AI aids with psychology in internet era
The brain-machine interactive control system developed by the Sichuan Institute for Brain Science and Brain-Inspired Intelligence on August 5, 2021 Photo: Lyu Guoying/RED STAR NEWS
Human psychology is changing as society has entered the information age and artificial intelligence flourishes. Grasping this trend in time and understanding the new law of human behavior is an expanding research topic for psychological researchers. Among them, the use of AI technology to transform the research methods and means of psychology and deepen the research content is one of the directions guiding current psychological researchers.
Mutual promotion
The progress of artificial intelligence has further facilitated the development of psychology and brain science, said Luo Fang, a professor from the Faculty of Psychology at Beijing Normal University. AI technology provides brain science research with a simulation means, system, and platform, and allows researchers to more systematically observe brain activities, promising a more comprehensive understanding of human sensations and perceptions, attention, memory, executive function, and feelings and emotion. Meanwhile, deep learning technology in the artificial intelligence field has remarkably elevated the processing efficiency of brain data. With highly efficient analysis and modeling of massive and complex brain data, researchers can more precisely identify psychological problems related to feelings or attention, promoting the large-scale application of brain science research results in practical tasks.
The multi-modal data collection and modeling methods assisted by AI technology have greatly enriched the research methods of psychology, Luo continued. Given that the psychological process is often accompanied by a series of physiological and behavioral responses, researchers conduct studies through wearable equipment to gather data on EEG, eye movement, and fine movement. Artificial intelligence is the core supporting technology of wearable devices. Virtual reality, augmented reality, and mixed reality technology help create a task environment for simulation and obtain more real information, which is crucial for recognizing implicit psychological motivation systems. At the same time, the multi-modal modeling method in the AI area makes integrative data analysis possible. Studies show that models with multi-modal data tend to achieve optimal prediction results. At present, studies based on multimodal data have seen applications in such psychological fields as psychological health, consumer psychology, cognitive effect, talent selection, and organizational management.
AI and psychological research in China are becoming ever more deeply integrated, forming a trend of mutual promotion, said Peng Yujia, a research fellow from the School of Psychological and Cognitive Sciences at Peking University. Psychologists can use AI technology for research. For instance, in the field of cognitive neuroscience, the multi-voxel pattern analysis (MVPA) introduces machine learning into psychology to conduct pattern recognition on human brain activities. In terms of application, MVPA helps identify patterns of brain activities and conduct decoded neurofeedback (DecNef), so as to achieve clinical interventions in the field of mental disorders, such as the regulation of addiction, anxiety disorders, and phobias.
In turn, the theories and findings of psychology are also driving the development of artificial intelligence, Peng added. Many algorithms of machine learning derive from psychological theories. For example, the reinforcement learning theory was inspired by the behaviorist theory in psychology, which discusses how organisms, under rewards or punishments given by the environment, gradually form expectations of stimulus and generate habitual behaviors to get maximum benefits. In addition, human learning, reasoning, and other advanced cognitive functions constantly derive inspirations from psychology and cognitive science, to develop more intelligent artificial algorithms.
Further integration
The development of artificial intelligence will widely expand the research scope of psychology, said Zhu Tingshao, a research fellow from the Institute of Psychology at the Chinese Academy of Sciences. On the basis of data collected and analyzed by AI techniques, researchers can finish previously hard to conduct studies, such as seeking correlations between various factors and expanding the temporal and spatial scope of psychological research.
For a long time, psychological research has involved a variety of factors, such as individual innate traits, family factors, and education experience, making it difficult to separate the interaction among these factors. Traditional psychological research often relies on precisely designed experimental paradigms, namely, controlling all other possible factors to examine the influence of a specific factor, which makes it difficult to fully reveal the multi-factor mechanism behind the research subject.
In the AI era, people’s neural activity in natural situations, including physiological responses, movement, facial expressions, and other multi-modal data, can be collected in a large scale and with fine granularity, which helps analyze the complex rules behind high-dimensional data and reveal possible statistical laws. In other words, traditional psychological research tends to be driven by the top-down theory, which first entails a good theoretical hypothesis and then verifies the hypothesis through data acquisition. Unlike traditional psychological research methods, after artificial intelligence obtains data, researchers can be driven by data and find the correlations among variables in a bottom-up way, thus discovering new theories and directions that were obscured in the past.
Peng noted that the interdisciplinary integration of AI and psychology is expected to show mainly in four fields: big data medical treatment, human-computer interaction, brain-computer interface, and artificial general intelligence. First, based on multimodal data, such as blood, saliva, and other biological samples, machine learning can be used to extract high-dimensional data to serve medical diagnosis and treatment. In terms of human-computer interaction, through combining human-factors engineering, cognitive science, and artificial intelligence, breakthroughs can be made in context awareness, eye movement tracking, gesture recognition, 3D input, speech recognition, and facial expression recognition, bringing forth the ultimate user experience and personalized service.
On a medical level, brain signals can be used to drive artificial limbs and create certain mobility for paralyzed patients, Peng continued. As for artificial general intelligence, AI development currently remains at the stage of artificial narrow intelligence. In the future, through fusing psychology and artificial intelligence, we can further the artificial general intelligence, endow computers with real intelligence to simulate advanced human cognitive abilities, and enable computers with learning and thinking abilities, thereby recognizing feelings, understanding human emotions, and even achieving dialogue and empathy with humans.
Luo suggested psychologists should represent an interdisciplinary talent pool with compound knowledge of natural language processing, computer vision, mechanical design and automation, psychology, biology, and other disciplines. Researchers need to rebuild a brand-new psychological science system as they strive to deeply excavate the rules behind human behaviors. Developmental psychology, personality and social psychology, organization and management psychology, cognitive ergonomics, and other branches will undergo substantial changes in their research content.
Edited by YANG LANLAN