The time course and characteristics of procedural learning in schizophrenia patients and healthy individuals

Schizophrenia-associated variation at ZNF804A correlates with altered experience-dependent dynamics of sleep slow waves and spindles in healthy young adults

The rs1344706 polymorphism in ZNF804A is robustly associated with schizophrenia and schizophrenia is, in turn, associated with abnormal non-rapid eye movement (NREM) sleep neurophysiology. To examine whether rs1344706 is associated with intermediate neurophysiological traits in the absence of disease, we assessed the relationship between genotype, sleep neurophysiology, and sleep-dependent memory consolidation in healthy participants. We recruited healthy adult males with no history of psychiatric disorder from the Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort. Participants were homozygous for either the schizophrenia-associated 'A' allele (N = 22) or the alternative 'C' allele (N = 18) at rs1344706. Actigraphy, polysomnography (PSG) and a motor sequence task (MST) were used to characterize daily activity patterns, sleep neurophysiology and sleep-dependent memory consolidation. Average MST learning and sleep-dependent performance improvements were similar across genotype groups, albeit more variable in the AA group. During sleep after learning, CC participants showed increased slow-wave (SW) and spindle amplitudes, plus augmented coupling of SW activity across recording electrodes. SW and spindles in those with the AA genotype were insensitive to learning, whilst SW coherence decreased following MST training. Accordingly, NREM neurophysiology robustly predicted the degree of overnight motor memory consolidation in CC carriers, but not in AA carriers. We describe evidence that rs1344706 polymorphism in ZNF804A is associated with changes in the coordinated neural network activity that supports offline information processing during sleep in a healthy population. These findings highlight the utility of sleep neurophysiology in mapping the impacts of schizophrenia-associated common genetic variants on neural circuit oscillations and function.

Sequence Learning in Minimally Verbal Children With ASD and the Beneficial Effect of Vestibular Stimulation

People with autism spectrum disorder (ASD) and especially the minimally verbal, often fail to learn basic perceptual and motor skills. This deficit has been demonstrated in several studies, but the findings could have been due to the nonoptimal adaptation of the paradigms. In the current study, we sought to characterize the skill learning deficit in young minimally verbal children with ASD and explore ways for improvement. For this purpose, we used vestibular stimulation (VS) whose beneficial effects have been demonstrated in the typical population, but the data regarding ASD are limited. We trained 36 children ages 6-13 years, ASD (N = 18, 15 of them minimally verbal) and typical development (TD, N = 18), on a touch version of the visual-motor Serial-Reaction-Time sequence-learning task, in 10 short (few minutes) weekly practice sessions. A subgroup of children received VS prior to each training block. All the participants but two ASD children showed gradual median reaction time improvement with significant speed gains across the training period. The ASD children were overall slower (by ~250 msec). Importantly, those who received VS (n = 10) showed speed gains comparable to TD, which were larger (by ~100%) than the ASD controls, and partially sequence-specific. VS had no effect on the TD group. These results suggest that VS has a positive effect on learning in minimally verbal ASD children, which may have important therapeutic implications. Furthermore, contrary to some previous findings, minimally verbal children with ASD can acquire, in optimal conditions, procedural skills with few short training sessions, spread over weeks, and with a similar time course as non-ASD controls. Autism Res 2020, 13: 320-337. © 2019 International Society for Autism Research, Wiley Periodicals, Inc. LAY SUMMARY: Minimally verbal children with ASD who received specially adjusted learning conditions showed significant learning of a visual-motor sequence across 10 practice days. This learning was considerably improved with vestibular stimulation before each short learning session. This may have important practical implications in the education and treatment of ASD children.

What Mechanisms Underlie Implicit Statistical Learning? Transitional Probabilities Versus Chunks in Language Learning

In a prior review, Perrruchet and Pacton (2006) noted that the literature on implicit learning and the more recent studies on statistical learning focused on the same phenomena, namely the domain-general learning mechanisms acting in incidental, unsupervised learning situations. However, they also noted that implicit learning and statistical learning research favored different interpretations, focusing on the selection of chunks and the computation of transitional probabilities aimed at discovering chunk boundaries, respectively. This paper examines the state of the debate 12 years later. The link between contrasting theories and their historical roots has disappeared, but a number of studies were aimed at contrasting the predictions of these two approaches. Overall, these studies strongly question the still prevalent account based on the statistical computation of pairwise associations. Various chunk-based models provide much better predictions in a number of experimental situations. However, these models rely on very different conceptual frameworks, as illustrated by a comparison between Bayesian models of word segmentation, PARSER, and a connectionist model (TRACX).

Stabilizing patterns in time: Neural network approach

Recurrent and feedback networks are capable of holding dynamic memories. Nonetheless, training a network for that task is challenging. In order to do so, one should face non-linear propagation of errors in the system. Small deviations from the desired dynamics due to error or inherent noise might have a dramatic effect in the future. A method to cope with these difficulties is thus needed. In this work we focus on recurrent networks with linear activation functions and binary output unit. We characterize its ability to reproduce a temporal sequence of actions over its output unit. We suggest casting the temporal learning problem to a perceptron problem. In the discrete case a finite margin appears, providing the network, to some extent, robustness to noise, for which it performs perfectly (i.e. producing a desired sequence for an arbitrary number of cycles flawlessly). In the continuous case the margin approaches zero when the output unit changes its state, hence the network is only able to reproduce the sequence with slight jitters. Numerical simulation suggest that in the discrete time case, the longest sequence that can be learned scales, at best, as square root of the network size. A dramatic effect occurs when learning several short sequences in parallel, that is, their total length substantially exceeds the length of the longest single sequence the network can learn. This model easily generalizes to an arbitrary number of output units, which boost its performance. This effect is demonstrated by considering two practical examples for sequence learning. This work suggests a way to overcome stability problems for training recurrent networks and further quantifies the performance of a network under the specific learning scheme.

The ability to learn and execute actions in fine temporal resolution is crucial, as many of our day to day actions require such temporal ordering (e.g. limb movement and speech). Indeed, generating stable time-varying outputs, using neural networks has attracted a lot of attention over the last years. One of the core problems, when facing such a task, is the solution stability, hence it was only possible to produce the sequence for a limited number of cycles. Here we propose a robust approach for the task of learning time-varying sequences.

Self-reported sleep disturbances associated with procedural learning impairment in adolescents at ultra-high risk for psychosis

Sleep disturbance contributes to impaired procedural learning in schizophrenia, yet little is known about this relationship prior to psychosis onset. Adolescents at ultra high-risk (UHR; N=62) for psychosis completed the Pittsburgh Sleep Quality Index (PSQI) and a procedural learning task (Pursuit Rotor). Increased self-reported problems with sleep latency, efficiency, and quality were associated with impaired procedural learning rate. Further, within-sample comparisons revealed that UHR youth reporting better sleep displayed a steeper learning curve than those with poorer sleep. Sleep disturbances appear to contribute to cognitive/motor deficits in the UHR period and may play a role in psychosis etiology.

Redefining "Learning" in Statistical Learning: What Does an Online Measure Reveal About the Assimilation of Visual Regularities?

From a theoretical perspective, most discussions of statistical learning (SL) have focused on the possible "statistical" properties that are the object of learning. Much less attention has been given to defining what "learning" is in the context of "statistical learning." One major difficulty is that SL research has been monitoring participants' performance in laboratory settings with a strikingly narrow set of tasks, where learning is typically assessed offline, through a set of two-alternative-forced-choice questions, which follow a brief visual or auditory familiarization stream. Is that all there is to characterizing SL abilities? Here we adopt a novel perspective for investigating the processing of regularities in the visual modality. By tracking online performance in a self-paced SL paradigm, we focus on the trajectory of learning. In a set of three experiments we show that this paradigm provides a reliable and valid signature of SL performance, and it offers important insights for understanding how statistical regularities are perceived and assimilated in the visual modality. This demonstrates the promise of integrating different operational measures to our theory of SL.

Offline Optimization of the Relative Timing of Movements in a Sequence Is Blocked by Retroactive Behavioral Interference

Acquisition of motor skills often involves the concatenation of single movements into sequences. Along the course of learning, sequential performance becomes progressively faster and smoother, presumably by optimization of both motor planning and motor execution. Following its encoding during training, "how-to" memory undergoes consolidation, reflecting transformations in performance and its neurobiological underpinnings over time. This offline post-training memory process is characterized by two phenomena: reduced sensitivity to interference and the emergence of delayed, typically overnight, gains in performance. Here, using a training protocol that effectively induces motor sequence memory consolidation, we tested temporal and kinematic parameters of performance within (online) and between (offline) sessions, and their sensitivity to retroactive interference. One group learned a given finger-to-thumb opposition sequence (FOS), and showed robust delayed (consolidation) gains in the number of correct sequences performed at 24 h. A second group learned an additional (interference) FOS shortly after the first and did not show delayed gains. Reduction of touch times and inter-movement intervals significantly contributed to the overall offline improvement of performance overnight. However, only the offline inter-movement interval shortening was selectively blocked by the interference experience. Velocity and amplitude, comprising movement time, also significantly changed across the consolidation period but were interference -insensitive. Moreover, they paradoxically canceled out each other. Current results suggest that shifts in the representation of the trained sequence are subserved by multiple processes: from distinct changes in kinematic characteristics of individual finger movements to high-level, temporal reorganization of the movements as a unit. Each of these processes has a distinct time course and a specific susceptibility to retroactive interference. This multiple-component view may bridge the gap in understanding the link between the behavioral changes, which define online and offline learning, and the biological mechanisms that support those changes.

Implicit motor sequence learning in schizophrenia and in old age: reduced performance only in the third session

Although there still is conflicting evidence whether schizophrenia is a neurodegenerative disease, cognitive changes in schizophrenia resemble those observed during normal aging. In contrast to extensively demonstrated deficits in explicit learning, it remains unclear whether implicit sequence learning is impaired in schizophrenia and normal aging. Implicit sequence learning was investigated using a computerized drawing task, the 'implicit pattern learning task (IPLT)' in 30 stable patients with schizophrenia, 30 age-matched controls and 30 elderly subjects on two consecutive days and after 1 week (sessions 1, 2 and 3). Fixed sequence trials were intermixed with random trials, and sequence learning was assessed by subtraction of the response time in fixed sequence trials from random trials. Separate analyses of response times and movement accuracy (i.e., directional errors) were performed. Explicit sequence knowledge was assessed using three different awareness tasks. All groups learned equally during sessions 1 and 2. In session 3, control subjects showed significantly larger learning scores than patients with schizophrenia (p = .012) and elderly subjects (p = .021). This group difference is mainly expressed in movement time and directional errors. Patients with schizophrenia demonstrated less subjective sequence awareness, and both patients with schizophrenia and elderly subjects had less explicit sequence recall. Explicit recall was positively correlated with task performance in all groups. After a short 24 h interval, all subjects showed similar improvements in implicit sequence learning. However, no benefit of prior task exposure 1 week later was observed in patients with schizophrenia and elderly subjects compared to controls. As patients with schizophrenia and elderly both display less explicit sequence recall, the control group superiority after 1 week could be explained by an explicit learning component. The few patients with schizophrenia and elderly subjects who had some sequence recall could possibly utilize this explicit knowledge to improve their task performance but did this by distinct mechanisms.

Event-related potentials demonstrate deficits in acoustic segmentation in schizophrenia

Segmentation of the acoustic environment into discrete percepts is an important facet of auditory scene analysis (ASA). Segmentation of auditory stimuli into perceptually meaningful and localizable groups is central to ASA in everyday situations; for example, separation of discrete words from continuous sentences when processing language. This is particularly relevant to schizophrenia, where deficits in perceptual organization have been linked to symptoms and cognitive dysfunction. Here we examined event-related potentials in response to grouped tones to elucidate schizophrenia-related differences in acoustic segmentation. We report for the first time in healthy subjects a sustained potential that begins with group initiation and ends with the last tone of the group. These potentials were reduced in schizophrenia, with the greatest differences in responses to first and final tones. Importantly, reductions in sustained potentials in schizophrenia patients were associated with greater negative symptoms and deficits in IQ, working memory, learning, and social cognition. These results suggest deficits in auditory pattern segmentation in schizophrenia may compound deficits in many higher-order facets of the disorder.

On Sequence Learning Models: Open-loop Control Not Strictly Guided by Hick's Law

According to the Hick’s law, reaction times increase linearly with the uncertainty of target stimuli. We tested the generality of this law by measuring reaction times in a human sequence learning protocol involving serial target locations which differed in transition probability and global entropy. Our results showed that sigmoid functions better describe the relationship between reaction times and uncertainty when compared to linear functions. Sequence predictability was estimated by distinct statistical predictors: conditional probability, conditional entropy, joint probability and joint entropy measures. Conditional predictors relate to closed-loop control models describing that performance is guided by on-line access to past sequence structure to predict next location. Differently, joint predictors relate to open-loop control models assuming global access of sequence structure, requiring no constant monitoring. We tested which of these predictors better describe performance on the sequence learning protocol. Results suggest that joint predictors are more accurate than conditional predictors to track performance. In conclusion, sequence learning is better described as an open-loop process which is not precisely predicted by Hick’s law.