Login / Signup

Feedforward prediction error signals during episodic memory retrieval.

Rafi U HaqueSara K InatiAllan L LeveyKareem A Zaghloul
Published in: Nature communications (2020)
Our memories enable us to form expectations for our future experiences, yet the precise neural mechanisms underlying how we compare any experience to our memory remain unknown. Here, using intracranial EEG recordings, we show that episodic memories formed after a single visual experience establish expectations for future experience within neocortical-medial temporal lobe circuits. When subsequent experiences violate these expectations, we find a 80-120 Hz prediction error signal that emerges in both visual association areas and the medial temporal lobe. Critically, this error signal emerges in visual association areas first and then propagates to the medial temporal lobe. This error signal is accompanied by alpha coherence between the two regions. Our data therefore suggest that internal models formed from episodic memories are generated throughout the visual hierarchy after just a single exposure, and that these internal models are then used for comparison with future experiences.
Keyphrases
  • current status
  • mental health
  • working memory
  • big data
  • electronic health record
  • machine learning
  • functional connectivity
  • artificial intelligence
  • deep learning
  • optic nerve