Temporal Sequence Learning Research Articles

Multitasking Induced Contextual Blindness.

To examine the impact of secondary task performance on contextual blindness arising from the suppression and masking of temporal and spatial sequence learning. Dual-task scenarios can lead to a diminished ability to use environmental cues to guide attention, a phenomenon that is related to multitasking-induced inattentional blindness. This research aims to extend the theoretical understanding of how secondary tasks can impair attention and memory processes in sequence learning and access. We conducted three experiments. In Experiment 1, we used a serial reaction time task to investigate the impact of a secondary tone counting task on temporal sequence learning. In Experiment 2, we used a contextual cueing task to examine the effects of dual-task performance on spatial cueing. In Experiment 3, we integrated and extended these concepts to a simulated driving task. Across the experiments, the performance of a secondary task consistently suppressed (all experiments) and masked task learning (experiments 1 and 3). In the serial response and spatial search tasks, dual-task conditions reduced the accrual of sequence knowledge and impaired knowledge expression. In the driving simulation, similar patterns of learning suppression from multitasking were also observed. The findings suggest that secondary tasks can significantly suppress and mask sequence learning in complex tasks, leading to a form of contextual blindness characterized by impairments in the ability to use environmental cues to guide attention and anticipate future events. These findings have implications for both skill acquisition and skilled performance in complex domains such as driving, aviation, manufacturing, and human-computer interaction.

Glove-Net: Enhancing Grasp Classification with Multisensory Data and Deep Learning Approach.

Grasp classification is pivotal for understanding human interactions with objects, with wide-ranging applications in robotics, prosthetics, and rehabilitation. This study introduces a novel methodology utilizing a multisensory data glove to capture intricate grasp dynamics, including finger posture bending angles and fingertip forces. Our dataset comprises data collected from 10 participants engaging in grasp trials with 24 objects using the YCB object set. We evaluate classification performance under three scenarios: utilizing grasp posture alone, utilizing grasp force alone, and combining both modalities. We propose Glove-Net, a hybrid CNN-BiLSTM architecture for classifying grasp patterns within our dataset, aiming to harness the unique advantages offered by both CNNs and BiLSTM networks. This model seamlessly integrates CNNs' spatial feature extraction capabilities with the temporal sequence learning strengths inherent in BiLSTM networks, effectively addressing the intricate dependencies present within our grasping data. Our study includes findings from an extensive ablation study aimed at optimizing model configurations and hyperparameters. We quantify and compare the classification accuracy across these scenarios: CNN achieved 88.09%, 69.38%, and 93.51% testing accuracies for posture-only, force-only, and combined data, respectively. LSTM exhibited accuracies of 86.02%, 70.52%, and 92.19% for the same scenarios. Notably, the hybrid CNN-BiLSTM proposed model demonstrated superior performance with accuracies of 90.83%, 73.12%, and 98.75% across the respective scenarios. Through rigorous numerical experimentation, our results underscore the significance of multimodal grasp classification and highlight the efficacy of the proposed hybrid Glove-Net architectures in leveraging multisensory data for precise grasp recognition. These insights advance understanding of human-machine interaction and hold promise for diverse real-world applications.

Reconfigurable stochastic neurons based on strain engineered low barrier nanomagnets

Stochastic neurons are efficient hardware accelerators for solving a large variety of combinatorial optimization problems. ‘Binary’ stochastic neurons (BSN) are those whose states fluctuate randomly between two levels +1 and −1, with the probability of being in either level determined by an external bias. ‘Analog’ stochastic neurons (ASNs), in contrast, can assume any state between the two levels randomly (hence ‘analog’) and can perform analog signal processing. They may be leveraged for such tasks as temporal sequence learning, processing and prediction. Both BSNs and ASNs can be used to build efficient and scalable neural networks. Both can be implemented with low (potential energy) barrier nanomagnets (LBMs) whose random magnetization orientations encode the binary or analog state variables. The difference between them is that the potential energy barrier in a BSN LBM, albeit low, is much higher than that in an ASN LBM. As a result, a BSN LBM has a clear double well potential profile, which makes its magnetization orientation assume one of two orientations at any time, resulting in the binary behavior. ASN nanomagnets, on the other hand, hardly have any energy barrier at all and hence lack the double well feature. That makes their magnetizations fluctuate in an analog fashion. Hence, one can reconfigure an ASN to a BSN, and vice-versa, by simply raising and lowering the energy barrier. If the LBM is magnetostrictive, then this can be done with local (electrically generated) strain. Such a reconfiguration capability heralds a powerful field programmable architecture for a p-computer whereby hardware for very different functionalities such as combinatorial optimization and temporal sequence learning can be integrated in the same substrate in the same processing run. This is somewhat reminiscent of heterogeneous integration, except this is integration of functionalities or computational fabrics rather than components. The energy cost of reconfiguration is miniscule. There are also other applications of strain mediated barrier control that do not involve reconfiguring a BSN to an ASN or vice versa, e.g. adaptive annealing in energy minimization computing (Boltzmann or Ising machines), emulating memory hierarchy in a dynamically reconfigurable fashion, and control over belief uncertainty in analog stochastic neurons. Here, we present a study of strain engineered barrier control in unconventional computing.

Implementation of Field-Programmable Gate Array Platform for Object Classification Tasks Using Spike-Based Backpropagated Deep Convolutional Spiking Neural Networks.

This paper investigates the performance of deep convolutional spiking neural networks (DCSNNs) trained using spike-based backpropagation techniques. Specifically, the study examined temporal spike sequence learning via backpropagation (TSSL-BP) and surrogate gradient descent via backpropagation (SGD-BP) as effective techniques for training DCSNNs on the field programmable gate array (FPGA) platform for object classification tasks. The primary objective of this experimental study was twofold: (i) to determine the most effective backpropagation technique, TSSL-BP or SGD-BP, for deeper spiking neural networks (SNNs) with convolution filters across various datasets; and (ii) to assess the feasibility of deploying DCSNNs trained using backpropagation techniques on low-power FPGA for inference, considering potential configuration adjustments and power requirements. The aforementioned objectives will assist in informing researchers and companies in this field regarding the limitations and unique perspectives of deploying DCSNNs on low-power FPGA devices. The study contributions have three main aspects: (i) the design of a low-power FPGA board featuring a deployable DCSNN chip suitable for object classification tasks; (ii) the inference of TSSL-BP and SGD-BP models with novel network architectures on the FPGA board for object classification tasks; and (iii) a comparative evaluation of the selected spike-based backpropagation techniques and the object classification performance of DCSNNs across multiple metrics using both public (MNIST, CIFAR10, KITTI) and private (INHA_ADAS, INHA_KLP) datasets.

Discrimination and learning of temporal input sequences in a cerebellar Purkinje cell model.

Temporal information processing is essential for sequential contraction of various muscles with the appropriate timing and amplitude for fast and smooth motor control. These functions depend on dynamics of neural circuits, which consist of simple neurons that accumulate incoming spikes and emit other spikes. However, recent studies indicate that individual neurons can perform complex information processing through the nonlinear dynamics of dendrites with complex shapes and ion channels. Although we have extensive evidence that cerebellar circuits play a vital role in motor control, studies investigating the computational ability of single Purkinje cells are few. We found, through computer simulations, that a Purkinje cell can discriminate a series of pulses in two directions (from dendrite tip to soma, and from soma to dendrite), as cortical pyramidal cells do. Such direction sensitivity was observed in whatever compartment types of dendrites (spiny, smooth, and main), although they have dierent sets of ion channels. We found that the shortest and longest discriminable sequences lasted for 60 ms (6 pulses with 10 ms interval) and 4,000 ms (20 pulses with 200 ms interval), respectively. and that the ratio of discriminable sequences within the region of the interesting parameter space was, on average, 3.3% (spiny), 3.2% (smooth), and 1.0% (main). For the direction sensitivity, a T-type Ca2+ channel was necessary, in contrast with cortical pyramidal cells that have N-methyl-D-aspartate receptors (NMDARs). Furthermore, we tested whether the stimulus direction can be reversed by learning, specifically by simulated long-term depression, and obtained positive results. Our results show that individual Purkinje cells can perform more complex information processing than is conventionally assumed for a single neuron, and suggest that Purkinje cells act as sequence discriminators, a useful role in motor control and learning.

Memory-Based Prediction Deficits and Dorsolateral Prefrontal Dysfunction in Schizophrenia

BackgroundTheories suggest that people with schizophrenia (SZ) have problems generating predictions based on past experiences. The dorsolateral prefrontal cortex (DLPFC) and hippocampus participate in memory-based prediction. We used functional magnetic resonance imaging to investigate DLPFC and hippocampal function in healthy control (HC) subjects and people with SZ during memory-based prediction. MethodsPrior to scanning, HC subjects (n = 54) and people with SZ (n = 31) learned 5-object sequences presented in fixed or random orders on each repetition. During scanning, participants made semantic decisions (e.g., “Can this object fit in a shoebox?”) on a continuous stream of objects from fixed and random sequences. Sequence prediction was demonstrated by faster semantic decisions for objects in fixed versus random sequences because memory could be used to anticipate and more efficiently process semantic information about upcoming objects in fixed sequences. Representational similarity analyses were used to determine how each sequence type was represented in the posterior hippocampus and DLPFC. ResultsSequence predictions were reduced in individuals with SZ relative to HC subjects. Representational similarity analyses revealed stronger memory-based predictions in the DLPFC of HC subjects than people with SZ, and DLPFC representations correlated with more successful predictions in HC subjects only. For the posterior hippocampus, voxel pattern similarity was increased for fixed versus random sequences in HC subjects only, but no significant between-group differences or correlations with prediction success were observed. ConclusionsIndividuals with SZ are capable of learning temporal sequences; however, they are impaired using memory to predict upcoming events as efficiently as HC subjects. This deficit appears related to disrupted neural representation of sequence information in the DLPFC.

Choice-selective sequences dominate in cortical relative to thalamic inputs to NAc to support reinforcement learning.

SUMMARYHow are actions linked with subsequent outcomes to guide choices? The nucleus accumbens, which is implicated in this process, receives glutamatergic inputs from the prelimbic cortex and midline regions of the thalamus. However, little is known about whether and how representations differ across these input pathways. By comparing these inputs during a reinforcement learning task in mice, we discovered that prelimbic cortical inputs preferentially represent actions and choices, whereas midline thalamic inputs preferentially represent cues. Choice-selective activity in the prelimbic cortical inputs is organized in sequences that persist beyond the outcome. Through computational modeling, we demonstrate that these sequences can support the neural implementation of reinforcement-learning algorithms, in both a circuit model based on synaptic plasticity and one based on neural dynamics. Finally, we test and confirm a prediction of our circuit models by direct manipulation of nucleus accumbens input neurons.

Auditory memory for random time patterns in cochlear implant listeners.

Learning about new sounds is essential for cochlear-implant and normal-hearing listeners alike, with the additional challenge for implant listeners that spectral resolution is severely degraded. Here, a task measuring the rapid learning of slow or fast stochastic temporal sequences [Kang, Agus, and Pressnitzer (2017). J. Acoust. Soc. Am. 142, 2219-2232] was performed by cochlear-implant (N = 10) and normal-hearing (N = 9) listeners, using electric or acoustic pulse sequences, respectively. Rapid perceptual learning was observed for both groups, with highly similar characteristics. Moreover, for cochlear-implant listeners, an additional condition tested ultra-fast electric pulse sequences that would be impossible to represent temporally when presented acoustically. This condition also demonstrated learning. Overall, the results suggest that cochlear-implant listeners have access to the neural plasticity mechanisms needed for the rapid perceptual learning of complex temporal sequences.

The effect of visual statistical learning in RSVP: Implicit learning or stream location artifact?

Visual statistical learning (VSL) occurs when participants are exposed to spatially or temporally ordered stimuli, and become increasingly sensitive to them without explicitly realizing the hidden regularities. In the temporal domain of VSL, participants are usually exposed to shape-triplets, followed by the use of familiarity judgments and recognition tasks to directly probe VSL. Other methods, such as the rapid serial visual presentation (RSVP) task, indirectly probe the learning of temporal sequences, and provide evidence of learning within the triplets. Despite the RSVP's success, however, previous implementations of the RSVP task have only pseudorandomized the triplets to form the test sequence, such that the task permits only two stream locations for a given target shape (belonging to a certain within-triplet position), out of six available locations. These fixed locations may result in confounded response time (RT) findings and potentially lead to an overestimation of a weak (or nonexistent) VSL effect. In this study, we conduct 4 experiments and show that the previously reported VSL effect in RSVP is eliminated when the RSVP stream is fully counterbalanced (Experiment 1), and resurfaces again when the RSVP stream returns to its original pseudorandomized design (Experiment 2). Importantly, in both, we observe a "stream location effect," where RT gradually becomes faster to reflect the hazard rate of target appearance, which may have been the factor driving previously reported VSL effects. Follow-up experiments show that results from Experiment 1 are not due to the absence of implicit learning (Experiment 3), although the stream location effect can occur without any learning, suggesting that it is something inherent to the RSVP task (Experiment 4). Together, these results identify an influential effect, the stream location effect, inherently present in the RSVP task, and show how misinforming the RSVP task can be, when not used with properly controlled parameters, and demand a reexamination of the effects found in previous studies which could have been the effect of detection stream locations, misconstrued as VSL. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Synergetic Learning of Heterogeneous Temporal Sequences for Multi-Horizon Probabilistic Forecasting

Time-series is ubiquitous across applications, such as transportation, finance and healthcare. Time-series is often influenced by external factors, especially in the form of asynchronous events, making forecasting difficult. However, existing models are mainly designated for either synchronous time-series or asynchronous event sequence, and can hardly provide a synthetic way to capture the relation between them. We propose Variational Synergetic Multi-Horizon Network (VSMHN), a novel deep conditional generative model. To learn complex correlations across heterogeneous sequences, a tailored encoder is devised to combine the advances in deep point processes models and variational recurrent neural networks. In addition, an aligned time coding and an auxiliary transition scheme are carefully devised for batched training on unaligned sequences. Our model can be trained effectively using stochastic variational inference and generates probabilistic predictions with Monte-Carlo simulation. Furthermore, our model produces accurate, sharp and more realistic probabilistic forecasts. We also show that modeling asynchronous event sequences is crucial for multi-horizon time-series forecasting.

Disrupted Modulation of Alpha and Low Beta Oscillations Mediates Temporal Sequence Memory Deficits in People With Schizophrenia

BackgroundPeople with schizophrenia (SZ) exhibit impaired episodic memory when relating objects to each other in time and space. Empirical studies and computational models suggest that low-frequency neural oscillations may be a mechanism by which the brain keeps track of temporal relationships during encoding and retrieval, with modulation of oscillatory power as sequences are learned. It is unclear whether sequence memory deficits in SZ are associated with altered neural oscillations. MethodsUsing electroencephalography, this study examined neural oscillations in 51 healthy control subjects and 37 people with SZ during a temporal sequence learning task. Multiple 5-object picture sequences were presented across 4 study-test blocks in either fixed or random order. Participants answered semantic questions for each object (e.g., living/nonliving), and sequence memory was operationalized as faster responses for fixed versus random sequences. Differences in oscillatory power between fixed versus random sequences provided a neural index of temporal sequence memory. ResultsAlthough both groups showed reaction time differences in late blocks (blocks 3 and 4), this evidence of sequence memory was reduced in people with SZ relative to healthy control subjects. Decreases in globally distributed prestimulus alpha (8–12 Hz) and beta 1 (13–20 Hz) power for fixed versus random sequences in late blocks were also attenuated in people with SZ relative to healthy control subjects. Moreover, changes in oscillatory power predicted individual reaction time differences and fully mediated the relationship between group and sequence memory. ConclusionsDisrupted modulation of alpha and beta 1 electroencephalography oscillations is a candidate mechanism of temporal sequence memory deficits in people with SZ.

The Differential Effects of Auditory and Visual Stimuli on Learning, Retention and Reactivation of a Perceptual-Motor Temporal Sequence in Children With Developmental Coordination Disorder.

This study investigates the procedural learning, retention, and reactivation of temporal sensorimotor sequences in children with and without developmental coordination disorder (DCD). Twenty typically-developing (TD) children and 12 children with DCD took part in this study. The children were required to tap on a keyboard, synchronizing with auditory or visual stimuli presented as an isochronous temporal sequence, and practice non-isochronous temporal sequences to memorize them. Immediate and delayed retention of the audio-motor and visuo-motor non-isochronous sequences were tested by removing auditory or visual stimuli immediately after practice and after a delay of 2 h. A reactivation test involved reintroducing the auditory and visual stimuli after the delayed recall. Data were computed via circular analyses to obtain asynchrony, the stability of synchronization and errors (i.e., the number of supplementary taps). Firstly, an overall deficit in synchronization with both auditory and visual isochronous stimuli was observed in DCD children compared to TD children. During practice, further improvements (decrease in asynchrony and increase in stability) were found for the audio-motor non-isochronous sequence compared to the visuo-motor non-isochronous sequence in both TD children and children with DCD. However, a drastic increase in errors occurred in children with DCD during immediate retention as soon as the auditory stimuli were removed. Reintroducing auditory stimuli decreased errors in the audio-motor sequence for children with DCD. Such changes were not seen for the visuo-motor non-isochronous sequence, which was equally learned, retained and reactivated in DCD and TD children. All these results suggest that TD children benefit from both auditory and visual stimuli to memorize the sequence, whereas children with DCD seem to present a deficit in integrating an audio-motor sequence in their memory. The immediate effect of reactivation suggests a specific dependency on auditory information in DCD. Contrary to the audio-motor sequence, the visuo-motor sequence was both learned and retained in children with DCD. This suggests that visual stimuli could be the best information for memorizing a temporal sequence in DCD. All these results are discussed in terms of a specific audio-motor coupling deficit in DCD.

Statistical learning of unbalanced exclusive-or temporal sequences in humans.

A pervasive issue in statistical learning has been to determine the parameters of regularity extraction. Our hypothesis was that the extraction of transitional probabilities can prevail over frequency if the task involves prediction. Participants were exposed to four repeated sequences of three stimuli (XYZ) with each stimulus corresponding to the position of a red dot on a touch screen that participants were required to touch sequentially. The temporal and spatial structure of the positions corresponded to a serial version of the exclusive-or (XOR) that allowed testing of the respective effect of frequency and first- and second-order transitional probabilities. The XOR allowed the first-order transitional probability to vary while being not completely related to frequency and to vary while the second-order transitional probability was fixed (p(Z|X, Y) = 1). The findings show that first-order transitional probability prevails over frequency to predict the second stimulus from the first and that it also influences the prediction of the third item despite the presence of second-order transitional probability that could have offered a certain prediction of the third item. These results are particularly informative in light of statistical learning models.

Analog Signal Processing Using Stochastic Magnets

We present a low energy-barrier magnet based compact hardware unit for analog stochastic neurons (ASNs) and demonstrate its use as a building-block for neuromorphic hardware. Networks assembled from these units are particularly suited for temporal inferencing and pattern recognition. We demonstrate example applications of these ASNs including multi-layer perceptrons, convolutional neurons, and reservoir computers showing tasks such as temporal sequence learning, processing, and prediction tasks which prove that these units can be used to build efficient, scalable, and adaptive neural network based signal-processors. We also provide an illustrative comparison with digital CMOS based circuits that implement similar functionality with networks built using the presented units, demonstrating a possible two orders of magnitude reduction in component-count and concomitant increase in energy efficiency. Efficient non von-Neumann hardware implementation of such signal-processors can open up a pathway for integration of hardware based cognition in a wide variety of emerging systems such as IoT, industrial controls, bio- and photo-sensors, self-driving automotives, and unmanned aerial vehicles.

Procedural learning and retention of audio-verbal temporal sequence is altered in children with developmental coordination disorder but cortical thickness matters.

Rhythmic abilities are impaired in developmental coordination disorder (DCD) but learning deficit of procedural skills implying temporal sequence is still unclear. Current contradictory results suggest that procedural learning deficits in DCD highly depend on learning conditions. The present study proposes to test the role of sensory modality of stimulations (visual or auditory) on synchronization, learning, and retention of temporal verbal sequences in children with and without DCD. We postulated a deficit in learning particularly with auditory stimulations, in association with atypical cortical thickness of three regions of interesting: sensorimotor, frontal and parietal regions. Thirty children with and without DCD (a) performed a synchronization task to a regular temporal sequence and (b) practiced and recalled a novel non-regular temporal sequences with auditory and visual modalities. They also had a magnetic resonance imaging to measure their cortical thickness. Results suggested that children with DCD presented a general deficit in synchronization of a regular temporal verbal sequence irrespective of the sensory modality, but a specific deficit in learning and retention of auditory non-regular verbal temporal sequence. Stability of audio-verbal synchronization during practice correlated with cortical thickness of the sensorimotor cortex. For the first time, our results suggest that synchronization deficits in DCD are not limited to manual tasks. This deficit persists despite repeated exposition and practice of an auditory temporal sequence, which suggests a possible alteration in audio-verbal coupling in DCD. On the contrary, control of temporal parameters with visual stimuli seems to be less affected, which opens perspectives for clinical practice.

Temporal Sequence Learning in People With Schizophrenia

Temporal Sequence Learning in People With Schizophrenia

Consistent within-group covariance of septal and temporal hippocampal neurogenesis with behavioral phenotypes for exploration and memory retention across wild and laboratory small rodents

Striking differences in the septo-temporal distribution of neurogenesis are found in small rodents. Here, we assessed the association of adult hippocampal neurogenesis with behavioral responses to novelty, temporal and spatial sequence and reversal learning in wild, wild-derived and laboratory rodents using an automated testing apparatus, the IntelliCage. Behaviorally, DBA/2 and wild-derived house mice were quickest to explore a novel environment, wild wood mice and bank voles were slowest, and C57BL/6 intermediate. Rule learning (temporal and spatial) was fastest in wood mice and bank voles, while DBA/2 and house mice performed poorer. C57BL/6 performed similar to the house mice in the temporal task and similar to wild rodents in the spatial task. Using the number of DCX-positive neurons and proliferating, Ki67-positive cells in the septal, intermediate and temporal hippocampus as a proxy, an ANCOVA was used to test for within-group relations between neurogenesis and behavior. We found that higher numbers of DCX-positive cells in the temporal hippocampus were associated with an increased latency and a lower frequency to explore a novel environment. Temporal and spatial sequence learning was not associated with neurogenesis. In the spatial reversal task however, animals with higher septal neurogenesis showed a persevering phenotype and slower re-learning. Our findings provide strong evidence of septo-temporally segregated neurogenesis effects on behavior across five rodent strains and species. While temporal neurogenesis covaries with behavioral responses to novelty, septal neurogenesis relates to perseverance of a successfully learned spatial rule. Importantly, these associations were independent of species or strain and can be found in both wild and domesticated rodents.

Multimodal imaging of brain connectivity reveals predictors of individual decision strategy in statistical learning.

Successful human behaviour depends on the brain's ability to extract meaningful structure from information streams and make predictions about future events. Individuals can differ markedly in the decision strategies they use to learn the environment's statistics, yet we have little idea why. Here, we investigate whether the brain networks involved in learning temporal sequences without explicit reward differ depending on the decision strategy that individuals adopt. We demonstrate that individuals alter their decision strategy in response to changes in temporal statistics and engage dissociable circuits: extracting the exact sequence statistics relates to plasticity in motor corticostriatal circuits, while selecting the most probable outcomes relates to plasticity in visual, motivational and executive corticostriatal circuits. Combining graph metrics of functional and structural connectivity, we provide evidence that learning-dependent changes in these circuits predict individual decision strategy. Our findings propose brain plasticity mechanisms that mediate individual ability for interpreting the structure of variable environments.

Theta oscillations promote temporal sequence learning.

Many theoretical models suggest that neural oscillations play a role in learning or retrieval of temporal sequences, but the extent to which oscillations support sequence representation remains unclear. To address this question, we used scalp electroencephalography (EEG) to examine oscillatory activity over learning of different object sequences. Participants made semantic decisions on each object as they were presented in a continuous stream. For three "Consistent" sequences, the order of the objects was always fixed. Activity during Consistent sequences was compared to "Random" sequences that consisted of the same objects presented in a different order on each repetition. Over the course of learning, participants made faster semantic decisions to objects in Consistent, as compared to objects in Random sequences. Thus, participants were able to use sequence knowledge to predict upcoming items in Consistent sequences. EEG analyses revealed decreased oscillatory power in the theta (4-7 Hz) band at frontal sites following decisions about objects in Consistent sequences, as compared with objects in Random sequences. The theta power difference between Consistent and Random only emerged in the second half of the task, as participants were more effectively able to predict items in Consistent sequences. Moreover, we found increases in parieto-occipital alpha (10-13 Hz) and beta (14-28 Hz) power during the pre-response period for objects in Consistent sequences, relative to objects in Random sequences. Linear mixed effects modeling revealed that single trial theta oscillations were related to reaction time for future objects in a sequence, whereas beta and alpha oscillations were only predictive of reaction time on the current trial. These results indicate that theta and alpha/beta activity preferentially relate to future and current events, respectively. More generally our findings highlight the importance of band-specific neural oscillations in the learning of temporal order information.

Neural Signatures of Spatial Statistical Learning: Characterizing the Extraction of Structure from Complex Visual Scenes.

Behavioral evidence has shown that humans automatically develop internal representations adapted to the temporal and spatial statistics of the environment. Building on prior fMRI studies that have focused on statistical learning of temporal sequences, we investigated the neural substrates and mechanisms underlying statistical learning from scenes with a structured spatial layout. Our goals were twofold: (1) to determine discrete brain regions in which degree of learning (i.e., behavioral performance) was a significant predictor of neural activity during acquisition of spatial regularities and (2) to examine how connectivity between this set of areas and the rest of the brain changed over the course of learning. Univariate activity analyses indicated a diffuse set of dorsal striatal and occipitoparietal activations correlated with individual differences in participants' ability to acquire the underlying spatial structure of the scenes. In addition, bilateral medial-temporal activation was linked to participants' behavioral performance, suggesting that spatial statistical learning recruits additional resources from the limbic system. Connectivity analyses examined, across the time course of learning, psychophysiological interactions with peak regions defined by the initial univariate analysis. Generally, we find that task-based connectivity with these regions was significantly greater in early relative to later periods of learning. Moreover, in certain cases, decreased task-based connectivity between time points was predicted by overall posttest performance. Results suggest a narrowing mechanism whereby the brain, confronted with a novel structured environment, initially boosts overall functional integration and then reduces interregional coupling over time.