Nature Neuroscience, 2022
This paper addresses a fundamental question in memory research: how does the brain determine where a memory begins and ends? Understanding this boundary is essential for modeling human memory using neural networks and for building AI systems capable of forming and recalling memories in a human-like way.
New Turing-style tests reveal that AI trained on human memories can mimic human behavior in vision and language—but only some models truly pass as human.
A computational model using random memory reactivations to construct lifelong memories. This work models long-term memory in humans.
Our brains retain surprisingly few details—human memory stores abstract summaries, not full recordings. This work models what humans remember, and what they don't.
Some moments are simply more memorable—this study predicts which movie scenes we'll recall even a year later.