# Throwing away muscle memory to make brain backups smaller. Would it work?

Overlord is the best 5-episode anime, as I only acknowledge the existence of episodes 14-18, the rest is gorehound edge bait I can only recommend to people who are still flaming about the titan CGI. However, with what little I left on my hard drive, there was one thing still interesting:

The protagonist gets killed and then revived with bs. Now, revival via bs has the side effect of reducing your EXP and level, which means you lose skills.

While this is there because almost every damn isekai needs to have bs game mechanics in their setting (never The Hat Economy, though), I thought I'd make a reasonable explanation for that.

The idea is simple: since ISPs are rather stingy about giving me bandwidth, I decided to expand more on mind compression.

Initially, via specialized hardware, data structures, and lossless compression, and in the most recent step: lossy compression. Those ISP company executives weren't too happy about being lab rats, though.

Much like how you can still recognize an image of a rose with medium jpeg compression, you will still recognize your favorite athlete as the same person, even after they suffer a horrible injury and can only walk with crutches.

Of course, stripping muscle memory from the backup is only partial, giving resurrectees "steps" to climb back on.

However, I'm unsure if there is a point to that, or if muscle memory is too small to matter in compression ratios.

Do we know enough about the human brain to answer this question? If yes, would striping away some muscle memory significantly decrease the amount of data we would have to store?

• Do I have to learn to walk, talk, write, and hand-eye coordination all over again? Walking is one thing...but what about finely honed skills of people like musicians, dancers, artists, craftsmen, and athletes? – DKNguyen Jan 29 at 23:22
• The question spends eight paragraphs in meaningless technobabble. When finally an intelligible question is asked, it is asked without essential elements which would be required to make it answerable. Most importantly, the question asks about the size of a backup without telling what data is to be backed up. No, "the brain" cannot be backed up -- the brain is a physical object, and we cannot make backups of physical objects. We can only make backups of data; thus it is important to explain how the brain state is to be represented numerically, to take a backup the representation. – AlexP Jan 30 at 0:05
• (Nitpicks: Speaking as an IT professional: first, there is no such thing as a lossy backup. By definition, a backup must provide the possibility of restoring the data exactly as it was at the point when the backup was taken. Second, the entire point of lossy encodings such as JPEG or MP3 is that they exploit what is known of the operation of the human visual and auditory sensory systems, so that a human would consider the original and the compressed image or sound to be almost indistinguishable. About brains we know very very very much less than about psychoacoustics.) – AlexP Jan 30 at 0:07
• While I understand (kinda) the lyrical metaphor behind it, comparing muscle memory to a numerical compression on the science point of view doesn't really make sense x_x. I think the closest you can get brains to computers are deep-learning AIs, but as far as I understand it's more about calculations, than data storage. That's, uh, problematic since you tagged your question as science-based... If you go further for the "compression", you should expand on how it would work <(^^)> – Tortliena Jan 30 at 0:28
• imagine a punchcard actually no, think of a carpark coupon... a memory is one of these holes and muscle memory are those coupons you tried to reuse by taping back the holes ;D – user6760 Jan 30 at 0:40

To be determined

Computers and data are just the latest analogue for memory and other human functions. Although we have good ideas how things work, we don't know.

Problems with the comparison of computer to brain arise quickly. Example: looking at objects light up parts of the occipital lobe. More or less (mostly less) we can determine objects seen inside the brain by the pattern that lights up. Interestingly, looking at faces can light up identical areas. How can you still differentiate two brothers if the exact same area's are activated? Each brain part isn't just activating. It is activating in a certain rhythm. Akin to morse code representing information, this rhythm from the neurons helps convey information.

Now the problem. How do you store this data on a computer or whatever you want? Besides questions like what kind of data storage and compression you might use, how do you translate the brain patterns and rhythm into data? Muscle memory is stored not just in the brain as well, but also in the spine. Interestingly, if you think of grabbing a cup, we can see identical patterns in the motor area, regardless of positioning of the hand. That means that the concept of grabbing the cup is the same, but translated to the correct movements at a later stage.

The creator of XKCD wrote an example I can't find right now, so it's probably in his book "what if". He compares the processing power of humans vs machines. In one way, humans are way, way ahead in power. In another, we've been left behind with some of the simple calculators. This illustrates the point if menory. We don't know how it works, nor how it compares. If we don't know either, how can we tell if removing muscle memory is significant in reducing storage on a computer?

To still give you an answer, it might not be as bad as you think. Muscle menory doesn't need to be unique. You can store part of the muscle memory for many people. You can then start mixing and matching. Not all combinations can coexist, but the advantage is that you store all of them just once. So you don't need to store muscle menory for catching a ball for each person. You just save it once and add it to the mix if you want.

• Nice solution, somewhat presupposes compatible architecture across individuals - or a degree of adaptability of some sort, which we may well posses. – A Rogue Ant. Jan 30 at 2:04
• You're correct - "Human Computer" is in the book of What If but not on the website. – KerrAvon2055 Jan 30 at 4:14
• I like the idea of muscle memory transplants - if you were to implement them in a video game, I suggest you do it as a speed boost to XP gain rather than as a one-time XP bonus. In theory you now know how to make a roundhouse kick - but you still need to figure out how the legendary martial artist's brain called the maneuver so that you know how to access it - and then tweak some settings so that you don't tear your muscles each time you try the move, or undershoot each kick because the artist had longer legs than you do. – John Dvorak Jan 31 at 13:55
• @Tantalus'touch. Neuroplasticity should fix it. Neurosignals sometimes start choosing different paths. Besides, if you make a new guy, you can just fix that during the build. – Trioxidane Jan 31 at 13:57
• I think your answer sells modern neuroscience very short, as well as dismisses the idea of the human brain as a computer far too easily. Just because the areas that light up look the same doesn't mean they are behaving the same. Our current MRI capabilities are far too coarse to recognize the full extent of whats happening in a brain with 100 billion neurons. And dismissing the ability to do what OP suggests because we simply don't know how is not really much of an answer as to its possibility. – stix Feb 4 at 17:13

This idea actually makes sense.

The state of a neural network is both in its nodes and the connections among such nodes. The nodes are the neurons and the connections are the synapses. Reducing the amount of synapses reduces the raw amount of data to store.

Do notice though that human memory is highly associative. If you remove muscle memory you won't just remove the memory of how to perform a dance or a martial arts technique. If you are used to typing a password without thinking about it, for example, you may forget the password itself. If you knew how to play an instrument, you may forget not only how to olay but also the songs themselves.

Many associations with mirror neurons may be lost as well, so not only you forget how to do a roundhouse kick, you might also be unable to name the technique when you see someone doing it.

• Meaning a programmer or an electronics engineer might turn half useless after restoring. Maybe not forgetting the actual techniques and what not, but if forgetting how to type makes you forget the password, it could actually make one forget the programming language. – Jan Dorniak Feb 4 at 1:46
• @jandorniak typing some keywords is so 8ngrained into my muscle memory that what youbl said makes more than absolute sense. I might forget the usual kinds of variables in strongly typed languages. – The Square-Cube Law Feb 4 at 3:10
• yeah, you just think "I want a 32 bit integer here" and int32_t appears on screen – Jan Dorniak Feb 4 at 3:11

The majority of "muscle memory" is stored in the cerebellum and spinal cord, which is completely different from the main memory you have in your cortex, so yes, it's feasible if your resurrection Macguffin only handles cortex memory and ignores the cerebellum for whatever reason.

Essentially, the cerebellum and the spinal cord store "movement programs" that are triggered by the higher brain functions. This is the unconcious muscle memory you're talking about. After all, you don't think about all of the coordinated movements of your legs and body when walking, you just "do" and tell your body where to go. There's a lot of computation and coordination going on completely under the hood and far from the higher brain just to do what we take for granted every day.

Scientists have actually triggered the "walk program" of the spinal cord in paraplegics to give them some function back in their legs, however, without the cerebellum to coordinate, it's a very jerky/stiff sort of walk. There is also the very creepy "Lazarus sign," which is a reflex in brain dead patients where the spinal cord makes their arms rise up and fold across their chest, all without any higher brain or cerebellar input. All of these examples are muscle memory and have nothing to do with the "important" "person-generating" part of the brain.

It's also interesting to note that the cerebellum has more neurons than the whole rest of your brain, anywhere from 50% to 80% depending on the source of how many neurons are in the brain, yet it isn't conscious and can be cut and damaged without harming the "person software," so it does make a lot of sense that trying to backup the cerebellum would be more difficult than the cortex.

So your concern about muscle memory being "too small" to save much by ignoring it doesn't really hold, as muscle memory is arguably the majority of what's going on in the central nervous system as a whole. In other words, you could actually save a lot of space by ignoring muscle memory.

You could also make some case about it being difficult to backup neurons in the spinal cord simply because whatever Macguffin you use to do the backing up cant reach all the way down into the spinal cord (say it's a helmet that sits on top of the head or something).

• This is the technically correct answer. – chasly - supports Monica Feb 4 at 13:00
• @chasly-supportsMonica the best kind of correct. – stix Feb 4 at 16:56

Muscle memory is a form of procedural memory that involves consolidating a specific motor task into memory through repetition, ...

Muscle memory is integral to physical functioning: walking, talking, moving an arm, opening and closing a mouth, rolling over, tying shoe laces, kissing, holding a book, playing a musical instrument, scratching an itch, putting on clothes and disrobing, etc.

If it were possible to store all but muscle memory, the revived person would have to relearn how to physically function.

Muscle memory is not associated with particularly parts of the brain, it appears to be diffuse.

It is difficult to display cases of "pure" motor memory impairment because the memory system is so widespread throughout the brain that damage is not often isolated to one specific type of memory.

so, isolating a muscle memory to omit from storage is not going to be possible.

### We don't know enough

The storage required for a brain backup that can be restored to new hardware is filled with known unknowns and at least one unknowable unknown. The unknowable unknown comes from lack of understanding of how lipid layers and sugars in cells interact with each other and proteins. The lipid layers and sugars may or may be involved.

### Premature Optimization

You are advocating for the optimization of a very complex problem before the problem is understood. In software development this is considered a bad idea. This is a concern even for software in the 10k lines level of complexity. Brains are at least eight perhaps twenty orders of magnitude more complex. With the way every ones body has different configurations it is nearly guaranteed that where each persons muscle memory will have enough variation that one size will not fit all.

There is evidence to suggest that brain cells add more DNA, more likely in the frontal cortex. There probably is per neuron significant methylation. So likely the storage would include genomes of each and every neuron. Each neuron has however many interconnects plus weightings. That is unlikely to be the whole story, but that's a lot of data, even with compression.

### Unethical / Torture

Additionally even if you could somehow remove only muscle memory. What muscle memory? Don't forget things like how to shape your mouth for speech is included in there. I would say that any person created with a backup without any muscle memory at all is a victim of torture. For while they have a mouth, they need to learn how to scream.

### Summary: No

Storing a person's mind is a complex difficult problem. By the time that is solved storage volume would be a trivial technical issue. Deciding on a subset to store is an additional complex risk filled problem.

My answer: No. It makes no sense ethically, economically, or practically.

edit: removed claim of estimated ratio of muscle memory vs other.

• Given that the cerebellum is 70% of the brain by neuron count, and the cerebellum is 90% concerned with muscle memory, the majority of the total amount of things to store is going to be muscle memory, not less. – stix Feb 4 at 17:18