 OK, so at UCLA, I study how learning and memory processes affect eating behaviors and how eating is learned about and remembered. And so as a memory researcher, something that I've always been interested in is just this basic question of, why is some information better remembered than others? And so this is something that probably everyone here has some familiarity with. We know that there are some things that can be quite difficult to remember, such as maybe where you left your keys or the name of someone you just met, and yet there are other things that are really easy to remember. So this could be maybe the phone number that you haven't used in several years, but would easily come to mind. And so we know that there's this large variability in terms of memory performance. The question is, why? So if you were to consult your local memory textbook, you'd find a lot of diagrams and models that look something like this. And the basic idea with all these models is that information comes into the memory system. Depending on the model that you're choosing, the information is going to get processed slightly differently. But basically, how well the information makes it through these channels is going to dictate the strength of the memory as it's transferred to long-term storage. And so an issue that I've always had with these models of memory is just this basic question of, what is input? It's this very large blanket term, and it doesn't really lend itself to the possibility that the memory system might be biased towards remembering some kinds of information over others. And what this really speaks to is a larger debate that's been going on for several decades in cognitive science, and that is, are cognitive abilities, in this case memory, are they just this blank slate, or are we systematically biased, for memory at least, to remember certain information over others? And so some of the work that I and a number of people have been doing for the past several years has been trying to document that there are these systematic biases to the human memory system that reflect evolutionary pressures. So a really nice and recent example of this is that we find better recall for objects when they're described as being touched by someone who's sick compared to someone who's described as being healthy. And so the way these experiments work is participants come into the lab and there are these different trials, and in these trials there are objects and the objects are just described as being touched by different people. And so in the first trial we have a coffee cup, which is described as being touched by someone who has a high fever, but then we also have trials like an object like this hat that's described as being touched by someone who has green eyes. And so participants do a number of these trials and then they're just asked to remember all of the different objects that they saw and of course this is all nicely counterbalanced so you've got all the different variations of objects and descriptions. And what we find is that we have better memory for objects when they're described as being touched by someone who's sick compared to someone who's described as healthy. And you can get the exact same effect if you get rid of the descriptions and you just put pictures of faces and you say this person touched this object but then you vary some of the faces just look like healthy individuals but then some of the faces are showing signs of sickness. So maybe like a runny nose or watery eyes, you get a similar effect. And so the idea here is that it looks like the memory system is biased towards tracking potential sources of contamination. And this makes perfect sense from an evolutionary perspective. The first sort of documentation of these biases in the human memory system is known as a survival processing effect. And so here what the researchers did was they had participants come into the lab again and they told them that they're gonna see a wordless and they had to rate those words based on their relevance to this imagined survival scenario. And they're told that, you know, they're supposed to imagine themselves stranded in the grasslands of a foreign land without any basic survival materials. And over the next few months you need to find a steady supply of food and water and protect yourself from predators. So that's the encoding scenario that they're given. Then they're shown a series of just regular concrete nouns but hopefully they're imagining some sort of situation like this and they're thinking, how am I gonna use all these different words in this survival scenario? And then the control group that this gets run against is a scenario where participants are asked to imagine that they're planning moving to a foreign land. They need to locate a new home and purchase new belongings. And so you've got very similarly worded scenarios but hopefully the participants are thinking of something that looks a little bit more like this. And of course the emphasis on this scenario isn't really anything of fitness value. And the nice thing about this design is that all the participants are given the exact same words in the exact same order. Everything's exactly the same. It's just the encoding scenario that we give to them. So then we ask them how many words that they saw and just write them all down and we get much better memory performance when items are encoded in the survival scenario compared to the moving scenario and I'll get to the pleasant scenario in a second. But it's not anything, the important thing to keep in mind it's not anything about the words themselves. So these are just random nouns and they're not any more relevant to the survival task than to this moving task. It seems to be something about the way people are encoding this information which is yielding the better retention. And so this kind of made a big splash in the memory world because we know about the different encoding scenarios that can yield really good memory performance. And what you see when you compare those different encoding scenarios to the survival scenario, you get sort of this off the chart memory performance when words are processed in this evolutionarily relevant manner. And so my contribution to this work, this is what I did as an undergraduate, was I reasoned if there's a survival processing effect and the idea behind it is that there are these evolutionary pressures that have affected the memory system, then there should be some sort of memory bias towards raising children. That's a very important part of evolution. And so we should see a similar benefit if we put the emphasis on raising a child. And so we just created a very similarly worded encoding scenario, but now we say that you've just parented a child and you have to provide care and nourishment to that child and we compare that to the survival group and the pleasantness control group. And the pleasantness control group, it's just sort of a gold standard way of encoding information to yield good memory. And what we see is we get a similar benefit to memory when words are encoded in this parenting scenario, it's basically just as strong as the survival processing scenario. So this was a neat finding, but we wanted to take it a little bit further and we asked this question of how general is this parent processing scenario? Is it specific to just your own child? Or if we describe the child as being adopted, and thus it's a little bit less fitness relevant, will we see any differences in memory performance? So we just took our original parenting scenario and we tweaked a few words and now it read that it was an adopted child and we compare the two parenting scenarios and then we also had participants imagine that either their child or an adopted child was doing the survival situation, so we call it third person survival processing. And what we find is it's pretty interesting, we see this main effect of biological relatedness affecting memory performance on these two tasks. So this seems to be a memory bias when these scenarios are relevant to your own children compared to an adopted child. And so I think these findings can be nicely summarized by this idea that the evolutionary fitness relevance of a stimulus, a behavior event, we can just say information, it seems to potentiate the ability of that information to be later remembered. And so if we're thinking about modeling human memory from a computational standpoint, it would be nice to incorporate this factor, we've decided to call it omega, that somehow quantifies the fitness relevance of information and then we'd predict there's this linear relationship between the omega value, which is again, this is the quantified fitness relevance of an event and the likelihood that that information will be remembered. And so you could use potentially this omega scaling factor to scale existing models of memory and you'd be able to explain a lot of the existing data that I just reviewed. But of course, anytime you introduce any new type of theory or model, not only do you wanna explain existing data, but you also wanna generate novel hypotheses and then test them with empirical methods. And so it's nice about the scaling factors that it makes very clear predictions about the way memory will proceed. And so basically as long as we can vary information on its fitness relevance, we should see differences in memory outcomes. So that got us interested in looking at memory of eating because eating is this really obviously evolutionary important behavior that people do every day. The other reason why we're interested in eating is that while these imagined scenarios are interesting for theoretical reasons, it's sort of hard to know what to make of these imagined scenarios. They're a little bit weird and it's not entirely clear what, how they relate to real world behavior. So we decided let's just actually look at real world behavior. And suggestive that there might be evolutionary pressures on the human memory system that bias it towards remember eating events or food relevant information. There's a really neat study done by researchers at Santa Barbara a few years ago where they had participants walk through a farmer's market and they basically sampled all the stands. They got a little bit of food at each stand. And then the researchers brought the participants into a tent and they had them just recreate all the different stand locations of the farmer's market. And what they find is this really interesting result where the memory accuracy, so on the y-axis we have memory error. So lower values means more accurate memory. More accurate memory for the location of the stands was associated with the caloric density of the food that was sold at each stand. And so this actually fits exactly the prediction that our omega scaling factor would make as you vary the fitness relevance of information in this instance caloric density of a food item we're seeing better memory performance. So that's nice and encouraging. But there's another reason why we were interested in looking at memory of eating. And that's because it turns out that memory and eating processes are highly intertwined. So memory actually turns out to play an important role in moderating future food consumption. And so if I were to ask a lot of people what causes eating, probably a standard answer would be something that reflects a homeostatic model where there's some sort of lack of nutrients. There's a detection of that lack of nutrients and so then a signal gets sent to encourage eating. Then as you eat, there's the detection that nutrients are coming into the system so there's some sort of stop signal that will stop eating and that's a nice homeostatic model. And certainly we know that such a model exists and we've worked out the physiology quite well in both rodent and human models. And yet a lot of people are starting to realize that a purely homeostatic model of eating can't explain all eating behaviors. And specifically it can't explain eating behaviors that we find particularly problematic. So if we eat just because we're lacking nutrients that doesn't explain why people overeat or why they eat when they're not hungry or lacking nutrients. So a number of people have started to suggest that there are other factors that influence eating and one of the recent ideas that's been gaining popularity is that perhaps memory is motivating eating behavior. And so this idea got started by looking at patients with amnesia like the famous patient HM and a phenomenon that they would observe with these patients is that you could serve them a meal and then they'd finish the meal, get rid of it and then they'd very quickly forget about the meal because of their condition. And then what you could do is you could serve them the exact same meal and they would readily accept it and start eating. And you could do this over and over again in a cyclic fashion. Similar to this, their self-reported hunger levels didn't seem to correlate strongly with their time since their last meal. And so this seems to suggest that something about having an intact memory of your most recent meal is impacting how hungry you are and how much you eat at a subsequent meal. Related to this, we can get similar effects in healthy individuals. So this isn't just some quirk of patients with amnesia. So you can do things that manipulate the strength of someone's meal memory by either having them watch TV or play video games, essentially anything to distract them. And not only do they have a reduced memory of the eating event, but they'll also eat more food in the future. And this is obviously concerning because we know this type of distracted eating is rising in prevalence. People often eat while distracted. In fact, we have entire food genres that are dedicated towards this sort of eating in front of a television screen. And yet the empirical data is quite clear. This style of eating can promote future overeating. We've also known for a long time that memory processes and obesity are associated with each other. So you can put rats on a Western diet and not only do they quickly gain weight, but you also very quickly can observe these deficits in memory performance on a large number of different tasks. Conversely, you can lesion the hippocampus, which is an important brain region for memory formation. And we see increases in food consumption and then weight gain. And you get the exact same types of effects in humans. So BMI is negatively correlated with episodic memory performance. And then in children and adults with obesity, we see reduced gray matter in the hippocampus and prefrontal cortex, which are areas that are associated with memory formation. Now there's this obvious issue of causality here. So is it that people with just poor memory performance, assuming there's some sort of distribution, is it that those who have just generally poor memory performance are more susceptible to gaining weight and obesity, or is obesity causing these deficits in memory performance? There seems to be a little bit of evidence of both, but there also seems to be evidence that these things are working together in a cyclic fashion such that eating a Western diet is disrupting hippocampal function. We know the hippocampus is important for memory formation and that memory moderates future eating. And so if that is now messed up, you're eating more of that Western diet. And so you've got this vicious cycle of obesity and cognitive decline. So bringing this all together, we got really interested in this question of just our meal memory special. And the idea behind this is that if memory of a recent meal is so important for moderating future food consumption, is it better remembered than other similar procedural tasks? And while you might think that memory of eating might be poor because you have that experience of having difficulty, maybe remembering what you ate a few days ago, those of us who study procedural memories know that most behaviors aren't all that well remembered. So if I were to ask you how many stop signs you pass on your way home from work every day, you'd probably give me a fairly inaccurate answer. That's just the way the procedural memory works. And so our question was, is memory for eating no different or because of this important role that memory plays in moderating future consumption, is it actually an enhanced form of memory? And then the other reason we were interested in this question is what I alluded to earlier and that's this larger theoretical question of, is the human memory system shaped by evolution to preferentially remember some types of behaviors over others? So our hypothesis is that meal memory should be particularly strong. And so the way we have to test this is comparing memory of eating compared to some sort of similar task that it isn't eating. So the way that we do this is we bring participants into the lab and they sit at a setup that looks something like this. So we've got a cardboard box with these different symbols around it, a bowl filled with either M&Ms or beads and then this narrow glass container. And we tell our participants that we're interested in studying memory of verbal information while distracted. And so while they watch this film, they're gonna have to either eat an M&M or move an M&M from one bowl to the container or move a bead from the bowl to the container every time a tone is sounded. And so as they're watching this film, a tone gets sounded on this random schedule, happens 30 times, and they either are picking up the M&M and eating it or they're picking up the M&M or the bead and putting it in this little glass container and it makes a nice little rattling sound as it goes down. And so then participants do this task. We then move them to a new context and they do a little distractor task. And then we ask them questions about memory for information that they heard during the video. We ask them to recreate this task context and we ask them critically, how many times did they perform this task? And what's nice about this procedure is that we know every single person did their respective tasks exactly 30 times. So we know that because they're only toned, they're only 30 tones throughout the video, but we can also just weigh the bowls as these people leave and confirm that they actually did the task exactly 30 times. So we know how many times they perform the task, they report to us some sort of number, and then we take the absolute value of the difference of those two values. And so we're looking now at memory error for how many times they did the task. And what we see quite strongly is that you get much better memory when you're doing this eating behavior compared to these nearly identical behaviors that aren't eating. And so they're lacking the fitness relevance. Similar to this, when we ask participants to recreate the task context, we get better memory performance. So again, this is error for recreating the context, better memory when subjects are eating instead of these other two similar tasks. Now this isn't quite significant, it's sort of hovering around statistically significant. So we're not entirely sure how reliable this effect is, but the general direction is fitting our hypothesis. So the important evolutionary aspects of this eating task, where you're doing it, and specifically what you're doing are well remembered, but then questions about how long the video lasted or information that was presented during the video, we don't see any differences between the groups. So that really nicely fits our hypotheses. Another really interesting finding that we found was that BMI was negatively correlated with contextual memory performance only in the eating condition. So only when participants were in that condition where they're eating food, we actually see better memory for recreating the context with participants of higher BMI. Now I should preface this with, we didn't actively recruit high BMI individuals, this just fell from recruiting a large number of subjects, and so we need to replicate this further. But an interesting theoretical understanding of these results is that perhaps overweight individuals are encoding more of their eating environments, which then can make them more susceptible to cues that are associated with that eating environment. And so this is known as the Incentive Motivation Theory, it's quite popular these days. And so the way this works is that we know every time you have an eating event, there are these cues that come right before the food that you eat. And what can happen is that these cues and that food outcome can enter these Pavlovian relationships with each other, such that this cue, the McDonald's sign, gets learned as a reliable predictor of a food outcome. And as a result of learning this relationship, the next time you encounter that stimulus or a stimulus that resembles a conditioned stimulus, it's gonna activate a memory of the food outcome and that can motivate desire to consume food. And so we know this is how a lot of addiction-like behaviors happen. This nicely explains why addicts can go through a rehab program somewhere else, they feel really good, no desire to use the drug, and then they come home to their hometown and they start experiencing all these cues that were associated with the drug outcome and now all of a sudden they're triggered to obtain the drug outcome. So we think that might be happening with eating behaviors. And so what our results suggest is that possibly eating, sorry, obesity is related to stronger cue food outcome learning. And again, we have this really interesting question of causality. So is there just some sort of distribution of people's ability to learn these cue food outcome associations and those that are really good at it are now more susceptible to gaining weight? Or is obesity somehow changing the learning preferences towards learning and being better at learning these cue food outcome associations? So these are interesting studies that we'd like to look at in the future. Now one issue that we had with our findings was a concern that our M&Ms contain glucose and it turns out that glucose is actually a really important substrate for the hippocampus. It sort of fuels its performance. And so there have been studies that shown that large amounts of glucose ingestion either before or after a task can improve memory performance. Now this is usually a very large dose of glucose nowhere near what we're using in our M&Ms but nonetheless this is still a potential physiological explanation of our results. So we're arguing that there's something about this eating behavior that's really evolutionarily important and that's what's causing this benefit in memory but it's possible that maybe it's just because the participants who are eating are getting this physiological ingestion of glucose. So what we did was we brought new participants back into the lab and we had them all perform the bead moving task. So again a task that's of no fitness relevance but we have some of them drink a liquid solution before and after the task that contains sugar and it contains the exact same amount of sugar that they would have had if they would have consumed those 30 M&Ms and then we compare that to a group that drinks Stivia so this is a sweetener that doesn't have glucose or just a water control group. So now we have three conditions. All of them are performing a behavior that we think is low on a fitness value but some of them are getting quite a bit of glucose as they do the task. And what we see in line with our predictions is no differences in memory performance for these three groups. And we also when we look at the group that was drinking the sugar, we don't see this association anymore between BMI and memory for the eating environment. And we don't get any differences either for the content of the video or the duration of it. So putting this all together, we've developed what we're calling the memory of eating task or the meat. And what we like about this is that it seems to be this nice simple behavioral paradigm that will enable us and others to systematically study memory of eating behaviors. And so, yeah, this is our setup. What's nice about this procedure, not only do I think we've already asked and answered a few interesting questions but will enable us to answer a large number of new and novel questions. So things like how is memory of eating influenced by the food item that's being consumed? So we can do things like look at is eating 30 peanuts just as well remembered or better remembered than eating 30 pieces of popcorn. So this is a question that memory researchers probably wouldn't consider because again, they're relying on these models of just information coming into the memory system and there's not much discussion about what is that information but our omega scaling factor makes this prediction that well, the peanuts are more calorically dense, they're of higher fitness value and so we should see better memory performance when you do this task and you're eating peanuts compared to popcorn. And of course, we'll see how the data plans out. We can also look at how pace of eating affects memory of eating. And so by using this task because we cue participants on when to eat, we're effectively getting to set the eating pace and so what we can do either by spacing out or collapsing the distribution of the tones, we can look at how fast versus slow eating affects memory of eating and that might have downstream effects on future food consumption. And again, all these materials are really cheap, they're easy to use. We've put the Python code that controls the video and the beeps all online and made it freely available and so we're hoping that other memory researchers or eating researchers will adopt this paradigm and we can study it together. So in conclusion, I hope I've showed you that the human memory system is biased towards remembering fitness relevant information and this should inform how we formalize models of memory and how we conceptualize the structure of the human mind. We've shown that meal memories seem to be particularly strong and they may be contributing to weight regulation and obesity may be related to an increased ability to learn cue food associations but of course we need to run replication studies with higher populations of, greater populations of high BMI individuals to confirm or deny these findings. So with that, I'd like to thank Aaron Blaisdell and Janet Tamayama who are my advisors, a team of undergrads who have helped with a lot of this work, my funding source, UCLA and the Dish Lab for their support and I'd be happy to take any questions. I'll kick off a question. Sure. Given that this is the ancestral health symposium and a lot of us are interested in taking what we understand scientifically about the evolutionary mismatch of our modern environment and how our physiology functions including our mental physiology, what could be some implications of this research for how we can then use this in our own daily lives? Yeah, so some of the memory researchers, especially the ones that have pioneered the work that memory of eating seems to influence how much you eat in the future, have been trying to promote eating practices that basically improve meal memories and so they've done some work just basically at a psychological level, queuing people just to focus on their eating events so the opposite of distracting them, having them really focus on the sensory aspects of their eating events seems to reduce how much they eat in the future and they've also looked into if you can choose food items that perhaps you need to chew for longer, that might result in better memory of eating and so I would say any type of manipulation that you can do that makes you more focused on your eating event will probably likely make you feel fuller and eat less in the future. Based upon your research, do you think that dessert, so having a meal but then having dessert, would that be the infusion of sugar to make that meal more memorable? That's what I haven't thought about it that way. Yeah, that's possible. Given we know that glucose ingestion can improve memory, so that's possible. You could eat something and then ingest a lot of glucose and then that might make the meal memory more stronger but of course now you've ingested a lot of glucose which has its own consequences. But what's nice is this area of memory of eating, like we really don't know the factors that influence memory of eating and so we're hoping with this paradigm and other researchers are now getting on board with this, we can actually start to determine the factors that influence memory of eating. I found the results in the HM case really surprising but also very persuasive. Do you, can you tell me more about that? I mean, are there some limits? How many times could he eat a meal? So from those papers, they're old papers but I've read that they can give him about 1,000 calories so I think they do 300 calories per meal and by the third one or fourth one they just stop it. At some point he'll not eat anymore but the surprising thing is, is that they will continue to eat in this way. Right, thank you. Related to that actually I'll just throw in a slide. So this was looked at in a really nice controlled model with modern neuroscience techniques. So basically what we have the ability to do now is basically zap memories from a living mouse and so what they did was they had some mice eat and then they got rid of the memory for eating and what that did was it resulted in quicker initiation of a future meal and then they ate much more on that future meal. And so this is a really nice controlled model of that old effects with HM and the amnesiac patients. And when you say zap the memory here, are you just doing something to prevent the long-term potential? Yeah, you're disrupting the consolidation of the memory. Wow, cool, thank you. The person's impression of the fitness relevance of some food can vary from person to person. So if you ask someone here how healthy the healthy whole grains in front of them are, you might get a different answer, very different answer to somebody off the street. Is there any indication of someone's understanding of the fitness relevance versus the actual fitness relevance of the food stuff? Yeah, that's a really good question. So when we've been thinking about the scaling factor, something that I've decided is at first it was just going to be the fitness relevance of an event or information, but it seems like it has to be the sort of perceived fitness relevance, like it might have to do with actually individual perception of the fitness relevance of an event. And so there's actually the finding I was talking about about you show pictures of a face of someone who touched an object and sometimes the people look sick. If you describe the people who look sick as actors who are wearing makeup, you get rid of the effect. And so to me, that suggests that there's something about perceiving it as being of more greater fitness value that's affecting the memory. What I think, so we've had this question too, is we want to vary food items based on their fitness relevance and see if we get differences in memory. The question is like, well, which food items do we choose? We've thought a lot about this. I think ultimately the approach that we're going to take is just trying a lot of different food items, seeing if we get any differences in memory, just seeing differences in memory for different food items would be really interesting. And then we can start to look at the characteristics that emerge of the food items that are very well remembered. So perhaps it's something about salt content or fat content could be a number of things. I think we're going to see the way the data plays out and then we'll arise conclusions from that. That's lunch. Thank you.