Electrophysiology of statistical learning: Exploring the online learning process and offline learning product

A continuous stream of syllables is segmented into discrete constituents based on the transitional probabilities (TPs) between adjacent syllables by means of statistical learning. However, we still do not know whether people attend to high TPs between frequently co‐occurring syllables and cluster them together as parts of the discrete constituents or attend to low TPs aligned with the edges between the constituents and extract them as whole units. Earlier studies on TP‐based segmentation also have not distinguished between the segmentation process (how people segment continuous speech) and the learning product (what is learnt by means of statistical learning mechanisms). In the current study, we explored the learning outcome separately from the learning process, focusing on three possible learning products: holistic constituents that are retrieved from memory during the recognition test, clusters of frequently co‐occurring syllables, or a set of statistical regularities which can be used to reconstruct legitimate candidates for discrete constituents during the recognition test. Our data suggest that people employ boundary‐finding mechanisms during online segmentation by attending to low inter‐syllabic TPs during familiarization and also identify potential candidates for discrete constituents based on their statistical congruency with rules extracted during the learning process. Memory representations of recurrent constituents embedded in the continuous speech stream during familiarization facilitate subsequent recognition of these discrete constituents.     

Hippocampal and prefrontal contributions to memory retrieval: Examination of immediate early gene,NMDA receptor and environmental interactions

Animals can use a range of strategies to recall important locations. These include simple stimulus–response strategies and more complex spatial (place) strategies, which are thought to have distinct neural substrates. The hippocampus—and NMDA receptor activation therein—is considered to be crucial for spatial, but not response strategies. The medial prefrontal cortex has also been implicated in memory retrieval; however, evidence concerning its specific role is equivocal. Both hippocampal and prefrontal regions have been associated with flexible behavioural responding (e.g. when task demands change). Here, we investigated the use of spatial and non‐spatial strategies in the Morris water maze and their associated brain areas in rats using immediate early gene (IEG) imaging of Zif268 and c‐Fos. Specifically, we charted the involvement of hippocampal and prefrontal subregions during retrieval of spatial and non‐spatial memories. Behavioural flexibility was also examined using intact and partial cue configurations during recall. Results indicated that regions of both the hippocampus (area CA3) and prefrontal cortex (anterior cingulate cortex) were preferentially engaged in spatial memory recall compared to response learning. In addition, both spatial and non‐spatial memories were dependent on NMDA receptor activation. MK801 impaired recall performance across all groups and reduced IEG activation across hippocampal and prefrontal regions. Finally, IEG results revealed divergent patterns of Zif268 and c‐Fos activity and support the suggestion that Zif268 plays a functional role in the recall of long‐term memories.     

Hippocampal ripples as a mode of communication with cortical and subcortical areas

Hippocampal sharp wave‐ripple complexes are transient events of highly synchronous neuronal activity that typically occur during “offline” brain states. This endogenous surge of activity consists of behaviorally relevant spiking patterns, describing spatial trajectories. They have been shown to play a critical role in memory consolidation during sleep and in navigational planning during wakefulness. Beyond their local impact on the hippocampal formation, ripples also exert direct and indirect effects on target cortical and subcortical areas, which are thought to play a key role in information processing and semantic network reconfiguration. We review research into the function of hippocampal sharp waves‐ripples, with a special focus on information flow between the hippocampus and its cortical and subcortical targets. First, we briefly review seminal work establishing a causal role of ripple‐related activity in cognitive processes. We then review evidence for a functional interplay between hippocampal ripples and specific patterns of cortical and subcortical activity. Finally, we discuss the critical role of the functional coupling between ripples and other sleep rhythms, including the cortical slow oscillation and thalamocortical sleep spindles.     

Multimodal fusion of structural and functional brain imaging in depression using linked independent component analysis

Previous structural and functional neuroimaging studies have implicated distributed brain regions and networks in depression. However, there are no robust imaging biomarkers that are specific to depression, which may be due to clinical heterogeneity and neurobiological complexity. A dimensional approach and fusion of imaging modalities may yield a more coherent view of the neuronal correlates of depression. We used linked independent component analysis to fuse cortical macrostructure (thickness, area, gray matter density), white matter diffusion properties and resting‐state functional magnetic resonance imaging default mode network amplitude in patients with a history of depression (n = 170) and controls (n = 71). We used univariate and machine learning approaches to assess the relationship between age, sex, case–control status, and symptom loads for depression and anxiety with the resulting brain components. Univariate analyses revealed strong associations between age and sex with mainly global but also regional specific brain components, with varying degrees of multimodal involvement. In contrast, there were no significant associations with case–control status, nor symptom loads for depression and anxiety with the brain components, nor any interaction effects with age and sex. Machine learning revealed low model performance for classifying patients from controls and predicting symptom loads for depression and anxiety, but high age prediction accuracy. Multimodal fusion of brain imaging data alone may not be sufficient for dissecting the clinical and neurobiological heterogeneity of depression. Precise clinical stratification and methods for brain phenotyping at the individual level based on large training samples may be needed to parse the neuroanatomy of depression.     

Generalizability of machine learning for classification of schizophrenia based on resting‐state functional MRI data

Machine learning has increasingly been applied to classification of schizophrenia in neuroimaging research. However, direct replication studies and studies seeking to investigate generalizability are scarce. To address these issues, we assessed within‐site and between‐site generalizability of a machine learning classification framework which achieved excellent performance in a previous study using two independent resting‐state functional magnetic resonance imaging data sets collected from different sites and scanners. We established within‐site generalizability of the classification framework in the main data set using cross‐validation. Then, we trained a model in the main data set and investigated between‐site generalization in the validated data set using external validation. Finally, recognizing the poor between‐site generalization performance, we updated the unsupervised algorithm to investigate if transfer learning using additional unlabeled data were able to improve between‐site classification performance. Cross‐validation showed that the published classification procedure achieved an accuracy of 0.73 using majority voting across all selected components. External validation found a classification accuracy of 0.55 (not significant) and 0.70 (significant) using the direct and transfer learning procedures, respectively. The failure of direct generalization from one site to another demonstrates the limitation of within‐site cross‐validation and points toward the need to incorporate efforts to facilitate application of machine learning across multiple data sets. The improvement in performance with transfer learning highlights the importance of taking into account the properties of data when constructing predictive models across samples and sites. Our findings suggest that machine learning classification result based on a single study should be interpreted cautiously.     

Short‐term plasticity following motor sequence learning revealed by diffusion magnetic resonance imaging

Current noninvasive methods to detect structural plasticity in humans are mainly used to study long‐term changes. Diffusion magnetic resonance imaging (MRI) was recently proposed as a novel approach to reveal gray matter changes following spatial navigation learning and object‐location memory tasks. In the present work, we used diffusion MRI to investigate the short‐term neuroplasticity that accompanies motor sequence learning. Following a 45‐min training session in which participants learned to accurately play a short sequence on a piano keyboard, changes in diffusion properties were revealed mainly in motor system regions such as the premotor cortex and cerebellum. In a second learning session taking place immediately afterward, feedback was given on the timing of key pressing instead of accuracy, while participants continued to learn. This second session induced a different plasticity pattern, demonstrating the dynamic nature of learning‐induced plasticity, formerly thought to require months of training in order to be detectable. These results provide us with an important reminder that the brain is an extremely dynamic structure. Furthermore, diffusion MRI offers a novel measure to follow tissue plasticity particularly over short timescales, allowing new insights into the dynamics of structural brain plasticity.