Matter has memory, too. For example, a crumpled nickel-titanium paperclip spontaneously reconfigures into its remembered shape upon heating. Hypothesizing that memory dynamics in non-living systems might have more in common with brain memory than meets the eye, a new research project will bring together experimental neuroscience, soft matter physics and computational neural networks to deepen our understanding of the mechanisms behind the storage and recall of information. The 3-year project “Memory – from material to mind” (M2M) worth over 1 million USD has been awarded by the Human Frontier Science Program (HFSP) to a team composed of SISSA neuroscientist Mathew Diamond (project coordinator) along with physicist Nathan Keim of Pennsylvania State University and neural networks theorist Omri Barak of Technion - Israeli Institute of Technology.
M2M will focus on perceptual memory, the process of storing and retrieving information extracted from sensory stimuli. The team will investigate whether the different types of memory, for example when comparing incoming signals which are close to each other in time, or else distant in time, are due to a single flexible reconfigurable system rather than to multiple single-task systems, as proposed by current theories.
“We will begin by examining the operations underneath memory formation in the living, behaving brain,” Mathew Diamond explains. “Our novel approach, however, is to extend from the brain to physical materials.”
Inspired by Richard Feynman’s claim, “What I cannot create, I do not understand”, the researchers will see how rich a memory they can construct in a network of materials, outside the brain. “We will use knowledge gained through experimental neuroscience to build materials networks that are capable of extracting and storing information. More specifically, through the soft matter physics expertise in our team, we will chart out a general framework for memory storage and retrieval aimed at replicating key biological findings, such as the interaction between short-term and longer-term memories. Materials that extract and store information, such as driven suspensions, can be arranged in a flexibly interacting network to serve as a physical model of brain networks.”
Finally the researchers will use their computational neural networks expertise as the bridge back to biology: “Thanks to the use of neural networks language, we will try to import the rules of materials memory back to the brain.”
HFSP selected only the top 4% of applications for funding, providing a $33million USD overall budget over the next three years . The Program promotes international collaboration in basic research focused on the elucidation of the sophisticated and complex mechanisms of living organisms. It funds cutting-edge, risky projects and therefore appeals to the innovative and creative potential of the research teams.
(Image: iStock)