Periodic Reporting for period 2 - REPLAY (The Function of Hippocampal and Cortical Memory Replay in Humans)
Période du rapport: 2021-10-01 au 2023-01-31
Every experience influences our thoughts and actions. Imagine, for instance, that a neighbor's dog barks at you. Most likely, you will instantly stand back, and your memory will be updated immediately to encode that your neighbor has a scary dog. In other words: when we are awake and active, our brain usually reacts immediately to the events around us, initiates a reaction and updates our thoughts and memories. But that is not everything. Sometimes, we internally re-experience the same situation at a later point in time – for instance when we dream about it a few days later. The REPLAY project is all about this “second life of our experiences”, and how mentally reactivating a past experience influences our decision making and memory. Why does this happen, and what effect does such a “replay” of experience have on us?
Research in rodents has first discovered “memory replay” in the brain over two decades ago. In 1996, Skaggs and McNaughton found that while rats were sleeping, brain cells in their hippocampus were at times quite active, and seemed to retrace the animals’ previous trajectories in a maze. Curiously, this “mental travel” was much faster than it had occurred in real time. This phenomenon is known as replay, and has become a major focus of neuroscientists and even artificial intelligence (AI) researchers over the past decades. We know for instance that in rodents replay is prevalent during wakeful resting and sleep, and that it is related to memory, planning and reward processing. An interesting parallel has occurred in AI research, where machine learning algorithms not only learn from the present, but also replay the past, to enhance performance.
While it is generally thought that replay is an important aspect of brain function, little is known about replay in the human brain. Fast neural processes, such as relay of previously traveled paths as found in rodents, is hard to measure in humans, where we can’t surgically implant electrodes in the hippocampus, as it is done in animals. REPLAY will address this problem of little knowledge about replay in the human brain. We will use a novel method to analyze functional magnetic resonance imaging (fMRI) data, that for the first time allows finding evidence for replay in fMRI patterns that we can record during rest or sleep in healthy volunteers.
The project REPLAY is will enhance our understanding of cognitive aspects of brain function, in particular learning, memory and decision making. Basic science that can help us better understand the human brain is important in a number of ways. Learning, memory and decision making play a big role in our everyday lives. In the context of healthy cognition, in the long run our research could benefit the attempt to find tools that support memory function, or even have an impact on how schools try to foster consolidation of knowledge that children acquired during a school day. It could also help to better understand the role of sleep for memory and incorporate this knowledge in health policy. One project also aims to better understand how memory reactivation changes with age, and how it is connected to age-related pathology such as Alzheimer's disease. As our society ages, gaining a better understanding of the changes in brain function and possibly what influences it, has increasing importance. Finally, the project sits at the intersection of psychology, neuroscience and AI. What we can learn from AI algorithms about ourselves, and what AI can learn from our knowledge of the human brain is a field of growing interest and potential for society. In combination, insights gained from this research promise to enhance our understanding of how memories guide adaptive behavior in humans and, to some extent, also in machines.
The main objective of REPLAY is to make replay events in the human brain accessible through non-invasive fMRI methods, and then to use this approach to provide deeper insights into replay in humans. Specifically, we aim to further develop and strengthen fMRI-based approaches that make it possible to measure fast neural sequences, which are the key signature of memory replay. Using these insights, the proposed research seeks to further our understanding of four cognitive and computational aspects of replay in the human brain: 1) the coordination of hippocampal replay with activity in other brain areas, 2) the effects of reward and planning on content and direction of replay, 3) the role of replay during sleep and its relation to sleep spindles, 4) its role in memory aging
We will use fMRI to study the brains of healthy young adult volunteers while they perform cognitive tasks (designed such as to elicit replay in different brain areas during task play).The results of these studies will be presented at scientific conferences, published in peer reviewed open access journals, and communicated to the public as appropriate. To meet Objective 3, we will perform joint fMRI-EEG measurements on sleeping volunteers. To achieve Objective 4, we will perform an age-comparative study of memory reactivation during and following a memory task in younger and older adults. Older participants will be screened for any signs of early dementia pathology, and physiological markers of metabolic syndrome will be taken into account as mitigating factors for age-related hippocampal deterioration.
Research in rodents has first discovered “memory replay” in the brain over two decades ago. In 1996, Skaggs and McNaughton found that while rats were sleeping, brain cells in their hippocampus were at times quite active, and seemed to retrace the animals’ previous trajectories in a maze. Curiously, this “mental travel” was much faster than it had occurred in real time. This phenomenon is known as replay, and has become a major focus of neuroscientists and even artificial intelligence (AI) researchers over the past decades. We know for instance that in rodents replay is prevalent during wakeful resting and sleep, and that it is related to memory, planning and reward processing. An interesting parallel has occurred in AI research, where machine learning algorithms not only learn from the present, but also replay the past, to enhance performance.
While it is generally thought that replay is an important aspect of brain function, little is known about replay in the human brain. Fast neural processes, such as relay of previously traveled paths as found in rodents, is hard to measure in humans, where we can’t surgically implant electrodes in the hippocampus, as it is done in animals. REPLAY will address this problem of little knowledge about replay in the human brain. We will use a novel method to analyze functional magnetic resonance imaging (fMRI) data, that for the first time allows finding evidence for replay in fMRI patterns that we can record during rest or sleep in healthy volunteers.
The project REPLAY is will enhance our understanding of cognitive aspects of brain function, in particular learning, memory and decision making. Basic science that can help us better understand the human brain is important in a number of ways. Learning, memory and decision making play a big role in our everyday lives. In the context of healthy cognition, in the long run our research could benefit the attempt to find tools that support memory function, or even have an impact on how schools try to foster consolidation of knowledge that children acquired during a school day. It could also help to better understand the role of sleep for memory and incorporate this knowledge in health policy. One project also aims to better understand how memory reactivation changes with age, and how it is connected to age-related pathology such as Alzheimer's disease. As our society ages, gaining a better understanding of the changes in brain function and possibly what influences it, has increasing importance. Finally, the project sits at the intersection of psychology, neuroscience and AI. What we can learn from AI algorithms about ourselves, and what AI can learn from our knowledge of the human brain is a field of growing interest and potential for society. In combination, insights gained from this research promise to enhance our understanding of how memories guide adaptive behavior in humans and, to some extent, also in machines.
The main objective of REPLAY is to make replay events in the human brain accessible through non-invasive fMRI methods, and then to use this approach to provide deeper insights into replay in humans. Specifically, we aim to further develop and strengthen fMRI-based approaches that make it possible to measure fast neural sequences, which are the key signature of memory replay. Using these insights, the proposed research seeks to further our understanding of four cognitive and computational aspects of replay in the human brain: 1) the coordination of hippocampal replay with activity in other brain areas, 2) the effects of reward and planning on content and direction of replay, 3) the role of replay during sleep and its relation to sleep spindles, 4) its role in memory aging
We will use fMRI to study the brains of healthy young adult volunteers while they perform cognitive tasks (designed such as to elicit replay in different brain areas during task play).The results of these studies will be presented at scientific conferences, published in peer reviewed open access journals, and communicated to the public as appropriate. To meet Objective 3, we will perform joint fMRI-EEG measurements on sleeping volunteers. To achieve Objective 4, we will perform an age-comparative study of memory reactivation during and following a memory task in younger and older adults. Older participants will be screened for any signs of early dementia pathology, and physiological markers of metabolic syndrome will be taken into account as mitigating factors for age-related hippocampal deterioration.
Our work consisted of conducting several scientific studies, analysing the resulting data and disseminating the results to the scientific community and general public. So far we have conducted 4 large scale studies that involve functional magnetic resonance scanning of the brains of human participants while they engaged in cognitive tasks developed by us for this particular purpose. We have developed a theoretical framework for our work, and have prepared 3 additional experiments. This work resulted in new knowledge about when and where the brain reactivates past experiences and internal states. Results have been published in prominent scientific peer reviewed journals, and communicated to the public in radio interviews, panel discussion public engagement events and newspaper articles.
Specifically, we have (a) developed and validated a new replay analysis method, showing that it can detect replay events in humans with fMRI; (b) discovered that replay in humans can occur outside of, and independently from, the hippocampus following a task which does not require memory; (c) published two novel theoretical perspectives on replay, which provide links between replay in machine learning and replay in the brain, and (d) shown how the orbitofrontal cortex shapes the representational structure in hippocampus, which in turn guides memory replay. Ongoing work investigates the role of replay in generalization, cognitive aging, and in sleep.
Specifically, we have (a) developed and validated a new replay analysis method, showing that it can detect replay events in humans with fMRI; (b) discovered that replay in humans can occur outside of, and independently from, the hippocampus following a task which does not require memory; (c) published two novel theoretical perspectives on replay, which provide links between replay in machine learning and replay in the brain, and (d) shown how the orbitofrontal cortex shapes the representational structure in hippocampus, which in turn guides memory replay. Ongoing work investigates the role of replay in generalization, cognitive aging, and in sleep.
Over the remaining funding period, we plan to set up an ambitious joint EEG-fMRI experiment, which will allow us for the first time to get a detailed picture of cortex wide coordination of replay activity during sleep. In addition, we aim to investigate the putatively predictive effects of replay for early onset cognitive decline. Age-related cognitive decline is a major health burden for society, but which role replay, a putatively major process for memory formation, play, is unknown. Finally, we are investigating the coordination of reward processing areas with replay events, leveraging fMRI’s ability to image all brain areas in parallel.