Nat commun: machine learning helps to reveal the formation mechanism of brain memory

December 2, 2019 / BIOON / -- recently, researchers at the National University of Singapore (NUS) discovered the key to short-term memory encoding in the brain, and made a breakthrough in the field of cognitive computing neuroscience Camilo libedinsky, assistant professor of psychology at the National University of Singapore, and Shih Cheng Yen, senior lecturer of innovation and design program at the National University of Singapore, found that the neural population in the frontal lobe of the brain contains stable short-term memory information in the dynamic changes of neural activities This finding may have a profound impact on understanding how an organism uses a finite size brain to perform a variety of psychological operations (such as memory, attention and decision-making) at the same time (photo source: www Pixabay Com) the results of this study were recently published in the journal Nature communications In human brain, frontal lobe plays an important role in processing short-term memory The ability of short-term memory to store information is very low "It usually only holds 6 to 8 items For example, our ability to remember phone numbers in seconds - using short-term memory, "explains libendisky Here, NUS researchers study how the frontal lobe represents short-term memory information by measuring the activity of many neurons Previous research by researchers has shown that if there is interference during memory maintenance, it will change the code that frontal neurons encode memory "It's counterintuitive because memory is stable, but the code has changed In this study, we solved this problem, "libendisky said Researchers use machine learning derived tools to find stable information in the changing neural population code This means that the NUS team has demonstrated that memory information can be read from a population of neurons that change their code after presenting the distractor This simple discovery has a broader meaning, indicating that a single neural population may contain a variety of independent information that does not interfere with each other "This may be an important characteristic of organisms that exhibit cognitive flexibility," explains libendisky Researchers are now expanding these studies to explore how multiple brain regions interact with each other to transmit and process different types of information This can be achieved by measuring in biological networks and simulating the interaction between artificial neural networks that can simulate their functions Researchers are also exploring these processes in unhealthy brains, such as those with dementia Sources of information: researchers use machine learning tools to revive how memories are coded in the brain original sources: Aishwarya Parthasarathy, Cheng Tang, Roger herikstad, long FAH Cheong, Shih Cheng Yen, Camilo Libedinsky Time-invariant working memory representations in the presence of code-morphing in the lateral prefrontal cortex Nature Communications , 2019; 10 (1) DOI: 10.1038/s41467-019-12841-y