​Neuron︱New Discovery!

Written by Fu Huimin, edited by Huimin Wang, as the center of cognition, memory can help us store the representations of past experiences and use these representations to inform our future behavior
.

Studies have shown that both of these processes may depend on a pattern of hippocampal neural activity: hippocampal replay [1]
.

Neural activity patterns related to past experiences can be reactivated in the hippocampus playback
.

However, the special role of hippocampal playback in supporting memory functions and the principle that it determines the reactivation of past experiences are currently unclear
.

During sleep, hippocampal playback participates in the plasticity of the distributed network to promote the storage of long-term memory [2]
.

During the waking period, hippocampal playback also plays an important role in memory-related processes [3]
.

However, it is not clear how hippocampal playback promotes memory function during wakefulness
.

 Recently, postdoctoral researcher Anna K.
Gillespiebi (the first article, corresponding author) and Professor Loren M.
Frank (corresponding author) of the University of California, USA, published in Neuron entitled "Hippocampal replay reflects specific past experiences rather than a The research paper of "plan for subsequent choice" found that the content of the hippocampus replays preferentially select places that have been rewarded before and have not been visited recently, and have nothing to do with subsequent choices
.

This strongly suggests that hippocampal playback supports the memory process by facilitating the maintenance and long-term storage of memory rather than guiding subsequent behaviors
.

 The research team designed an eight-arm space model with a win-loss switch based on memory-oriented selection (Figure 1A).
The rat found the target arm in the "search phase", and then obtained a certain amount of data by accessing the target arm in the "repetition phase".
Reward (Figure 1B)
.

In a complete behavioral test process (60-90 minutes), the test rats completed several test blocks, namely the search and repetition phases of a given target location (Figure 1 C, D)
.

Next, by experimenting with 4 adult Long-Evans rats, the author found that the reward rate of the target arm of all subjects randomly started and increased (Figure 1E), and all subjects prioritized visiting the target location after finding the target location (Figure 1F)
.

Therefore, the rat can perform spatial memory tasks flexibly: the rat learns to alternate between finding a new reward position and repeating a remembered target
.

Therefore, this experimental design and results help this study describe the relationship between behavior and playback content
.

 Figure 1 Rats learn to perform flexible spatial memory tasks (Source: Gillespie et al.
, Neuron 2021) Next, in order to be more concise and effective in computer evaluation of spatial playback content, the research team first used the above-mentioned (Figure 1A) two-dimensional maze environment (Eight-arm space model) Linearized into a flat model of the center "box" shape and the surrounding 8 arms (Figure 2A, left), it is found that sharp-wave ripples (SWRs) mainly occur in the reward consumption of the target arm During the period (Figure 2A, right), repeated reward reception does not cause a deviation between the spatially predicted position on the target arm and the spatial real position (Figure 2B)
.

Considering that spatial continuous events represent a wide range of locations, including the current location (local) in the same maze segment and the real location (remote) of different maze segments [4]
.

Therefore, in this study, the spatial continuous SWRs are divided into local or remote.
Figure 2C-2E represents an example of remote playback and local playback of a subject.
In 4 rats, the researchers detected a considerable part of SWRs that include remote playback.
(Figure 2F)
.

Therefore, these studies show that most of the sharp wave ripples can explain the spatial playback content
.

 Figure 2 The playback content during decoding of SWRs (Source: Gillespie et al.
, Neuron 2021) Figure 3 The playback of the target arm in the past has been very rich (Source: Gillespie et al.
, Neuron 2021) Then, the research team found that the subject Before the rat chooses the model arm, most of the spatially continuous partial playback and remote playback occur in the box area of ​​the flat model (Figure 3A)
.

The long-range replay of the past target arm of all rats is consistent and strong, while the future long-range replay of the target arm is small and different (Figure 3 B, C)
.

By using Poisson generalized linear models (GLMs), the research team found that in the search and repeat stages, the past target arm is the most important factor driving playback (Figure 3D, E)
.

In addition, they found that the playback of the current target arm is only continuously enhanced in the later stages of the repetition phase (Figure 4 A, B).
As the subject accumulates reward experience at the current target location, the playback of this position will gradually increase (Figure 4A, B).
4C)
.

Therefore, these results show that the playback content is enriched in the past target location, and the current target location increases with the increase of experience
.

 Figure 4 The playback of the current target increases with the increase of experience (Source: Gillespie et al.
, Neuron 2021) Next, the researchers found that the average number of playbacks of the target arm of the 4 rats in the future is less than 1 time, and repeats The occurrence rate of replay in the stage is not as common as in the search stage (Figure 5A)
.

Further inspection of the repeated phase test, they found that there was no difference in the replay rate of the future arm of the two rats in the correct and wrong repeated trials, and the replay rate of the current target arm in the correct and wrong repeated trials of the three rats did not differ (Figure 5).
B, C)
.

In the error test, the difference in playback is unlikely to directly lead to subsequent errors (Figure 5D)
.

And they found that the playback of the past target arm does not necessarily cause the rats to continue to visit the previous target, because they will visit any other arm (Figure 5E)
.

These results indicate that the playback of the future arm, the current target arm, or the past target arm does not affect the subsequent choice of rats
.

 Figure 5 The upcoming behavior has nothing to do with the playback of future, current or past targets (Source: Gillespie et al.
, Neuron 2021) In addition, they found that in many trials after the target location stopped providing rewards, the playback rate was several times The test reached its peak (Figure 6A)
.

In the same experiment, after stopping the reward for the previously rewarded target arm, the number of visits to the target arm rapidly decayed and then returned to the baseline level (Figure 6B)
.

As multiple new target positions are discovered, the playback of the past target arm will continue in multiple test blocks.
In at least two test blocks, the playback of the past target position continues to increase (Figure 6C)
.

These results indicate that the playback of the previous target will continue for a long time after the behavior change
.

Figure 6 After the behavior change, the playback of the past target arm will continue for a long time (Source: Gillespie et al.
, Neuron 2021) In order to evaluate whether all past experiences have the same playback priority, the researchers tested those The playback rate of the arm that was never used as the target position during the current behavior.
It was found that the playback rate of the arm that was not visited in 5 trials was higher than that of the arm that was visited in 1 or 2 trials.
High (Figure 7)
.

Therefore, in addition to the previous reward locations, the remote events before the selection also give priority to non-recent past experiences
.

Figure 7 The playback of the unrewarded arm is biased towards the non-recent past (source: Gillespie et al.
, Neuron 2021).
Finally, in order to determine whether the observed important, non-recent experience occurred in the selected playback (outer arm) In ), the researcher found a rich replay of the current target arm through a reward experiment (where the past target arm was the same as the current target arm), and also found a strong enriched replay of the past target arm (Figure 8)
.

These results indicate that the past enrichment of target arms is not only specific to pre-selection replay, but also obvious throughout the mission
.

Figure 8 The replay takes place at the outer arm port (source: Gillespie et al.
, Neuron 2021).
Conclusion and discussion, inspiration and outlook Compared with previous studies, this research has several key advantages
.

First, the playback before selection and the playback after selection are separated, and different periods of playback related to the plan are provided
.

Second, the combination of changing the target location and searching on many arms makes it possible to eliminate the different effects of reward history and recent visits on playback
.

The study found that the experience reproduced during the playback process is usually the place where the subject has received multiple rewards before, and it usually appears in a location that has not been visited recently
.

During the playback process, the representation of past experience will be strengthened and updated
.

This research provides evidence for the role of awake playback in guiding upcoming behaviors in spatial memory, and further reveals the role of playback in memory maintenance and memory storage
.

 In addition, the research reveals the similarities between wakefulness and sleep playback: previous research pointed out that sleep playback controls date memory by promoting the storage of memories in the distributed cortex-hippocampus network [5], and this research found that wakefulness playback is also like sleep Playback plays this function
.

And the playback of these two states prioritizes specific experiences [2], indicating that playback is a general mechanism for preserving the neural representation of behavior-related experiences
.

Original link: https://doi.
org/10.
1016/j.
neuron.
2021.
07.
029 Selected previous articles [1] Mol Psychiatry︱ New discovery of biomarkers for depression-Mitochondrial proteins in exosomes [2 ] Science | Breakthrough! Astrocyte Ca2+ induces ATP release to regulate myelin axon excitability and conduction velocity [3] Neurosci Bull︱Shen Ying’s team reveals the three-dimensional heterogeneity of the cerebellar nucleus to thalamus projection [4] J Neurosci︱ Cao Junli’s group Revealing the loop mechanism of the anterior cingulate gyrus to regulate mirror pain [5] Nat Commun︱Non-human primate (monkey marmoset) autism model reveals the biological abnormalities in the early development of human diseases [6] Cell Discovery︱ Ma Yuanwu/Shen Bin’s team realized the precise editing of rat mitochondrial DNA for the first time [7] Dev Cell︱ Lactic acid promotes peripheral nerve damage and repair B side: Long-term axon lactate metabolism increases will lead to oxidative stress and axon degeneration [8] Nat Commun︱ Selective inhibition of microglia activation is expected to alleviate the pathological transmission of α-syn [9] Science︱ Serotonin helps overcome cocaine addiction? [10] Mol Psychiatry︱ Gao Tianming’s research group reveals the different roles of astrocytes and neurons in synaptic plasticity and memory [11] Sci Transl Med︱ Xiang Xianyuan and others reveal the brain’s immune cells crazy sugar phagocytosis, helping nerves Early diagnosis of degenerative diseases [12] A new mechanism of Mol Cell︱ Alzheimer's disease: Tau protein oligomerization induces nuclear cell transport of RNA binding protein HNRNPA2B1 and mediates enhancement of m6A-RNA modification [13] Cereb Cortex | Li Tao project The group reported the abnormality of the cortical myelin covariation network with the deep characteristics of the cerebral cortex in schizophrenia [14] Cell︱ hand in hand, advance and retreat together! Microglia form a cellular connection network and work together to degrade pathological α-syn.
Recommended high-quality scientific research training courses [1] Discount countdown ︱ Near-infrared brain function data processing class (online: 11.
1~11.
14) [2] Data graphs help guide! How good is it to learn these software? 【3】JAMA Neurol︱Attention! Young people are more likely to suffer from "Alzheimer's disease"? [4] Patch clamp and optogenetics and calcium imaging technology seminar (October 30-31) References (slide up and down to view) 1.
Buckner, R.