Periodic Reporting for period 4 - SolutionSleep (Understanding creativity and problem solving through sleep-engineering)
Período documentado: 2021-08-01 hasta 2023-07-31
Innovative problem solving is critical for all spheres of organised endeavour, including science and industry, and thus forms the cornerstone of a successful society. Such creative thinking often re-quires suppression of preconceptions and restructuring of existing knowledge. Pioneering work has shown that sleep facilitates problem solving, but exactly how, and which sleep characteristics are important, remain to be determined. We know that recent experiences are replayed in sleep, and that in Slow Wave Sleep (SWS) this replay integrates new knowledge with old. The role of such replay in Rapid Eye Movement (REM) sleep, a stage which is strongly linked to creativity, is unknown. I have proposed a model which combines physiology, behavioural studies, and computational modelling to make testable predictions about the complimentary contributions of memory replay in REM and SWS to problem solving. SolutionSleep has tested this model through explicit manipulation of memory replay in sleep using a recently developed technique to trigger memory replay and a pioneering method for quantifying triggered replay. This work has studied memory replays in both REM and SWS in detail and described many of their characteristics, for instance finding that in SWS such replays are temporally compressed and tend to occur at the peaks of slow oscillations. We have also shown that triggered memory replay in SWS can facilitate both gist abstraction and the formation of seemingly difficult connections between distant concepts, while triggering replay in REM can help emotionally arousing information to become less upsetting. Our computational model of memory replay in sleep has revealed that there are specific windows of opportunity when triggering replay is more likely to be effective. Immediate consequences of this work include a translational project harnessing memory replay to develop treatments for Depression and PTSD. Longer range consequences will include a project studying how replay can be harnessed to facilitate problem solving. Overall, the findings of SolutionSleep have revolutionised the understanding of how sleep impacts upon some of our most important cognitive abilities—memory and problem solving, and how these interact with mental health.
We first published the theoretical model ‘BiOtA’ which sets out the basic theoretical framework that explains how we believe non-REM and REM sleep are respectively acting to extract gist and create novel connections, thus fostering creativity.
In WP 2.1.1 We reviewed studies detecting neural reply with EEG classifiers then used EEG classifiers to examine the properties of neural replay during sleep. We thus characterised the time-course of memory reactivation after TMR in both REM and NREM sleep, and showed that TMR is most effective when applied in the down-to-up phase of the slow oscillation and only reactivations occurring immediately after TMR lead to behavioural benefit. We also showed that TMR cued reactivations in NREM sleep are temporally compressed compared to wake, and that memory encoding strength is an important determinant of whether reactivation provides benefits, with only medium strength encoding benefitting.
In WP 2.1.2: We used TMR to test the differential impact of memory reactivation in SWS and REM upon abstraction and the formation of unexpected connections. This showed that TMR of overlapping memories can lead to hierarchical knowledge (WP 2.1.2a) but only when applied in the up phase of the slow oscillation. We also found that TMR can promote abstraction of rules from overlapping information (WP 2.1.2b) and demonstrated the importance of sleep for integration across multiple elements in a learned hierarchy. We examined the long-term impacts of TMR in terms of behaviour and brain structure and function. We also performed a connectivity analysis showing how brain connectivity alters during sleep as a result of such stimulation. On the theoretical side, we published an analysis of how REM replay benefits consolidation.
In WP 2.2: We built a model of how memory replay facilitates consolidation using the MINERVA framework and used this to examine the impact of different types of replay upon consolidation. We also built a two-part model that has a putative ‘hippocampus’ and ‘neocortex’ which interact in a realistic way during consolidation via replay and used this to examine the impact of different types of replay. We compared the models in terms of how well they capture real-life characteristics of the response to TMR.
In WP 2.3: We explained how neural oscillations interact to boost consolidation. We explicitly manipulated neural oscillations and examined the impact on abstraction/integration and analogical reasoning tasks. We published a careful analysis of how closed loop auditory stimulation in SWS impacts on oscillations and developed a method to predict which SOs will respond to stimulation online , as well as assisting in an analysis of which tones work best in this type of closed loop stimulation, and how many stimuli should be delivered. We explored a method for extending REM sleep using auditory stimulation locked to the eye movements. We took the decision not to use electrical stimulation due to the highly conflicting literature, and instead used the time to explore the use of eyemasks for manipulating sleep, and to examine the neural structures involved in inferences and how these change with time / sleep.
In WP 2.1.1 We reviewed studies detecting neural reply with EEG classifiers then used EEG classifiers to examine the properties of neural replay during sleep. We thus characterised the time-course of memory reactivation after TMR in both REM and NREM sleep, and showed that TMR is most effective when applied in the down-to-up phase of the slow oscillation and only reactivations occurring immediately after TMR lead to behavioural benefit. We also showed that TMR cued reactivations in NREM sleep are temporally compressed compared to wake, and that memory encoding strength is an important determinant of whether reactivation provides benefits, with only medium strength encoding benefitting.
In WP 2.1.2: We used TMR to test the differential impact of memory reactivation in SWS and REM upon abstraction and the formation of unexpected connections. This showed that TMR of overlapping memories can lead to hierarchical knowledge (WP 2.1.2a) but only when applied in the up phase of the slow oscillation. We also found that TMR can promote abstraction of rules from overlapping information (WP 2.1.2b) and demonstrated the importance of sleep for integration across multiple elements in a learned hierarchy. We examined the long-term impacts of TMR in terms of behaviour and brain structure and function. We also performed a connectivity analysis showing how brain connectivity alters during sleep as a result of such stimulation. On the theoretical side, we published an analysis of how REM replay benefits consolidation.
In WP 2.2: We built a model of how memory replay facilitates consolidation using the MINERVA framework and used this to examine the impact of different types of replay upon consolidation. We also built a two-part model that has a putative ‘hippocampus’ and ‘neocortex’ which interact in a realistic way during consolidation via replay and used this to examine the impact of different types of replay. We compared the models in terms of how well they capture real-life characteristics of the response to TMR.
In WP 2.3: We explained how neural oscillations interact to boost consolidation. We explicitly manipulated neural oscillations and examined the impact on abstraction/integration and analogical reasoning tasks. We published a careful analysis of how closed loop auditory stimulation in SWS impacts on oscillations and developed a method to predict which SOs will respond to stimulation online , as well as assisting in an analysis of which tones work best in this type of closed loop stimulation, and how many stimuli should be delivered. We explored a method for extending REM sleep using auditory stimulation locked to the eye movements. We took the decision not to use electrical stimulation due to the highly conflicting literature, and instead used the time to explore the use of eyemasks for manipulating sleep, and to examine the neural structures involved in inferences and how these change with time / sleep.
1) We have demonstrated that triggered memory reactivation in REM can facilitate gist abstraction [3], while similar triggering in NREM can facilitate the formation of conceptual hierarchies [8] through the integration of different types of information.
2) We have developed EEG classifiers which can accurately detect cued memory reactivation during non-REM and REM sleep.
3) We have related the memory reactivation detected by the above classifiers to performance improvements/deficits in the above listed abilities.
4) We have developed a computational model which captures the concept of memory replay in sleep. This includes representations of the hippocampus and neocortex and how replay is coordinated between these two structures in non-REM sleep, but decoupled in REM sleep.
5) We ensured that our model captures main properties of the system, such as the fact that new information is learned in wake, then this is followed by offline replay and consolidation across a night of sleep with REM and non-REM iteratively interleaved.
6) We have used our computational model and its outputs to design further experiments using triggered memory reactivation and manipulation of oscillations.
7) We have optimised a method for boosting slow oscillations via applying sounds in sleep.
8) We have measured the long term effects (behavioural and neural) of a single night of non-REM reactivation.
9) We have shown the behavioural benefits of increasing sleep by wearing an eye mask.
10) We have shown that TMR of emotional memories in REM leads to a reduction in how subjectively arousing they are rated to be. Related to this, we showed that TMR of emotionally arousing memories in non-REM lead to a downregulation of responses in orbitofrontal cortex and insula (both of which mediate arousal responses).
2) We have developed EEG classifiers which can accurately detect cued memory reactivation during non-REM and REM sleep.
3) We have related the memory reactivation detected by the above classifiers to performance improvements/deficits in the above listed abilities.
4) We have developed a computational model which captures the concept of memory replay in sleep. This includes representations of the hippocampus and neocortex and how replay is coordinated between these two structures in non-REM sleep, but decoupled in REM sleep.
5) We ensured that our model captures main properties of the system, such as the fact that new information is learned in wake, then this is followed by offline replay and consolidation across a night of sleep with REM and non-REM iteratively interleaved.
6) We have used our computational model and its outputs to design further experiments using triggered memory reactivation and manipulation of oscillations.
7) We have optimised a method for boosting slow oscillations via applying sounds in sleep.
8) We have measured the long term effects (behavioural and neural) of a single night of non-REM reactivation.
9) We have shown the behavioural benefits of increasing sleep by wearing an eye mask.
10) We have shown that TMR of emotional memories in REM leads to a reduction in how subjectively arousing they are rated to be. Related to this, we showed that TMR of emotionally arousing memories in non-REM lead to a downregulation of responses in orbitofrontal cortex and insula (both of which mediate arousal responses).