AI Interfaces for Augmenting Episodic Memory

Abstract

Episodic memory, the memory of personal experiences, is a core component of human cognition. It functions within the neural substrate to store progress towards personal goals. Thus, it influences human behavior by enriching social interactions, forming a personal narrative, and facilitating personal growth. With the rise of challenges such as poor sleep, aging and dementia, and fragmented attention, people experience difficulties with episodic memory retrieval. These difficulties range from momentary lapses such as forgetting previous interactions during conversations, to recalling multiple events during reminiscing and decision-making.

In this work, we explore artificially intelligent (AI) systems that augment episodic memory by enabling people to interact with their memories effectively. We design, develop, and evaluate two systems: (i) Memoro, a wearable audio-based memory assistant that presents concise suggestions in real-time while minimizing disruption to the user’s primary task, and (ii) Resonance, a web-based reflective memory assistant that offers actionable suggestions to help users savor their past, present, and future experiences for mental health benefits. By conducting an in-person user study for Memoro and a longitudinal online user study for Resonance, we investigate the effects of these systems on users, measure their technical efficacy, and gather feedback on user experiences. Recent advances in artificial intelligence offer novel opportunities to enhance episodic memory. Therefore, exploring interfaces that seamlessly integrate with human behavior is crucial to ensure that AI-based systems enrich everyday experiences.