Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster D in Poster Session D: Wednesday, August 5, 2:00 – 3:45 pm, Kimmel Center, Shorin & Rosenthal Rooms

Humans Can Rapidly Learn What to Encode in Memory

Qihong Lu¹, Kenneth A. Norman², Daphna Shohamy³; ¹City University of Hong Kong, ²Princeton University, ³Columbia University

Presenter: Qihong Lu

Episodic memory (EM) enables humans to rapidly encode experiences and later retrieve them to guide behavior. However, it remains unclear how humans learn a policy for what to encode, because the usefulness of an encoded memory is often revealed only after a long delay. We hypothesized that retrieved memories can act as a skip connection to support such encoding policy learning. To test this, we developed a behavioral paradigm in which subjects had to learn item–value associations in two scene contexts. These scenes served as encoding indicators that covertly signaled whether encoding the co-presented stimuli would be useful later. The results show that human subjects were able to learn to selectively encode task-relevant memories in response to task demands. An EM-augmented neural network reproduced these behavioral patterns. Collectively, these findings suggest that retrieval can bypass long encoding–retrieval delays to facilitate rapid learning of encoding policies.

Topic Area: Memory, Learning & Knowledge Structures