Cookie Consent by Free Privacy Policy Generator Secrets of Memory: How AI Models Mimic the Human Brain's Learning and Recall Processes | Review Space



Cover Image

Secrets of Memory: How AI Models Mimic the Human Brain's Learning and Recall Processes

The Intricate Dance of the Hippocampus and Neocortex in Memory Formation and Imagination

NEWS  Science  January 21, 2024  Reading time: 2 Minute(s)

mdo Max (RS editor)


In a study funded by the Wellcome foundation, researchers from University College London (UCL) delve into the fascinating realm of memory, employing artificial intelligence to unravel the intricacies of how our brains absorb, recall, and reimagine information. This exploration into the synergy between the hippocampus and neocortex sheds light on the mechanisms underlying human cognition.

Understanding Human Memory

Human memory, a cornerstone of cognitive processes, is indispensable for learning, reliving past experiences, and constructing imaginative scenarios. Recent advances in generative neural networks used in artificial intelligence have provided insights into the extraction of information from experience. Eleanor Spens, lead author and Ph.D. student at UCL, emphasizes the parallel between remembering and imagining the past based on concepts, melding stored details with expectations.

Computational Mimicry

The researchers employed a generative neural network, a computational model mirroring our brains' observation and memory of diverse events. With a focus on the hippocampus and neocortex, they sought to comprehend how these brain components collaborate in memory formation and spatial navigation.

Two Brain Components Under the Microscope

The hippocampus, vital for memory consolidation and spatial navigation, and the neocortex, responsible for computations related to attention, thought, perception, and episodic memory, were the focal points of the study. The research involved exposing the AI model to an array of 10,000 pictures, akin to simple scenes in a story.

Learning and Replay

The hippocampus network rapidly assimilated each scene, replaying it to instruct the neocortex. Through repetitive playback, the neocortex network learned to reconstruct these scenes by organizing information, emphasizing crucial aspects such as the layout of walls and objects. This efficient process enables the brain to remember and create entirely new scenes by focusing on distinctive elements.

Efficiency in Memory Recall

The study proposes that the brain, when replaying memories during rest, discerns patterns from past experiences. This ability is instrumental in predicting future scenarios, aiding in tasks like avoiding danger or locating sustenance. The brain's capacity to prioritize essential details over minutiae contributes to its efficiency in memory recall.

Reconstructing Memories

Professor Neil Burgess, senior author of the study, highlights how memories are reconstructed, underscoring the blending of experience meanings with unique details. This dynamic process results in biases in how we recall events, challenging the notion of memories as veridical records of the past.

Published Findings

The study, providing profound insights into the workings of memory and imagination, was published in the journal Nature Human Behaviour on January 19 and it is freely available for reading here.

As artificial intelligence continues to unlock the mysteries of the human mind, the study at UCL offers a glimpse into the intricate dance between the hippocampus and neocortex. Understanding how memories are formed, recalled, and reconstructed provides a foundation for future advancements in AI and neuroscience, paving the way for innovative approaches to enhancing memory and cognitive processes.

 COVER IMAGE BY KJPARGETER ON FREEPIK 

SHARE THIS ARTICLE



 COMMENTS


Currently there are no comments, so be the first!

*Our pages may contain affiliate links. If you buy something via one of our affiliate links, Review Space may earn a commission. Thanks for your support!
spacer

SPONSORED



SPONSORED


CATEGORIES



banner

Buy Me a Coffee at ko-fi.com