Summary: A recent study has found a way for the mind to effectively encode and recall physical memories by categorizing them into categories and using the exact timing of cerebral activity. Researchers used machine learning to decipher the type of individual being remembered by recording cortical spikes in epilepsy patients as they watched images from five different object categories.
This demonstrates that the hippocampus classifies sensory inputs to make it easier to store memory information rather than store each object separately. These findings may provide individuals with memory loss with novel brain-computer interfaces and storage prostheses.
Important Information
- Category-Based Storage: The head makes it easier to store images in groups rather than store each photo separately.
- Temporal Coding: Important data about memory categories is transmitted when timing neurological spikes, never really firing rate.
- Clinical Potential: Research may lead to the development of recollection implants to assist people with dementia-related neurological conditions.
Origin: USC
Researchers at USC  have made a major advance in understanding how the human brain organizes, shops, and retains physical thoughts.
A new study, which was published in Advanced Science, combines the strong machine learning model and human person head recordings to shed new light on the inner code that categorizes memories of objects. Think of it like the body’s submitting cabinet of imagery.
The results demonstrated that the study team could effectively read people’s minds by identifying the physical image category that was recalled, based solely on the subject’s brain’s accurate timing.
The research addresses a key issue in biology and opens up new avenues for developing brain-computer interfaces, including memory implants, to recover lost memory in people with dementia-related neurological disorders.
The study was led by associate professor in the Department of Neurological Surgery and the Alfred E. Mann Department of Biomedical Engineering, Dong Song. and professor of biological engineering at , USC , Viterbi School of Engineering, as well as professor of biological engineering at , USC , Liu, the , USC , and Charles , Liu, the , USC ,
Xiwei She, the second artist, is a former doctor. A graduate student from the Song Lab who is presently a Stanford University doctoral scholar.
How is sensory information stored in the brain?
The brain is a crucial mental region known for its part in generating fresh acute thoughts, including the what, where, and when of past occasions. Although its role in encoding spatial ( “where” ) and temporal ( “when” ) information is well understood, how it manages to encode the vast and complex world of objects ( “what” ) has remained a mystery.
Scientists speculated that the brain may reduce this complexity by encoding objects into categories because it’s just not practical for the hippocampus to save every object separately.
With possible medical applications for people with dementia and Alzheimer’s disease, Song, who is Director of the , USC , Neural Modeling and Interface Laboratory, has been conducting pioneering research in the field of memory prosthetics. He has created products that resemble and restore cognitive performance.
We have thoroughly screened our recollection implants on a large number of people. We developed the prosthetics and published several papers that demonstrate its ability to improve memory recall, Song said.
” But I even wanted to take this opportunity to respond to some important questions about science. And this is just one of them.
Head recordings of people with epilepsy reveal insight.
The most recent job by Song,  , Liu, and their team uses head recordings from 24 seizures patients who have been implanted in their brains to localize seizures.
Hippocampal neurons ‘ recordings allowed the researchers to understand how difficult visual information is encoded by their exact timing, not just by firing rate alone.
It was “extraordinary interesting to see the recent reports reveal a design for the neural basis of memory formation,” Liu, Liu, said while working with people patients with memory function.
To understand this complex process, the research team created an impressive experimental-modeling methodology. In the seizure people, the group captured electrical activity from the cortical CA3 and CA1 neurons, specifically” spikes.”
The recordings were taken while patients engaged in a “delayed match-to-sample” ( DMS ) task, a well-known neuroscience method to test visual short-term memory.
” We allow the patients to see the images of an animal, a flower, a building, a car, and a small tool,” the doctor said. The cortical signal was therefore recorded, Song said. Therefore, using our machine learning approach, we asked ourselves a problem based on the sign. Can we learn what type image they are recalling solely based on their brain signals?
The findings supported the idea that the human mind can actually understand physical objects by categorizing them, and that these physical memory categories that the patients were considering were decodable based on their brain signals.
” It’s like reading your hippocampus to see what kind of memory you’re trying to form,” Song said. We discovered that we are capable of doing that. We can surprisingly decode what kind of category of image the patient was attempting to recall.
An effective method for storing various memories
The research team’s interpretable memory decoding model serves as the research team’s central point of the discovery. This advanced model analyzes the” spatio-temporal patterns” of spikes from an entire ensemble of neurons, in contrast to previous methods that frequently rely on averaging neuronal activity over multiple trials or using pre-selected temporal resolutions.
Additionally, the study provides proof that the hippocampus interprets visual memory categories using a temporal code. This implies that the precise timing of individual neuron spikes, which are frequently milliseconds, contains important information.
Compared to earlier studies that focused on individual neurons, this study found that hippocampal neuron ensembles encode memory categories in a distributed manner.
This implies that while a significant number of neurons ( 70 to 80 % ) were involved in granting a category a visual memory, only brief, specific moments within each neuron contributed to this encoding. This effective approach reduces energy consumption while preserving diverse memories.
With this information, we can begin developing clinical tools to repair memory loss and make better lives, including memory prostheses and other neurorestorative techniques, Liu said.
Although this finding may be significant for all memory-related patients, it has significant relevance for the epilepsy patients who participated in the studies, many of whom have hippocampal dysfunction, which results in both seizures and cognitive/memory disorders.
About this information on research into visual memory
Author: Amy Blumenthal
Source: USC
Contact: Amy Blumenthal – USC
Image: The image is credited to Neuroscience News
Original research: Free of charge.
Dong Song et al.,” Distributed Temporal Coding of Visual Memory Categories in Human Hippocampal Neurons Revealed by an Interpretable Decoding Model.” Advanced Science
Abstract
Distributed Temporal Coding of Visual Memory Categories in Human Hippocampal Neurons Revealed by an Interpretable Decoding Model
For creating fresh episodic memories, the hippocampus is essential. Although it is well understood, how it encodes objects ( what ) is a mystery due to the high dimensionality of object space.
The hippocampus may encode object categories to reduce complexity rather than encoding each object separately.
Here, we develop an experimental-modeling approach to understand how the hippocampus of humans encodes visual memory categories.
In 24 epilepsy patients who were subjected to a delayed match-to-sample task involving five image categories, spikes were recorded from hippocampal CA3 and CA1 neurons.
To identify the spatio-temporal characteristics of hippocampal encoding and decode memory categories from hippocampal spiking activity, an interpretable memory decoding model is used.
The optimal temporal resolutions for each visual memory category per neuron are estimated using this model.
The results support the existence of category-specific coding because they demonstrate that visual memory categories can be decoded from hippocampal spike patterns.
Compared to a population code, hippocampal neuron ensembles use a distributed method to encode memory categories, while individual neurons use a temporal code.
Additionally, CA3 and CA1 neurons exhibit similar and redundant memory category information, which is most likely caused by CA3 to CA1 regions ‘ strong and diffuse feedforward synaptic connections.
