A Tale of Two Walks: Episodic Memory and Semantic Representations Both Guide Memory Retrieval

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Memory retrieval is guided by two complementary learning systems: an episodic memory system storing individual experiences, and a semantic knowledge system representing ‘things which go together’. While much is known about these structures in isolation, their interplay has been comparatively less studied. In this work we propose a spotlight-cascade model, which characterizes the interplay between these systems through two walks: a cascading walk in episodic space and an autocorrelated walk in semantic space. We combine these two walks in a spotlight-cascade model. This model is then validated through an online experiment in which participants travel to a fictive island and interact with 16 fantasy animals, dubbed Fennimals. Some of these Fennimals are encountered in the same region of the island (creating episodic groups), whilst some Fennimals share semantically similar heads and names (creating semantic groups). Participants are then asked to recall the names of these Fennimals. The experiment consisted of two between-subjects conditions. In the semantic spotlight condition, the semantic and episodic groups overlapped. In line with a predicted semantic spotlight effect, participants sequentially retrieved Fennimals along the imposed semantic structure. In the episodic jump condition, we then mixed these two groups, such that Fennimals in the same episodic group contained features from distinct semantic groups. In line with theoretical predictions based on our spotlight-cascade model, participants were now less likely to follow the semantic grouping structure when sequentially retrieving names – instead following a mixture of both semantic and episodic ties. These findings highlight that our novel model bridges both complementary learning systems and sheds a new light on previous models of memory retrieval. In addition, our findings suggest promising new directions of research into various topics in cognitive science.

Article activity feed