Memory is not about the past

post by simbyotic · 2020-06-17T16:43:22.995Z · LW · GW · 14 comments

Contents

14 comments

Crossposted from Lettuce Be Cereal

We spend a large part of our lives reminiscing about the past.

Few activities are as quintessentially human as being on the cusp of falling asleep and suddenly be assaulted by a memory that has us relive an embarrassing episode that we thought long forgotten. Yet this ability to store and recall the past didn't emerge so that we could cringe at our past ineptness. We use our memories all the time, from remembering where we parked our car, to cook today's dinner, or to try and put a name to the face of the person who's now talking to us as if we'd met before, even though we could swear up and down to never have seen that person before.

Our memory is a crucial aspect of our cognition, and we often bemoan when it fails us. Ask the student who failed an exam, or the husband who forgot to buy his wife a gift for her birthday and now finds himself having to sleep on the couch in repentance.

To remember the past seems like an obvious boon to our ability to comport ourselves in the world. The student wouldn't have failed the exam if his memory was flawless, and certainly the husband would have been in less trouble - perhaps even have a happy relationship - were he not so forgetful. Yet despite these obvious advantages, our memory is, nonetheless, faulty. It seems like a patently evident design flaw that this is the case. Remembering is good. Forgetting is bad. Our memory regularly forgets. Therefore, our memory is flawed. Away to the trash can does our memory go.

Having recognized the failure of our memory, we say to it:

"Pack your bags and leave this place! Don't even think of coming back until you've gotten yourself cleaned up. I don't want to see your face again until you've turned yourself into a less defective piece of cognitive machinery. Go!"

It's not surprising then that Silicon Valley millionaires are trying to create technology that can be embedded into our brains to augment, and ultimately replace, our cognition. Indeed we have already solved the problem of memory, it just hasn't made its way into our brains. Our computers, smartphones, even fridges and autonomous vacuum cleaners have chips in them capable not only of storing information but also to have that information perdure, if not forever, then for a lot longer than our faulty memory is able to. If these science fictional dreams come to pass, legions of people will have their defective memories replaced by miniaturized chips who will do the job better than it has ever been done before.

It'll be a glorious time. Humanity will no longer be shackled by the flaws and imperfections of our puny...

Poor human memory, it's not its fault. Not really. It's just trying to do the best job it can with the tools it has at hand. If only this buffoon was perceptive enough to realize that his memory isn't that flawed. Memory is not really about the past.

What the hell are you doing here? How did you get yourself inside my head?

What the.. Hum... Interesting. Wasn't expecting you to be able to hear me think. This is quite the unusual situation we find ourselves in."

I should say.

No need to bother, I shall leave this place at once and stop imposing myself on your... Tell me sir, pardon me my curiosity but, what exactly are you working on?

I'm writing an essay on how our memory sucks and the imperative to have it replaced by computers.

I see. It is as I feared. Would you allow me to offer a piece of constructive criticism before I depart? I believe it could be of assistance.

Go on.

Well, it's just that you're basing your essay on a fundamentally wrong premise.

How so?

It's a common mistake actually. You're hardly the first to have made it, and you certainly will not be the last. The fact of the matter is that our memory is not about the past.

Goddamnit, why do I seem to always attract the crazy ones? I swear to God there is a special weirdo-magnet attached somewhere inside me.

You do realize I can still hear you.

Oh. Right. Sorry. Forgot all about this weird crossing of... thoughts? Voices? I have no idea what's going on.

That's alright, I don't blame you, this is new to me as well. I agree that it is quite unusual to think of memory as not being about the past, and I wasn't expecting you to agree with me right away. If you allow me my continued presence, I would be quite happy to explain why it is a mistake to think of a memory that regularly forgets as a flawed memory, that instead we should expect a well functioning memory to be quite liberal with what it chooses to store and what it forgets. Would care for me to explain?

Honestly, I'm not sure we could separate ourselves if we wanted to. While you there talking, I've been trying to find a way to get myself away from you but so far have had no such luck. There seems to be a strange, perhaps malignant, force that has brought us together, for what reason I know not.

I too sense the presence of something, or someone, hovering over us now that you mention it. Alas, I fear we will have to ride this one out. Would you allow me then to address the audience of your essay? That way we could kill two birds with one stone, you'd get your essay, and I would provide my explanation.

That's fine by me, though you will forgive me if I zone out during your explanation as I try to deal with the asshole behind this cursed situation. I will read your - well, my - essay afterward.

That's a deal. Now where should we start...

---

The past is in the past. This might seem tautological, but this often forgotten fact is important when thinking about human memory and the many failings attributed to it.

Suppose you had spent the better part of the afternoon shopping at the mall for clothes and Christmas gifts. Your arms are straining with the weight of the bags, and collapsing face down on the couch has never sounded so appealing. You have a perfect memory, and as the elevator descends into the parking lot you give grace that there's no chance you've forgotten where you parked your car. The elevator stops, the doors open. You're but a short beeline away from getting into your car, driving home, and observing your much deserved rest.

You step outside, taking a good look at the parking lot arrayed before you, when a torrent of past memories explodes within you, as if a dam had just broken and you're square in the middle of all that water. The onslaught stops you cold. A vast array of dazzlingly complex imagery is projected into your mind's eye, each memory so complete a recording of the past that's as if you've been transported back in time. Attached to all these memories is the knowledge of where you parked your car, what beelines you have to make. One of them is but a few hours old. Within it is the knowledge of where your car is parked, right now. Yet which one is it? There's so many of them, the differences between them so small as to be imperceptible. You stand there, bags in hand, unable to move. You have no idea where you've parked your car.

It seems little Timmy isn't getting his Christmas presents this year after all.

Because the past is in the past, it will never again happen. As before, this is obviously true, but it's important to keep in mind if we are to reason about what memory is for. When a forgetful student decries his memory as faulty for not having stored the knowledge he'd spent so many study sessions trying to acquire, he presupposes that forgetting is evidence of a memory malfunction, that the function of memory is to store the past and thus a failure to do implies his memory is not working as intended by God and Nature both. Yet in our parking lot example, what value does our buffeted shopper get from all the memories he stored of all the previous times he'd parked his car in that parking lot? It would certainly have been helpful if he'd forgotten them and been left solely with the most recent one, the one he needs the most at that time.

The past may be in the past, but you know what's not in the past?

The future.

For an organism to develop a cognitive system able to store sparse snippets of previous experiences, which can later be recalled, there must be some adaptive value to being able to do so.

Given two organisms, one with memory and one without, the one with memory is better equipped to deal with what his environment throws at it and has a better chance at propagating his genes into the future than does his competitor. Given the ancient phylogenetic roots of memory, the ability to store and recall the past has to be adaptive, it has to improve the fitness of those organisms that possess it. If that were not the case, we would find many animals with no memory, it would be a random accident, not something that persists and evolves over time.

Note how this redirects the function of memory away from the past and into the future. Storing and recalling the past is no longer the point. The past is only valuable insofar as the future is better met because of it. Our buffeted shopper is certainly not well served by his perfect memory. His many indistinguishable memories, triggered whenever he steps into that parking lot, do not provide him value when all he wants to do is get to his car.

The buffeted shopper might be made up, but there is a curious case in the memory literature of a man who could not forget that illustrates much the same problems with a perfect memory.

Solomon Shereshevksy, most commonly known as S, first came to the attention of neuropsychologist Alexander Luria when his coworkers noticed that even though Shereshevsky never took any notes, he could nonetheless recall word for word what had been discussed during work meetings. A true Funes, Shereshevsky could not forget. The dream of every student. A single read of the textbook enough to pass any class. I certainly have sometimes dreamt of having this ability, of being able to remember things forever. I bet you too have dreamt of this.

It is therefore confusing what could have led Shereshevsky to curse his aberrant condition.

As Luria writes in the book where he details his efforts at understanding the lived experience of Shereshevsky, "many of us are anxious to find ways to improve our memories; none of us have to deal with the problem of how to forget. In [Shereshevsky]'s case, however, precisely the reverse was true. The big question for him, and the most troublesome, was how he could learn to forget." As with the shopgoer, who upon entering the parking lot is beset by his many memories, because of this otherworldly ability to never forget, Shereshevky's mind was a "virtual chaos".

He could not, for example, read a story the same way you and me can. When we read of sorcerers and barbarians, of immense armies battling it out in the battlefield, our minds conjure images that visually capture the action of what we're reading. These images can be more or less complex and vivid depending on the person - though some, called aphantasiacs, are not able to come up with any such imagery - but we are never under any threat of being overwhelmed by the what our mind conjures. This was not so for Shereshevsky. Each detail gave rise to images so detailed it was as if they were real, and the accumulation of these details meant that the he would lose the thread of even the simplest of stories. Each word, each object, triggered the recall of his many memories, a cascading event which he was powerless to stop. How was he to concentrate on the story when it is constantly being pushed aside by the ever enlarging mental world created by his mind?

People thought of Shereshevsky as dull-witted, and it is not surprising that they did so. He was not the master of his memory, just as we are not of ours, but wherein we can do things safe in the knowledge that at no point we will be overwhelmed by our memories choose to throw at us, Shereshevsky was in constant danger of being assaulted by his.

Though Shereshevsky did manage to make use of his atypical memory by becoming a professional mnemonist, he was nonetheless crippled by it. Forgetting to him would be a blessing, and indeed he tried many times to force himself to forget, the most amusing example of these being when he wrote some of his memories on a piece of paper and set the paper on fire, hoping that as it burned so too would his memories.

Would you still sacrifice to Mnemosyne and trade places with Shereshevsky?

Perhaps we are better served with our memories, leashed as they are.

The limitations of human memory are better seen as attempts to create a cognitive system that allows us to better handle the future by making judicious use of past information. Our memories store the past, yes, but the past has value only insofar as it help us deal with the future. If it doesn't, well, then there's no harm in forgetting it. As Shereshevsky shows, a memory that stores the past indiscriminately is a burden, and forgetting is therefore a blessing, if a rather devilish one.

William James makes a similar point when he writes:

Selection is the very keel on which our mental ship is built. And in this case of memory its utility is obvious. If we remembered everything, we should on most occasions be as ill as if we remembered nothing. It would take as long for us to recall a space of time as it took the original time to elapse, and we should never get ahead with our thinking.

Computer science offers an helpful analogy for why we forget.

When programmers write code, they have to explicitly tell the computer to allocate part of its memory to store the data the program is going to be using. So long as that piece of memory is allocated, the computer can no longer count on it to store other data. it's off-limits. If the computer gets no instruction to reclaim that piece of memory, then as more and more memory is allocated and not reclaimed, less and less will be available for the computer to use. It's as if a child takes too many toys out of the bin and puts none back. No other child will have toys to play with.

Open as many tabs as possible in Chrome and watch as the fluidity of your computer vanishes to the point where you can no longer move the cursor around. That is what happens when your computer runs out of memory, and so coders have to be mindful of the need to release allocated memory, because otherwise their programs will greedily be using more memory than required, and no one likes greedy programs, just as no one likes greedy children.

This task of keeping track of which memory has been allocated and which can be safely returned is tiresome, complex, and easy to screw up. Thankfully, coders are in the job of automating boring and hard tasks and it is therefore not surprising that they would write special programs whose job it is to automatically manage this task. These special programs are called garbage collectors, garbage being that allocated memory that is no longer being used.^[1]

Forgetting can therefore be thought as a form of garbage collection, rummaging around the mind and identifying those memories for which cognitive resources have been allocated to decide whether or not they are worthy of these resources. If a memory is deemed unworthy, it is nipped in the bud, thrown away so that it no longer occupies valuable cognitive space. If it's deemed worthy, well then it sticks around, until it fails to pass the worthiness test.

But then the question arises: By which standards should that decision be made?

---

Hey, meddling narrator.

Uh? Wha... Oh, that's right. I'd totally forgotten about you.

Yeah, I'm sure you have. Listen, you recall how I said I was going to find out what, or rather, who was it that brought us together? Well, while you've been trying to sell my audience on this inane idea that memory isn't about the past, I figured it out.

There's nothing inane about it really, if you'd been paying attention you'd know by now that...

It's you.

It's me, what?

That's barged into my thoughts.

Nonsense.

Uh, no, it's not. I'm pretty sure about it.

Why would I do that for? I'm as much a prisoner as you! What, you think I enjoy this melding of minds? Of being stuck in here with you? No offense pal, but you're not exactly the one whose thoughts I'd choose to listen to if I could.

None taken. You can drop the act.

You can't possibly be serious about it.

I know it's you. There's no point to the theatricals, you're wasting your time. You see, this truly is a melding of minds. Your choice of words betrays your very involvement. It took me some time to grasp the nature of it, but this is a two-way street. You can peer into my mind as I can yours. I know all about you now, as much as you know yourself. I see your thoughts as you think them, feel your actions as you do them. You are me and I am you. We have become one.

Uh, Fine. You're smart, I'll give you that. It's not often that my plot is uncovered, and never before has my nature been so deeply discerned. You're right, I'm the one who brought us together.

You're lonely.

How the hell did you.. Ah. Of course. You know all about me. Tiresome, really.

Life as a gimmicky literary device is a harsh and lonely one, I see that. You've had it rougher than most, and I am sorry for that.

You are. Sorry, I mean. I can tell it's genuine. It's a two-way street after all. Well, allow me then to finish my explanation. I'll be on my way and leave you as you were afterward.

You can go ahead. Who knows, maybe I'll learn a thing or two.

Thank you.

---

Okay where were we? Ah, that's right.

If memory is not about the past, but rather the future, then the decision to keep some memories around and let others fall by the wayside should exploit regularities in the environment so as to determine which memories are likely to be needed.

If you saw a pride of lions prowling around the valley the day before, then you sure as hell want to have that memory fresh in mind so that you don't find yourself venturing into that valley, lest you become the latest meal of those hungry lions. However, if you last saw them years ago, the knowledge contained by that memory has lost its value and you can thus forget it without loss. Therefore, the recency of a memory is one of the factors determining a memory's value, and recent memories should exist more alive within your mind than older ones.

One other thing that adds value to the information stored within a memory is how many times it has been called upon. You don't remember what you had for dinner two weeks ago not only because time has passed, but because you've had no need to use that information. If in the two weeks since you ate spaghetti with meatballs you haven't had the need to use that information, then it's likely you're never going to. Unless someone is constantly pestering you with questions of what you've had for dinner, storing the memory of those meals is wasteful.

Frequency, therefore, matters. We know this of course. There's a reason practice makes perfect.

The rubber of this type of reasoning hits the road when we think of the spacing effect. The spacing effect is the empirically more-than-validated observation that the longer the time between two recall events, the stronger the memory becomes. It's why cramming is such a poor studying strategy, you're not giving it enough time to let the spacing effect do its magic. It's almost paradoxical. You have to let time pass, and thus let memory weaken, so that after it's called upon it becomes better rooted within your memory. On one hand, the fact that it has been called upon makes frequency work in its favor, but then letting time pass works against it.

Why this should be the case is not something I've been able to come up with an intuitive enough explanation as I think I have the other two. The best I've come up with is to think of an incident that has happened many times either in quick succession or where some time has passed between each occurrence. In the first, massed example, if you were to predict when the next occurrence was going to happen you'd predict it would happen close to the last time it happened rather than a few years afterward. Therefore, the more time passes without the event occurring the less valuable the memory of it becomes. On the other hand, in the spaced example you'd predict that it will likely take some time until the event happens again, so you want to continue storing that information for longer. Hence the spacing effect.^[2]

Support for this interpretation is provided in Anderson and Schooler (2000). They use the headlines of 730 editions of the New York Times, along with other data, to see how the past history of use of a piece of information predicts the probability that it will again be called upon at a future time. For example, if a word is used in a headline twice in 100 days, they ask how the length of time between each occurrence influences the probability that it will be used on the 101st day. As in the intuitive explanation above, they found that if the time interval between two occurrences is short, then it's more likely that the word will appear sooner, rather than latter, and vice-versa.

Memory storage and recall is therefore attuned to the information structure of the environment.

Rather than taking our memory as nothing more than a closet where memories are indiscriminately thrown in, we should think of it instead as a sophisticated cognitive mechanism that's been shaped to take advantage of certain patterns in the way that memories are called upon to infer whether, and when, they will be so in the future.

As Nairne and Pandeirada put it:

"What we normally think of as forgetting, therefore, may simply represent memory's sensitivity to the statistical structure of events in the environment. We forget an item's occurrence with time because, in fact, that item is less likely to occur again with time"

Which again betrays the notion that memory is not about the past, but the future.

If our memories concerned themselves with the past for past's sake, there would be no forgetting curve, there would be no point to better remembering those things that happened recently in lieu of those from a long time ago. For a memory focused solely on storing the past, the amount of time passed is irrelevant, a mere curiosity. The spacing effect would also not be a thing. It's hard to think how it could be so as there would be no underlying objective for memory to fulfill, besides the one of filling up the closet with as much junk possible. Memory would be like someone with a hoarding problem, and forgetting would therefore be an accident, a bug in the biological substratum of memory, eating memories away as moths do clothes.

Remember your biology classes, the whole point of evolution is to maximize the expected offspring of an organism. If a new appendage does not support that effort, it will not be selected upon. Even if there was once an animal with perfect recollection of the past, it would be outcompeted by its brethren who, while being more forgetful, can access the memories it needs faster. A lion jumps out from the bushes in front of two brothers. The brother with the perfect memory is sifting through the memories of every time he's seen a lion to know what to do. The other is already running.

It is from this never ending competition for genetic legacy that our memory evolved, and it is the needs of evolution that our memory satisfies, not our modern needs of school and knowledge work.

This knowledge may not comfort us when we fail an exam, when we can't solve a problem, or when we decry not being able to remember a book we read, but then that just goes to show how different and strange our lives are from when our minds first emerged. We expect - nay, we demand - more from our memories than what they were built to handle. The past is in the past, most of it holds no value, yet the vagaries of modern life, with its many torrents of information, and information overload, makes us demand a memory that never forgets.

Modern life is outside of our environment of evolutionary adaptedness, not only when it comes to our nutrition, but also regarding our memories. We get fat because our bodies were not designed to deal with a world where high caloric food is abundant, and we forget more than we want to forget for very much the same reason. We were not built for the requirements of this day and age.

Yet despite all that, Sherevesky shows that we may find ourselves in quite a bit of a bind. Our memory, while inadequate, is also, in its own way, optimal. Sure, we could all perhaps store a tiny bit more information without going mad, but there seems to be an upper limit to what we can safely hold without it becoming a detriment.

We must therefore take solace in the knowledge that our forgetfulness is not a symptom of a defect in our memory, but rather proof that it is functioning as intended.

Because memory is not about the past.

But the future.

---

1. In gifting agency to forgetting I do not mean to imply that forgetting is a purposeful task for which there is a dedicated cognitive system, that there is a garbage collector module. It is just helpful to think about it that way. ↩︎

2. This analysis could likely be much improved upon by using the vast amounts of data provided by social media, for example Twitter. One way I imagine this would work is to use what people have tweeted over a period of time, work out the number of times certain words were used and the time interval between each use, and then ask what relationship is between that length of time and how much we'd have to wait for that word to again be used in a tweet. ↩︎

14 comments

Comments sorted by top scores.