The Academic Salon of the Department of Psychological and Cognitive Sciences is held every Thursday afternoon. Welcome to all students and faculty members from every department!
Time: June 5 (Thursday) afternoon 15:30
Location: Room 1103, 11th floor, Lv Dalong Building
Department of Psychology and Cognitive Sciences Fei Wang Research Group Academic Salon
Report One
Metacognitive-Sensory Conflict in Self-Voice Recognition: ERP Research Evidence
Presenter: Nan Li
Content:
The proliferation of digital technology has significantly increased individuals' exposure to their own recorded voices, with many reporting discomfort when hearing their own recordings. Behavioral research indicates that this discomfort stems from cognitive dissonance caused by conflict between metacognition (voice source) and sensory cognition (voice physical attributes) (i.e., perceived voice does not match self-expectation), but its neural basis remains unclear. This study (n=30) used voice transformation technology in a laboratory environment to establish new self-voice and other-voice identity associations for participants, thereby separating metacognition and sensory cognition. Using an event-related potential (ERP) paradigm, a 64-channel EEG system was used to record participants' brain electrical responses when listening to different voice-transformed speech. Results observed the classic N2-P3 component combination related to cognitive control in the frontal-parietal region. Specifically, when metacognition was self, the N2 peak negative amplitude at frontal-parietal electrodes under sensory cognition inconsistent conditions was significantly greater than under consistent conditions; when metacognition was other, N2 peak differences were not significant. Regardless of whether metacognition was self or other, compared to conditions consistent with the learning phase voice transformation, P3 peaks in the central-parietal region significantly decreased under inconsistent conditions. The above effects showed polarity reversal at occipital electrodes. The ERP results of this study provide neurophysiological evidence for the cognitive dissonance theory of metacognitive and sensory cognition conflict found in previous behavioral research, deepening understanding of the neural mechanisms of self-perception.
Report Two
The Core of Self: True Self Advantage Effect in Shape-Label Matching Paradigm
Presenter: Yue Zhao
Content:
Psychological research has long focused on the distinction between self and others. However, within the self contains multiple factors—do all parts of the self have equal importance, or are some parts more central? Recent research indicates that even within the self, people select a subset and regard it as the "true self." The true self reflects an individual's beliefs about who they truly are, or the most authentic aspect of self-concept. This study used an adapted shape-label matching paradigm to examine the priority of true self-related stimuli in cognitive processing. Research results showed that compared to general self and strangers, true self-related stimuli exhibited processing advantages, and the processing level of general self-related stimuli was comparable to that of strangers (Experiment 1). However, in the absence of true self stimuli, general self-related stimuli showed processing advantages compared to friend and stranger stimuli (Experiment 2). Further research results showed that the processing advantage difference between true self and general self was not significant, and ruled out interference caused by neglect of the "general self" definition (Experiment 3). These findings provide empirical evidence for the true self advantage effect, indicating that even within the self, certain components still have cognitive priority.
Report Three
Optimization of Matching Paradigm for Self-Prioritization Effect: An Attempt at Word Association Learning
Presenter: Changlin Luo
Content:
The Self-prioritization Effect (SPE) refers to self-related stimuli usually having processing advantages, such as the classic cocktail party effect. The perceptual matching paradigm (shape-label matching paradigm) proposed by Sui et al. (2012) solved the interference of familiarity on SPE, however, the association between graphic stimuli and identity labels established by this paradigm is difficult to generalize, making it difficult to further explore the cognitive mechanisms of SPE. Therefore, this study proposes using word association learning methods to optimize the association between graphic stimuli and identity labels in the matching paradigm, and then explore cognitive patterns of SPE in other paradigms.
This study used a 2 (learning method: classic matching vs. word association) × 3 (identity label: self vs. friend vs. stranger) mixed design, with learning method as a between-subjects variable. The study recruited 48 participants (32 females), with 24 participants in each group needing to complete classic matching learning or word association learning, entering the formal experimental phase after testing; participants needed to judge whether graphic stimuli and identity labels on the screen matched and press keys, with reaction time and accuracy recorded for subsequent data analysis.
Linear mixed model results showed that participants in both the classic matching group and word association group exhibited significant self-prioritization effects, with participants responding faster and more accurately when judging "self" labels. However, the interaction between learning method and label was not significant, the accuracy of the word association group showed no significant difference from the classic matching group, and the reaction time of the word association group was significantly slower than the classic matching group. Further HDDM analysis showed that the increased reaction time in the word association group was due to larger decision boundaries (boundary separation, a), meaning that compared to the classic matching group, participants in the word association group needed to accumulate more evidence before making decisions, which reflects a meta-strategic change.
Overall, this study found that using word association learning methods can establish associations between neutral stimuli and self-concept, and exhibit self-prioritization effects. However, this learning method did not make participants have more obvious processing advantages for self-related stimuli (no difference in drift rate v was found), but only made participants' decision style more cautious (significant increase in decision boundary a); which learning method to choose for studying self-prioritization effects still requires more exploration.
Report Four
Mapping the Landscape of AI-empowered Psychology: A topic modeling-based bibliometric analysis
Presenter: Songlin Jia
Content:
With the rapid development of artificial intelligence technology, AI plays an increasingly prominent role in psychological research. AI technology not only provides new methods for cognitive modeling and psychological measurement, but also brings profound changes in fields such as clinical intervention and digital health. However, there is currently a lack of systematic review and analysis of the overall landscape of "AI-empowered psychology" research. This study aims to fill this gap by providing reference and insights for academic communication and future research directions in this field through large-scale bibliometric and topic modeling.
This study selected 10,079 papers on "AI-empowered psychology" from the Web of Science database between 2000-2024, using the all-mpnet-base-v2 model to convert abstracts into 768-dimensional semantic vectors, supplemented by UMAP dimensionality reduction, HDBSCAN clustering, and c-TF-IDF keyword extraction, identifying 27 specific research topics and categorizing them into seven major directions: cognitive modeling, AI-driven psychological measurement, computational psychiatry, brain-computer interface research and applications, AI-assisted learning science, human-computer interaction, and digital psychological health intervention.
Temporal evolution analysis showed that this field experienced three stages: 2000-2015 was dominated by artificial neural networks and cognitive models; 2015-2020 saw the rise of deep learning technology in clinical detection and prediction directions; 2021-2024 witnessed explosive growth in generative AI and digital psychological health. Among these, topics such as "human cognitive science," "AI-assisted depression treatment," and "machine learning-based personality assessment" were most active.
Bibliometric results showed that "Frontiers in Psychology," "Frontiers in Psychiatry," and "Frontiers in Human Neuroscience" were high-output journals, with the United States, mainland China, and the United Kingdom leading in research output. The international collaboration network presented multiple clustering patterns of the US-Canada-Israel, European powers, China and Asian countries, and Latin America-Europe.
Although AI technology has made significant progress in psychological measurement, clinical intervention, and human-computer interaction, it still faces challenges such as insufficient exploration of causal mechanisms, poor interpretability of black-box models, lack of unified standards for data preprocessing and modeling processes, and insufficient data diversity between regions. Future research should focus on combining causal inference with explainable AI, promoting standardization of analytical processes, and strengthening international cooperation and data sharing globally, especially in underdeveloped regions, to promote more equitable and reliable AI-empowered psychology practice.
