Periodic Reporting for period 4 - REPLAY_DMN (A theory of global memory systems)
Reporting period: 2024-03-01 to 2024-08-31
The aim of this project was a comprehensive theoretical and experimental investigation of the role of global brain activity, interactions between cortex and hippocampus, spontaneous activity during rest and the links to memory processes.
In recent years, two major experimental findings have highlighted the importance of intrinsic brain dynamics. At the global brain dynamics level, the default mode network (DMN), a set of brain areas whose activity is most markedly correlated during periods in which the subject performs no active task. At the level of neural ensembles and codes, memory replay, the spontaneous reactivation of activity configurations (sequentially or simultaneously activated neuronal groups) previously occurred during active experience. From the functional point of view, DMN has been linked to, for example, mental imagery, memory, task preparation, evaluation of the emotional state, whereas memory replay has been seen as supporting systems memory consolidation, i.e. the rearrangement of memory information onto new anatomical substrates, and the extraction of novel memory representations reflecting the statistical structure of the world.
We tested the hypothesis that the DMN plays a key role in memory replay processes. This theory, if confirmed, would bring important conceptual advances:
- To memory studies, as it would provide a mechanism supporting the formation and consolidation of complex memory representations.
- To the Default Mode Network field, as replay can be used as the “Rosetta Stone” to decipher the computations the DMN performs, moving beyond the connectivity, dynamics, and cognitive correlates, typical focus of DMN research.
For this we used a combination of recordings of a large number of neurons, and high time resolution imaging of large portions of the brain, techniques that are eaas, which most easily performed in rodent preparations, combined with innovative analytical and theoretical tools, based on statistical physics and machine learning, to decode their dynamics and information content.
We characterized DMN activation at fast time scales during sleep and wakefulness, in terms of their statistics and structure
We have also characterized neural activity in detail in a number of cortical areas, extracting patterns that relate to memory, and making significant inroad in understanding how sequential patterns of activity ("neural sequences") are produces.
We believe that we have made substantial inroad towards the goals that we stated years ago, which summarized into a novel theory on “cascaded memory systems”.
In recent years, two major experimental findings have highlighted the importance of intrinsic brain dynamics. At the global brain dynamics level, the default mode network (DMN), a set of brain areas whose activity is most markedly correlated during periods in which the subject performs no active task. At the level of neural ensembles and codes, memory replay, the spontaneous reactivation of activity configurations (sequentially or simultaneously activated neuronal groups) previously occurred during active experience. From the functional point of view, DMN has been linked to, for example, mental imagery, memory, task preparation, evaluation of the emotional state, whereas memory replay has been seen as supporting systems memory consolidation, i.e. the rearrangement of memory information onto new anatomical substrates, and the extraction of novel memory representations reflecting the statistical structure of the world.
We tested the hypothesis that the DMN plays a key role in memory replay processes. This theory, if confirmed, would bring important conceptual advances:
- To memory studies, as it would provide a mechanism supporting the formation and consolidation of complex memory representations.
- To the Default Mode Network field, as replay can be used as the “Rosetta Stone” to decipher the computations the DMN performs, moving beyond the connectivity, dynamics, and cognitive correlates, typical focus of DMN research.
For this we used a combination of recordings of a large number of neurons, and high time resolution imaging of large portions of the brain, techniques that are eaas, which most easily performed in rodent preparations, combined with innovative analytical and theoretical tools, based on statistical physics and machine learning, to decode their dynamics and information content.
We characterized DMN activation at fast time scales during sleep and wakefulness, in terms of their statistics and structure
We have also characterized neural activity in detail in a number of cortical areas, extracting patterns that relate to memory, and making significant inroad in understanding how sequential patterns of activity ("neural sequences") are produces.
We believe that we have made substantial inroad towards the goals that we stated years ago, which summarized into a novel theory on “cascaded memory systems”.
- The theoretical framework that was sketched in the project proposal has been significantly expanded and has been spelled out in a review article co-authored with project members Karola Kaefer and Federico Stella and with collaborator Prof. Bruce McNaughton, that has been accepted for publications in Nature Reviews Neuroscience (Kaefer et al., 2022).
- We demonstrated that spontaneous activity in the cerebral cortex during sleep follows the statistics of critical phenomena (one of the key hypothesis in the project proposal). Additionally, we discovered that cortical activity links to hippocampal activation not only through the known “sharp-wave/ripples” activity patterns, but with an involvement of slow gamma oscillations (30-50 Hz). This results highlights how communication between the hippocampus and the neocortex is bi-directional, which is already an update with respect to the hypotheses in the project proposal. (Pedrosa et al. 2022).
- Based on previous work on resting state activity in the mouse neocortex (Pedrosa et al., PNAS 2022), we characterized global cortical activity in the mouse during a simple Classical Conditioning task, the Appetitive Auditory Trace Conditioning (AATC) task, using a combination of voltage sensitive imaging and electrophysiology. In this task, two different tones(CS+ and CS–) are played. CS+ predicts a reward 1 second later, CS– is not followed by reward. Mice acquire a classical conditioning to licking the reward spout when they hear the CS+, prior to reward delivery. We observe a distinct cortical response sequence, that starts from auditory cortex and extends to parietal, retrosplenial cortex and the hippocampus, which develops with training. This response is replayed in the resting state concomitantly with hippocampal sharp waves. This results provides support for the basic hypothesis in the project. These results are currently being submitted for publication
- We characterized resting state activity in the developing mouse, by extending the Pedrosa (2022) paradigm to a longitudinal examination of cortical activity from postnatal day 18 (P18) and P28, the developmental period in which spatial memory and the related neural patterns of activity starts to emerge. We find that the statistics of the activity change radically, and become similar to the statistics in the adult, around P25. The results are being prepared for publication
- We developed a task looking at the formation of spatial memories involving information from different senses (eg vision, hearing), and we find novel types of neural responses in cortical in the medial temporal lobe. This work is being continued with other funding.
- A new direction, that was not anticipated in the research proposal is the use of socio-emotional stimuli to test memory, developed together with Dr. Arie Kim, postdoc on the project is currently being funded by a NWO-XL project.
- We have recorded ~100 sessions with two-photon recording in the mouse neocortex, in a number of paradigms. Analysis of this data is under way (funded by other sources), along the lines that have been defined in the project
- We demonstrated that spontaneous activity in the cerebral cortex during sleep follows the statistics of critical phenomena (one of the key hypothesis in the project proposal). Additionally, we discovered that cortical activity links to hippocampal activation not only through the known “sharp-wave/ripples” activity patterns, but with an involvement of slow gamma oscillations (30-50 Hz). This results highlights how communication between the hippocampus and the neocortex is bi-directional, which is already an update with respect to the hypotheses in the project proposal. (Pedrosa et al. 2022).
- Based on previous work on resting state activity in the mouse neocortex (Pedrosa et al., PNAS 2022), we characterized global cortical activity in the mouse during a simple Classical Conditioning task, the Appetitive Auditory Trace Conditioning (AATC) task, using a combination of voltage sensitive imaging and electrophysiology. In this task, two different tones(CS+ and CS–) are played. CS+ predicts a reward 1 second later, CS– is not followed by reward. Mice acquire a classical conditioning to licking the reward spout when they hear the CS+, prior to reward delivery. We observe a distinct cortical response sequence, that starts from auditory cortex and extends to parietal, retrosplenial cortex and the hippocampus, which develops with training. This response is replayed in the resting state concomitantly with hippocampal sharp waves. This results provides support for the basic hypothesis in the project. These results are currently being submitted for publication
- We characterized resting state activity in the developing mouse, by extending the Pedrosa (2022) paradigm to a longitudinal examination of cortical activity from postnatal day 18 (P18) and P28, the developmental period in which spatial memory and the related neural patterns of activity starts to emerge. We find that the statistics of the activity change radically, and become similar to the statistics in the adult, around P25. The results are being prepared for publication
- We developed a task looking at the formation of spatial memories involving information from different senses (eg vision, hearing), and we find novel types of neural responses in cortical in the medial temporal lobe. This work is being continued with other funding.
- A new direction, that was not anticipated in the research proposal is the use of socio-emotional stimuli to test memory, developed together with Dr. Arie Kim, postdoc on the project is currently being funded by a NWO-XL project.
- We have recorded ~100 sessions with two-photon recording in the mouse neocortex, in a number of paradigms. Analysis of this data is under way (funded by other sources), along the lines that have been defined in the project
We demonstrated bidirectional communication involving both sharp wave ripples and slow gamma as we demons represents a major shift from the view of memory consolidation processes that we sketched in our Nature Reviews paper. (which was already an update on the mainstream theory. Understanding the consequences of this will be a major research direction will represent a major challenge, based on follow-up analysis of the data we gathered and further experimental work.
We have shown global cortical replay at the mesoscopic level, bridging the gap between the views from neural ensemble recording in animals and from the human neuroimaging literature.
In Guardamagna et al. we find a dissociation between two modes of organization of hippocampal activity (theta phase precession and theta sequences) that were previously thought to be connected, this will spur theoretical and experimental work to update our view of the mechanisms of generation of neural sequences in the hippocampus
We have shown global cortical replay at the mesoscopic level, bridging the gap between the views from neural ensemble recording in animals and from the human neuroimaging literature.
In Guardamagna et al. we find a dissociation between two modes of organization of hippocampal activity (theta phase precession and theta sequences) that were previously thought to be connected, this will spur theoretical and experimental work to update our view of the mechanisms of generation of neural sequences in the hippocampus