Sequence Learning Task Research Articles

Deep Multi-Kernel Convolutional LSTM Networks and an Attention-Based Mechanism for Videos

Action recognition greatly benefits motion understanding in video analysis. Recurrent networks such as long short-term memory (LSTM) networks are a popular choice for motion-aware sequence learning tasks. Recently, a convolutional extension of LSTM was proposed, in which input-to-hidden and hidden-to-hidden transitions are modeled through convolution with a single kernel. This implies an unavoidable trade-off between effectiveness and efficiency. Herein, we propose a new enhancement to convolutional LSTM networks that supports accommodation of multiple convolutional kernels and layers. This resembles a Network-in-LSTM approach, which improves upon the aforementioned concern. In addition, we propose an attention-based mechanism that is specifically designed for our multi-kernel extension. We evaluated our proposed extensions in a supervised classification setting on the UCF-101 and Sports-1M datasets, with the findings showing that our enhancements improve accuracy. We also undertook qualitative analysis to reveal the characteristics of our system and the convolutional LSTM baseline.

Plasticity induction in the pre-supplementary motor area (pre-SMA) and SMA-proper differentially affects visuomotor sequence learning

BackgroundBoth pre-supplementary motor area (pre-SMA) and SMA-proper (SMA) must play important roles in visuomotor sequence learning. However, functional differences between the pre-SMA and SMA have not been well studied in humans. ObjectiveTo elucidate the functional differences between the pre-SMA and SMA in sequence learning in humans. MethodsTo induce LTP/LTD, we administered quadripulse transcranial magnetic stimulation (QPS) with an inter-stimulus interval of 5 or 50 ms (QPS-5/50) over the pre-SMA or SMA in healthy volunteers. The sham stimulation was also done as a control. We studied the effects of LTP/LTD in the pre-SMA/SMA on a new sequence learning and the performance of well-learned sequence by using sequence learning task called the “2 × 10 task”. Effects on the simple choice reaction time task were also studied for comparison. ResultsQPS-5 over the pre-SMA increased the error rate without any changes in movement speed. When administered over the SMA, QPS-5 decreased, and QPS-50 increased the rate of reaction time reduction across trials without changes in the error rate. QPS over neither the pre-SMA nor SMA affected the performances of a well-learned sequence or a simple choice reaction time task. ConclusionsOur findings that QPS over the pre-SMA correlated with sequence learning performance and that over the SMA with execution speed are consistent with the previous results in animals and humans. Our results lend further support to the utility of QPS for modulating motor learning in humans.

Dyslexia-related impairments in sequence learning predict linguistic abilities

Dyslexia is often characterized by disordered word recognition and spelling, though dysfunction on various non-linguistic tasks suggests a more pervasive deficit may underlie reading and spelling abilities. The serial-order learning impairment in dyslexia (SOLID) hypothesis proposes that sequence learning impairments fundamentally disrupt cognitive abilities, including linguistic processes, among individuals with dyslexia; yet only some studies report sequence learning deficits in people with dyslexia relative to controls. Evidence may be mixed because traditional sequence learning tasks often require strong motor demands, working memory processes and/or executive functions, wherein people with dyslexia can show impairments. Thus, observed sequence learning deficits in dyslexia may only appear to the extent that comorbid motor-based processes, memory capacity, or executive processes are involved. The present study measured sequence learning in college-aged students with and without dyslexia using a single task that evaluates sequencing and non-sequencing components but without strong motor, executive, or memory demands. During sequencing, each additional link in a sequence of stimuli leading to a reward is trained step-by-step, until a complete sequence is acquired. People with dyslexia made significantly more sequencing errors than controls, despite equivalent performance on non-sequencing components. Mediation analyses further revealed that sequence learning accounted for a large portion of the variance between dyslexia status and linguistic abilities, particularly pseudo-word reading. These findings extend the SOLID hypothesis by showing difficulties in the ability to acquire sequences that may play an underlying role in literacy acquisition.

Dissociable implicit sequence learning mechanisms revealed by continuous theta-burst stimulation.

The primary motor area (M1) has been implicated in visuomotor sequence learning. However, it has been suggested there are multiple neural networks that undertake visuomotor sequence learning. The role of M1 in sequence learning may be specific to learning simple sequences comprising predictable associations between adjacent movements. This study aimed to investigate the role of M1 in learning simple ("first-order conditional") and more complex ("second-order conditional") sequences. It was hypothesized that continuous theta burst stimulation (cTBS) over M1 would result in poorer learning of the simple sequence only. Forty-eight healthy adults received cTBS to either M1 or the parietal lobe or received sham cTBS before immediately completing 2 visuomotor sequence learning tasks. The tasks only differed in relation to the structure (i.e., simple vs. complex) of the sequence. The group who received cTBS over M1 demonstrated significantly poorer learning of the simple sequence in comparison to the more complex sequence. The parietal lobe stimulation and sham stimulation did not affect learning of either sequence. This is the first study to show differential involvement of M1 in visuomotor sequence learning, dependent on sequence structure. The study provides new evidence that sequence learning might be supported by different networks in the brain. Specifically, M1 sequence learning appears to be important for learning simple item-to-item associations but not for more complex sequences. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Is there more room to improve? The lifespan trajectory of procedural learning and its relationship to the between- and within-group differences in average response times.

Characterizing the developmental trajectories of cognitive functions such as learning, memory and decision making across the lifespan faces fundamental challenges. Cognitive functions typically encompass several processes that can be differentially affected by age. Methodological issues also arise when comparisons are made across age groups that differ in basic performance measures, such as in average response times (RTs). Here we focus on procedural learning–a fundamental cognitive function that underlies the acquisition of cognitive, social, and motor skills–and demonstrate how disentangling subprocesses of learning and controlling for differences in average RTs can reveal different developmental trajectories across the human lifespan. Two hundred-seventy participants aged between 7 and 85 years performed a probabilistic sequence learning task that enabled us to separately measure two processes of procedural learning, namely general skill learning and statistical learning. Using raw RT measures, in between-group comparisons, we found a U-shaped trajectory with children and older adults exhibiting greater general skill learning compared to adolescents and younger adults. However, when we controlled for differences in average RTs (either by using ratio scores or focusing on a subsample of participants with similar average speed), only children (but not older adults) demonstrated superior general skill learning consistently across analyses. Testing the relationship between average RTs and general skill learning within age groups shed light on further age-related differences, suggesting that general skill learning measures are more affected by average speed in some age groups. Consistent with previous studies of learning probabilistic regularities, statistical learning showed a gradual decline across the lifespan, and learning performance seemed to be independent of average speed, regardless of the age group. Overall, our results suggest that children are superior learners in various aspects of procedural learning, including both general skill and statistical learning. Our study also highlights the importance to test, and control for, the effect of average speed on other RT measures of cognitive functions, which can fundamentally affect the interpretation of group differences in developmental, aging and clinical psychology and neuroscience studies.

E-LSTM: An Efficient Hardware Architecture for Long Short-Term Memory

Long Short-Term Memory (LSTM) and its variants have been widely adopted in many sequential learning tasks, such as speech recognition and machine translation. Significant accuracy improvements can be achieved using complex LSTM model with a large memory requirement and high computational complexity, which is time-consuming and energy demanding. The low-latency and energy-efficiency requirements of the real-world applications make model compression and hardware acceleration for LSTM an urgent need. In this paper, several hardware-efficient network compression schemes are introduced first, including structured top- $k$ pruning, clipped gating, and multiplication-free quantization, to reduce the model size and the number of matrix operations by 32 $\times $ and 21.6 $\times $ , respectively, with negligible accuracy loss. Furthermore, efficient hardware architectures for accelerating the compressed LSTM are proposed, which support the inference of multi-layer and multiple time steps. The computation process is judiciously reorganized and the memory access pattern is well optimized, which alleviate the limited memory bandwidth bottleneck and enable higher throughput. Moreover, the parallel processing strategy is carefully designed to make full use of the sparsity introduced by pruning and clipped gating with high hardware utilization efficiency. Implemented on Intel Arria10 S $\times $ 660 FPGA running at 200MHz, the proposed design is able to achieve 1.4–2.2 $\times $ energy efficiency and requires significantly less hardware resources compared with the state-of-the-art LSTM implementations.

Comparison of high-definition and conventional tDCS on visuomotor sequence learning

Introduction Conventional transcranial Direct Stimulation (tDCS) has the disadvantage of spreading the current beyond the underlying cortical targets. More recent High-Definition tDCS (HD) is known to increase the spatial focality of the current. While anodal stimulation is generally thought to enhance cortical excitability and cathodal to reduce excitability, polarity-specificity of tDCS effects have been observed not to follow a simple general rule. The impact of tDCS at a network-level of brain modulation appears as a critical factor to explain its behavioral effects. Here, we propose to compare the effect of conventional-tDCS and HD-tDCS on a sequence-learning task (Serial-Response-Time-Task, SRTT) involving a network of visual and motor brain regions. Method The study involved 10 healthy participants. The SRTT generates fast (when the sequence is predicted in advance) and slow (when not) reaction times (RTs). During the task, participants were blinded for four 2 mA tDCS configurations (types: conventional, HD; conditions: active, sham). The active electrode was visual-POz cathodal. Continuous-EEG was acquired during tDCS. Results We observed a significant tDCS type condition on trial-by-trial RTs (F(1.46529) = 68.17, P Conclusions Visuomotor sequence learning improvement caused by tDCS involves a specific shift from a slow to a fast mode of processing. The visual HD-tDCS is superior over conventional-tDCS on this improvement and may be due to a focusing effect of the stimulation to the motion-perception network involved in the task. This argues for a network-level of tDCS-induced neuroplasticity rather than a local-level effect. The results raise the possibility of the usefulness of HD-tDCS in neurorehabilitation strategies for visuomotor dysfunctions.

P79-T Sensory-motor integration during a visuomotor sequence learning task

Background Sensory-motor integration (SMI) is an important process in the learning and execution of voluntary movements. This study aimed to test the electrophysiological course of SMI along with fMRI during visuomotor sequence learning (VMSL). Methods Nineteen healthy participants were included. They performed the VMSL task with their left hand. On the first day, they completed one training run and a test run. They repeated daily training tasks and were re-tested on the fifth day. Ipsilateral electrical median nerve stimulation followed by transcranial magnetic stimulation of the contralateral motor cortex with various interstimulus intervals (ISI) were performed. Recordings were made from left forearm flexor and extensor muscles. SMI studies were performed at before (T0) and after (T1) 1st-day training, and after 5th-day training (T2). Motor response (MEP) ratios for each ISI were calculated. Mean reaction time (RT) and mean error rate were measured. Results RTs (training: 283.37 ± 65.76; test: 235.20 ± 51.00; re-test: 194.26 ± 50.26 ms) and mean error rates (0.19 ± 0.16; 0.06 ± 0.06; 0.02 ± 0.03, respectively) improved significantly in the early and late stages. MEP ratios showed early afferent inhibition (SAI, ISIs 20–35 ms), late afferent inhibition (LAI, ISIs 80–200 ms) and intervening afferent facilitation (AF) for forearm flexors. Although MEP ratios at T0 andT1 were very similar, less SAI and more AF were observed at T2. Conclusions There was a tendency for decreased SAI and increased AF towards the end of the learning period. This may be caused by the learning task which requires a more prominent sensorimotor integration towards the end of the study.

Theta-Phase Dependent Neuronal Coding During Sequence Learning in Human Single Neurons

The ability to maintain a sequence of items in short-term memory (STM) is a fundamental cognitive function. Sequential item encoding in STM has been linked to a spike-theta-phase code for sequentially organized spatial locations observed in the rodent hippocampus (phase precession). The timing of neuronal activity relative to different brain oscillations is postulated to play a key role in maintaining the sequence order. We recorded single neuron and local field potential activity in the human brain during a sequence-learning task. Spikes for two consecutive items in the sequence were phase-locked at distinct phases of the theta oscillation. Surprisingly, the order of phases in the sequence-learning task was the opposite of that observed in phase precession during navigation. These results suggest that a spike-phase code is employed in the human brain during sequence learning, but with important differences compared to the rodent spike-theta-phase-dependent coding scheme.

Time-dependent motor memory representations in prefrontal cortex

How memories evolve over time is fundamental for understanding memory. Hippocampus-dependent episodic memories are generally assumed to undergo a time-dependent neural reorganization involving an increased reliance on neocortical areas. Yet, whether other forms of memory undergo a similar reorganization over time remains unclear. Here, we examined whether the neural underpinnings of motor sequence memories change over time. Participants were trained on a motor sequence learning task. Either 1d or 28d later, they performed a retention test for this task in the fMRI scanner. Sequence-specific motor memory was observed both 1d and 28d after initial training. Bayesian second-level fMRI analyses suggested a higher probability for task activity in the middle frontal gyrus and frontal pole 28d compared to 1d after initial motor learning. Searchlight representational similarity analysis indicated that areas in middle and superior frontal cortex were more involved in differentiating between multivariate activity patterns for old motor sequence memories and newly learned motor sequences in the 28d-group compared to the 1d-group. This increased involvement of lateral frontal areas during the task after 28 days was not paralleled by a decrease in those areas that were involved in performing the motor sequence retention task after 1d. These novel findings provide insights into how memories beyond the hippocampus evolve over time.

Sleep preferentially enhances memory for a cognitive strategy but not the implicit motor skills used to acquire it

Sleep is known to be beneficial to the strengthening of two distinct forms of procedural memory: memory for novel, cognitively simple series of motor movements, and memory for novel, cognitively complex strategies required to solve problems. However, these two types of memory are intertwined, since learning a new cognitive procedural strategy occurs through practice, and thereby also requires the execution of a series of simple motor movements. As a result, it is unclear whether the benefit of sleep results from the enhancement of the cognitive strategy, or the motor skills required to execute the solution. To disentangle the role of sleep in these aspects of procedural memory, we employed two tasks: (1) the Tower of Hanoi (ToH), and, (2) a modified version of the ToH, akin to an implicit Motor Sequence Learning (MSL) task. The MSL task involved the identical series of motor movements as the ToH, but without access to the information necessary to execute the task according to the underlying cognitive procedural strategy. Participants (n = 28) were trained on the 3-disk ToH, then retested on 5-disk versions of both ToH and MSL tasks. Half (n = 15) were trained and immediately tested at 8 PM and retested at 8 AM after a night of sleep. They were retested again at 8 PM after a day of wake (PM-AM-PM condition). The other half (n = 13) were trained and immediately tested at 8 AM, retested at 8 PM after a day of wake, and retested again at 8 AM after a night of sleep (AM-PM-AM condition). ToH performance only improved following a period of sleep. There was no benefit of sleep to implicit MSL. Our results show that sleep, but not wake, allowed individuals to extrapolate what was learned on a simpler 3-disk version of the task to the larger 5-disk problem, which included new elements to which they had not yet been exposed. Here, we isolate the specific role sleep plays for cognitive procedural memory: sleep benefits the cognitive strategy, rather than strengthening implicitly acquired motor sequences required to learn and execute the underlying strategy itself.

The effect of frontoparietal paired associative stimulation on decision-making and working memory

Previous single-site neurostimulation experiments have unsuccessfully attempted to shift decision-making away from habitual control, a fast, inflexible cognitive strategy, towards goal-directed control, a flexible, though computationally expensive strategy. We employed a dual-target neurostimulation approach in 30 healthy participants, using cortico-cortical paired associative stimulation (ccPAS) to target two key nodes: lateral prefrontal cortex (LPFC) and intraparietal sulcus (IPS), to test whether decision-making can be artificially shifted from habitual toward goal-directed control. Participants received three active stimulations, delivered at least six days apart (each involving 100 paired pulses over the IPS and LPFC, varying the interstimulus interval): two interventional, time-relevant ccPAS (10 msec interval) and one control, non-time-relevant ccPAS (100 msec interval). Following stimulation, participants completed a sequential learning task, measuring goal-directed/habitual control, and a working memory task. IPS→LPFC ccPAS (stimulating IPS, then LPFC with a 10 msec interval) shifted decision-making from habitual toward goal-directed control, compared to control ccPAS. There was no effect of LPFC→IPS ccPAS, nor an effect of any PAS condition on working memory. Previous studies have shown ccPAS effects outside the motor domain targeting prefrontal regions on response inhibition, attentional bias, and alpha asymmetry. The present study measures the behavioural effects of parietal-prefrontal PAS, focusing on a highly complex decision-making task and working memory. If confirmed in larger studies, this would be the first instance of neurostimulation successfully shifting decision-making from habitual to goal-directed control, putatively via inducing long-term potentiation between the IPS and LPFC. However, we found no effect in the other direction (LPFC→IPS ccPAS), and no effect on working memory overall. PAS is a relatively new neuromodulatory technique in the cognitive arsenal, and this study could help guide future approaches in healthy and disordered decision-making.

Cortical reactivations during sleep spindles following declarative learning

Increasing evidence suggests that sleep spindles are involved in memory consolidation, but few studies have investigated the effects of learning on brain responses associated with spindles in humans. Here we used simultaneous electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) during sleep to assess haemodynamic brain responses related to spindles after learning. Twenty young healthy participants were scanned with EEG/fMRI during (i) a declarative memory face sequence learning task, (ii) subsequent sleep, and (iii) recall after sleep (learning night). As a control condition an identical EEG/fMRI scanning protocol was performed after participants over-learned the face sequence task to complete mastery (control night). Results demonstrated increased responses in the fusiform gyrus both during encoding before sleep and during successful recall after sleep, in the learning night compared to the control night. During sleep, a larger response in the fusiform gyrus was observed in the presence of fast spindles during the learning as compared to the control night. Our findings support a cortical reactivation during fast spindles of brain regions previously involved in declarative learning and subsequently activated during memory recall, thereby promoting the cortical consolidation of memory traces.

Differential effects of left and right prefrontal cortex anodal transcranial direct current stimulation during probabilistic sequence learning.

Left and right prefrontal cortex and the primary motor cortex (M1) are activated during learning of motor sequences. Previous literature is mixed on whether prefrontal cortex aids or interferes with sequence learning. The present study investigated the roles of prefrontal cortices and M1 in sequence learning. Participants received anodal transcranial direct current stimulation (tDCS) to right or left prefrontal cortex or left M1 during a probabilistic sequence learning task. Relative to sham, the left prefrontal cortex and M1 tDCS groups exhibited enhanced learning evidenced by shorter response times for pattern trials, but only for individuals who did not gain explicit awareness of the sequence (implicit). Right prefrontal cortex stimulation in participants who did not gain explicit sequence awareness resulted in learning disadvantages evidenced by slower overall response times for pattern trials. These findings indicate that stimulation to left prefrontal cortex or M1 can lead to sequence learning benefits under implicit conditions. In contrast, right prefrontal cortex tDCS had negative effects on sequence learning, with overall impaired reaction time for implicit learners. There was no effect of tDCS on accuracy, and thus our reaction time findings cannot be explained by a speed-accuracy tradeoff. Overall, our findings suggest complex and hemisphere-specific roles of left and right prefrontal cortices in sequence learning. NEW & NOTEWORTHY Prefrontal cortices are engaged in motor sequence learning, but the literature is mixed on whether the prefrontal cortices aid or interfere with learning. In the current study, we used anodal transcranial direct current stimulation to target left or right prefrontal cortex or left primary motor cortex while participants performed a probabilistic sequence learning task. We found that left prefrontal and motor cortex stimulation enhanced implicit learning whereas right prefrontal stimulation negatively impacted performance.

Learning Where to See: A Novel Attention Model for Automated Immunohistochemical Scoring.

Estimating over-amplification of human epidermal growth factor receptor 2 (HER2) on invasive breast cancer is regarded as a significant predictive and prognostic marker. We propose a novel deep reinforcement learning (DRL)-based model that treats immunohistochemical (IHC) scoring of HER2 as a sequential learning task. For a given image tile sampled from multi-resolution giga-pixel whole slide image (WSI), the model learns to sequentially identify some of the diagnostically relevant regions of interest (ROIs) by following a parameterized policy. The selected ROIs are processed by recurrent and residual convolution networks to learn the discriminative features for different HER2 scores and predict the next location, without requiring to process all the sub-image patches of a given tile for predicting the HER2 score, mimicking the histopathologist who would not usually analyze every part of the slide at the highest magnification. The proposed model incorporates a task-specific regularization term and inhibition of return mechanism to prevent the model from revisiting the previously attended locations. We evaluated our model on two IHC datasets: a publicly available dataset from the HER2 scoring challenge contest and another dataset consisting of WSIs of gastroenteropancreatic neuroendocrine tumor sections stained with Glo1 marker. We demonstrate that the proposed model outperforms other methods based on state-of-the-art deep convolutional networks. To the best of our knowledge, this is the first study using DRL for IHC scoring and could potentially lead to wider use of DRL in the domain of computational pathology reducing the computational burden of the analysis of large multi-gigapixel histology images.

Modeling of Basal Ganglia to Incorporate the Procedural Memory

In this research work, a computational model is proposed for the procedural memory, and tried to mimic the functionalities of the Basal Ganglia, which is the brain region responsible for the procedural memory. In this proposed model, the procedural memory is further divided into the two sub memories. First, is the cognitive procedural memory that learns the sequence of decisions/actions, afterward it predicts the action in context of the previous occurred actions and the current sensory activations. Second, is the motor procedural memory which performs the skill learning task in context to the selected action of the cognitive procedural memory. The Learned action sequences in the cognitive procedural memory may have some overlapping which requires the identification of the current context during the sequence learning task, that helps the model in prediction of action in the current context and reduces the ambiguity. Model selects the action corresponding to the semantic activations of the semantic memory and the current context of the previous occurred actions and the gestures get invoked by the selected cognitive actions which then translated by the real working memory parameters to generate the real low-level action.

Visual sequential processing and language ability in children who are deaf or hard of hearing

This study investigated the role of sequential processing in spoken language outcomes for children who are deaf or hard of hearing (DHH), ages 5;3-11;4, by comparing them to children with typical hearing (TH), ages 6;3-9;7, on sequential learning and memory tasks involving easily nameable and difficult-to-name visual stimuli. Children who are DHH performed more poorly on easily nameable sequencing tasks, which positively predicted receptive vocabulary scores. Results suggest sequential learning and memory may underlie delayed language skills of many children who are DHH. Implications for language development in children who are DHH are discussed.

The ease and sureness of a decision: evidence accumulation of conflict and uncertainty

The likelihood of an outcome (uncertainty or sureness) and the similarity between choices (conflict or ease of a decision) are often critical to decision-making. We often ask ourselves: how likely are we to win or lose? And how different is this option's likelihood from the other? Uncertainty is a characteristic of the stimulus and conflict between stimuli, but these dissociable processes are often confounded. Here, applying a novel hierarchical drift diffusion approach, we study their interaction using a sequential learning task in healthy volunteers and pathological groups characterized by compulsive behaviours, by posing it as an evidence accumulation problem. The variables, Conflict (difficult or easy; difference between reward probabilities of the stimuli) and Uncertainty (low, medium or high; inverse U-shaped probability-uncertainty function) were then used to extract threshold ('a', amount of evidence accumulated before making a decision) and drift rate ('v', information processing speed) parameters. Critically, when a decision was both difficult (high conflict) and uncertain, relative to other conditions, healthy volunteers unexpectedly accumulated less evidence with lower decision thresholds and accuracy rates at chance levels. In contrast, patients with obsessive-compulsive disorder had slower processing speeds during these difficult uncertain decisions; yet, despite this more cautious approach, performed suboptimally with poorer accuracy relative to healthy volunteers below that of chance level. Thus, faced with a difficult uncertain decision, healthy controls are capable of rapid possibly random decisions, displaying almost a willingness to 'walk away', whereas those with obsessive compulsive disorder become more deliberative and cautious but despite appearing to learn the differential contingencies, still perform poorly. These observations might underlie disordered behaviours characterized by pathological uncertainty or doubt despite compulsive checking with impaired performance. In contrast, alcohol-dependent subjects show a different pattern relative to healthy controls with difficulties in adjusting their behavioural patterns with slower drift rates or processing speed despite decisions being easy or low conflict. We emphasize the multidimensional nature of compulsive behaviours and the utility of computational models in detecting subtle underlying processes relative to behavioural measures. These observations have implications for targeted behavioural interventions for specific cognitive impairments across psychiatric disorders.

Abstract TMP39: The neural and behavioral basis of Transcranial Direct Current Stimulation

Non-invasive brain-stimulation has been reported to modulate cortical excitability, synaptic plasticity, and interhemispheric interactions of homotop brain regions. All of these processes could play a role in facilitating stroke recovery. Modulation in cortical excitability covaries with changes in regional cerebral blood flow. Our aim was to determine whether varying parameters of transcranially applied direct current stimulation would modulate regional cerebral blood flow and behavioral measures such as motor sequence learning in a differential effect. Twelve right-handed subjects participated in all behavioral and imaging studies. We used three different dose levels (Sham, 2mA and 4mA corresponding current densities: 0mA/cm2, 0.156mA/cm2, and 0.318mA/cm2) and two different electrode montages (unihemispheric or bihemispheric) to examine tDCS effects on a finger sequence learning task and on regional cerebral blood flow (using Arterial Spin Labeling as a Blood Flow MR sequence) using an MR compatible NeuroConn DCMC stimulator. An anodal electrode (diameter of 4 cm) was placed over the right motor region (C4) while the cathodal electrode (diameter of 5 cm) was either placed over the left supraorbital region (unihemispheric montage) or over the left motor region (C3; bihemispheric montage). Each stimulation was of 10 min duration (ON condition) and preceded by 6 minutes of no stimulation and followed by 8 minutes of no stimulation (OFF conditions). A significant difference (p<0.05) in cerebral blood flow between ON and OFF conditions was seen for the group in a region of interest including the perirolandic region on the right (under the anodal electrode) showing an increase in rCBF between sham, 2mA and 4mA. There were trends for stronger bihemispheric blood flow changes for the bihemispheric montage than the unihemispheric montage. In the finger sequence learning task, subjects learned to type a randomly chosen sequence of 7 numbers with digits 2-5 of each hand as accurately and as fast as possible, before and immediately after a 10 minute stimulation phase. Higher numbers of correct finger sequence were seen in the higher current conditions for the unihemispheric condition. Results were more mixed for the bihemispheric montages.

Glucocorticoid response to stress induction prior to learning is negatively related to subsequent motor memory consolidation.

Hippocampal activity during early motor sequence learning is critical to trigger subsequent sleep-related consolidation processes. Based on previous evidence that stress-induced cortisol release modulates hippocampal activity, the current study investigates whether exposure to stress prior to motor sequence learning influences the ensuing learning and overnight consolidation process. Seventy-four healthy young adults were exposed to a stressor (i.e., the socially evaluated cold pressor test, SECPT) or a control procedure before initial training on a bimanual motor sequence learning task. Participants were retested on the motor task 24 h (including a night of sleep) after training to assess memory consolidation. Our results indicate that the SECPT, as compared to the control condition, induced significant physiological stress responses as evidenced by increased heart rate and blood pressure as well as elevated salivary cortisol concentrations. Cortisol concentration in the stress group reached peak levels immediately before and stayed significantly elevated for the full duration of initial motor learning before returning to baseline during the consolidation period. Stress induction prior to learning did not, on average, influence initial performance nor subsequent motor memory consolidation as indicated by similar overnight gains in performance in both groups. However, higher levels of stress-induced cortisol prior to training were correlated to smaller overnight gains in performance speed. These results indicate that the glucocorticoid response to a stressful encounter experienced prior to hippocampal-mediated motor learning is negatively related to subsequent memory consolidation processes.