Your question got me thinking about Temple Grandin. (She's an academic, but probably best known because she was featured in Oliver Sacks's book An Anthropologist on Mars). She describes herself as thinking entirely in pictures. For her, memory is like replaying video clips or flipping through photographs -- but even these visual memories aren't exact recordings. They're still reconstructions, just primarily visual ones.
Different people think (and therefore reconstruct memories) in different ways. Some people report having almost no visual imagery at all (a condition called aphantasia), while others, like Temple, think almost entirely in pictures. We all manage to remember and reconstruct experiences -- we just do it differently.
What about you? Do you find yourself thinking more in pictures?
Well I guess I still recognise much if not all of this - I am surprised by this more than I can convey.
It is a marvellous and succinct treatment which gives the new student a great framework on which to build. Like I remember saying previously, I do wish that I had read or heard you lecture (presumably through some space time anomaly - like Kevin Costner in the film about the radio broadcast across time!) when I was a student way back. It would have saved me a lot of trouble and confusion (with some errors thrown in for good measure).
Filmic fantasies and allusions aside, I do wonder how you manage to maintain such a high standard and communicate so well. Just luck, brains and toil, I guess. Thanks Suzi.
Hi John! Your comments always make me smile -- I love how you blend scientific curiosity with your personal reflection.
Thanks so much for your kind words. I'm trying to remember that movie you mentioned - was it Field of Dreams? (Though now I'm second-guessing myself about which Kevin Costner time-travel movie it was!)
I have to smile at your "luck, brains and toil" comment. Sure, it takes some time to write these posts. Some weeks, I'm not sure if I'll make it, but I enjoy it so much. This intersection between philosophy of mind, the science of consciousness, and artificial intelligence is endlessly fascinating to me.
On wishing you had this framework as a student -- I get that! I often feel the same way, too. But then I wonder whether I would have got the same out of it then as I would now. Maybe it's our years of thinking and learning that help us see things differently.
I’m going to have to admit that the film is a false recollection on my part - that’ll be the atrophy! I think in my mind it’s a fusion of “Frequency” with Dennis Quaid and a French or Spanish film I once saw. But I can visualise the radio set in my memory - a total confection! Somewhat apropos I suppose. KC was starring too - guess he didn’t get paid 😉.
This is fascinating! When you consider how much a person’s personality and daily function depends on their memory, you have to ponder the mystery of what a person actually is at the most basic level…
What struck me while writing this article was that H.M. still maintained his personality and sense of humour, even though he couldn't form new memories. But then you have Phineas Gage (the guy who had that tamping iron shot through his head, obliterating his frontal lobes), and his friends said that after the accident, "Gage was no longer Gage."
Of course, H.M. didn't lose all of his memories, so that makes sense... but it really makes you wonder -- if memory makes up who we are, and memory changes each time we recall it... what does that mean for self?
It reminds me of that old philosophical puzzle: if we gradually replace every plank in a wooden ship, is it still the same ship? With memory, we're doing this reconstruction all the time.
I enjoyed this description of the mechanisms of memory.
I think the best way to think about memory is in terms of what we're going to do, whether that be immediate action or simulation of possible actions. For immediate action, recognizing past stimuli and responses seems like a crucial evolutionary function. But it's the simulations which seem to give us our concept of both imagination and episodic memory, with an asterisk attached to the latter as a past event.
Which, I think, explains why episodic memory is lousy as an archive of past events. That's just not its evolved role, which is to help in planning future actions. Note that "future" here might be in the next second or two.
It also helps, I think, in explaining why we sometimes remember the episode and other times just the semantic memory. It's all in what we (often unconsciously) think we need for the future. If we frequently rerun the simulation of our first kiss, which we may have emotional reasons for doing, then we're going to remember it better. If we frequently pull up the semantic memory of Earth's orbit, without focusing on the classroom or whatever setting we learned it from, then that setting will fade away, probably at best blending into more of a generic equivalent over time, or disappearing completely.
Of course, this doesn't always work right. Sometimes we end up ruminating over stuff we should just let go. Evolved mechanisms don't always fire in the way they should at the right time.
I so often find myself agreeing with you. And here I am again -- this is such a great point.
We often think that memory is so we can reminisce -- but from an evolutionary perspective memory is more likely about prediction. It's intriguing to me that the same neural networks that we use for remembering are the same ones that were used during experiencing and also the same involved in imagining the future.
On your point about rumination -- it's interesting that the most effective therapy for this kind of stuck pattern is CBT, which is all about recognizing these patterns and actively reshaping how we think about past experiences. Maybe that's because it works with, rather than against, memory's future-oriented nature.
I always value your evolutionary perspective on these topics, Mike.
In making things up, what if the brain got them wrong?
Let me answer my own question. What the brain is doing, it seems to me, is following clues. The clue following process is self-correcting, the reason codebreakers can do what they do: zeroing in on the hidden message.
Love this! We are detectives! And just like good detectives (or codebreakers), our brains have ways of cross-checking these clues against each other to build a coherent picture that helps us figure out what to do next.
Yet: trouble can come if the (metaphorical) detective analysing clues is more like The Defective Detective than Sherlock Holmes(*).
If the brain creates memories as it retrieves those clues, then tries to force them to cohere with a picture that is incorrect, I'd say trouble lies ahead.
(I'm not thinking of nasty old "false memory syndrome" - a different matter.)
I'm talking about someone (a specific someone I know) who at least seems to genuinely 'remember' things that just aren't true - but that do serve to reinforce what are, in effect, delusions. To the point that saying things like "I wrote it down, you agreed with it at the time, and here is your agreement" fails to sway them. "That's not how I recall it, you are lying, the document is fabricated" is the typical response (language toned down for a family newsletter). No amount of seemingly independent and objective evidence can convince them that their recall is incorrect (nor their delusion false).
It seems to me that this might well be a failure of "self-correction" during recall of those clues to (re)create a memory (and might make up at least part of their condition).
(*) There's a wonderful scene in George Macdonald Fraser's story "Flashman and the Tiger" which pokes fun at Sherlock Holmes' methods. It's worth the chuckle if you can find it.
Have to say, however, that after all these years, I’ve come to challenge this structure. Not the biological theory which is solid, but focus on functional memory and tracing it back, like first kiss.
So, my view differs significantly here. First kiss is so emotionally laden that it spreads itself into many corners of the mind and is quite a pervasive, large memory. The size matters here. Factual memories also have size which increases with repetition, and also if there is an emotional glue that makes it stick here and there. Think of a dungeon full with sticky notes.
Jeff Hawkins’ work was very aligned with how I perceive it.
Memory is very analog. It does not simply exist or doesn’t exist. It has so very many axis.
I’ve trained myself to trigger emotionally suppressed memories, like that orchestra, when on THC.
Thanks for sharing your perspective! Your sticky note analogy offers an interesting way to visualise memory. Though I suspect the relationship between emotion and memory might be a bit more complex than that.
I'm curious to hear more about how you see these ideas connecting with Hawkins' work -- maybe you could expand on that in a future discussion/article?
Thank you. Expanded opinion on this is a serious work that I’m not suited for yet.
Briefly, I’m inspired by Hawkins’ theorybof cortical systems with consensus-like connectivity, where most “likely” or most “matching” decisions win.
Dungeon analogy is flawed.
Maybe another one: we may have an event that we take a lot of photos and notes of, and another of which very few records exist. First kiss is firmer, emotions nailed down those memories. Trip to the garage to get more paper towels was latter, it barely “happened” as far as memory is concerned.
Memory is an *amount* of records of a certain state. Records themselves are quite abstract, just neural stimulation patterns really.
As usual Suzi you have written a fascinating issue, nice to read and with a great ability of scientific communication. It is an inspiration every time to read issues written in this way!
This article is of particular interest to me because I know that I haven’t delved deeply enough into the technicals of memory and how modern science classifies this sort of thing. Where I might add something to this discussion however is by providing a more full explicit model of the system which incorporates it. Unfortunately modern science has found it challenging to adopt coherent models of what we are, or essentially to “connect the dots”.
We begin with an entirely non-conscious brain which accepts input information, processes it algorithmically through the AND, OR, and NOT relationship that exists between synaptically connected neurons, and then provides output function. One of these output functions of the brain is to create consciousness, which is to say a value driven form of computer that seeks to feel as good as it can from moment to moment. Beyond the value input itself there is also a sense input. A sore toe for example doesn’t just provide negative value, but also sense data about where the problem happens to be located.
A third and I think final form of conscious input is memory, or the topic of Suzi’s article. It’s essentially past consciousness that in some sense remains for present use. Though much of the long term sort is non-declarative, such as teaching the to brain to vocally speak the words we want to say, the more personal element concerns our semantic facts and episodic events. Why do some of them last in us while others do not? Because in a value sense they’re more important to us. We evolved such that horrible experiences, for example, scar us as motivation to not let ourselves get into those situations again.
In any case with value, sense, and memory forms of conscious input, we try to figure out how to promote our value based interests. Theoretically the end goal of each of us is to become as happy as we can for as long as we can.
Your comment about happiness really got me thinking. Particularly this sentence: "Theoretically the end goal of each of us is to become as happy as we can for as long as we can." It's an intriguing idea, but I keep scratching my head about a few things.
Take parents, for example, they're constantly doing things that make them less happy in the moment (getting up at 3am with a crying baby, etc). An evolutionary explanation would be that we do this (make ourselves less happy) because we are wired to prioritise passing on our genes (not necessarily happiness). I'm curious what you think about this -- do you see the pursuit of happiness as opposed to evolutionary theory?
And then there's also a question I have about time. What about people who sign up to run a marathon. Training for that marathon is definitely not going to make that person the happiest they could be in the moment of training. The bet seems to be that their future self will be happier. Though honestly, who knows? Maybe future-them will have completely different ideas about what brings them joy.
What's your take on all this? How do you think about these trade-offs between now-happiness and later-happiness? And do you see evolutionary drives as working with or against our pursuit of happiness?
Hey Suzi! I do stand behind everything I said above. There are some further clarifications that ought to help you grasp this model better though. I consider this perspective extremely consistent with evolution, or as you said, that we are wired to pass on our genes. From this perspective, happiness right now is actually all that matters.
Theoretically the conscious form of function exists as an instantaneous entity or “self”. An instantaneous self’s only connection with past selves is memory, or the topic of your current article. So a loss of memory should bring a loss of connection with the selves we used to be. You’ve instead asked about the future however. Why might someone consciously decide to have children, run a marathon, or you might even have asked why someone might douse themselves with gasoline and light themselves up?
Theoretically when we think about what we might do, some options seem more hopeful to us and thus feel good in that sense while others seem more worrying and thus feel bad in that sense. So it could be that when the decision is made to have a child, this type of life seems more hopeful and less worrying than the converse. Of course later one might consider that choice horribly mistaken, but amending it can be both legally and emotionally off the table . And of course to cause human babies evolution merely needed to make sex feel good to the instantaneous self that I propose. A quick google search suggests that 45% of American pregnancies were unplanned.
On training for and running a marathon, again it’s not about making future selves feel good but rather the present self. Of course not everyone runs marathons and I think marathoners tend to like it this way — their dedication probably makes them feel more respectable and so good in that sense. While I suspect than many or most parents would have more happy lives if they wouldn’t have had kids (which I say as a parent who has always been grateful to be a parent (even though that baby business sucked!)), I doubt that running marathons causes people to have unhappy lives. It shouldn’t be too difficult to quit if it does. An extremity would be how you can’t just quit the implications of self immolation. Because the instantaneous self only cares about how it feels presently however, horrible personal miscalculations are sometimes made.
This is simply the free will argument. Why might you do something you don't want to do? Only because there is a stronger motivation than the one you're thinking of. Otherwise it would be effect without cause which is both impossible in a causal universe, and semantically impossible, cos w/o cause it wouldn't be an effect.
Wow Malcolm, it sounds like you agree with my basic model of how conscious beings function! This surprises me because my understanding is that modern psychologists in general neither accept this model nor any competing models — they set aside fundamental explanations to instead work on peripheral matters. If we’re in agreement regarding the basic mechanics of conscious function however then there are several other ways that we might be aligned!
I'm not a psychologist. I have a biological and computer background so incline towards mechanistic models. What other ways did you have in mind? (Perhaps we ought to take this to CHAT)
Let’s just see how things go Malcolm. If you ever have questions or comments about the sorts of things that I say here, then I hope to hear from you about this. There are several ways that philosophy and mental/behavioral sciences disappoint me. I tend to present radical potential solutions, but this often irritates people in the business because it suggests that they aren’t doing their jobs well enough. So I must walk a fine line here to potentially help institute change, while not simply being dismissed as an enemy. I’m only an enemy to those who have become too invested in the status quo to earnestly consider the various problems that these fields have inherited.
Pfffft, I don't need my stinkin' memory to make up bullshit facts on my behalf! I'm perfectly capable of doing that myself, thank you very much!
But upon reading your post, it struck me how much more similar our memory creation/retrieval process is to LLMs than to e.g. file storage software.
Creating new memories is a bit like pre-training an LLM - you're not loading every single piece of information into your brain and storing it in its entirety. Instead, you encode relationships between people, events, smells, etc., much like LLMs encode relationships between tokens.
And when you're retrieving a memory, you're not pulling it out of a file cabinet wholesale, but you're "generating" a plausible version of events based on the encoded connections, much like LLMs generate different but broadly similar responses to the same query.
The same goes for LLM hallucinations and our memory hallucinations, which explains why my wife and I have very different recollections of the events of the day of our first kiss. That, and she's objectively wrong....according to my memory, that is.
Ha! Love it! I think I'm perfectly capable of making up my own nonsense, too! 😄
Great point! The parallels between human memory and LLMs is striking. Both systems are more about reconstructing from patterns and relationships than retrieving exact copies. I guess the difference is that with LLMs if we have the right seed we can reproduce exactly the same response -- but with brains, by the time we get around to trying to reproduce a memory, the paths that were made during the last recall have already changed.
And that last bit about you and your wife's different memories of your first kiss made me laugh! (of course, your version is _definitely_ the correct one... 😉)
Good point about the seed! Although we do also use the term "plant a seed in someone's head," so I say it's all a conspiracy to hide the fact that we're basically powered by LLMs. That, or we live inside a simulation. Someone should try to make a movie about that. I bet it'd be quite popular.
Marketing runs on impressions - the pseudo science says the more you see something like a brand or a logo, the stronger the imprint on your memory where you either consciously or subconsciously pull it out (recreate it?) when you are making a purchase decision.
I looked into this a bit for running campaigns and here's what I found. I wonder what you think? BS or legit?
---
Humans have two kinds of memory: Short Term and Long Term.
Humans can keep 7 things +/- 2 in our short-term memory. Chunking or grouping helps extend our short term memory so, for example, you remember a phone number as 3 groups even though it's 10 numbers.
Long term memory gets more complicated and here's where we need our impressions to create an entry, because short term memory is fleeting and measured in minutes.
There are three ways you can get something into a humans long term memory:
- Urgency - stressful situations can imprint into long term memory based on strong emotion
- Association - an impression that aligns with something you already know
- Repetition - seeing the same or similar messaging repeatedly moves something to long term memory that becomes more vivid with more impressions
I like these three categories (urgency, association, repetition). I can imagine most of the science would fall into one of these three buckets (although the science might use different words).
On the "7 plus or minus 2" rule for short-term memory -- while it's a useful rule of thumb, we now know there are multiple types of shorter memory for different kinds of information. So some forms (like visual sensory memory) might have a higher capacity than other forms (like working memory -- the type of memory we use to figure things out in our head). And the capacity will depend on what else we are doing at the time.
The 7 plus or minus 2 rule usually applies mostly to simple, short-term storage of unprocessed items (like sensory memory). But you're spot on about chunking! It was later work that suggested the limit of working memory capacity is closer to 4 chunks. This "cap of 4" is the more modern understanding of working memory (when considering active manipulation and chunking).
Limits depend on whether the task is passive (sensory) or active (working memory), and whether the items are familiar, complex, or organised. Things get complicated when we consider how emotions, attention, and meaning all interact with these limits.
I have a brain tumour in six lobes and my different kinds of memory are degrading one by one. My vocabulary has suffered the most — I can’t think of words, especially nouns and, even more, proper nouns (tumour in my temporal lobe, I suppose).
My short-term memory is on its way out now too - which especially sucks because I am studying for a degree in philosophy at the moment. I can build and remember an argument still but I can't remember new facts. I still have a very good memory for long-ago stories too — those are episodic memories, I guess.
By an odd coincidence, my girlfriend died of the same tumour (astrocytoma) 33 years ago. Rita’s tumour pushed on her hippocampus and her short-term memory practically vanished. We would be having a conversation and she would totally forget what we were talking about in the middle of it and I would have to start over.
I have a little story about Rita’s memory loss on my Substack if you are interested.
I'm so moved by your sharing this. What an incredibly difficult journey you're on. But here you are being an inspiration - studying philosophy and writing beautiful things.
I didn't get very far into your story about Rita before I had to take a break -- my eyes were too filled with tears to read the words. The image of you singing Billie Holiday to her... I can't imagine what it was like to watch he go through what she went through, and now to find yourself facing similar challenges. It must give you a very unique perspective.
Thank you for sharing your experience with us. Please keep writing and sharing. You write beautifully ❤️
That's fascinating! My memory is a mystery to me. I have almost no memory of when things happened (in a year, for example) but can frequently remember where they happened (in which home office I had a conversation).
Then there are the things that make no sense. I was talking to a friend, and recommend a book that I had begun but not finished - in around 1985. I knew the title, author, and immediately found it on Amazon - I can feel the book in my hands even now.
Memory, as you rightly say, is far more than just recalling facts. It's a very complex and mysterious process.
Oh, this is really interesting! You're not alone in finding your memory seems to anchor more strongly to places than times. Our brain loves to represent space. Time, not so much.
I love the way you can recall the physical sensation of holding that book. Memory is amazing like that, isn't it!? So rich and multi-layered for some things but so vague and fleeting for others.
I'm curious -- do you use any spatial anchoring techniques to help you remember? Like the memory palace or otherwise do you deliberately think about where you were when trying to remember things? I've been wondering whether people differ on the extent to which they do this?
That's really interesting - I thought the spacial side of memory was just me. My memory of (and feelings about) time have always seemed very idiosyncratic - but perhaps they're not as unusual as I'd thought.
I've never really tried the memory palace, though the idea interests me. My memory as I get older hasn't really changed - except in one area. I used to be able to memorize language vocabulary very quickly. That's gotten slower over the years. Perhaps using some visualization for vocabulary items could help - I may give it a try.
One thing I've realized is that I think I have a memory of my life - but really, it's a relatively few incidents with a feeling of roughly what was going on - and that feeling is probably somewhat illusory. It's rather like remembering a dream; we only remember a few incidents and then fill in the rest...
Hi Suzi, great as always. I wonder if you're familiar with the movie, Ghajini? It's apparently a remake of an American movie, Memento (I haven't seen that one). In Ghajini the main character receives a head injury such that, every 15 minutes he loses all previous memories (but not his basic abilities like fighting). If you can tolerate the violence (very considerable) and don't mind a subtitled movie, I found it a fascinating movie: https://www.youtube.com/watch?v=vZ4-QcO5uUU. Your post brought it to mind, obviously.
This one is a beautifully cut jewel, Suzi. One thing I admire about your writing here—it’s clear you are skimming the surface of your knowledge, that at any moment you could pull the rug out from under your reader, yet you never do. You’re able to get a bead on readers and lead them to re-examine assumptions and, in some cases, defend them they’ve been so scientifically shaken. I actually went through a grief process when I had to admit to myself John Searle’s take on Daniel Dennett was wrong.
You might be interested in Lev Vygotsky’s theory of informal concepts, tantamount to semantic memory. Vygotsky, of course, was interested in the relationship between thought and language, hence the title of his seminal book in 1934.
Vygotsky used observational methods to study the emergence of language in children not so much as intersubjective communication but as a psychological tool chest, cognitive tools. He called language the “master tool.”
One study I’ve thought about seems like forever I call “red feet.” I’ve even written poems using red feet as images. He found a patient in a medical facility who had a terrible time walking in a straight line. The guy would bounce from one wall to the other.
Vygotsky painted red feet on the floor of a hallway so that they mirrored the patient’s walking pattern. The man could use these feet to make his way down the hall close to normally. Vygotsky included semiotic materials like images as loosely “language like.”
Informal concepts are constructed by way of psychological tools afforded by language deployed as an inner voice. When thought carries language traces as a feature of a memory into long term memory, language becomes an integral part of the memory together with episodic sensations and apprehensions. If we think of memory consisting of informal concepts, truncated and private inner speech can be red feet guiding consciousness toward a sort of rehydrated thought. Informal concepts are personal, idiosyncratic.
Formal concepts are very different. Constructed internally by way of the master tool under the direction of the executive agent, these concepts are learned and taxonomized or narrativized—not exactly like a filing cabinet, but systematic nonetheless.
Another term he used for these is “scientific concepts.”
Formal concepts are often built from an academic schema in a formal setting. Unlike informal concepts that involve random or granular information picked up through the senses, formal concepts require a disciplined application of language and thought in interaction. Locating a new piece of information is done carefully if more quickly as expertise strengthens. It’s as if information is linguistically tagged and, yes, filed away for easy retrieval in novel contexts. We construct wheelhouses.
What role do Wernicke and Broca play in memory? I believe Wernicke is the area where language comprehension occurs and Broca is the more technical decoder. I could have this backwards or totally garbled as I did with the amygdalae. From this memory post I see the hippocampi have something to with emotions as well.
Re getting episodic memories in isn’t the same as getting them out—the guy who couldn’t create new episodic memories but could recall the pre surgery experiences—could he construct formal concepts that begin and end (often) in the head through external and internal language?
Thanks as always. Your generosity in sharing your wisdom will come back to you forever as a force for good.
Thank you so much for your kind words. Your feedback is always wonderful and I appreciate it more than I can say.
I resonate with that grief you're talking about when you change your mind. I've had many of those moments myself. Sometimes it took me a while to drop the emotional attachment I felt to a particular idea. It's fascinating how emotionally invested we can get.
I'm vaguely familiar with Vygotsky's work. Although I'm not sure why. Perhaps during educational psychology or developmental psychology studies? The red feet study is great. The idea that external tools (which I imagine could be goals or instructions, too) can scaffold learning. The way you connect this to memory and language really got me thinking about how we use various forms of these sorts of tools to support thinking, learning, and memory. Now you've got me going down rabbit holes!
On your specific questions:
- Broca's area is indeed involved in speech production while Wernicke's area handles comprehension (you had it right!). Both play interesting roles in memory, particularly in how we encode and retrieve memories that involve language.
- About H.M. - he could actually learn new procedural skills and even some limited semantic information, but had trouble integrating these into coherent conceptual frameworks.
Thanks for your thought-provoking comment -- I'm going to be thinking about these ideas for days. This relationship between language, memory, and concept formation is fascinating!
Ah, thanks for that clarification! That makes a lot of sense -- informal concepts emerge from direct experience (like episodic memories), while formal concepts align more with that structured, factual knowledge we call semantic memory.
Hi JT,
Your question got me thinking about Temple Grandin. (She's an academic, but probably best known because she was featured in Oliver Sacks's book An Anthropologist on Mars). She describes herself as thinking entirely in pictures. For her, memory is like replaying video clips or flipping through photographs -- but even these visual memories aren't exact recordings. They're still reconstructions, just primarily visual ones.
Different people think (and therefore reconstruct memories) in different ways. Some people report having almost no visual imagery at all (a condition called aphantasia), while others, like Temple, think almost entirely in pictures. We all manage to remember and reconstruct experiences -- we just do it differently.
What about you? Do you find yourself thinking more in pictures?
Well I guess I still recognise much if not all of this - I am surprised by this more than I can convey.
It is a marvellous and succinct treatment which gives the new student a great framework on which to build. Like I remember saying previously, I do wish that I had read or heard you lecture (presumably through some space time anomaly - like Kevin Costner in the film about the radio broadcast across time!) when I was a student way back. It would have saved me a lot of trouble and confusion (with some errors thrown in for good measure).
Filmic fantasies and allusions aside, I do wonder how you manage to maintain such a high standard and communicate so well. Just luck, brains and toil, I guess. Thanks Suzi.
Hi John! Your comments always make me smile -- I love how you blend scientific curiosity with your personal reflection.
Thanks so much for your kind words. I'm trying to remember that movie you mentioned - was it Field of Dreams? (Though now I'm second-guessing myself about which Kevin Costner time-travel movie it was!)
I have to smile at your "luck, brains and toil" comment. Sure, it takes some time to write these posts. Some weeks, I'm not sure if I'll make it, but I enjoy it so much. This intersection between philosophy of mind, the science of consciousness, and artificial intelligence is endlessly fascinating to me.
On wishing you had this framework as a student -- I get that! I often feel the same way, too. But then I wonder whether I would have got the same out of it then as I would now. Maybe it's our years of thinking and learning that help us see things differently.
I’m going to have to admit that the film is a false recollection on my part - that’ll be the atrophy! I think in my mind it’s a fusion of “Frequency” with Dennis Quaid and a French or Spanish film I once saw. But I can visualise the radio set in my memory - a total confection! Somewhat apropos I suppose. KC was starring too - guess he didn’t get paid 😉.
I agree! It’s astounding.
Thanks so much, Terry!
This is fascinating! When you consider how much a person’s personality and daily function depends on their memory, you have to ponder the mystery of what a person actually is at the most basic level…
Hey Glen!
Oh! Absolutely -- great point!
What struck me while writing this article was that H.M. still maintained his personality and sense of humour, even though he couldn't form new memories. But then you have Phineas Gage (the guy who had that tamping iron shot through his head, obliterating his frontal lobes), and his friends said that after the accident, "Gage was no longer Gage."
Of course, H.M. didn't lose all of his memories, so that makes sense... but it really makes you wonder -- if memory makes up who we are, and memory changes each time we recall it... what does that mean for self?
It reminds me of that old philosophical puzzle: if we gradually replace every plank in a wooden ship, is it still the same ship? With memory, we're doing this reconstruction all the time.
I enjoyed this description of the mechanisms of memory.
I think the best way to think about memory is in terms of what we're going to do, whether that be immediate action or simulation of possible actions. For immediate action, recognizing past stimuli and responses seems like a crucial evolutionary function. But it's the simulations which seem to give us our concept of both imagination and episodic memory, with an asterisk attached to the latter as a past event.
Which, I think, explains why episodic memory is lousy as an archive of past events. That's just not its evolved role, which is to help in planning future actions. Note that "future" here might be in the next second or two.
It also helps, I think, in explaining why we sometimes remember the episode and other times just the semantic memory. It's all in what we (often unconsciously) think we need for the future. If we frequently rerun the simulation of our first kiss, which we may have emotional reasons for doing, then we're going to remember it better. If we frequently pull up the semantic memory of Earth's orbit, without focusing on the classroom or whatever setting we learned it from, then that setting will fade away, probably at best blending into more of a generic equivalent over time, or disappearing completely.
Of course, this doesn't always work right. Sometimes we end up ruminating over stuff we should just let go. Evolved mechanisms don't always fire in the way they should at the right time.
But maybe I'm simulating this all wrong.
Excellent as always Suzi!
Hi Mike!
I so often find myself agreeing with you. And here I am again -- this is such a great point.
We often think that memory is so we can reminisce -- but from an evolutionary perspective memory is more likely about prediction. It's intriguing to me that the same neural networks that we use for remembering are the same ones that were used during experiencing and also the same involved in imagining the future.
On your point about rumination -- it's interesting that the most effective therapy for this kind of stuck pattern is CBT, which is all about recognizing these patterns and actively reshaping how we think about past experiences. Maybe that's because it works with, rather than against, memory's future-oriented nature.
I always value your evolutionary perspective on these topics, Mike.
In making things up, what if the brain got them wrong?
Let me answer my own question. What the brain is doing, it seems to me, is following clues. The clue following process is self-correcting, the reason codebreakers can do what they do: zeroing in on the hidden message.
Love this! We are detectives! And just like good detectives (or codebreakers), our brains have ways of cross-checking these clues against each other to build a coherent picture that helps us figure out what to do next.
Lovely! Thank you for sharing this with us.
Yet: trouble can come if the (metaphorical) detective analysing clues is more like The Defective Detective than Sherlock Holmes(*).
If the brain creates memories as it retrieves those clues, then tries to force them to cohere with a picture that is incorrect, I'd say trouble lies ahead.
(I'm not thinking of nasty old "false memory syndrome" - a different matter.)
I'm talking about someone (a specific someone I know) who at least seems to genuinely 'remember' things that just aren't true - but that do serve to reinforce what are, in effect, delusions. To the point that saying things like "I wrote it down, you agreed with it at the time, and here is your agreement" fails to sway them. "That's not how I recall it, you are lying, the document is fabricated" is the typical response (language toned down for a family newsletter). No amount of seemingly independent and objective evidence can convince them that their recall is incorrect (nor their delusion false).
It seems to me that this might well be a failure of "self-correction" during recall of those clues to (re)create a memory (and might make up at least part of their condition).
(*) There's a wonderful scene in George Macdonald Fraser's story "Flashman and the Tiger" which pokes fun at Sherlock Holmes' methods. It's worth the chuckle if you can find it.
Impressive overview.
Have to say, however, that after all these years, I’ve come to challenge this structure. Not the biological theory which is solid, but focus on functional memory and tracing it back, like first kiss.
So, my view differs significantly here. First kiss is so emotionally laden that it spreads itself into many corners of the mind and is quite a pervasive, large memory. The size matters here. Factual memories also have size which increases with repetition, and also if there is an emotional glue that makes it stick here and there. Think of a dungeon full with sticky notes.
Jeff Hawkins’ work was very aligned with how I perceive it.
Memory is very analog. It does not simply exist or doesn’t exist. It has so very many axis.
I’ve trained myself to trigger emotionally suppressed memories, like that orchestra, when on THC.
There is a lot more work needed here.
Thanks for sharing your perspective! Your sticky note analogy offers an interesting way to visualise memory. Though I suspect the relationship between emotion and memory might be a bit more complex than that.
I'm curious to hear more about how you see these ideas connecting with Hawkins' work -- maybe you could expand on that in a future discussion/article?
Thank you. Expanded opinion on this is a serious work that I’m not suited for yet.
Briefly, I’m inspired by Hawkins’ theorybof cortical systems with consensus-like connectivity, where most “likely” or most “matching” decisions win.
Dungeon analogy is flawed.
Maybe another one: we may have an event that we take a lot of photos and notes of, and another of which very few records exist. First kiss is firmer, emotions nailed down those memories. Trip to the garage to get more paper towels was latter, it barely “happened” as far as memory is concerned.
Memory is an *amount* of records of a certain state. Records themselves are quite abstract, just neural stimulation patterns really.
memory is rehearsal
As usual Suzi you have written a fascinating issue, nice to read and with a great ability of scientific communication. It is an inspiration every time to read issues written in this way!
You've made my day, Riccardo. Thank you so much for your kind words.
Thank you Suzi for the great issue!
This article is of particular interest to me because I know that I haven’t delved deeply enough into the technicals of memory and how modern science classifies this sort of thing. Where I might add something to this discussion however is by providing a more full explicit model of the system which incorporates it. Unfortunately modern science has found it challenging to adopt coherent models of what we are, or essentially to “connect the dots”.
We begin with an entirely non-conscious brain which accepts input information, processes it algorithmically through the AND, OR, and NOT relationship that exists between synaptically connected neurons, and then provides output function. One of these output functions of the brain is to create consciousness, which is to say a value driven form of computer that seeks to feel as good as it can from moment to moment. Beyond the value input itself there is also a sense input. A sore toe for example doesn’t just provide negative value, but also sense data about where the problem happens to be located.
A third and I think final form of conscious input is memory, or the topic of Suzi’s article. It’s essentially past consciousness that in some sense remains for present use. Though much of the long term sort is non-declarative, such as teaching the to brain to vocally speak the words we want to say, the more personal element concerns our semantic facts and episodic events. Why do some of them last in us while others do not? Because in a value sense they’re more important to us. We evolved such that horrible experiences, for example, scar us as motivation to not let ourselves get into those situations again.
In any case with value, sense, and memory forms of conscious input, we try to figure out how to promote our value based interests. Theoretically the end goal of each of us is to become as happy as we can for as long as we can.
Hey Eric!
Your comment about happiness really got me thinking. Particularly this sentence: "Theoretically the end goal of each of us is to become as happy as we can for as long as we can." It's an intriguing idea, but I keep scratching my head about a few things.
Take parents, for example, they're constantly doing things that make them less happy in the moment (getting up at 3am with a crying baby, etc). An evolutionary explanation would be that we do this (make ourselves less happy) because we are wired to prioritise passing on our genes (not necessarily happiness). I'm curious what you think about this -- do you see the pursuit of happiness as opposed to evolutionary theory?
And then there's also a question I have about time. What about people who sign up to run a marathon. Training for that marathon is definitely not going to make that person the happiest they could be in the moment of training. The bet seems to be that their future self will be happier. Though honestly, who knows? Maybe future-them will have completely different ideas about what brings them joy.
What's your take on all this? How do you think about these trade-offs between now-happiness and later-happiness? And do you see evolutionary drives as working with or against our pursuit of happiness?
Hey Suzi! I do stand behind everything I said above. There are some further clarifications that ought to help you grasp this model better though. I consider this perspective extremely consistent with evolution, or as you said, that we are wired to pass on our genes. From this perspective, happiness right now is actually all that matters.
Theoretically the conscious form of function exists as an instantaneous entity or “self”. An instantaneous self’s only connection with past selves is memory, or the topic of your current article. So a loss of memory should bring a loss of connection with the selves we used to be. You’ve instead asked about the future however. Why might someone consciously decide to have children, run a marathon, or you might even have asked why someone might douse themselves with gasoline and light themselves up?
Theoretically when we think about what we might do, some options seem more hopeful to us and thus feel good in that sense while others seem more worrying and thus feel bad in that sense. So it could be that when the decision is made to have a child, this type of life seems more hopeful and less worrying than the converse. Of course later one might consider that choice horribly mistaken, but amending it can be both legally and emotionally off the table . And of course to cause human babies evolution merely needed to make sex feel good to the instantaneous self that I propose. A quick google search suggests that 45% of American pregnancies were unplanned.
On training for and running a marathon, again it’s not about making future selves feel good but rather the present self. Of course not everyone runs marathons and I think marathoners tend to like it this way — their dedication probably makes them feel more respectable and so good in that sense. While I suspect than many or most parents would have more happy lives if they wouldn’t have had kids (which I say as a parent who has always been grateful to be a parent (even though that baby business sucked!)), I doubt that running marathons causes people to have unhappy lives. It shouldn’t be too difficult to quit if it does. An extremity would be how you can’t just quit the implications of self immolation. Because the instantaneous self only cares about how it feels presently however, horrible personal miscalculations are sometimes made.
This is simply the free will argument. Why might you do something you don't want to do? Only because there is a stronger motivation than the one you're thinking of. Otherwise it would be effect without cause which is both impossible in a causal universe, and semantically impossible, cos w/o cause it wouldn't be an effect.
Wow Malcolm, it sounds like you agree with my basic model of how conscious beings function! This surprises me because my understanding is that modern psychologists in general neither accept this model nor any competing models — they set aside fundamental explanations to instead work on peripheral matters. If we’re in agreement regarding the basic mechanics of conscious function however then there are several other ways that we might be aligned!
I'm not a psychologist. I have a biological and computer background so incline towards mechanistic models. What other ways did you have in mind? (Perhaps we ought to take this to CHAT)
Let’s just see how things go Malcolm. If you ever have questions or comments about the sorts of things that I say here, then I hope to hear from you about this. There are several ways that philosophy and mental/behavioral sciences disappoint me. I tend to present radical potential solutions, but this often irritates people in the business because it suggests that they aren’t doing their jobs well enough. So I must walk a fine line here to potentially help institute change, while not simply being dismissed as an enemy. I’m only an enemy to those who have become too invested in the status quo to earnestly consider the various problems that these fields have inherited.
Pfffft, I don't need my stinkin' memory to make up bullshit facts on my behalf! I'm perfectly capable of doing that myself, thank you very much!
But upon reading your post, it struck me how much more similar our memory creation/retrieval process is to LLMs than to e.g. file storage software.
Creating new memories is a bit like pre-training an LLM - you're not loading every single piece of information into your brain and storing it in its entirety. Instead, you encode relationships between people, events, smells, etc., much like LLMs encode relationships between tokens.
And when you're retrieving a memory, you're not pulling it out of a file cabinet wholesale, but you're "generating" a plausible version of events based on the encoded connections, much like LLMs generate different but broadly similar responses to the same query.
The same goes for LLM hallucinations and our memory hallucinations, which explains why my wife and I have very different recollections of the events of the day of our first kiss. That, and she's objectively wrong....according to my memory, that is.
Upvote ⬆️
Hi Daniel!
Ha! Love it! I think I'm perfectly capable of making up my own nonsense, too! 😄
Great point! The parallels between human memory and LLMs is striking. Both systems are more about reconstructing from patterns and relationships than retrieving exact copies. I guess the difference is that with LLMs if we have the right seed we can reproduce exactly the same response -- but with brains, by the time we get around to trying to reproduce a memory, the paths that were made during the last recall have already changed.
And that last bit about you and your wife's different memories of your first kiss made me laugh! (of course, your version is _definitely_ the correct one... 😉)
Good point about the seed! Although we do also use the term "plant a seed in someone's head," so I say it's all a conspiracy to hide the fact that we're basically powered by LLMs. That, or we live inside a simulation. Someone should try to make a movie about that. I bet it'd be quite popular.
I agree! It might be so popular they make it into a tetralogy.
Marketing runs on impressions - the pseudo science says the more you see something like a brand or a logo, the stronger the imprint on your memory where you either consciously or subconsciously pull it out (recreate it?) when you are making a purchase decision.
I looked into this a bit for running campaigns and here's what I found. I wonder what you think? BS or legit?
---
Humans have two kinds of memory: Short Term and Long Term.
Humans can keep 7 things +/- 2 in our short-term memory. Chunking or grouping helps extend our short term memory so, for example, you remember a phone number as 3 groups even though it's 10 numbers.
Long term memory gets more complicated and here's where we need our impressions to create an entry, because short term memory is fleeting and measured in minutes.
There are three ways you can get something into a humans long term memory:
- Urgency - stressful situations can imprint into long term memory based on strong emotion
- Association - an impression that aligns with something you already know
- Repetition - seeing the same or similar messaging repeatedly moves something to long term memory that becomes more vivid with more impressions
Hey Andrew!
I like these three categories (urgency, association, repetition). I can imagine most of the science would fall into one of these three buckets (although the science might use different words).
On the "7 plus or minus 2" rule for short-term memory -- while it's a useful rule of thumb, we now know there are multiple types of shorter memory for different kinds of information. So some forms (like visual sensory memory) might have a higher capacity than other forms (like working memory -- the type of memory we use to figure things out in our head). And the capacity will depend on what else we are doing at the time.
The 7 plus or minus 2 rule usually applies mostly to simple, short-term storage of unprocessed items (like sensory memory). But you're spot on about chunking! It was later work that suggested the limit of working memory capacity is closer to 4 chunks. This "cap of 4" is the more modern understanding of working memory (when considering active manipulation and chunking).
Limits depend on whether the task is passive (sensory) or active (working memory), and whether the items are familiar, complex, or organised. Things get complicated when we consider how emotions, attention, and meaning all interact with these limits.
Thanks Suzi; good to know it's based in science!
Excellent again, Suzi. I always learn so much!
I have a brain tumour in six lobes and my different kinds of memory are degrading one by one. My vocabulary has suffered the most — I can’t think of words, especially nouns and, even more, proper nouns (tumour in my temporal lobe, I suppose).
My short-term memory is on its way out now too - which especially sucks because I am studying for a degree in philosophy at the moment. I can build and remember an argument still but I can't remember new facts. I still have a very good memory for long-ago stories too — those are episodic memories, I guess.
By an odd coincidence, my girlfriend died of the same tumour (astrocytoma) 33 years ago. Rita’s tumour pushed on her hippocampus and her short-term memory practically vanished. We would be having a conversation and she would totally forget what we were talking about in the middle of it and I would have to start over.
I have a little story about Rita’s memory loss on my Substack if you are interested.
https://raggedclown.substack.com/p/solitude
I'm so moved by your sharing this. What an incredibly difficult journey you're on. But here you are being an inspiration - studying philosophy and writing beautiful things.
I didn't get very far into your story about Rita before I had to take a break -- my eyes were too filled with tears to read the words. The image of you singing Billie Holiday to her... I can't imagine what it was like to watch he go through what she went through, and now to find yourself facing similar challenges. It must give you a very unique perspective.
Thank you for sharing your experience with us. Please keep writing and sharing. You write beautifully ❤️
Thank you, Suzi. I love your writing too.
That's fascinating! My memory is a mystery to me. I have almost no memory of when things happened (in a year, for example) but can frequently remember where they happened (in which home office I had a conversation).
Then there are the things that make no sense. I was talking to a friend, and recommend a book that I had begun but not finished - in around 1985. I knew the title, author, and immediately found it on Amazon - I can feel the book in my hands even now.
Memory, as you rightly say, is far more than just recalling facts. It's a very complex and mysterious process.
Oh, this is really interesting! You're not alone in finding your memory seems to anchor more strongly to places than times. Our brain loves to represent space. Time, not so much.
I love the way you can recall the physical sensation of holding that book. Memory is amazing like that, isn't it!? So rich and multi-layered for some things but so vague and fleeting for others.
I'm curious -- do you use any spatial anchoring techniques to help you remember? Like the memory palace or otherwise do you deliberately think about where you were when trying to remember things? I've been wondering whether people differ on the extent to which they do this?
That's really interesting - I thought the spacial side of memory was just me. My memory of (and feelings about) time have always seemed very idiosyncratic - but perhaps they're not as unusual as I'd thought.
I've never really tried the memory palace, though the idea interests me. My memory as I get older hasn't really changed - except in one area. I used to be able to memorize language vocabulary very quickly. That's gotten slower over the years. Perhaps using some visualization for vocabulary items could help - I may give it a try.
One thing I've realized is that I think I have a memory of my life - but really, it's a relatively few incidents with a feeling of roughly what was going on - and that feeling is probably somewhat illusory. It's rather like remembering a dream; we only remember a few incidents and then fill in the rest...
Hi Suzi, great as always. I wonder if you're familiar with the movie, Ghajini? It's apparently a remake of an American movie, Memento (I haven't seen that one). In Ghajini the main character receives a head injury such that, every 15 minutes he loses all previous memories (but not his basic abilities like fighting). If you can tolerate the violence (very considerable) and don't mind a subtitled movie, I found it a fascinating movie: https://www.youtube.com/watch?v=vZ4-QcO5uUU. Your post brought it to mind, obviously.
Hey Jack! No, I am not familiar with Ghajini, but I'm very interested! Thanks for the recommendation! I'm always looking for great movies to watch.
This one is a beautifully cut jewel, Suzi. One thing I admire about your writing here—it’s clear you are skimming the surface of your knowledge, that at any moment you could pull the rug out from under your reader, yet you never do. You’re able to get a bead on readers and lead them to re-examine assumptions and, in some cases, defend them they’ve been so scientifically shaken. I actually went through a grief process when I had to admit to myself John Searle’s take on Daniel Dennett was wrong.
You might be interested in Lev Vygotsky’s theory of informal concepts, tantamount to semantic memory. Vygotsky, of course, was interested in the relationship between thought and language, hence the title of his seminal book in 1934.
Vygotsky used observational methods to study the emergence of language in children not so much as intersubjective communication but as a psychological tool chest, cognitive tools. He called language the “master tool.”
One study I’ve thought about seems like forever I call “red feet.” I’ve even written poems using red feet as images. He found a patient in a medical facility who had a terrible time walking in a straight line. The guy would bounce from one wall to the other.
Vygotsky painted red feet on the floor of a hallway so that they mirrored the patient’s walking pattern. The man could use these feet to make his way down the hall close to normally. Vygotsky included semiotic materials like images as loosely “language like.”
Informal concepts are constructed by way of psychological tools afforded by language deployed as an inner voice. When thought carries language traces as a feature of a memory into long term memory, language becomes an integral part of the memory together with episodic sensations and apprehensions. If we think of memory consisting of informal concepts, truncated and private inner speech can be red feet guiding consciousness toward a sort of rehydrated thought. Informal concepts are personal, idiosyncratic.
Formal concepts are very different. Constructed internally by way of the master tool under the direction of the executive agent, these concepts are learned and taxonomized or narrativized—not exactly like a filing cabinet, but systematic nonetheless.
Another term he used for these is “scientific concepts.”
Formal concepts are often built from an academic schema in a formal setting. Unlike informal concepts that involve random or granular information picked up through the senses, formal concepts require a disciplined application of language and thought in interaction. Locating a new piece of information is done carefully if more quickly as expertise strengthens. It’s as if information is linguistically tagged and, yes, filed away for easy retrieval in novel contexts. We construct wheelhouses.
What role do Wernicke and Broca play in memory? I believe Wernicke is the area where language comprehension occurs and Broca is the more technical decoder. I could have this backwards or totally garbled as I did with the amygdalae. From this memory post I see the hippocampi have something to with emotions as well.
Re getting episodic memories in isn’t the same as getting them out—the guy who couldn’t create new episodic memories but could recall the pre surgery experiences—could he construct formal concepts that begin and end (often) in the head through external and internal language?
Thanks as always. Your generosity in sharing your wisdom will come back to you forever as a force for good.
Hi Terry!
Thank you so much for your kind words. Your feedback is always wonderful and I appreciate it more than I can say.
I resonate with that grief you're talking about when you change your mind. I've had many of those moments myself. Sometimes it took me a while to drop the emotional attachment I felt to a particular idea. It's fascinating how emotionally invested we can get.
I'm vaguely familiar with Vygotsky's work. Although I'm not sure why. Perhaps during educational psychology or developmental psychology studies? The red feet study is great. The idea that external tools (which I imagine could be goals or instructions, too) can scaffold learning. The way you connect this to memory and language really got me thinking about how we use various forms of these sorts of tools to support thinking, learning, and memory. Now you've got me going down rabbit holes!
On your specific questions:
- Broca's area is indeed involved in speech production while Wernicke's area handles comprehension (you had it right!). Both play interesting roles in memory, particularly in how we encode and retrieve memories that involve language.
- About H.M. - he could actually learn new procedural skills and even some limited semantic information, but had trouble integrating these into coherent conceptual frameworks.
Thanks for your thought-provoking comment -- I'm going to be thinking about these ideas for days. This relationship between language, memory, and concept formation is fascinating!
Ooops—informal concepts are tantamount to episodic memory… formal concepts are semantic memory
Ah, thanks for that clarification! That makes a lot of sense -- informal concepts emerge from direct experience (like episodic memories), while formal concepts align more with that structured, factual knowledge we call semantic memory.