Tongran Liu

Response preparation and cognitive control of highly intelligent children: a Go-Nogo event-related potential study

A cued Go-Nogo task was employed to explore the neural correlation among response preparation, cognitive control and intelligence in two groups of early adolescents with different intellectual levels using event-related potential (ERP) technique. Behavioral results indicated that the gifted children had better cognitive control performances with higher correct hit rate and lower commission error rate than the average children. Electrophysiological results further showed that the gifted children elicited efficient cue-P2 response for automatic cue detection and stronger cue-P3 activation for cue evaluation. Moreover, gifted children induced faster N2 and Nogo-P3 responses for conflict monitoring and inhibition processing and stronger P3 activation for attentional control. The current results supported the neural efficiency hypothesis of intelligence and further shed light on the close relationship among response preparation, cognitive control and human intelligence.

Conflict control of children with different intellectual levels: An ERP study

Conflict control is an important cognitive ability in human behavioral regulation. The Eriksen flanker task was employed to explore the neural correlation between conflict control and intelligence with the aid of event-related potential (ERP) techniques. Two groups of early adolescents with different intellectual levels participated in the current study (an intellectually gifted group of 20 children vs. an intellectually average group of 21 children, with mean scores of 43 vs. 35.7 in Cattells Culture Fair Test, respectively). Behavioral results indicate that the gifted children had better conflict control performances, with increased accuracy and faster response speeds than the intellectually average children. Electrophysiological results further show that the gifted children had more efficient N2 activations during conflict monitoring processing, faster P3 responses over frontal regions, and stronger P3 activations over central-parietal regions during attentional control processing. The difference waveform analysis showed that the gifted children had the weakest N2d activations when elicited by multiple conflicts. N2d amplitudes can be used to distinguish a stimulus conflict from a response conflict, and P3d amplitudes can be used to separate multiple conflicts from a single conflict. The results support the neural efficiency hypothesis of intelligence and shed light on the close relationship between conflict control ability and human intelligence.

Neural mechanisms of auditory sensory processing in children with high intelligence

To investigate the differences in event-related potential parameters related to childrens intelligence, 18 intellectually gifted children and 18 average children participated in this study. The electroencephalograms were recorded the auditory sensory memory that elicited the mismatch negativity (MMN) and late discriminative negativity (LDN), as well as involuntary attention switch that elicited the P3a and early MMN were analyzed. The results indicated that children with high intelligence had comparatively larger MMN, LDN, early MMN, P3a amplitudes, and earlier peak latency in LDN than average children. The enhanced neural function of the intellectually gifted children might be due to more spatially and temporally coordinated neural network, faster neural processing speed and more efficient neural activation functions.

Response inhibition, preattentive processing, and sex difference in young children: an event-related potential study

Response inhibition and preattentive processing are two important cognitive abilities for child development, and the current study adopted both behavioral and electrophysiological protocols to examine whether young children’s response inhibition correlated with their preattentive processing. A Go/Nogo task was used to explore young children’s response inhibition performances and an Oddball task with event-related potential recordings was used to measure their preattentive processing. The behavioral results showed that girls committed significantly fewer commission error rates, which showed that girls had stronger inhibition control abilities than boys. Girls also achieved higher d′ scores in the Go/Nogo task, which indicated that they were more sensitive to the stimulus signals than boys. Although the electrophysiological results of preattentive processing did not show any sex differences, the correlation patterns between children’s response inhibition and preattentive processing were different between these two groups: the neural response speed of preattentive processing (mismatch negativity peak latency) negatively correlated with girls’ commission error rates and positively correlated with boys’ correct hit rates. The current findings supported that the preattentive processing correlated with human inhibition control performances, and further showed that girls’ better inhibition responses might be because of the influence of their preattentive processing.

Sensory gating, inhibition control and child intelligence: an event-related potentials study

The current study explored the relationship among sensory gating, inhibition control and human intelligence in two groups of children with different intellectual levels. A Go-Nogo task was adopted to investigate childrens behavioral performances in inhibition control processing, and a paired-click paradigm with event-related potentials (ERP) recording was used to explore childrens neural activation during sensory gating processing. The behavioral results showed that the intellectually gifted children committed significantly less commission error rate, which indicated that gifted children had better inhibition control than their average peers. The electrophysiological results showed that the gifted group had lower S2P50/S1P50 amplitude ratio than the average group and illustrated that gifted children had stronger sensory gating. The results of correlation analysis between inhibition control performances and sensory gating showed that children with stronger P50 suppression (lower S2/S1 latency ratio) in the fronto-central area and stronger N100 suppression (lower S2/S1 amplitude ratio) in the frontal and fronto-central areas had shorter reaction time in the Go-Nogo task. Moreover, the correlation patterns between sensory gating and inhibition control were different between two groups of children. The present findings further demonstrated the close relationship among sensory gating, inhibition control and human intelligence in children.

Neural Correlates of Conflict Control on Facial Expressions with A Flanker Paradigm

Conflict control is an important cognitive control ability and it is also crucial for human beings to execute conflict control on affective information. To address the neural correlates of cognitive control on affective conflicts, the present study recorded event-related potentials (ERPs) during a revised Eriksen Flanker Task. Participants were required to indicate the valence of the central target expression while ignoring the flanker expressions in the affective congruent condition, affective incongruent condition and neutral condition (target expressions flanked by scramble blocks). Behavioral results manifested that participants exhibited faster response speed in identifying neutral target face when it was flanked by neutral distractors than by happy distractors. Electrophysiological results showed that happy target expression induced larger N2 amplitude when flanked by sad distractors than by happy distractors and scramble blocks during the conflict monitoring processing. During the attentional control processing, happy target expression induced faster P3 response when it was flanked by happy distractors than by sad distractors, and sad target expression evoked larger P3 amplitude when it was flanked by happy distractors comparing with sad distractors. Taken together, the current findings of temporal dynamic of brain activity during cognitive control on affective conflicts shed light on the essential relationship between cognitive control and affective information processing.

Automatic Change Detection to Facial Expressions in Adolescents: Evidence from Visual Mismatch Negativity Responses

Adolescence is a critical period for the neurodevelopment of social-emotional processing, wherein the automatic detection of changes in facial expressions is crucial for the development of interpersonal communication. Two groups of participants (an adolescent group and an adult group) were recruited to complete an emotional oddball task featuring on happy and one fearful condition. The measurement of event-related potential was carried out via electroencephalography and electrooculography recording, to detect visual mismatch negativity (vMMN) with regard to the automatic detection of changes in facial expressions between the two age groups. The current findings demonstrated that the adolescent group featured more negative vMMN amplitudes than the adult group in the fronto-central region during the 120–200 ms interval. During the time window of 370–450 ms, only the adult group showed better automatic processing on fearful faces than happy faces. The present study indicated that adolescent’s posses stronger automatic detection of changes in emotional expression relative to adults, and sheds light on the neurodevelopment of automatic processes concerning social-emotional information.

The event-related low-frequency activity of highly and average intelligent children

Using time‐frequency analysis techniques to investigate the event‐related low‐frequency (delta: 0.5–4 Hz; theta: 4–8 Hz) activity of auditory event‐related potentials (ERPs) data of highly and average intelligent children, 18 intellectually gifted children, and 18 intellectually average children participated the present study. Present findings show that intellectually gifted children had significantly larger delta activity than their normal peers in mismatch negativity (MMN) component, while in P3a component, which originates from stimulus‐driven frontal attention mechanisms during task processing, intellectually gifted children had both significantly larger delta and theta activities than their normal peers. The present findings further supported that low‐frequency brain activity could be regarded as the basis of intelligence and cognitive functions, and spectral EEG time‐frequency analysis technique should be used to explore some new aspects of brain activity relating to intelligence.

An event-related brain potential study of children's conservation

To investigate the relationship between cortical activation and conservation ability, 22 children were divided into two groups based on their performance on a standard Piagetian Conservation test. Visual evoked potentials were recorded while children performed a weight conservation task. A bilateral, frontal-distributed, broad late positive component at 900 ms differed between non-conserving and conserving children, with non-conservers having a larger amplitude. The significant interaction between conservation ability and hemisphere on the amplitude of this component suggests that inferior generators gradually move from central to right frontal-central while conservation level increases. The results indicate the existence of an ERP component that reflects weight conservation ability in children and a possible relationship between conservation ability and brain activation.

Fast Parallel Training of Neural Language Models

Training neural language models (NLMs) is very time consuming and we need parallelization for system speedup. However, standard training methods have poor scalability across multiple devices (e.g., GPUs) due to the huge time cost required to transmit data for gradient sharing in the backpropagation process. In this paper we present a sampling-based approach to reducing data transmission for better scaling of NLMs. As a “bonus”, the resulting model also improves the training speed on a single device. Our approach yields significant speed improvements on a recurrent neural network-based language model. On four NVIDIA GTX1080 GPUs, it achieves a speedup of 2.1+ times over the standard asynchronous stochastic gradient descent baseline, yet with no increase in perplexity. This is even 4.2 times faster than the naive single GPU counterpart.