Memory lanes: using neural circuit architectures for predicting recognition behavior
How do our brains recognize things? Savin’s on the case
What sparks recognition? Neuroscientists speculate that it’s usually either “due to recollection (‘here comes my old school buddy’),” as Cristina Savin explains in her most recent co-authored paper, “or due to a vague sense of familiarity (‘I’m sure I’ve met this person before, but I have no idea when and why’).”
There are presently two theories in neuroscience that are used to analyze neural data about recognition memory — dual module theories (DM) and single module theories (SM). DM theories believe that recollection and familiarity are independent aspects, while SM theories believe that recollection and familiarity work together to spark recognition.
“Despite over 30 years of memory recognition research,” the researchers explain, “no consensus exists about which class of models provides a more satisfying account of the data.” Savin and her co-author, Máté Lengyel, however, have developed a new neural circuit architecture that combines both SM and DM models. The neural network provides a more wholesome analysis of neural data, and can “efficiently operate in the face of trace strength-ambiguity.”
by Cherrie Kwok