No abstract

Researchers classify critical neural events during sleep called spindles that are related to memory consolidation using the method of scalp electroencephalography (EEG). Manual classification is time consuming and is susceptible to low inter-rater agreement. This could be improved using an automated approach. This study presents an optimized filter based and thresholding (FBT) model to set up a baseline for comparison to evaluate machine learning models using naïve features, such as raw signals, peak frequency, and dominant power. The FBT model allows us to formally define sleep spindles using signal processing but may miss examples most human scorers would agree are spindles. Machine learning methods in theory should be able to approach performance of human raters but they require a large quantity of scored data, proper feature representation, intensive feature engineering, and model selection. We evaluate both the FBT model and machine learning models with naïve features. We show that the machine learning models derived from the FBT model improve classification performance. An automated approach designed for the current data was applied to the DREAMS dataset [1]. With one of the expert’s annotation as a gold standard, our pipeline yields an excellent sensitivity that is close to a second expert’s scores and with the advantage that it can classify spindles based on multiple channels if more channels are available. More importantly, our pipeline could be modified as a guide to aid manual annotation of sleep spindles based on multiple channels quickly (6–10 seconds for processing a 40-minute EEG recording), making spindle detection faster and more objective.

How the brain representation of conceptual knowledge varies as a function of processing goals, strategies and task-factors remains a key unresolved question in cognitive neuroscience. In the present functional magnetic resonance imaging study, participants were presented with visual words during functional magnetic resonance imaging (fMRI). During shallow processing, participants had to read the items. During deep processing, they had to mentally simulate the features associated with the words. Multivariate classification, informational connectivity and encoding models were used to reveal how the depth of processing determines the brain representation of word meaning. Decoding accuracy in putative substrates of the semantic network was enhanced when the depth processing was high, and the brain representations were more generalizable in semantic space relative to shallow processing contexts. This pattern was observed even in association areas in inferior frontal and parietal cortex. Deep information processing during mental simulation also increased the informational connectivity within key substrates of the semantic network. To further examine the properties of the words encoded in brain activity, we compared computer vision models—associated with the image referents of the words—and word embedding. Computer vision models explained more variance of the brain responses across multiple areas of the semantic network. These results indicate that the brain representation of word meaning is highly malleable by the depth of processing imposed by the task, relies on access to visual representations and is highly distributed, including prefrontal areas previously implicated in semantic control.

There is growing interest in understanding how specific neural events that occur during sleep, including characteristic spindle oscillations between 10 and 16 Hz (Hz), are related to learning and memory. Neural events can be recorded during sleep using the well-known method of scalp electroencephalography (EEG). While publicly available sleep EEG datasets exist, most consist of only a few channels collected in specific patient groups being evaluated overnight for sleep disorders in clinical settings. The dataset described in this Data in Brief includes 22 participants who each participated in EEG recordings on two separate days. The dataset includes manual annotation of sleep stages and 2528 manually annotated spindles. Signals from 64-channels were continuously recorded at 1 kHz with a high-density active electrode system while participants napped for 30 or 60 min inside a sound-attenuated testing booth after performing a high- or low-load visual working memory task where load was randomized across recording days. The high-density EEG datasets present several advantages over single- or few-channel datasets including most notably the opportunity to explore spatial differences in the distribution of neural events, including whether spindles occur locally on only a few channels or co-occur globally across many channels, whether spindle frequency, duration, and amplitude vary as a function of brain hemisphere and anterior-posterior axis, and whether the probability of spindle occurrence varies as a function of the phase of ongoing slow oscillations. The dataset, along with python source code for file input and signal processing, is made freely available at the Open Science Framework through the link https://osf.io/chav7/.