Shaping memories during sleep - how learning-related changes in the sleeping brain impact long-term memory

Why do we sleep? Considering that we spend roughly a third of our lives asleep, seemingly unproductive and vulnerable, there must be an immense evolutionary benefit to sleep. Yet we know remarkably little about the precise function of sleep. Perhaps one of the most intriguing effects of sleep is that it strengthens our memories. That is, it has been established for almost a hundred years that learning followed by sleep leads to better retention of the learning material than learning followed by the same amount of time spent awake. But how does our brain accomplish this feat of strengthening new memory traces while we are asleep?

The project will tackle the question of how lasting, learning-related changes are induced in the sleeping brain and how this process impacts long-term memory performance. We will focus on a particular electrophysiological signature of sleep, so-called sleep spindles. These are oscillations observable when recording ‘brain waves’ during sleep with a method called ‘Electroencephalography (EEG)’. The past years have provided converging evidence for a link between sleep spindles and post-sleep memory performance. However, what exactly it is that spindles do in service of memory retention is still unknown.

One prominent model argues for an active role of spindles in the transfer of individual memory traces from a short-term storage site (the hippocampus) to dedicated neocortical sites for more permanent storage, thereby distributing and solidifying the memory trace (akin to a data backup system for digital information). To test this hypothesis, we will first create different learning scenarios that engage distinct parts of the brain: In a first session, a participant will learn a motor sequence task in the evening (similar to learning to play a short melody on a piano and known to rely on the primary motor cortex) and in a second session, one week later, they will learn image pairs (engaging dedicated neocortical modules for objects and scenes). If sleep spindles indeed bolster memories in a content-specific manner, there should be a greater prevalence of spindles at motor cortex after the first learning session and a greater prevalence at object/scene sites after the second learning session.

Ultimately, the outcome of successful learning is a change in brain structure that reflects a stable memory trace. A second goal of the project is to assess whether (i) such structural changes can be observed in the healthy human brain after a single night of sleep and (ii) whether spindles play a direct role in inducing these changes. To this end, we will capitalise on recent developments in structural magnetic resonance imaging (MRI) which allow detection of tissue growth after short learning intervals.

As mentioned above, it is thought that memory traces are initially stored in the hippocampus. This region has its own electrophysiological sleep signature, called ripples (>80 Hz oscillations). If learning information is transferred from the hippocampus to neocortical target sites for more permanent storage, we should see a systematic interaction between hippocampal ripples and spindles. We will use a rare method available only at a few centres worldwide to test this hypothesis. In particular, we will use – in collaboration with Birmingham’s Queen Elisabeth Hospital - intracranial electroencephalography (iEEG) in pre-surgical epilepsy patients to obtain simultaneous sleep recordings from the human hippocampus and neocortical sites involved in learning.

Together, the project will link behaviour, physiology and structural changes to provide a unified and mechanistic understanding of how we consolidate memory traces during sleep.