In fact theres evidence proving we actively and passively forget alot of information throughout the day, and this information dump of actively and passively forgetting things we deem unimportant, is precisely what allows us to learn. The balance between active forgetting during the day and memory consolidation during the night is what makes the brain so effective.
Fast-reverse memory replay reminds me of how eligibility traces and TD(lambda) in reinforcement learning work, by updating the Q value of each past visited state after receiving a reward in the "current" state, and how the reward filters back essentially in reverse order of visitation. Obviously it's not at all the same, but it's cool how even though TD(lambda) wasn't created with this biological process in mind, it still forms a loose analogue to how the "real thing" works.
Hey Artem, I just wanted to say that I absolutely love the content you make, specially the videos about specific neuroscience topics. Please never stop!! :)
I had no idea the hippocampus was doing that while we were sleeping. I knew we had memory consolidation during REM, but not this hippocampal cell spiking. This is so fascinating wow. My god 😮
I like how Artem had 9k while asking us to follow and now he has gotten ~x10 times more. He will grow to the million. This channel and the content is brilliant. Doing Quantum AI + theoretical physics related research right now. It is funny how biology inspires new models in AI a lot. Keep it up! Greetings by a Saint Petersburg’s 🇷🇺physicist from Brazil 🇧🇷🙌🏼
This is such a fascinating topic, and I had never heard of the reverse order replay that you explained. Just an excellent video and explanation! Your animations are beautiful. By the way, I also make neuroscience videos so if you would ever be interested in collaborating, let me know!
This was crazy. Reminds me of a book by Eric Kandel: 'In search of memory' where he discussed the role of the hippocampus in memory. Also coincides with Jordan Peterson × Huberman Interview recently where they talked about the hippocampus when it comes to "trying on" different potential versions of your self via imagination and through the prefrontal cortex.
Peterson and Huberman have brought stunning examples of paradigm-remodeling insights from the neuroscience frontier to the layman. They have helped me with real practical problems as well as making the study field so much more engaging.
"Trajectories the mouse was taking during the day at night recapitulate during the shark wave ripple. And that means that the same place cells were activated one atfer another as if representing the same trajectory only their order was reversed and the time scale of the entire thing was much smaller."
Год назад
I wonder... If the reverse play is to learn what behavior led to either a positive or negative reinforcement event. Like the birds in Skinner's box, desperately trying to figure out what made the food come out, begin to show "superstitions" and repeat all sorts of odd behaviors (like turning around to the left, for example) in hopes of triggering a new feeding. It makes sense to me that it would be examined in reverse in order to suss out what behavior led to the positive or negative event in order to either suppress or reinforce those behaviors.
Omg this was amazingly exciting and done exceptionally ❤ One question: does the brain send those “reverse replay packages” during memory consolidation process only during specific phases of sleep?
If forwards is sensory sending stuff to the hippocampus, and the backward replay memories are being sent out to the sensory organs. They may not be a reverse replay, but a regular replay.
Sounds exactly like Backpropagation. I have believed this the moment I learned how to write Neural networks and even named some of my Backprop functions simply " sleep(double _delta){ "
Very interesting video. If the hippocamal replay occurs during daytime, what is so special about sleep? My understanding is that information in short-term memory that is deemed as unimportant will be forgotten so sleep could hinder lessons just before sleep, on the other hand they also reinforce memories being encoded into long-term memory. Your thoughts?
It is hypothesized that forward awake replay is used for planning ahead whereas reverse awake replay is used for encoding valuable behavioral sequences since it mainly occurs when the animal encounters an unexpected reward. As for sleep replays, they are thought to serve information transfer from hippocampal-dependent episodic memory to cortex-dependent semantic memory. Contrary to what is said in the video, I believe the replay of sequences during sleep are usually in the forward order.
Very interesting question! For starters, sleep has other important functions as well, such as clearing metabolic waste products from the brain. As for memory consolidation, the frequency of the ripples is much larger during slow-wave sleep, than during waking. Also, during sleep the balance of various neurotransmitters (such as dopamine, noradrenaline, acetylcholine) is changed, which could potentially affect the "effectiveness" of the ripple, providing a special surrounding for upregulation of synapses But this is just my own thoughts. There are still a lot of blind spots in our knowledge on this matter 🙂
Brilliant explanation! Thank you. I have already subscribed to your channel. That was in mouse. The same thing happens in humans? Do we plan the future with Ripley waves when sleeping? thanks Hugo
Your voice is hard to understand, because of the reflection of your table into the mic. Try to get it closer to you or use a Lavalier microphone. Love your vids. Keep it coming :) ♥