Dr. Jiaxiang Zhang

Voluntary decision-making in the human brain

We can all feel the freedom of making choices to fulfil our goals and desires, even when all the options have equal or no expected values. This talk will discuss our recent work on using brain imaging, electrophysiology and modelling to understand the cognitive processes underlying such voluntary choices. First, we used model-based fMRI analysis to show that when making a voluntary choice, a decision network centred in the medial frontal cortex accumulates action intention until a response threshold is reached. Second, I will present recent results on endogenous and exogenous factors that influence voluntary choice behaviour, including tissue microstructure, neurodegeneration, reward certainty and perceptual load. Last, I will show that the randomness in a sequence of voluntary choices fluctuates over time and correlated with both fMRI and MEG activity in the frontopolar cortex. This suggests a spatial distribution of cortical regions sensitive to information regularities, which may concurrently monitor and regulate decision-making. Our results highlight the potential and challenges of establishing a neurobiological theory of voluntary behaviour

刘威

Event segmentation and integration in the human hippocampus contribute to episodic memory formation

How we form memories of our experiences is a fundamental scientific question with broad implications. In the past two decades, human neuroimaging and electrophysiology studies have implicated a distinct set of brain regions including the hippocampus and parahippocampal gyrus as well as the prefrontal cortex in processes associated with successful memory formation. However, real-world memories are formed based on a continuous stream of information instead of sequentially presented, isolated items used in these and most following studies. Potentially, the stream of life is segmented into distinct events and then integrated into a coherent, episode-based representation of their sequential relationship. Supporting this model, a hippocampal “chunking code” has been recently described in rodents that represents an episode by segmenting events and integrating them within a sequential, meaningful stream (Sun, Yang, Martin, & Tonegawa, 2019). However, relatively little is known about the precise neural correlates underlying segmentation and integration in the human brains. Here, we bolster preliminary evidence that similar neuro-computations also exist in the human hippocampus during memory encoding of real-life experience and can be measured non-invasively by fMRI. More specifically, we reanalysed an existing dataset (Baldassano et al., 2017; Chen et al., 2017) that is based on an experiment that participants watched a movie while brain activity was monitored by fMRI and were afterwards instructed to retell the full story without freely. We applied a distinct set of univariate and multi-voxel pattern analyses (MVPA) to quantify neural responses to event boundaries during movie watching and related them to subsequent memory recall. We demonstrate that distinct activations patterns of hippocampus and subregions of prefrontal cortex for the mnemonic representations of events and the beneficial effect of distinct representations. By contrast, we found that similar within-region connectivity patterns across event boundaries of these regions facilitate memory formation and are critical for the temporal order of subsequent events retrieval. We propose that distinct activation patterns are the neural implementations of event segmentation while similar connectivity patterns act as the “neural string” for event integration. Our results reveal different mnemonic functions of activation and connectivity pattern and provided novel evidence to link them with hippocampal event segmentation and integration during episodic memory formation of real-life experience.

覃恺洋

MEASURE THE CUE DRIVEN BEHAVIOR IN A NOVEL RT-TASK：THE ROLE OF MOTIVATION

People’s behavior can be induced by the external cue associated with an incentive outcome. The present study implemented a novel RT-tasks and investigated whether the variation of motivation can moderate this effect. Experiment 1 found a significant interference effect when participants were primed a Pavlovian cue which is associated with a higher value monetary reward, but this interference effect was not significant on a lower reward Pavlovian cue. Experiment 2 observed the same phenomenon by changing the higher monetary reward to 50 cents. Experiment 3 further replicated the same effect by adding a neutral Pavlovian cue which is not linked with any reward value. Experiment 4 found the moderating role of delay discounting on cue-driven responses and Experiment 5 further revealed that self-interest cue interfered participants’ responses more compared to pro-social cue. These findings provide initial evidence that RT-tasks are sensitive to detect the strength of specific Pavlovian-to-Instrumental Transfer (PIT) effect, which can be dependent on the associated subjective value.

杨金骉

No More FFT High-pass: A noise-insensitive detrend algorithm for EEG/MEG signal

EEG data can tell us the neural dynamics of language cognition because of its high temporal resolution. However, researchers often worry about the low SNR (signal-to-noise ratio) of their EEG data. The noise in EEG data can be caused by the recording device, motion of participant (e.g. eye blinks), and even task-unrelated brain activity. In practice, researchers will pre-process the data to reduce the noise, but the current denoising algorithms are not as enough as we wish them to be. We designed some new algorithms and would like to introduce them to you:

1. No More FFT High-pass: A noise-insensitive detrend algorithm for EEG/MEG signal

2. Keep The Signal Alive: EOG artifact (eye-movements) denoiser based on regression.

3. Make The Machine Learning Result More Robust: A trick for trial-based analysis.

宋玥

Prosocial Behavior in Young Preschoolers: A Cross-Cultural Study Across The Netherlands, India, and China

Although prosocial behaviors have long been regarded as universal features that are immune to socialization until the preschool age or beyond (e.g., Hamlin, 2013; Wynn, 2009), an emerging group of studies are beginning to show early socialization effects (for reviews, see Brownell et al., 2017; Dahl, 2018; Kärtner, 2018). These associations have not been consistent across cultures, however. Cultural specific relationships between parental practices and toddlers’ instrumental helping were found, with punitive practices being positively correlated with helping behavior in collectivistic cultures, but negatively in individualistic cultures (Giner Torréns & Kärtner, 2017). In addition, parents socialize young children’s prosocial behavior within a cultural context through their own values, the goals they set for the child, and the practices they use to specifically encourage prosociality. The present study compared the prosocial behavior of Dutch, Indian and Chinese preschoolers and determined whether parental values, socialization goals, and practices could explain any between- and within-country differences in prosociality.

In total 101 Dutch toddlers (M = 34.11 months, SD = 3.94 months, 55 boys), 37 Indian toddlers (M = 34.71 months, SD = 7.82 months, 15 boys), and 89 Chinese toddlers (M = 48.54 months, SD = 6.15 months, 44 boys) completed three experimental tasks (one for instrumental helping, sharing, and empathic helping, respectively), in which their prosocial behaviors (spontaneous and/or requested) towards the experimenter were observed. For these participants, 55 Dutch parents, all Indian parents and 75 Chinese parents also completed questionnaires on their values (self-enhancement and self-transcendence); socialization goals for their children (autonomous and relational goals); and parent socialization practices related to helping and sharing.

For prosocial behaviors, after controlling for age, binary logistic regressions showed that Chinese preschoolers were more likely to share on request than Dutch and Indian preschoolers, but there were no differences for spontaneous sharing, instrumental helping or empathic helping. In addition, ANOVA analyses showed no cultural differences for the percentage of treats shared, or the readiness of instrumental/empathic helping. Moreover, for parental values, goals and practices, ANOVA analyses showed that Dutch parents valued self-enhancement more strongly than both Indian and Chinese parents, and Indian parents valued self-transcendence more than Dutch and Chinese parents. In addition, Indian parents had stronger relational goals for their children than both Dutch and Chinese parents. Both Indian and Chinese parents used more practices to foster sharing than Dutch parents. Additionally, for the associations between parental values, goals and practices and prosocial behaviors, significant correlations were only found among the Indian sample. Specifically, parent’s value of self-enhancement was negatively associated with requested sharing, though a positive relationship existed between the level of autonomous socialization goals and the frequency of instrumental helping. Across all three cultures (i.e., the total sample), linear regression showed that parental socialization of helping practices was positively associated with children’s readiness to provide empathic helping.

In sum, our findings indicate that culture-specific pathways of parenting can help explain young children’s individual differences in prosocial behavior within cultures, but might not explain differences between cultures.

张磊

Using reinforcement learning models in social neuroscience: Frameworks, pitfalls, and suggestions

Recent years have witnessed a dramatic increase in the use of reinforcement learning models in social, cognitive and affective neuroscience. This approach, in combination with neuroimaging techniques such as functional magnetic resonance imaging, enables quantitative investigations into the latent mechanistic processes underlying social decision-making. However, increased use of relatively complex computational approaches has led to a range of misconceptions and imprecise interpretations of model outcomes. In this article, we present a conceptual framework for the examination of social decision-making with reinforcement learning models. We then discuss common pitfalls in the application of this approach and provide practical suggestions. We aim to provide simple and scalable explanations of reinforcement learning models in order to assist readers in better implementing and interpreting their model-based analyses.

Xiangbin Teng 滕相斌

Musical Phrase Tracking and Phase Precession at Ultra-low Frequencies of EEG Recording

Music matches speech for its hierarchical and complex structure; previous research in music neuroscience, tough provided enormous insights on music processing of note and beat structures, has not yet touched the core of music complexity – how does the brain establish high-level music structures online and segment continuous music stream? Here, we presented 10 Bach chorales to 32 participants undergoing EEG recording and investigated how the brain tracks the high-level music structures – music phrases. We removed acoustic cues indicating phrasal structures so that listeners can only rely on harmonic structures to parse music streams; we disrupted phrasal structures as in speech studies by locally or globally reversing the music pieces so that the finding on the original music can be controlled and compared. We replicated previous findings on music tracking of note and beat structures and on the positive correlation between music training and the neural tracking. Furthermore, we unprecedentedly discovered a neural signature at the ultra-low frequency range around 0.1 Hz (temporal modulations of EEG power at the beat rate) that reliably tracks music phrase structures. We further innovatively quantified phrase phase precession index (PPPi) of the EEG power and revealed that the music phrase tracking is an active operation involving predictive processes. Collectively, our findings demonstrate that the brain establishes complex music structures online at a large timescale and segments, with a predictive process, continuous music streams into units of ‘musical’ meanings, in a comparable manner of language processing.

Peiying Zuo

Trait Self‐control and Relationship Satisfaction among Heterosexual Couples: How Strong Is the Effect Really?

Previous research has demonstrated that self-control is linked to a range of positive romantic relationship processes, suggesting that self-control should be positively and robustly linked to one’s own and the partner’s relationship satisfaction. However, empirical evidence is limited and mixed. With three datasets of heterosexual couples (S1: N = 195 couples, longitudinal; S2: N = 249 couples, longitudinal; S3: N = 929 couples, cross-sectional), the present pre-registered studies examined: 1) the dyadic effects of self-control on relationship satisfaction both cross-sectionally and longitudinally; 2) whether these effects hold when controlling for relationship commitment and attachment insecurity; and 3) explored the potential moderating role of relationship commitment and attachment insecurity on the association between self-control and relationship satisfaction. To address the first two research questions, actor–partner interdependence models (APIMs) were tested across the three studies. Actor–partner interdependence moderation models (APIMoMs) were tested to answer the third research question. The results indicated 1) a cross-sectional positive actor effect, some but non-consistent support for cross-sectional partner effects, and very little support for longitudinal effects. 2) After controlling for commitment and attachment, all effects of self-control diminished. 3) Both relationship commitment and attachment insecurity moderated some associations between self-control and relationship satisfaction (particularly for men’s satisfaction), but without any consistent moderating patterns. Potential explanations for the current results, and implications for theory and practice, are discussed.

Qizhang Sun

Unpacking the Intention-Behavior Gap in Privacy Decision Making for the Internet of Things (IoT) Using Aspect Listing

In e-commerce and social networks, researchers have observed an intention-behavior gap that has been labeled the “privacy paradox”: people disclose personal information (behavior) despite expressing negative sharing intentions (in surveys). However, this phenomenon has not been studied in the Internet of Things (IoT) in which users’ personal information sharing is crucial for the functionality of the technology. We explore this phenomenon by comparing participants’ intentions via a survey or their actual behavior via a privacy-setting interface. Furthermore, we explore the decision processes underlying these privacy decisions by measuring and manipulating these processes using an aspect listing task. We find a reversed intention-behavior gap in IoT: participants disclosed less (rather than more) information in the behavior than in the intention condition, an effect that was associated with fewer benefits than risk aspects listed in the behavioral condition. The number and type of aspects listed fully mediated the effect of decision type (intention versus behavior) on the decision, which suggests that a risk-benefit calculation guided the privacy decision-making. Moreover, this reversed intention-behavior gap vanishes if we ask participants to think about positive and negative aspects of the decision, as this allows them to consider both risks and benefits, irrespective of decision type.

Chao Zhang

A Sequential Sampling Approach to the Integration of Habits and Goals

The phenomenon that habits often conflict with goal-directed behaviors is ubiquitous in people’s daily lives, and has been convincingly demonstrated in instrumental learning experiments. Recent computational models explain habit-goal conflicts as the competition between two learning systems, arbitrated by a central unit. In this paper, we show that habit-goal conflicts can be more parsimoniously explained by the dynamic integration of habit and goal values in a sequential sampling model, without any external arbitration. A computational model was developed by extending the multialternative decision field theory with the assumptions that habit values bias starting positions of preference accumulation, and that goal importance and goal relevance determine attribute sampling probabilities. Simulation studies demonstrated our approach’s ability to qualitatively reproduce empirical findings from three paradigms – reversal learning, classical devaluation, and devaluation with a concurrent schedule, and also to predict gradual changes in decision time. We discuss the implications of our work for a more unified approach to theorize the cognitive processes in instrumental learning and decision-making.

Stella Cheng Qian

How does attention modulate the switch frequency of binocular rivalry?

Binocular rivalry (BR) is observed when the two eyes receive conflicting information, leading to perceptual switches between the eyes’ images. Previous studies have examined the effect of reduced attention to a BR stimulus. Computational models predict that attention reduction should increase switch frequency, yet empirical work shows the opposite. To resolve this inconsistency, we aimed to verify the empirical finding using a design that, contrary to existing work, does not pose observers with the challenge of reporting switches and performing an attention task simultaneously. Instead, while observers performed an auditory attention task we used reflexive eye movements to track perceptual switches of task-irrelevant, moving rivalry stimuli.

Our results show that BR switch frequency decreases as the attention task becomes more challenging, thus confirming the existing empirical result. Further analysis shows that this decrease in switch frequency under reduced attention to the BR stimulus is attributable to an increased proportion of non-exclusive (piecemeal) perception. We propose that attention influences BR by facilitating grouping of the percept across space, and that a reduction of this grouping under conditions of reduced attention is responsible for the observed reduction of switch frequency as well as the observed increase in the proportion of piecemeal perception.

梁希同

Neural mechanisms of Drosophila circadian rhythms

Animals have circadian rhythms in a variety of physiological functions and behaviors, such as locomotor, sleep, and mating behaviors. In Drosophila melanogaster, these behavioral rhythms are driven by circadian clock genes oscillating in ~150 circadian pacemaker neurons. To explain how pacemaker neurons encode time and regulate different behavioral rhythms, we performed whole-brain calcium imaging in vivo for 24 hours using light-sheet microscopy. Firstly, we found that five major groups of pacemaker neurons display synchronous clock gene oscillations, yet each exhibits a distinct phase of daily neural activation. The activation phases of pacemaker groups that were associated with the morning or evening locomotor activities occurred ~4 hours before their respective behaviors. We further asked how synchronous clock gene oscillations generate nonsynchronous multi-phasic neural activity pattern. The pacemaker neural activities with a proper phase pattern required external light inputs and internal inhibitions between pacemakers. Light and inhibitory neuropeptides PDF and sNPF act dynamically at distinct hubs of the circuit to produce several multi-hour delays that create the proper tempo and sequence of pacemaker neural activities. Finally, we asked how pacemaker neural activities regulate different behavioral outputs. The circadian regulations were mediated by several downstream output circuits. These output circuits showed pacemaker-driven circadian rhythms in their neural activities that were associated with locomotor, sleep, and mating behavioral rhythms.

Sizhu Han

The dissociation of covert spatial attention during visual working memory

Covert spatial attention can be divided into endogenous attention and exogenous attention. It is well known that these two types of attention exert their functions distinctively in perceptual field. However, whether such dissociation appears in working memory field remains unclear yet. To address this issue, we conducted 4 experiments (N=132, 50 males, aged 18-28) using the retro-cue paradigm in combination with the StandardMixture Model and neuroimaging techniques. The general design was that one of 2 or 4 gabors in memory display would be cued either by an endogenous cue or by an exogenous cue during the middle of delay and then probed at the test phase. These retro-cue trials were randomly mixed with no cue trials, serving as a baseline. In experiment 1 (N=64, 21 males) and 2 (N=20, 6 males), pupillary responses and EEG were recorded respectively. We consistently found that both retro-cues similarly enhanced precision under load-4, while exogenous cues led to larger retro-cue benefits than endogenous cues by decreasing the guess rate under load-2. This distinction could be explained by a larger pupil changes as well as a shorter latency of anterior directing attention negativity (ADAN) during 300-500ms after the exogenous cue onset. To determine the critical brain areas where retro-cues were processed and how memory representation was changed over time, we conducted a MEG study (N=27, 14 males) combined with Multivariate Pattern Analysis (MVPA) and Granger Causality Analysis (GCA). The behavioral pattern duplicated the above results and we additionally found that a stronger activation at left prefrontal cortex (LPFC) for exogenous cues than endogenous cues during the post-cue 300-500ms could positively predict their behavioral differences. MVPA results further indicated that exogenous cues maintained the target representation throughout the whole post-cue period. GCA results confirmed this by showing that the top-down control from LPFC to lateral occipital cortex (LOC) started within 0-200ms after the exogenous cue onset. To examine the causal role of LPFC in retro-cue processing, we conducted a TMS study (N=21, 9 males). We found that a single pulse at LPFC could eliminate the retro-cue benefits by exogenous cues alone when the stimulation was given at post-cue 100ms. Together, these findings suggested a lower level of internal noise as well as a faster and stronger LPFC activation induced by exogenous than endogenous retro-cues under the low load. The present findings may shed light on the interplay between attention and working memory and further reveal the differentiation between perceptual attention and mnemonic attention.

Yiguang Liu

Why Smoggy Days Suppress Our Mood? - The Embodied Association Between Physical Clarity and Valence

The intuition of clarity–valence association seems to be pervasive in daily life, however, whether there exists a potential association between clarity (i.e., operationalized as visual resolution) and affect in human cognition remains unknown. The present study conducted five experiments, and demonstrated the clarity–valence congruency effect, that is, the evaluations showed performance advantage in the congruent conditions (clear-positive, blurry-negative). Experiments 1 through 3 demonstrated the influence of the perception of clarity on the conceptualization of affective valence, while Experiments 4 and 5 verified the absence of the influence of conceptualization on perception, thus the unidirectionality of clarity–valence association in cognition is confirmed. The findings extend the affective perceptual–conceptual associations into the dimension of clarity, thus providing support for the ideas of embodied cognition as well as implications for our preference for clarity and aversion to blur.

刘拓 Liu Tuo

Assimilation or Contrast: How the sense of embodiment influence the effect of thin-ideal avatars on body image in Virtual Reality

A large number of previous studies have examined the effects of operating thin-ideal avatars in virtual reality on users’ body image, although no consensus has been reached so far. One possible explanation of the inconsistent findings is that individuals’ sense of embodiment can influence the avatar-self relationship. Then, the avatar-self relationship exerts an impact on the effects of using thin ideal avatars on body image. Using implicit and explicit measures, the current study explores how the sense of embodiment (high or low) affects body image when using thin-ideal avatars in virtual reality. Eighty female participants complete a VR task using a thin-ideal avatar. Their sense of embodiment is manipulated by adjusting the body continuity of the avatar, which is a paradigm rarely used in prior studies. The result shows participants with a high sense of embodiment are more likely to have a more positive actual body image instead of ideal body image than that with a low sense of embodiment, no matter implicit or explicit measures are used. This result is robust when the alternative explanation is ruled out. These findings extend our understanding of avatar use effects in virtual reality, thereby provide new insights into the ways that the avatar in virtual reality can be used in a variety of meaningful contexts.

Yuxuan Cai

Attention modulates numerosity responses in human parietal cortex

Numerosity, the set size of visual items, guides human behavior and decisions. We previously described numerosity-selective neural populations throughout human association cortex, organized in systematic topographic maps (Harvey el al, 2013, 2017). Here, we investigate how neural responses to numerosity are affected by attention.

We used ultra-high field fMRI at 7T to measure responses elicited by viewing systematically varying numerosities. The numerosity stimuli contained black and white dots, presented simultaneously. When the white-dots systematically varied from 1 to 7, the black-dots systematically varied from 26 to 20. In this fashion the total numerosity remained constant (27). This stimulus sequence then repeated with colors switched. Subjects attended either black or white, detecting a subtle dot shape change. No numerosity judgements were required. We summarized the fMRI signals using a logarithmic Gaussian function with two parameters, preferred numerosity and tuning width, responding to the attended group’s numerosity.

Each cortical location responded specifically to the attended numerosity. Neural populations at the same cortical location preferred similar numerosities when participants either attended to white or black dots.

Identical numerosity stimuli can give rise to different numerosity responses depending on the task the subject performs. Neural responses underlying perception of numerosity are strongly modulated by attention and task demands.