From Cognition and Brain Plasticity Unit, Dynamics of Memory Formation Group
will present a paper titled
Shifting neural activity towards a stable state mediates the retrieval of real-life episodic memories
Abstract
Memories are thought to be retrieved by attractor dynamics whereby, if a given input is sufficiently similar to an internal representation, neural activity tends to settle into a stable pattern (i.e., the attractor state) [1-5]. In humans, attractor-like neural activity has been reported in perceptual categorization [6-7], in working memory [8] and in spatial navigation [9]. However, it remains elusive whether episodic memory retrieval is driven by attractor-like dynamics. Here, we leveraged the fine-grained temporal resolution of scalp electroencephalography (EEG) activity to test whether the retrieval of real-life episodic memories triggered neural pattern dynamics that rapidly evolved towards a stable state. We asked participants to use a wearable camera to automatically capture pictures depicting real-life episodes taking place at “home” and at “office” contexts, and then we used these pictures to cue the retrieval of these past episodes in the lab, while EEG was recorded. We show that pictures depicting past episodes elicited neural response activity that shifted towards a stable state after ~800 ms from picture onset. Importantly, we found that a similar stable neural pattern was elicited by pictures depicting experienced, but not novel, episodes from the same context (i.e., at home or at the office) and that it was not shared across individuals, thereby supporting the theoretical prediction that attractor-like states are content-specific in individual’s memory [10]. Finally, these findings were replicated on data from a second experiment with different participants and extended them by showing that the degree of context-specific neural stability was associated to participants’ ability to retrieve contextual information from the cued episode. These results indicate that the recollection of autobiographical memory episodes is supported by neural activity patterns that rapidly shift towards stable states.