memory experiment can be very carefully designed to achieve this blind randomization ideal that minimizes the chance of experimenter bias, so that the person measuring something like “freezing behavior” does not know whether the animal in question is or is not a member of the control group. Such a thing is a very important guard against experimenter bias. But such a thing is almost never done is any of the memory experiments purporting to show evidence of a brain storage of memories. And it wasn't done in the Ramirez study I am discussing. So the “gold standard” isn't being met. 

Perhaps the biggest problem in the study is its inadequate sample sizes. The study used study groups with 11 or 12 animals. It is well known that a minimum of 15 animals per study group is needed to achieve a modestly reliable result in an animal study. A 2017 study  entitled "Effect size and statistical power in the rodent fear conditioning literature -- A systematic review" looked at what percentage of 410 experiments used the standard of 15 animals per study group. The study found that only 12 percent of the experiments met such a standard. What this basically means is that 88 percent of the experiments had low statistical power, and are not compelling evidence for anything.

In her post “Why Most Published Neuroscience Findings Are False,” Kelly Zalocusky PhD calculates (using Ioannidis’s data) that the median effect size of neuroscience studies is about .51. She then states the following, talking about statistical power:

To get a power of 0.2, with an effect size of 0.51, the sample size needs to be 12 per group. This fits well with my intuition of sample sizes in (behavioral) neuroscience, and might actually be a little generous. To bump our power up to 0.5, we would need an n of 31 per group. A power of 0.8 would require 60 per group.

If we describe a power of .5 as being moderately convincing, it therefore seems that 31 animals per study group is needed for a neuroscience study to be moderately convincing. But most experimental neuroscience studies involving rodents and memory (including this week's Ramirez study) use far fewer than 15 animals per study group.

Zalocusky states the following:

If our intuitions about our research are true, fellow graduate students, then fully 70% of published positive findings are “false positives”.This result furthermore assumes no bias, perfect use of statistics, and a complete lack of “many groups” effect. (The “many groups” effect means that many groups might work on the same question. 19 out of 20 find nothing, and the 1 “lucky” group that finds something actually publishes). Meaning—this estimate is likely to be hugely optimistic.

In short, we have multiple reasons to suspect that the latest memory study by Ramirez is simply yet another neuroscience false positive. The study doesn't do anything to show that memory is stored in mouse brains; and even if you showed such a thing, it wouldn't prove that memories are stored in human brains. 

Study #2: Zapping Human Brains to Try to Improve Recall

Now let's look a very different type of study announced this week.  A UCLA press release announced a study in which people supposedly had improved memory recall after getting electrical stimulation in a brain region called "the left rostrolateral prefrontal cortex."

Each study group consisted of 24 people, who were shown 80 words on a computer screen, and asked to memorize them. The next day they were asked to recall the words, after some gizmo had been attached to their head.  One study group received 30 minutes of real electrical stimulation to their head. The other study group received only an instant of electrical stimulation at the beginning of the 30 minutes.  Supposedly the group that received the prolonged electrical stimulation scored 15% better at recalling the words. 

But it is very easy to explain this result without believing the prefrontal cortex of the brain had anything to do with it. The press release tells us that the brain stimulation produces "a warm, mild tingling sensation."  So from the way the study was designed, the people getting the actual stimulation could easily have been able to tell that they were in the "real brain stimulation" group and not the fake stimulation group (the control group). 

Imagine you're a participant in the study, and you feel minutes of real brain stimulation.  You may think to yourself something like: wow, I'm getting the real stuff, I'd better think hard, they're going to be expecting me to recall better.  But if you were in the control group, getting only an instant of brain stimulation, you might think, "Oh, I'm just in the control group, no need to recall that hard."  This could easily account for a 15% difference in the scores. 

Another possibility is that the brain stimulation did nothing at all to increase actual memory recall, but simply increased alertness. So the brain stimulation might have acted like a large cup of coffee acts to increase the alertness of someone in the morning.  We can imagine all kinds of reasons why alertness might have been increased by someone who was receiving an actual electrical stimulation of the brain.  Extra alertness could easily account for a 15% difference in scores. 

I may note that the results are also entirely compatible with the idea that memories are not stored in your brain. It could be that rather than being a device for retrieving memories, that the brain is largely a device for restricting our memory and consciousness, and that the brain acts kind of like a valve. Now suppose you electrically stimulate some part of the brain. That may somehow reduce this "valve effect," resulting in an increase in memory recall. 

If the brain actually were a device for retrieving memories, there is no reason why electrical stimulation in some part of the brain would cause recall of memory to speed up, just as there would be no reason why electrically zapping some part of a computer would cause it to increase the speed at which it retrieved information.  In previous posts I have specified the kind of reasons why the brain is not a plausible candidate for a memory recall system. These include reasons such as the lack of any addressing system in the brain allowing the precise location of a memory to be instantly found,  the fact that brains and synapses are very noisy (too noisy for reliable signal transmission during memory retrieval, as discussed here), and the fact that signal transmission in the brain (greatly slowed by synaptic delays and synaptic fatigue) is actually far too slow to account for instantaneous recall of memories (as discussed here).  Sending in some extra electricity to the brain would not help with any of these problems.  In fact, it would make one of them much worse (the noise problem), because the added electricity would just be more noise for a brain that already has way too much electrical noise inside it for it to be a plausible candidate for a device that instantly retrieves memories reliably.  

What brain electrical stimulation really is