Expert-level sleep staging using an electrocardiography-only feed-forward neural network.

Reliable classification of sleep stages is crucial in sleep medicine and neuroscience research for providing valuable insights, diagnoses, and understanding of brain states. The current gold standard method for sleep stage classification is polysomnography (PSG). Unfortunately, PSG is an expensive and cumbersome process involving numerous electrodes, often conducted in an unfamiliar clinic and annotated by a professional. Although commercial devices like smartwatches track sleep, their performance is well below PSG. To address these disadvantages, we present a feed-forward neural network that achieves gold-standard levels of agreement using only a single lead of electrocardiography (ECG) data. Specifically, the median five-stage Cohen's kappa is 0.725 on a large, diverse dataset of 5 to 90-year-old subjects. Comparisons with a comprehensive meta-analysis of between-human inter-rater agreement confirm the non-inferior performance of our model. Finally, we developed a novel loss function to align the training objective with Cohen's kappa. Our method offers an inexpensive, automated, and convenient alternative for sleep stage classification-further enhanced by a real-time scoring option. Cardiosomnography, or a sleep study conducted with ECG only, could take expert-level sleep studies outside the confines of clinics and laboratories and into realistic settings. This advancement democratizes access to high-quality sleep studies, considerably enhancing the field of sleep medicine and neuroscience. It makes less-expensive, higher-quality studies accessible to a broader community, enabling improved sleep research and more personalized, accessible sleep-related healthcare interventions.

Cellular composition and circuit organization of the locus coeruleus of adult mice.

The locus coeruleus (LC) houses the vast majority of noradrenergic neurons in the brain and regulates many fundamental functions, including fight and flight response, attention control, and sleep/wake cycles. While efferent projections of the LC have been extensively investigated, little is known about its local circuit organization. Here, we performed large-scale multipatch recordings of noradrenergic neurons in adult mouse LC to profile their morpho-electric properties while simultaneously examining their interactions. LC noradrenergic neurons are diverse and could be classified into two major morpho-electric types. While fast excitatory synaptic transmission among LC noradrenergic neurons was not observed in our preparation, these mature LC neurons connected via gap junction at a rate similar to their early developmental stage and comparable to other brain regions. Most electrical connections form between dendrites and are restricted to narrowly spaced pairs or small clusters of neurons of the same type. In addition, more than two electrically coupled cell pairs were often identified across a cohort of neurons from individual multicell recording sets that followed a chain-like organizational pattern. The assembly of LC noradrenergic neurons thus follows a spatial and cell-type-specific wiring principle that may be imposed by a unique chain-like rule.

Phantom flashes caused by interactions across visual space.

Studies regarding the effects of context on the perception of a visual target's temporal properties have generally addressed the cross-modal integration of auditory context, within a functional or ecological (e.g., Bayesian) framework. A deeper understanding of contextual effects in temporal vision may be gained by drawing connections with the rich models of signal processing developed in the field of spatial vision. To bridge this gap, we investigate a purely visual version of the cross-modal "double-flash" illusion (L. Shams, Y. Kamitani, & S. Shimojo, 2000; J. T. Wilson & W. Singer, 1981). Here, a single target flash can be perceived as several flashes if it is presented in the context of multiple visual inducers. This effect is robust across conditions where the target and inducers are of opposite contrast polarity, in different hemifields, are non-collinear, are presented dichoptically, or are high-frequency Gabor patches. The effect diminishes when target-inducer distance is increased or when the target is moved toward the fovea. When the target is foveated, the effect can still be recovered if the inducers are placed at 3° distance. Finally, we find that multiple target flashes are not "merged" into a smaller number of perceived flashes when presented with singular inducers. These results suggest a cortical mechanism based on isotropic propagation of transient signals or possibly based on higher level event detection. Finally, we find that multiple target flashes are not "merged" into a smaller number of perceived flashes when presented with singular inducers. These results suggest a mechanism based on the propagation of transient signals and argue against the relevance of the cue integration model developed for the cross-modal version of the effect.

What Makes an Image Interesting and How Can We Explain It.

Here, we explore the question: What makes a photograph interesting? Answering this question deepens our understanding of human visual cognition and knowledge gained can be leveraged to reliably and widely disseminate information. Observers viewed images belonging to different categories, which covered a wide, representative spectrum of real-world scenes, in a self-paced manner and, at trial's end, rated each image's interestingness. Our studies revealed the following: landscapes were the most interesting of all categories tested, followed by scenes with people and cityscapes, followed still by aerial scenes, with indoor scenes of homes and offices being least interesting. Judgments of relative interestingness of pairs of images, setting a fixed viewing duration, or changing viewing history - all of the above manipulations failed to alter the hierarchy of image category interestingness, indicating that interestingness is an intrinsic property of an image unaffected by external manipulation or agent. Contrary to popular belief, low-level accounts based on computational image complexity, color, or viewing time failed to explain image interestingness: more interesting images were not viewed for longer and were not more complex or colorful. On the other hand, a single higher-order variable, namely image uprightness, significantly improved models of average interest. Observers' eye movements partially predicted overall average interest: a regression model with number of fixations, mean fixation duration, and a custom measure of novel fixations explained >40% of variance. Our research revealed a clear category-based hierarchy of image interestingness, which appears to be a different dimension altogether from memorability or awe and is as yet unexplained by the dual appraisal hypothesis.

Perceiving a discontinuity in motion.

Studies have shown that the position of a target stimulus is misperceived owing to ongoing motion. Although static forces (fixation, landmarks) affect perceived position, motion remains the overwhelming force driving estimates of position. Motion endpoint estimates biased in the direction of motion are perceptual signatures of motion's dominant role in localization. We sought conditions in which static forces exert the predominant influence over perceived position: stimulus displays for which target position is perceived backward relative to motion. We used a target that moved diagonally with constant speed, abruptly turned 90° and continued at constant speed; observers localized the discontinuity. This yielded a previously undescribed effect, "turn-point shift," the tendency of observers to estimate the position of orthogonal direction change backward relative to subsequent motion direction. Display and mislocalization direction differ from past studies. Static forces (foveal attraction, repulsion by subsequently occupied spatial positions) were found to be responsible. Delayed turn-point estimates, reconstructed from probing the entire trajectory, shifted the horizontal coordinate forward in the direction of motion. This implies more than one percept of turn-point position. As various estimates of turn-point position arise at different times, under different task demands, the perceptual system does not necessarily resolve conflicts between them.

Orthopaedic and brain injuries over last 10 seasons in the National Football League (NFL): number and effect on missed playing time.

OBJECTIVE: To examine trends in number and seriousness of major injuries in the National Football League (NFL) over seasons 2010-2019 and the effect of rule changes to injuries to the leg, back, arm and head. METHODS: We calculated, from publicly available weekly injury reports, the number of players that were injured and playing time missed, that is, the number of weeks on average that an injured player had to sit out, as a function of injury to a specific body part. Using classical time series analysis techniques, we fitted injury data with linear and non-linear functions. RESULTS: The number of major injuries to the leg, back, arm and head has not declined over the last 10 years. During this time period, time missed because of injuries to the head has shown a significantly increasing trend. Rule changes designed specifically to protect arm or head have, respectively, succeeded in shortening the time that the injured player misses, but the impact lasts only over a single season. CONCLUSIONS: Overall, our data support the argument that new, well-intentioned rules adopted every season by the NFL have been proven to be too weak to make the NFL game safer. Broad-based management of brain and orthopaedic injuries and adoption of preventative measures to reduce the number of players injured and the seriousness of their injuries are required in the modern NFL.

Adapting to an aftereffect.

We report a new type of orientation-contingent color aftereffect in which the color aftereffect is opposite to the classical McCollough effect, i.e., the perceived color of the aftereffect is the same as the inducer's color. Interleaved exposure to red, horizontal and achromatic (gray), horizontal gratings led to a long-lasting aftereffect in which achromatic horizontal gratings appeared reddish. The effect, termed the anti-McCollough effect, although weaker than the classical aftereffect, remained stable for a moderate duration of time (24 hours). Unlike the classical aftereffect, which is known to not transfer interocularly, the new after-aftereffect transferred 100%, suggesting that its locus in the brain was downstream of the classical effect. It is likely that neurons in a higher-order area adapted to the classical color aftereffect that was represented in a lower-order area, thus forming an aftereffect of an aftereffect, i.e., an after-aftereffect. Our finding has implications as to how neural activity in lower- and higher-level areas in the brain interacts to yield conscious visual experience.

Dynamical evolution of motion perception.

Motion is defined as a sequence of positional changes over time. However, in perception, spatial position and motion dynamically interact with each other. This reciprocal interaction suggests that the perception of a moving object itself may dynamically evolve following the onset of motion. Here, we show evidence that the percept of a moving object systematically changes over time. In experiments, we introduced a transient gap in the motion sequence or a brief change in some feature (e.g., color or shape) of an otherwise smoothly moving target stimulus. Observers were highly sensitive to the gap or transient change if it occurred soon after motion onset (< or =200 ms), but significantly less so if it occurred later (> or = 300 ms). Our findings suggest that the moving stimulus is initially perceived as a time series of discrete potentially isolatable frames; later failures to perceive change suggests that over time, the stimulus begins to be perceived as a single, indivisible gestalt integrated over space as well as time, which could well be the signature of an emergent stable motion percept.

What drives slow wave activity during early non-REM sleep: Learning during prior wake or effort?

What is the function of sleep in humans? One claim is that sleep consolidates learning. Slow wave activity (SWA), i.e. slow oscillations of frequency < 4 Hz, has been observed in electroencephalograms (EEG) during sleep; it increases with prior wakefulness and decreases with sleep. Studies have claimed that increase in SWA in specific regions of the sleeping brain is correlated with overnight improved performance, i.e. overnight consolidation, on a demanding motor learning task. We wondered if SWA change during sleep is attributable to overnight consolidation or to metabolic demand. Participants executed out-and-back movements to a target using a pen-like cursor with their dominant hand while the target and cursor position were displayed on a screen. They trained on three different conditions on separate nights, differing in the amount and degree of rotation between the actual hand movement direction and displayed cursor movement direction. In the no-rotation (NR) condition, there was no rotation. In the single rotation (SR) condition, the amount of rotation remained the same throughout, and performance improved both across pre-sleep training and after sleep, i.e. overnight consolidation occurred; in the random rotation (RR) condition, the amount of rotation varied randomly from trial to trial, and no overnight consolidation occurred; SR and RR were cognitively demanding. The average EEG power density of SWA for the first 30 min. of non-rapid eye movement sleep after training was computed. Both SR and RR elicited increase in SWA in the parietal region; furthermore, the topographic distribution of SWA in each was remarkably similar. No correlation was found between the overnight performance improvement on SR and the SWA change in the parietal region on measures of learning. Our results argue that regulation of SWA in early sleep is associated with high levels of cognitive effort during prior wakefulness, and not just overnight consolidation.

Two Visual Pathways in Primates Based on Sampling of Space: Exploitation and Exploration of Visual Information.

Evidence is strong that the visual pathway is segregated into two distinct streams-ventral and dorsal. Two proposals theorize that the pathways are segregated in function: The ventral stream processes information about object identity, whereas the dorsal stream, according to one model, processes information about either object location, and according to another, is responsible in executing movements under visual control. The models are influential; however recent experimental evidence challenges them, e.g., the ventral stream is not solely responsible for object recognition; conversely, its function is not strictly limited to object vision; the dorsal stream is not responsible by itself for spatial vision or visuomotor control; conversely, its function extends beyond vision or visuomotor control. In their place, we suggest a robust dichotomy consisting of a ventral stream selectively sampling high-resolution/focal spaces, and a dorsal stream sampling nearly all of space with reduced foveal bias. The proposal hews closely to the theme of embodied cognition: Function arises as a consequence of an extant sensory underpinning. A continuous, not sharp, segregation based on function emerges, and carries with it an undercurrent of an exploitation-exploration dichotomy. Under this interpretation, cells of the ventral stream, which individually have more punctate receptive fields that generally include the fovea or parafovea, provide detailed information about object shapes and features and lead to the systematic exploitation of said information; cells of the dorsal stream, which individually have large receptive fields, contribute to visuospatial perception, provide information about the presence/absence of salient objects and their locations for novel exploration and subsequent exploitation by the ventral stream or, under certain conditions, the dorsal stream. We leverage the dichotomy to unify neuropsychological cases under a common umbrella, account for the increased prevalence of multisensory integration in the dorsal stream under a Bayesian framework, predict conditions under which object recognition utilizes the ventral or dorsal stream, and explain why cells of the dorsal stream drive sensorimotor control and motion processing and have poorer feature selectivity. Finally, the model speculates on a dynamic interaction between the two streams that underscores a unified, seamless perception. Existing theories are subsumed under our proposal.

Sound-aided recovery from and persistence against visual filling-in.

Disappearance phenomena, in which salient visual stimuli do not register consciously, have been known to occur. Recovery from such phenomena typically occurs through change in some visual attribute, such as increase in luminance contrast or stimulus duration. Thus far, there have been no reports of cross-modal modulation of disappearance phenomena. In particular, what effect a cross-modal attentional cue has on sensory suppression is unknown. Here, we show that an adapted, flickered visual target that is synchronous with a brief sound appears more vivid than a similarly adapted, otherwise identical, visual target that is offset in time by more than 200 ms from the auditory cue. We argue that the brief auditory stimuli momentarily boost the concurrent signal of the adapted visual stimulus at a site downstream of the visual adaptation, thus causing the transient recovery from the visual adaptation. Repetitive visual cues cause significantly less recovery from visual adaptation than repetitive auditory cues, implying that there are functions a cross-modal cue can perform that a cue of the same modality cannot. Moreover repetitive auditory cues selectively prevent synchronous visual targets from undergoing visual adaptation. Ours is the first report of cross-modal modulation of a disappearance phenomenon.

Stopping the motion and sleuthing the flash-lag effect: spatial uncertainty is the key to perceptual mislocalization.

A moving object is perceived to lie beyond a static object presented at the same time at the same retinal location (flash-lag effect or FLE). Some studies report that if the moving stimulus stops moving (flash-terminated condition or FTC) the instant the flash occurs, a FLE does not occur. Other studies, using different stimuli, report that the FLE does, in fact, occur in the FTC. The FTC is thus a crucial turning point in theories of flash-lag. Unraveling the mystery of the FLE in the FTC will help unravel the mechanisms underpinning flash-lag and perhaps even perceptual localization in general. Our experiments show that eccentricity of the moving stimulus was a contributing factor, as were eccentricity of the flashed stimulus and spatial separation between the two stimuli. Other factors, such as contrast and offset of moving stimulus, also modulate the magnitude of the FLE in the FTC. We surmise that uncertainty in determining the position in space of a moving stimulus is a key requirement for the lag-effect. A lag-effect in the FTC challenges influential models, such as differential latency, motion extrapolation, and postdiction. Based partly on the notion of an asymmetric spread of activity that arises because of the sheer nature of motion and from a combination of established physiological mechanisms, we propose a schematic account of the present findings that subsumes previous psychological models and scaffolds past experimental findings.

Sound-induced perturbations of the brain network in non-REM sleep, and network oscillations in wake.

During sleep, the brain network processes sensory stimuli without awareness. Stimulation must affect differently brain networks in sleep versus wake, but these differences have yet to be quantified. We recorded cortical activity in stage 2 (SII) sleep and wake using EEG while a tone was intermittently played. Zero-lag correlation measured input to pairs of sensors in the network; cross-correlation and phase-lag index measured pairwise corticocortical connectivity. Our analysis revealed that under baseline conditions, the cortical network, in particular the central regions of the frontoparietal cortex, interact at a characteristic latency of 50 ms, but only during wake, not sleep. Nonsalient auditory stimulation causes far greater perturbation of connectivity from baseline in sleep than wake, both in the response to common input and corticocortical connectivity. The findings have key implications for sensory processing.

Dynamic multi-channel TMS with reconfigurable coil.

Investigations of the causal involvement of particular brain areas and interconnections in behavior require an external stimulation system with reasonable spatio-temporal resolution. Current transcranial magnetic stimulation (TMS) technology is limited to stimulating a single brain area once in a given trial. Here, we present a feasibility study for a novel TMS system based on multi-channel reconfigurable coils. With this hardware, researchers will be able to stimulate multiple brain sites in any temporal order in a trial. The system employs a wire-mesh coil, constructed using x- and y-directional wires. By varying the current direction and/or strength on each wire, we can configure the proposed mesh-wire coil into a standard loop coil and figure-eight coil of varying size. This provides maximum flexibility to the experimenter in that the location and extent of stimulation on the brain surface can be modified depending on experimental requirement. Moreover, one can dynamically and automatically modify the site(s) of stimulation several times within the span of seconds. By pre-storing various sequences of excitation patterns inside a control unit, one can explore the effect of dynamic TMS on behavior, in associative learning, and as rehabilitative therapy. Here, we present a computer simulation and bench experiments that show the feasibility of the dynamically-reconfigurable coil.

Functional assays of local connectivity in the somatosensory cortex of individuals with autism.

Emerging evidence for differences between individuals with autism spectrum disorder (ASD) and neurotypical (NT) individuals in somatic processing and brain response to touch suggests somatosensory cortex as a promising substrate for elucidating differences in functional brain connectivity between individuals with and without autism. Signals from adjacent digits project to neighboring locations or representations in somatosensory cortex. When a digit is stimulated, i.e. touched, its representation in cortex is directly activated; local intracortical connections indirectly activate nonprimary cortical representations corresponding to adjacent digits. The response of the nonprimary cortical representations is thus a proxy for connection strength. Local overconnectivity in autism implies that the nonprimary/primary response ratios of the ASD group will be higher than those of the NT group. D1 and D2 of the dominant hand of the participant were individually stimulated while we recorded neural responses using magnetoencephalography. The cortical representations of D1 and D2 (somatosensory-evoked fields) were computed from the ensemble-averaged data using (a) dipole model fits and (b) singular value decomposition. Individual adjacent/primary response ratios were measured, and group response ratio data were fitted with straight lines. Local overconnectivity in autism implies steeper ASD vs. NT group slopes. Our findings did not support local overconnectivity. Slopes were found to be significantly shallower for the ASD group than the NT group. Our findings support the idea of local underconnectivity in the somatosensory cortex of the brains of individuals with ASD.

Interaction of finger representations in the cortex of individuals with autism: a functional window into cortical inhibition.

An established neural biomarker of autism spectrum disorder (ASD) has the potential to provide novel biological and pharmacological targets for treatment. Lower level of inhibition in brain circuits is a leading biomarker candidate. A physiological investigation of the functional levels of inhibition in the cortex of individuals with autism can provide a strong test of the hypothesis. The amplitude of cortical response to the stimulation of adjacent fingers is controlled by the level of cortical inhibition and provides just such a test. Using magnetoencephalography, we recorded the response of the somatosensory cortex to the passive tactile stimulation of the thumb (D1), and index finger (D2), and to the simultaneous stimulation of both fingers combined (D1,D2) of the dominant (right) hand of young subjects with and without autism. For each participant, we measured the response to the stimulation of both fingers combined (D1,D2) relative to the post hoc sum of the responses to the stimulation of each finger alone (D1+D2) in multiple different ways and linearly regressed the ASD and neurotypical (NT) groups' responses. The resulting slopes were then compared: Smaller slope values imply attenuated response to paired finger stimulation, and enhanced levels of inhibition. The short-latency M40 and mid-latency M80 response slopes of the group with autism obtained in different ways were either significantly smaller, or statistically indistinguishable from NT. The result does not support reduced inhibition in the somatosensory cortex of individuals with autism, contrary to the seminal hypothesis of reduced inhibition. Implications are discussed including refinements of current theory.

Sleep shelters verbal memory from different kinds of interference.

STUDY OBJECTIVES: Studies have shown that sleep shelters old verbal memories from associative interference arising from new, more recently acquired memories. Our objective is to extend the forms of interference for which sleep provides a sheltering benefit to non-associative and prospective interference, and to examine experimental conditions and memory strengths for which sleep before or after learning particularly affects verbal memory consolidation. DESIGN: Acquiring paired word associates, retention across intervening sleep and wake, training on new, interfering word associates, and test recall of both sets. SETTING: University laboratory. PARTICIPANTS: Healthy volunteers. INTERVENTIONS: N/A. MEASUREMENTS AND RESULTS: Comparing recall before and after intervening periods of sleep versus wake, we found that: (i) Sleep preferentially shields weakly encoded verbal memories from retroactive interference. (ii) Sleep immediately following learning helps shelter memory from associative and non-associative forms of retroactive interference. (iii) Sleep protects new verbal memories from prospective interference. (iv) Word associations acquired for the first time in the evening after a day spent in the wake state are encoded more strongly than word associations acquired in the morning following a night of sleep. CONCLUSIONS: The findings extend the known sleep protection from interference to non-associative as well as prospective interference, and limit the protection to weakly encoded word associations. Combined, our results suggest that sleep immediately after verbal learning isolates newly formed memory traces and renders them inaccessible, except by specific contextual cues. Memory isolation in sleep is a passive mechanism that can reasonably account for several experimental findings.

Sleep's influence on a reflexive form of memory that does not require voluntary attention.

STUDY OBJECTIVES: Studies to date have examined the influence of sleep on forms of memory that require voluntary attention. The authors examine the influence of sleep on a form of memory that is acquired by passive viewing. DESIGN: Induction of the McCollough effect, and measurement of perceptual color bias before and after induction, and before and after intervening sleep, wake, or visual deprivation. SETTING: Sound-attenuated sleep research room. PARTICIPANTS: 13 healthy volunteers (mean age = 23 years; age range = 18-31 years) with normal or corrected-to-normal vision. INTERVENTIONS: N/A. MEASUREMENTS AND RESULTS: ) ENCODING: sleep preceded adaptation. On separate nights, each participant slept for an average of 0 (wake), 1, 2, 4, or 7 hr (complete sleep). Upon awakening, the participant's baseline perceptual color bias was measured. Then, he or she viewed an adapter consisting of alternating red/horizontal and green/vertical gratings for 5 min. Color bias was remeasured. The strength of the aftereffect is the postadaptation color bias relative to baseline. A strong orientation contingent color aftereffect was observed in all participants, but total sleep duration (TSD) prior to the adaptation did not modulate aftereffect strength. Further, prior sleep provided no benefit over prior wake. Retention: sleep followed adaptation. The procedure was similar except that adaptation preceded sleep. Postadaptation sleep, irrespective of its duration (1, 3, 5, or 7 hr), arrested aftereffect decay. By contrast, aftereffect decay was arrested during subsequent wake only if the adapted eye was visually deprived. CONCLUSIONS: Sleep as well as passive sensory deprivation enables the retention of a color aftereffect. Sleep shelters this reflexive form of memory in a manner akin to preventing sensory interference.

Detecting social and non-social changes in natural scenes: performance of children with and without autism spectrum disorders and typical adults.

We probed differences in the ability to detect and interpret social cues in adults and in children and young adolescents with and without autism spectrum disorders (ASD) by investigating the effect of various social and non-social contexts on the visual exploration of pictures of natural scenes. Children and adolescents relied more on social referencing cues in the scene as compared to adults, and in the presence of such cues, were less able to use other kinds of cues. Typically developing children and adolescents were no better than those with ASD at detecting changes within the various social contexts. Results suggest children and adolescents with ASD use relevant social cues while searching a scene just as typical children do.

Extracting biomarkers of autism from MEG resting-state functional connectivity networks.

The present study is a preliminary attempt to use graph theory for deriving distinct features of resting-state functional networks in young adults with autism spectrum disorder (ASD). Networks modeled neuromagnetic signal interactions between sensors using three alternative interdependence measures: (a) a non-linear measure of generalized synchronization (robust interdependence measure [RIM]), (b) mutual information (MI), and (c) partial directed coherence (PDC). To summarize the information contained in each network model we employed well-established global graph measures (average strength, assortativity, clustering, and efficiency) as well as graph measures (average strength of edges) tailored to specific hypotheses concerning the spatial distribution of abnormalities in connectivity among individuals with ASD. Graph measures then served as features in leave-one-out classification analyses contrasting control and ASD participants. We found that combinations of regionally constrained graph measures, derived from RIM, performed best, discriminating between the two groups with 93.75% accuracy. Network visualization revealed that ASD participants displayed significantly reduced interdependence strength, both within bilateral frontal and temporal sensors, as well as between temporal sensors and the remaining recording sites, in agreement with previous studies of functional connectivity in this disorder.