If this is your first time reading, please check out the overview for Poor Ash’s Almanack, a free, vertically-integrated resource including a latticework of mental models, reviews/notes/analysis on books, guided learning journeys, and more.
Human Memory Mental Model: Executive Summary
If you only have two minutes, this introductory section will get you up to speed on the Human Memory mental model.
Human memory in one quote:
“An appreciation of the distortions of memory, a realization that even deeply felt memories might be wrong, might encourage people to hold their memories more lightly, to drop the certainty that their memories are always accurate, and to let go of the appealing impulse to use the past to justify problems of the present.”
Key takeaways/applications: relying on memory is a bad idea; various methods of improving it are unreliable. The better approach is to use structural problem solving or human-centered design to allow the things we’re good at to augment the things we don’t.
Three brief examples of human memory:
Disasters big and small. Distracted climbers have faced peril because they forgot to tie basic knots; fatal plane crashes have occurred because distracted pilots were too focused on responding to error messages on the instrument panel that they forgot to “FLY THE PLANE.” Every year, people are committed to jail due to faulty eyewitness testimony.
Business and investing mistakes. As I discuss in more depth in a later section, “hindsight bias” can lead us to view good or bad outcomes as “obvious” in hindsight, whereas beforehand it might have (correctly) looked very probabilistic, with a lot of luck involved. This has real consequences for investors and executives, because to learn, we need clear feedback,, and it’s hard to get clear feedback when it’s mediated by our unreliable memories.
Where are the !@#$ing keys?!?!?!?! Need I say more? 🙂
If this sounds interesting/applicable in your life, keep reading for deeper understanding and unexpected applications.
However, if this doesn’t sound like something you need to learn right now, no worries! There’s plenty of other content on Poor Ash’s Almanack that might suit your needs. Instead, consider checking out our learning journeys, our discussion of the vividness, path-dependency, or feedback mental models, or our reviews of great books like “ Ordinary Men” ( OrdM review + notes), “ Uncontainable” ( UCT review + notes), or “ Scale” ( SCALE review + notes).
Human Memory Mental Model: A Deeper Look
“Remembering the past is not merely a matter of activating or awakening a dormant trace or picture…
but instead involves a far more complex interaction between the current environment, what one expects to remember, and what is retained from the past.”
Relative to certain other mental models, human memory bears a little more conceptual discussion before we jump into the interactions. Below are some of the important, practical concepts I’ve learned about how our memories work… or don’t.
The fundamental principle of human memory is that not all information that we encounter is useful; as such, it would be energetically and attentionally inefficient to retain all of it. In other words, as psychologist Daniel Schacter argues convincingly in “The Seven Sins of Memory” (7SOM review + notes), our faulty memories are a generally- adaptive feature, not a bug.
While our attention, focus, and worldview (see: “schema” mental model) dictate what information we actually process, our memory, outside of our conscious control, dictates which of this information is retained for future use.
We have multiple kinds of memory: short-term memory (also known as working memory) allows us to remember very small amounts (single-digit – think five items) of information for very short periods of time, for the purpose of working with it.
Some, but not all, of the information that goes through our short-term memory is encoded into long-term memory, the capacity of which is likely not unlimited, but can store incredible amounts of information.
Although it’s popular to compare human memory to computers, it’s an inaccurate comparison. Long-term memory is not like a hard drive: while it rarely fails completely, it does slowly “leak” over time, very rapidly in the first hours/days but then more slowly after that.
This leakage accelerates with age, although age-related memory loss can be slowed by consistently high levels of cognitive activity (explains how Munger’s still sharp at 90something!)
Chunking. Making small “groups” out of memorization makes it easier to memorize. 5120101234 (a random string of digits) is harder to remember than (512) 010-1234 (the same string of digits as a phone number).
Meaning matters. Our brain generally seems to operate on a heuristic pattern which makes sense in an evolutionary-biology context: for example, highly emotional events are etched more deeply than unemotional events, with negative events being recalled more specifically than positive events.
Frequency + recency matter. Specifically, as any college professor can tell you, memory works elaboratively and repeated time-delayed re-encoding works better than cramming: the more time/effort/focus you spend on something, the more likely you are to remember it; for committing something to long-term memory, repeated use over time seems to be stronger than cramming.
This is why, for example, I still remember the Pythagorean formula or the quadratic equation despite years of non-use – I used them all the time – but can’t tell you anything about any of the stuff I memorized the night before my biochemistry and molecular biology tests – other than platitudes like “mitochondria are the energy powerhouse of the cell,” I don’t even remember what the topics of the exams were!
This is also why I don’t take notes while I read books, even if doing so might be a little more efficient – instead, as I discuss in The Seven Habits of Highly Effective Readers, I just write down the page numbers where I saw something interesting, draw a line in the margin to note the appropriate paragraphs, and come back to the book a few days later to take notes on those pages, etching the concepts into my memory twice.
Memory is both in the head and in the world. This may be surprising at first, but much of what we think of as being “in” our memory is actually “out there” in the world. In Peter Godfrey-Smith’s excellent Other Minds (OthM review + notes), he discusses the concept of “embodied cognition,” in which some of our “intelligence” is actually embedded in the structure of our bodies; the angle of our joints, for example, helps us move automatically.
Similarly, our memory is really an interaction between our brain and the world: Chapter Three of Don Norman’s phenomenal “The Design of Everyday Things” (DOET review + notes) discusses this at length.
Here’s a test: do you actually know what a penny looks like? Which way does Lincoln’s head face? Does the front of the penny say “E Pluribus Unum” or “In God We Trust” at the top? Or does it say “United States of America?” Or is that on the back? Where is the mint mark?
Here’s the punchline: regardless of whether or not you know the answer to any of those, you can still use the penny, because all the details I referenced above are completely meaningless for everyday purposes. all you have to remember is that it’s the little copper-colored coin.
Don’t feel bad: it turns out many Americans can’t pick the penny out of a lineup; my younger numismatist self also occasionally spotted Canadian pennies in circulation in the U.S. on occasion.
And if you’re European, you don’t have to feel left out either: Norman cites the example of France’s 10-franc coin, which was so similar to and easily confused with the half-franc coin (worth 20x less!) that public outrage ensued.
As Norman explains, this type of imperfect memorization isn’t the exception to the rule, but rather than the default:
“most things in the world have a sensible structure, which tremendously simplifies the memory task. When things make sense, they correspond to knowledge that we already have, so the new material can be understood, interpreted, and integrated with previously acquired material.”
Norman, elsewhere, discusses how various kinds of constraints make our lives easier. We don’t have to “remember” where to insert our car key because there’s only one way it’ll fit; on the other hand, if our car had a dashboard full of key slots and only one of them turned the car on, chances are it would take us a while.
Good design, whether of business policies or stairwells/fire escapes, can make appropriate use of constraints via “forcing functions” – constraints that prevent us from forgetting to do bad things. I discuss this more in the activation energy mental model.
General familiarity is generally retained more extensively than specific, detailed recall. As for specifics, memories are not static, but rather reconstructed every time, so over time we’re prone to stitch together various memories. I frequently find myself thinking that an example or anecdote was contained in one book because it fit with the general topic while, in fact, it was contained in a completely different book.
Cues can help recall, especially if they are based on location, although we have to be careful because our memories are also suggestible – under certain circumstances, we can confuse old memories, or convince ourselves that we experienced something which, in reality, we did not. Research demonstrates that children are particularly susceptible to this, but adults can be too.
Memory works associatively, and the more “links” between an item in memory and other items, the better you’ll be able to recall it – this is why, for example, you’ll often be able to remember facts about a person (we went to school together… he likes Watership Down… his wife works for E&Y…) but not their name – there are more “links” to that person’s qualities than there are to their name (which is only linked to one thing, i.e. your broader conception of them).
It would be hard to forget that our acquaintance John is African-American, or that our acquaintance James is so pale that he gets sunburn from household CFLs, but it might be equally hard to remember that John’s last name is “Black” and James’ last name is “White.” As Green Day put it in “Whatsername,”
“I remember the face but I can’t recall the name… now I wonder how Whatsername has been.”
Similarly, we’re more likely to remember that our old neighbor Sally is a baker (particularly if she’s baked us some delicious treats in the past) than that her last name is “Baker.”
This is also the foundational conception of mnemonics, which have utility in some cases (for example, remembering people by creating associative links between their name, features, and profession/details about them), but are generally (as will be discussed momentarily) too time-intensive to be useful for most applications.
Memory x Structural Problem Solving x Trait Adaptivity
Frequent readers know that one of my favorite models is structural problem solving. And memory is my favorite place to use structural problem solving: my #1 rule is “never try to remember stuff.” It’s much more effective to simply set up my life so I don’t have to.
When I’m cooking, I use a timer on my phone so I don’t forget to take the sweet potatoes out of the oven; when I’ve got a meeting, I use Google Calendar to keep it on the horizon… when I learn something interesting about a company, I write it down in full detail because by the time I’ve looked at another fifty, I’ll have forgotten literally everything.
A somewhat more thorough explanation of the structural-problem-solving approach to memory is available in the last section of the trait adaptivity mental model, as well as a section of the activation energy mental model; since this model is already long enough on its own, you can go read those there.
The punchline? Repeat National Memory Champion Tatiana Cooley can’t remember her grocery lists and lives by Post-It Notes.
Joshua Foer, the journalist who won the national memory championship and wrote about it in Moonwalking with Einstein (MwE review + notes), can’t remember where his keys are.
The confines of our working memory are largely biological in nature. We can try to overcome those limitations… we can also try, grade school style, to flap our arms hard enough to fly. Newsflash: the physics don’t work.
As such, Don Norman – rightly – views the glorification of memorization in the educational system and elsewhere as immature and inappropriate, delivering a phenomenal analysis thereof toward the end of “The Design of Everyday Things” (DOET review + notes). The book is worth buying just for those few pages (let alone the rest of the god-level material), but the highlight:
“In Ancient Greece […] Socrates complained about the impact of books, arguing that reliance on written material would diminish not only memory but the very need to think, to debate, to learn through discussion […] but over the years […] human intelligence has certainly not diminished […]
Does the fact that I can no longer remember my own phone number indicate my growing feebleness?
No, on the contrary, it unleashes the mind from the petty tyranny of tending to the trivial and allows it to concentrate on the important and the critical.
[…] the power of the unaided mind is highly overrated.
[…] Human intelligence is highly flexible and adaptive, superb at inventing procedures and objects that overcome its own limits. The real powers come from designing external aids that enhance cognitive abilities.”
Application/Impact: anyone celebrating mnemonics or willpower as a way to overcome memory limitations is ignorant of basic models like structural problem solving, trait adaptivity, and opportunity costs. We’re smarter with our technology than without it.
Memory x Confirmation Bias x Recency / Salience = “Hindsight Bias”
I love Kip Tindell, the cofounder and former CEO of The Container Store. “Uncontainable” (UCT review + notes) is one of my favorite business books. But Tindell gets one thing wrong in Uncontainable – on page 65, he states that:
“And yet we simply had no doubt that The Container Store would be a hit […] when students or aspiring entrepreneurs ask me how they can tell if their business idea is any good, I always use the analogy of musicians who made a hit record.
Almost every time, I’m sure, they knew […] they just knew.
There was no doubt it would be a huge hit from the moment they walked out of the studio.”
From an entrepreneur perspective, it’s understandable. As I discuss in the agency mental model, this may be an example of adaptive overoptimism: the sort of self-fulfilling prophecy that means only people who’ve decided to live survive life-or-death situations, as explored in Laurence Gonzales’s “Deep Survival” (DpSv review + notes).
But from a memory and business perspective, it’s absolutely wrong. One specific facet of human memory that pops up over and over the place is the “revisionist” tendencies of memory. Four pages earlier, in fact, Tindell notes – amusingly – that he and cofounder Garrett Boone have different recollections of how they came up with the concept for TCS:
“Suddenly a lightbulb went off,” Garrett recalls. “I was driving down the highway in Dallas and thought, “That’s it! We could create a store that sold the resources for people to organize all the stuff in their lies.” In truth, my own recollection is a more gradual process that evolved after countless conversations with Garrett […] and scores of other folks.”
True “aha” moments are rare; chances are that Tindell’s recollection is probably more accurate than Boone’s… of course, the latter makes for a better story (cough storytelling cough).
This “revisionist” tendency to view the past as obvious or deterministic rather than probabilistic has a name: hindsight bias, which I believe to be an intersection of memory with confirmation bias and availability/salience.
The underlying mechanism may well be that we overweight the information that is available to us; since long-ago memories are less available to us (or gone entirely), we place much more weight on the evidence we can see. Any situation where you say “I knew it” is usually a good candidate for examination: there’s a strong possibility hindsight bias may be creeping in!
Remember that 10-franc coin example from Don Norman I cited earlier? Well, here’s a line from a 1986 LA Times article discussing the fiasco. And, ahem:
In retrospect, the French decision seems so foolish that it is hard to fathom how it was made.
This isn’t unique, of course; in retrospect, “New Coke” is viewed as one of the biggest business fiascos of all time. And, no doubt, management got a lot wrong – but a real historical view, of the sort available in “For God, Country, and Coca Cola” (FGC review), makes it clear that whether or not it was a bad decision, it wasn’t so obviously terrible beforehand.
Hindsight bias, which psychologist Daniel Schacter calls “ubiquitous” in “The Seven Sins of Memory” (7SOM review + notes), means that we use the limited set of current circumstances to extrapolate that to what we thought, or should have thought, at a previous time: “oh, this stock doubled. It was so obvious that their new product was revolutionary. Why didn’t I invest more?”
As Schacter puts it,
“our memories of the past are often rescripted to fit with our present views and needs.”
It’s a problem for historians, among others: John Lewis Gaddis discusses in “The Landscape of History” (LandH review + notes) how we have a tendency to evaluate the past through the moral and technological lens of the present.
Three of my all-time favorite pages in any book are Michael Mauboussin’s discussion of the MusicLab experiment in “The Success Equation” (TSE review + notes). Researchers found that social proof and path dependency combined to make it completely unpredictable which songs would end up at the top of the charts in a simulated environment. Megan McArdle discusses similar points regarding movies in “The Up Side of Down” (UpD review + notes).
But it pops up everywhere. Back to science/medicine: in David Oshinsky’s medical history “Bellevue” (BV review + notes), he tells the story of Dr. Charles A. Leale, who treated President Abraham Lincoln immediately after he was shot by John Wilkes Booth:
Why did he choose to add [such vital details?] The likely answer […] is that Leale was determined to go down in history as having used all measures to save Lincoln’s life – whether he did or not. Since a procedure like chest compression was rarely employed in America in the 1860s, he probably did not.
In his 1865 report, moreover, Leale said that he placed his finger deep into Lincoln’s wound to look for dangerous bullet fragments […] Leale made little mention of this in his 1909 speech, for good reason. [… With the coming of germ theory,] no respectable physician in 1909 would have attempted such a potentially fatal maneuver.”
Oshinsky paints Leale out to be intentionally deceptive, but the scary thing is that it’s entirely possible that Leale really and truly believed the account he gave in 1909 – that he wasn’t intentionally lying to the audience, but rather just telling them what he remembered about what happened nearly 50 years prior.
A similar phenomenon is observed by Christopher Browning, commenting in “Ordinary Men” (OrdM review + notes) on the trial testimony of Reserve Police Battalion 101, a group of middle-aged working-class Germans who were tasked with carrying out some of the unthinkable horrors of the Holocaust. Separated from their crimes by decades, the men often found it difficult to reconcile their actions of the past with the current circumstances – and thus, in many cases, tried to explain it away or minimize their part.
This kind of self-justification is, again, more the rule than the exception. Tavris/Aronson’s “Mistakes were Made (but not by me)” (MwM review + notes) goes deep into this topic, analyzing how in contexts as diverse as politics, the legal system, and marriage, memory is an infallible guide. It’s a deeply insightful and often chilling book that will dispel any myths you have about the accuracy of your memory.
The authors note early on that the area between conscious lying and unconscious self-justification is “patrolled by that unreliable, self-serving historian – memory,” further noting that “we may come to believe our own lies, little by little.”
So what are we to do? The easy answer, again, is structural problem solving: writing things down, as Norman notes in “The Design of Everyday Things,” is in fact a powerful technology. In a probabilistic environment where luck and skill are intermingled, having a written process can generate clear feedback that we can use to improve our future decisions with the benefit of actual, true – rather than remembered – understandings of why and how we did things.
Application/impact: It doesn’t matter if it’s a Post-It, an email, a dictaphone, Louis Litt style, or a tattoo, Memento-style. If you want to remember something accurately, do not rely on your memory.
Memory x Activation Energy = Multitasking
As usual, I’ll keep the last one short-ish (since we’ve been here long enough). There’s this famous Richard Feynman quote (page 19 of “The Pleasure of Finding Things Out” – PFTO review + notes) that goes a little something like this:
“To do real good physics work, you do need absolutely solid lengths of time, so that when you are putting ideas together which are vague and hard to remember; it’s very much like building a house of cards and each of the cards is shaky, and if you forget one of them the whole thing collapses again.
But if you have got a job in administration, then you don’t have this solid time. So, I have invented another myth for myself- that I’m irresponsible. I tell everybody, I don’t do anything. If anybody asks me to be on a committee to take care of admissions, no, I’m irresponsible, I don’t give a damn about the students- of course I give a damn about the students but I know that somebody else’ll do it- and I take the view ” Let George do it”, a view which you’re not supposed to take, okay, because that’s not right to do, but I do that because I like to do physics and I want to see if I can still do it, and so I’m selfish, okay. I want to do my physics.“
As usual, Feynman was onto something.
“The fact that new information […] forces things out of working memory means that we can’t pay active attention to too many things at once […] in most people, the executive function can do one task at a time, and attempting to perform simultaneous tasks that involve a conflict begins to break it down.”
Gonzales goes on to cite the (nearly-tragic) example of a climber on the equivalent of an easy “bunny slope” who nearly died because she forgot to tie her rope to her harness due to being distracted by a conversation. In addition, pilots distracted by an emergency have been known to be so focused on dealing with that that they forget to “FLY THE PLANE” – hence, it’s now an item on checklists.
Finally, the dangers of texting and driving are so well established that we shouldn’t need a reminder – and yet we as a nation do: Warren Buffett (whose Berkshire Hathaway subsidiary GEICO is one of the largest American auto insurers) has pointed to distracted driving as an increasing rather than decreasing trend.
Returning to the Feynman example of less critical environments, the idea of focusing on a single task has picked up steam lately, perhaps as backlash against the overstimulating world of social media and 24/7 news. It’s not clear if Cal Newport’s “Deep Work” (DpWk review + notes) inspired much of this or was simply riding the zeitgeist, but either way it’s a thoughtful, worthwhile read.
Newport takes a somewhat different tack than Feynman: he recommends asking bosses straight up, “how much of my time do you want me to spend on shallow work?”
He cites some interesting research demonstrating that even at heavily client-service-focused firms like BCG (Boston Consulting Group), less “always-on” led to more productivity and happier clients.
Meanwhile, our house-favorite psychologist Shawn Achor has some thoughts too: around page 159 of “The Happiness Advantage” (THA review + notes), he notes the activation energy / frictional cost of multitasking:
“the average employee gets interrupted from their work every 11 minutes, and on each occasion experiences a loss of concentration and flow that takes almost as many minutes to recover from.”
And, returning to where we started, Don Norman’s “The Design of Everyday Things” takes a similar tack on pages 199 – 200:
“Multitasking […] erroneously appears to be an efficient way of getting a lot done. It is much beloved by teenagers and busy workers, but in fact, all the evidence points to severe degradation of performance, increased errors, and a general lack of both quality and efficiency. Doing two tasks at once takes longer than the sum of the times it would take to do each alone.”
Norman goes on to cite the dangerous role of memory slips in medical errors, a topic covered in a lot of depth in Atul Gawande’s frequently-cited but poorly-understood “The Checklist Manifesto” (TCM review + notes).
Application/impact: Multitasking, or any environment that creates frequent distractions (like an open office), simply doesn’t work because of the fundamental constraints of human memory.