How we remember, why we forget | Anthony Wagner

At some point in our lives, we all struggle with memory — learning a new name, remembering that book you were reading just yesterday or that word on the tip of your tongue.

So what can neuroscience teach us about why we remember, why we forget, and how we might even improve our memories?

To answer this question, I spoke with neuroscientist Anthony Wagner, a memory expert in Stanford's Department of Psychology.

Learn More
Wagner lab website

• Recent lab publications
• Anthony's new book: Brain Sciences for Lawyers, Judges, and Policymakers (2024). Jones, O. D., Schall, J. D., Shen, F. X., Hoffman, M. B., & Wagner, A. D. Oxford University Press. Order

Stress thwarts our ability to plan ahead by disrupting how we use memory, Stanford study finds (Stanford News 2020)

Stanford researchers link poor memory to attention lapses and media multitasking (Stanford News, 2020)

Episode credits
This episode was produced by Michael Osborne at 14th Street Studios, with production assistance by Morgan Honaker. Our logo is by Aimee Garza. The show is hosted by Nicholas Weiler at Stanford's Wu Tsai Neurosciences Institute. 

Thanks for listening! If you're enjoying our show, please take a moment to give us a review on your podcast app of choice and share this episode with your friends. That's how we grow as a show and bring the stories of the frontiers of neuroscience to a wider audience.

Learn more about the Wu Tsai Neurosciences Institute at Stanford and follow us on Twitter, Facebook, and LinkedIn.

Transcript

Nicholas Weiler:

At some point in our lives, we all struggle with memory — some of us more than others — learning a new name, that book you were reading just yesterday, that word on the tip of your tongue. 

So what can neuroscience teach us about why we remember, why we forget, and how we might even improve our memories? To answer this question, I spoke with neuroscientist Anthony Wagner, a memory expert in Stanford's Department of Neurology. 

This is From Our Neurons to Yours from the Wu Tsai Neurosciences Institute at Stanford University, bringing you to the frontiers of brain science. 

I started off asking Anthony to take me through exactly what we mean when we talk about memory.

Anthony Wagner:

There are numerous kinds of memories. The brain is plastic for a reason because of systems are trying to build sort of models of the world about the kinds of things that they represent. And these models of the world provide us predictions as we're going forward in life about what's likely to be occurring, what were likely to be seen, what thoughts we might have. And so all systems of the brain build particular kinds of memories and different systems build different kinds of memories.

Nicholas Weiler:

So there's sort of an overlap between when we talk about memory, just the general learning abilities of the brain is one way we could think about memory.

Anthony Wagner:

Absolutely. Learning is the plasticity, it's the changes that happen within our brains. And memories are the residues, the products of that plasticity, that influence, shape, and change our future cognition. That is memory in action. And so we think about episodic memories, sort of what are we going to focus on today, these are memories of life events. A key thing about episodic memories in contrast to other forms of memories are they're multidimensional, an event, an experience consists of many different things, the who, what, where, when features of an experience of an event and these different dimensions of an event, they're represented in different regions in our brain, different cortical regions, different subcortical regions.

Nicholas Weiler:

Right. When there's a moment you're remembering, you may remember the sounds, you may remember the smells, partly what you saw. And all those things are coming from different parts of the brain.

Anthony Wagner:

Absolutely. And so the time of the experience, you need a system that actually is aware, not in a conscious sense, but that receives the inputs from all these other systems that are representing different features, dimensions of the event and can glue it together. And in your earlier wonderful episode with Lisa Giocomo, you had a great discussion about the critical role that the hippocampus plays in episodic memory. And this is because the hippocampus is situated to receive inputs from these different cortical and subcortical regions that are representing the different features of an experience as it's unfolding. And the hippocampus has the circuitry and the ability to learn really rapidly.

Nicholas Weiler:

And not to get too far on a sidetrack, but if it's recording at all times, there must be something that triggers it to say, "This is something I'm going to keep".

Anthony Wagner:

Absolutely. So what are some of the factors that influence the likelihood that the hippocampus will build a memory, stabilize that memory and hang onto it over the long term? So we know that the hippocampus, it's fundamental for building these memory traces and it's this memory trace that sits in the hippocampus that then we know is fundamental to support remembering. At some point in the future, we are trying to see that acquaintance's face and we're trying to remember, "Oh yeah, I saw this person recently at a party, what was their name," and we're trying to use those cues to reactivate that hippocampal memory trace. And to the extent that we can do that, the hippocampus can then drive back out into these other cortical structures and subcortical structures, can reinstate versions of the patterns of activation that were present during that experience in the past. But it doesn't happen in isolation and that's one thing I want to stress.

Nicholas Weiler:

Right. Yeah. Well, we should definitely come back to that because I want to talk about what are the things that influence how well this system works and when it works and when it doesn't work. I mean, it's so interesting the way you describe it. I don't know if I've ever quite thought about it this way before. Basically the memory is the association between a bunch of different things that happened at the same time. It's the link between those things. And so to record a memory, you have to lay down that link and say all these different things were going on at the same time. And then to retrieve it, you need to maybe start with one of those things, the person's face for example. And that needs to reactivate, go to the hippocampus, and look up all of the other things or some subset of the other things that were happening the last time you saw that person's face, one of which may be their name.

Anthony Wagner:

That's a great summary, but it's not that the hippocampus has all that information packed into it. Its trace ends up being kind of like a pointer back to these representations on other parts of the brain.

Nicholas Weiler:

So you've done work showing that you can actually see when you've got someone in a brain scan showing that you can actually sort of predict in real time how good people's memory for a new piece of information will be. Many of us have had the experience of struggling to get something into our brains and then struggling to get it out again. What are you learning about what predicts whether we will be able to remember something when we're exposed to it?

Anthony Wagner:

Well, that's a great question, but let me just sort of note, as a neuroscientist and a psychological scientist who's been studying memory for about 30 some odd years now, we're now at the point as you just noted where we can witness in some sense the birth of a memory, the formation of a memory in a human brain. At the time of retrieval, we can witness the replay of the contents or aspects of the content of that memory back out into the cortex and we can kind of know is this person bringing back online the face of somebody or they're bringing back online about where they were, et cetera. We can kind of read that out in the individual brain at the individual memory trace, that moment, that individual experience that's been unfolding over the last sort of teen 20 years with advanced functional imaging of the human brain combined with advances in statistics, machine learning, and I'm actually... I think we're at the cusp of the next explosion of exciting work as modern neuroscience data gets combined with artificial intelligence and other kinds of deep neural networks, models of neural function and cognitive function.

Nicholas Weiler:

Well, I wanted to clarify one thing quickly. So when you say that we can see what people are remembering, I'm assuming we can see where they are reactivating, what parts of their brain they're reactivating, and so we can see them reactivating spatial areas or areas involved in faces or areas involved in smell or hearing or something so that we can say it looks like they're remembering a sound, but we don't yet have access to the specific contents of people's memory presumably.

Anthony Wagner:

Not yet, but we're a little further along than just the category now, or at least what you described sort of the sensory modality. And this is built on a wealth of work over the last sort of 20 some odd years or so where in the visual neurosciences as you're noting, there are patches of our cortex, parts of our brains that seem to prefer certain kinds of stimuli. It could be based on modality, auditory cortex versus visual cortex versus the sense of touch, somatosensory cortex. But within a modality, it also could be based on the categories of things. Take for example vision. There are patches of cortex that seem to be differentially responsive to when we perceive faces versus when we view common objects versus when we see landmarks and other spatial stimuli versus looking at visual words.

A number of years ago, Nancy Kanwisher did a nice sort of demonstration that, "Well, given that you've got these patches of cortex that seem to prefer faces say relative to places, can I look at where there's activation in the brain and can I know, are they looking at a face or a place? Are they imagining a face or a place? Are they remembering a face versus a place?"

And you can do this by looking at where there's activation present, but you can also just take the pattern of activation across say the entire visual cortex and you can feed that pattern of activation into a statistical classifier that teaches the classifier, this is the profile when somebody is seeing a face versus a place. And now you can ask, "Does that pattern of activation get reinstated when people are remembering?" In my lab amongst others, we've shown that we can know essentially the strength with which a prior pattern is being reinstated at that moment in time. This is what we mean by neural time travel. The extent to which the pattern when we're remembering looks more like the pattern during the experience where we perceive or we experience the memory as being more vivid or more confident in our memory and we can act more quickly based on that memory.

So to your question about is it just about categories, say faces or places? There are signals, one experience with one individual, one face versus a different experience with a different individual. There are small but detectable differences even within category. And it's there where I think the combination with sort of modern artificial intelligence, the emerging AI advances will help us make further progress here.

Nicholas Weiler:

Well, I'd love to turn to some of your recent research on sort of what determines how well these patterns get laid down and retrieved. And one of the key factors that you've published on is this idea of multitasking, which as I think many people have heard, we don't really multitask. We task switch and sometimes we task switch very rapidly and other times we task switch more slowly. So can you tell us, I mean, what have you found about how multitasking which we're all doing much more than we used to is affecting our ability to remember what we're doing?

Anthony Wagner:

Of course. That story relates to this broader fact about our memory systems, which is we know that the hippocampus is fundamental for building and therefore being able to allow us to retrieve memories for experiences. Absent the hippocampus, you will be densely amnesic. You will never be able to build a new episodic memory and be able to consciously remember a new experience. This is also the system that's affected in most individuals who have Alzheimer's disease and that's why one has essentially a memory deficit as the first cognitive symptom, but we know that the hippocampus isn't working in isolation. Part of our discussion already has been this notion of there's this kind of dance between the cortex and other brain regions that represent the features of events. Those features come into the hippocampus that get bound together cues at the time of retrieval into the hippocampus, drive reactivation of the hippocampal memory trace, and then the hippocampus drives reactivation back out in the cortex.

But that's only part of it. We also know that other systems modulate, interact, influence what gets stored in our episodic memories and also our ability to retrieve it. These are processes that heavily depend upon networks in our frontal and our parietal lobes, networks that support goal-directed attention, deciding to attend to certain dimensions of this experience, systems that allows to represent our goals. What am I trying to achieve at this moment in life?

And through that sort of representation of our goals, these frontal parietal networks can orchestrate, can shape what's going on in our minds and brains, selecting aspects of experience that are relevant to our goals and perhaps filtering out irrelevant aspects. And so this is why you and I can have what we think is a shared experience because what gets into your mind, your memory, and what gets into my memory is going to be influenced by what you happen to be attending to and selecting and relating to other knowledge you have and your theory about what's happening during that moment in that event. And that might overlap sometimes with what I'm doing in terms of my frontal parietal attention and goal-directed sort of networks, but my mind and brain are doing different things.

Nicholas Weiler:

Well, we talk about people having selective memory as a pejorative, but really we all have selective memory. That's the way it works.

Anthony Wagner:

We're all having selective memory. We'd be overwhelmed. Perception by definition, we're getting all of this sort of energy, this sort of sensory signals through our ears and our eyes and other sense systems. We can't perceive it all. We can't tend to it all. And through attentional selection, we're in essence gating what gets into memory. So these kinds of influences of attention and our goals and our theories about the world about what's happening, it also influences what we bring out of memory. Those same attentional and executive function is sort of a jargony term or cognitive control. These other processes, they actually influence what we bring online, bring to mind. And sometimes our prior knowledge distorts what we're bringing back to mind. This is where false memories in part come about.

Remembering is this reconstruction. You've got this reactivation and replay coming out of the hippocampus trying to drive patterns back out into the cortex of the information. But those patterns are also being influenced by what we're currently attending to, our theories about what might've happened in the past, our theory about who this individual is that I'm interacting with, are they friend or foe, or our memories are reconstructed and in that way somewhat vertical but often also quite distorted.

Nicholas Weiler:

How does this play into this question of multitasking? What's going on when we multitask and how does that affect our memory?

Anthony Wagner:

So I go into this because, and at the outset I want to note, and this is a line of work that really is kind of developed in my lab over the last almost 15 years now, but it really started with my late colleague Cliff Nass, who was a professor in communication and Cliff and I had a master's level student who was sitting between our two labs and Cliff and [inaudible 00:14:17] they were interested in this very question, how does interactions in everyday life with the modern digital media landscape... and this was work that was begun in around 2008 or so, pre-iPhone and pre a lot of social media and the other kinds of digital media we have. They were interested in this question of, is attention and changes in attention occurring or at least related to this media multitasking behavior? And so the three of us sort of conducted some initial work asking Stanford undergraduates to tell us a bit about their media use behavior and quantifying and determining which of the undergraduates tended to be heavier media multitaskers in any given hour.

They're not really multitasking. They're task switching between many media channels relative to others who on average tend to switch between media channels much less frequency. They might have one or two types of media, TV, audio, text, searching the web, et cetera. They might only have one or two types of media that they're engaging with in a given hour. And that early work provided some initial evidence that individuals who are heavier media multitaskers in every life tend to struggle with attentional control.

And this is the case even when in our experiments the individuals are asked to only do a single task, but the task requires attention to select things that are relevant and to filter out things that are irrelevant. And in the last sort of 10 years or so, my lab is sort of because we know that these mechanisms of attention impact learning and remembering, building and retrieving of episodic memories, we began to ask the question, are heavier media multitaskers less successful in building episodic memories, less successful in being able to retrieve episodic memories? We conducted a series of studies that actually make clear there are small but significant reductions in episodic memory performance in individuals who are heavier versus lighter media multitaskers.

Nicholas Weiler:

And this doesn't only hold true during multitasking. This is even if they're not multitasking.

Anthony Wagner:

This is when they're not multitasking. That's the key. You're in the lab, you're presented stimuli to learn. That's your only task. Your task is only to try to remember, to engage in episodic retrieval. And under those single tasking conditions when you're a heavier medium multitasker in everyday life, you perform a little bit worse than if you're a light medium multitasker.

Nicholas Weiler:

So what do you think is going on in the hippocampus or in the interaction between the hippocampus and the cortex?

Anthony Wagner:

Right. Why did I go into this narrative about attention and goal-directed behavior and the fact that episodic memory is not just a byproduct of hippocampal cortical interactions, but hippocampal function. It's those core mechanisms of memory supported by the hippocampus interacting with mechanisms of attention and goal-directed cognition. And so what we observe in most recent work, you can measure somebody's attentive state, arousal or whether or not they're sustaining attention on the task at hand.

Nicholas Weiler:

I love this. You're doing this with the pupil dilation, right?

Anthony Wagner:

With pupil dilation, scalp measures of electrical activity that is related to attention, sustained attention. And what we see is that in everybody, whether you're a heavier or light media multitasker, in the moment just prior to when you're asked to engage in a goal-directed task either to learn to build a new memory or to retrieve an already encoded memory, when there's a lapse of attention, you are less likely to learn or you're less likely to be able to use retrieval cues to reactivate a memory. We all have these sort of moments again where we feel like we're remembering quite well and other times we've got these periods of frustration and, "Boy, I can't seem to bring sort of this name back to mind or whatever." Moment-to-moment variability in our arousal and attentive state has small but significant impacts on how well we're learning, how well we're remembering.

So that's within individuals and now between individuals, those individuals who are heavier media multitaskers who were more likely to suffer a lapse of attention just prior to engaging in goal-directed learning or goal-directed retrieval relative to lighter multitaskers, lapses of attention seem to negatively impact our ability to represent, "What is my goal at this moment in time," and use that goal as experiences unfolding to build a memory or use that goal with retrieval cues to try to reactivate a memory. Lapses of attention have a cascading effect on other kinds of systems that then interact with learning and remembering. And our data suggests that part of the reason for this negative relationship between media multitasking and everyday life and worse, episodic learning and remembering, is through this increase in the probability of suffering a lapse of attention with cascading effects.

Nicholas Weiler:

And one thing that I thought worth bringing up is that you point out this effect is mostly true for the media multitaskers when they're doing simpler tasks compared to the light multitaskers. So people who don't multitask a lot do a lot better at these single focused memory tasks, but then when you ask people to do a more multitasking related task, they perform equally. Do I remember that correctly?

Anthony Wagner:

Well, so the data on whether heavier media multitaskers in everyday life actually multitask better... When I talk about this work to my undergraduate students, not all of them, but many of them look at me and say, "You're an old man. We're from a generation where we grew up constantly multitasking, task switching. Our minds and brains, we've kind of built up the neurocognitive sort of muscle, the ability to actually do this." The literature is kind of mixed to be honest.

When you multitask, you're task switching and when you switch between tasks, there's actually a cost. You're a little bit slower, you're a little bit less accurate. We call that the switch cost. Heavier media multitaskers in our study weren't better at multitasking. If anything, they were a little bit worse. There are a few other studies that don't see any difference in switch costs between heavy and light and there are a few studies that may here or there, maybe they're actually a little bit better, the jury is out. But back to your point about difficulty though, we've spent a lot of time talking about episodic memory, but there are other forms of memory that also interact with episodic memory. One form of memory is working memory.

Nicholas Weiler:

Sort of your mental workspace.

Anthony Wagner:

The mental workspace, exactly. And we know there are sort of capacity limits on working memory. We can't keep everything active in mind. Our attention is limited. You can only keep so many representations active before you fail to attend to one of them for some period of time and it's been lost. It's decayed, it's lost out of working memory. So to your question about difficulty and media multitasking, what we and others see is that heavier media multitaskers on average perform a bit worse on working memory tasks. They can keep fewer things in working memory relative to lighter media multitaskers. But this is the case when we're asking everybody to keep just a few things in working memory, not a lot. When you push all participants to their cognitive max, we give them a lot of things to try to hang on to in working memory. Turns out the heavier and the lighter media multitaskers don't differ in their working memory performance.

And this is actually what led us to this attention elapse idea, which is it's not that heavier media multitaskers memory abilities are changing in terms of their fundamental capacity or their ability to learn and remember, it's that when they're in a context in which there's some cognitive slack, they kind of know, "Okay, yeah, I can do this task and it's not that hard," it appears that they let their attention, not necessarily volitionally, but attention lapses more often in the heavier media multitaskers relative to the light. But when they and the light media multitaskers know they're in a context where, "I really need to focus. I need to harness my attentional abilities to really sort of achieve what I'm being asked to achieve," they actually do about the same. And so this gives rise to this hypothesis that there's a balance between goal-directed focused attention and more reactive, reflexive attention.

And this balance might be a balance between focusing on the things you want, sort of exploiting the current environment versus letting your mind explore, discover, be open to other signals and thoughts that might capture your attention and you can react to them and perhaps to get a reward because it's that social media sort of chime or some other sort of reinforcement. And so media multitasking might be changing this balance between exploratory versus exploitative behavior in a cognitive sense or between reflexive and goal-directed top-down attention. And in a situation where there's some cognitive slack, you pay a very small price for that. You perform a little bit worse, you learn a little bit worse, you remember a little bit worse, but it doesn't mean your core capabilities are in fact now at a lower capacity. It's just with some slack you let attention lapse and therefore you perform a little bit worse. But when push comes to shove, you perform comparably.

Nicholas Weiler:

Well, I do have one more question that I'd love to end with just briefly. So there are a couple of threads here. So one, I want to mention that there's still more work to do to understand if multitasking is causing these differences in processing or if individual differences in attention control and working memory and so on influence whether people tend to be heavy media multitaskers or lighter media multitaskers. Without going into detail on that, is that generally right?

Anthony Wagner:

That's generally right. There's very little data on what is the causal direction. The little bit of data out there suggests that both factors are in play.

Nicholas Weiler:

Right. So then the last thing I want to ask before we close is what's in our control? Is there a short list of things that we can think about to help protect our memories or optimize our ability to learn new memories and retrieve the memories we want to when we want to? What do we know about that?

Anthony Wagner:

Yeah, we know a lot. There's still more to be discovered, but there are many factors that are going to shape how likely it's that we're learning and learning effectively. And some of them are under our control. So if you just think about the experience, the event unfolding, the time of building that memory, how we're attending really matters. And so one thing we can do is be strategic in how we're harnessing attention, how we're using our attention as experience is unfolding. If it's really important to remember something about an event that really matters to you, be sure that's the focus of your attention. And there's all kinds of distractors out there and trying to minimize those distractors, maybe put them away, not have them present, but also just strategically harness attention. There's another very powerful thing that we can do. It's actually really still not taught as much as I think as a learning and memory scientist.

I think it should be taught. And many of my peers in the field kind of share this view, which is having had an experience, having laid down a memory trace, one of the most powerful ways to ensure that that experience is going to persist over time and you can access it and remember that event well into the future is actually to engage in retrieval practice. Test your memory, whether it's reminiscing about the event, telling a story with friends and family over dinner or at points in the future, reflect back on it, try to remember it, try to reconstruct it, bring it back to mind.

Within the context of education, we actually know there's now beautiful studies that show if you have students try to learn some new information like a text passage in one condition, they study that text passage three, four times and at each time they're reading the text passage, you ask them, "How well do you think you've learned this?" In the other group, you only read the text passage once. And the other three learning opportunities is just trying to retrieve, remember as much as you can from that text passage and you ask, "How well do you think you've learned this?" The group that reviewed the passage four times, they thought they learned really well. The group that only studied it once but then engaged in retrieval practice, they thought they didn't learn it as well.

You test memory for that passage now after the fourth study episode either minutes later or a week later. Minutes later, the group that cram that read the passage four times, they actually outperformed the read it once, retrieve it three times, they outperformed them a little bit. One week later, the group that only read the text passage, they'd lost about 50% of the information they were able to retrieve after the study episode. The group that only studied it once and then they tested their memory multiple times, they retrieved 80-ish, 90% of what they were able to remember minutes after the passage. They held onto it for a week and they well outperformed the group in the study only.

This science of learning and the power of retrieval practice for learning and long-term retention, it requires effort, it requires strategic behavior, but it's something that you and I can do that harness sort of the power of our episodic memories to really hang on to these meaningful life moments and be able to use them and remember them in the future, bringing us perhaps happiness, joy, but also perhaps being able to use that knowledge to behave optimally and to navigate our worlds.

Nicholas Weiler:

Well, Anthony, thank you so much. That's a wonderful place to end, and I hope that listeners, if you want to remember everything you learned here, you can go and reconstruct this for your friends. That'll help you remember what we've talked about today. So Anthony, thank you so much for joining us on the show.

Anthony Wagner:

Thanks, Nick. This was really fun.

Nicholas Weiler:

Thanks again to our guest, Anthony Wagner. We'll include links for you to learn more about his work in the show notes. If you're enjoying the show, please subscribe and share with your friends. It helps us grow as a show and bring more listeners to the Frontiers of Neuroscience. We'd also love to hear from you. Tell us what you love about the show or what you hate or share your suggestions about the next frontier of neuroscience you'd like to hear about. You can reach us on social media @StanfordBrain. From Our Neurons to Yours is produced by Michael Osborne at 14th Street Studios with production assistance by Morgan Honaker. I'm Nicholas Weiler at Stanford's Wu Tsai Neurosciences Institute. See you next time.