Doctoral thesis

Towards physiologically-driven human memory augmentation

  • 2025

PhD: Università della Svizzera italiana

English Despite technological advances in lifelogging devices—-capturing large amounts of photos, location data, and sensor logs, in high-quality formats—-, users face an ever-growing amount of data when they try to revisit past experiences. Sifting through these personal "memory vaults" is time-consuming and often unproductive, as many recorded moments may lack personal significance. One approach to this problem is given by memory cues, which are prompts—-like a photograph, a souvenir from a trip, or a scent—-that trigger the efficient recall of information with minimal user effort. However, users still face the challenge of selecting which moments they want to remember and, consequently, warrant cue extraction. Current systems typically rely on external context, such as time, location, or transcripts, to rank the user’s content. However, an underexplored path is the use of the internal context of the user, such as their affective state. Affective Computing and Human Activity Recognition research have shown the potential of physiological and inertial signals—-such as fluctuations in heart rate variability, spikes in electrodermal activity, and characteristic motion patterns—-captured from wearable devices to recognise the user’s affective state and activities. My research focuses on leveraging the physiological state of humans to augment their memory by detecting relevant moments for extracting memory cues. In particular, I focus on collecting physiological signals gathered from wearable devices during an experience, and analyse the relationship of those signals with the user’s posterior recall of the experience. A Human Memory Augmentation system could use this information to select and extract pertinent memory cues that the user can review and use to recall the experience. Through repeated exposure to these cues, the user will be able to remember the experience effortlessly.
Collections
Language
  • English
Classification
Computer science and technology
License
License undefined
Open access status
green
Identifiers
Persistent URL
https://n2t.net/ark:/12658/srd1332436
Statistics

Document views: 22 File downloads:
  • 2025INF009.pdf: 97