ISSN 0439-755X
CN 11-1911/B
25 March 2024, Volume 56 Issue 3 Previous Issue    Next Issue
For Selected: Toggle Thumbnails
Reports of Empirical Studies
The effect of task relevance on serial dependence in numerosity
LIU Yujie, LIU Chenmiao, ZHOU Liqin, ZHOU Ke
2024, 56 (3):  255-267.  doi: 10.3724/SP.J.1041.2024.00255
Abstract ( 1827 )   PDF (502KB) ( 2510 )   Peer Review Comments
Serial dependence refers to the phenomenon where current perception is influenced not only by the current stimulus input but also by preceding events in recent history. This effect plays a crucial role in the establishment of relatively stable perceptions in dynamically changing environments. Previous studies have shown that the extent and direction of serial dependence are related to the task relevance of stimulus features. It is still unclear, though, if task relevance in linearly distributed features affects this impact, given that the majority of these researches have mostly focused on experiments using circularly distributed features. The current study investigated the impact of task relevance of linearly distributed features on serial dependence by using estimation tasks with dot arrays as stimulus materials, which were varying orthogonally in two dimensions: number/area (Experiment 1) or number/size (Experiment 2).
The study employed a 7 (number of dots) × 7 (dot array area in Exp 1/average dot size in Exp 2) × 2 (task relevance: relevant vs. irrelevant feature) block design. In the number estimation task, participants were instructed to focus on the number of dots, thus prioritizing the number as a relevant feature, while deeming the field area irrelevant. Conversely, the field area estimation task directed attention to the field area of dot array, making it the relevant feature and relegating the number to irrelevance. Experiment 2 followed the same experimental paradigm as Experiment 1, with the key difference being that it replaced the field area estimation task with an average item size estimation task. Each participant underwent all experimental conditions, with the order of the two tasks balanced across them. Initially, a fixation cross was presented for 1350-1450 ms, followed by a dot array image shown at the center of the screen for 250 ms. Task instruction then appeared at the top of the screen, accompanied by an axis beneath. Participants were instructed to accurately estimate the number of dots in the array by selecting a point on the number line through mouse click. The mouse click would trigger the appearance of a white marker, indicating the selected position, and its corresponding numerical value was exhibited underneath. Participants then affirm their estimation by pressing the “Enter” key. A response window of 15 seconds was provided; failure to respond within this period led to a 'no response' recording (marked as “N/A”) for that trial, and the program automatically proceeded to the next trial.
Our findings revealed that the effect of a feature from previous trial on current perception consistently counteracted the influence of the same feature in the current trial, regardless of the feature's task relevance. From a serial dependence perspective, the effects of previous task-relevant features were always repulsive; however, whether the previous task-irrelevant features showed attractive or repulsive serial dependence effects, was depended on the specific feature. This highlights the dual influence of task relevance and feature characteristics on the serial dependence effect of linearly distributed features. Notably, the persistence of the serial dependence of the irrelevant features implies that serial dependence can also arise at the object level.
References | Related Articles | Metrics
Influence of group information on facial expression recognition
WANG Weihan, CAO Feizhen, YU Linwei, ZENG Ke, YANG Xinchao, XU Qiang
2024, 56 (3):  268-280.  doi: 10.3724/SP.J.1041.2024.00268
Abstract ( 2034 )   PDF (727KB) ( 2800 )   Peer Review Comments
Emotions surface during interaction between individuals. Thus, an accurate recognition of facial expressions is essential in the realm of social interactions. In recent years, numerous studies have revealed that individuals not only depend on facial configuration information for identifying facial expressions but also place considerable emphasis on contextual information extracted from external cues beyond the face. People’s behavior frequently unfolds within intricate social group dynamics, wherein individuals often perceive and interpret the facial expressions of their fellow group members during interaction. However, the impact of group information on facial expression recognition, being an essential social contextual factor, remains somewhat unclear. Hence, three experiments were conducted to investigate the influence exerted by group information on the recognition of facial expressions.
The stimuli used in the study were happy, fearful, and neutral face images selected from the NimStim set, including 15 pictures (seven females) of each of the aforementioned emotions. Group information was manipulated following the presentation of a fixation cross through perceptual cues. Subsequently, during the facial expression recognition phase, participants were instructed to recognize the facial expressions exhibited by target individuals. In the first experiment, participants were instructed to rate the intensity of target facial expressions on a six- emotion scale, and the surrounding facial expressions were always congruent with the target facial expressions. A total of 29 college students (16 females, mean age 20.00 ± 1.80 years) were recruited to participate in this experiment. In Experiments 2 and 3, we manipulated the emotional congruency between the surrounding faces and the target faces during the facial expression recognition phase. Additionally, we controlled for variations in physical characteristics across different experimental conditions. The task requirement of Experiment 2 was the same as that of Experiment 1. However, in Experiment 3, participants were instructed to judge the target facial expressions by pressing corresponding keys on the keyboard as quickly and accurately as possible. A total of 26 college students (14 females, mean age 21.15 ± 1.99 years) participated in Experiment 2, and 32 college students (15 females, mean age 21.20 ± 1.60 years) participated in Experiment 3.
Results revealed the following: (1) Compared with emotion-incongruent conditions, emotional congruency between target faces and surrounding faces resulted in shorter RTs and higher accuracy. (2) Group information regulated the influence of surrounding facial expressions on target facial expression recognition. Specifically, under group conditions, participants tended to recognize target facial expressions according to the emotional state of the surrounding faces. When the target facial expressions in line with the expectations established by the participants that group members have congruent emotional state, the recognition of target facial expressions was faster and more accurate than incongruent conditions. However, under nongroup conditions, participants recognized target facial expressions without reference to the emotional states of the surrounding faces. (3) Participants exhibited a faster and more accurate recognition of happy faces, indicating the recognition advantage effect for happy facial expressions.
Results revealed that group information influenced facial expression recognition, individuals recognized facial expressions based on the social relationship between the interactions, and understanding social interaction plays an important role in the process of emotion perception.
References | Related Articles | Metrics
Different roles of initial and final character positional probabilities on incidental word learning during Chinese reading
LIANG Feifei, FENG Linlin, LIU Ying, LI Xin, BAI Xuejun
2024, 56 (3):  281-294.  doi: 10.3724/SP.J.1041.2024.00281
Abstract ( 775 )   PDF (2098KB) ( 1030 )   Peer Review Comments
In natural unspaced Chinese reading, there are no salient visual word segmentation cues (like word spaces) to demark where words begin or end, yet Chinese skilled readers process a comparable amount of text content as efficiently as English readers, processing roughly 400 characters (equal to 260 words) per minute. This raises the question of how Chinese readers engage in such word segmentation processing efficiently and effectively. Liang et al (2015, 2017) have shown that the positional probability information associated with a character, might offer a cue to the likely positions of word boundaries during Chinese incidental word learning. Given that they simultaneously manipulated the positional probabilities of both word initial and word final characters to make their manipulations maximally effective, it is unclear whether the initial, the final, or both constituent characters’ positional probabilities contribute to the word segmentation and word identification effects during incidental word learning in Chinese reading. For this reason, in the present study, two parallel experiments were designed to directly investigate whether word initial, or word ending characters are more or less important for word segmentation word learning in Chinese reading.
Two-character pseudowords were constructed as novel words. Each novel word was embedded into six high-constraint contexts for readers to establish novel lexical representation. In Experiment 1, we examined how word’s initial character positional probability influenced word segmentation and word identification during Chinese word learning. The initial character’s positional probability of target words was manipulated as being either high or low, and the final character was kept identical across the two conditions. In Experiment 2, an analogous manipulation was made for the final character of the target word to check whether the final character positional probability of two-character words can be used as word segmentation cue. We also included “Exposure” as a continuous variable into the model to further examine how the process of initial and final character positional probabilities changed with exposure.
In both experiments, the participants spent shorter reading times and made fewer fixations on targets that comprised initial and final characters with high relative to low positional probabilities, suggesting that the positional probability of both the initial and final character of a word influences segmentation commitments in novel word learning in Chinese reading. Furthermore, both the effect of initial and final character positional probabilities of novel words decreased with exposure, showing the typical familiarity effect. To be somewhat different, the familiarity effect associated with the initial character had a slower time course relative to final character. This finding suggests that the role of word’s initial character positional probability is of more importance than that of final character’s, supporting the concurrent standpoint that word beginning constituents might be more influential than word final constituents during two-character word identification in Chinese reading.
Based on the findings above, the time course of the process of initial and final character positional probabilities of novel words is argued and summarized as follows. During the early stage of word learning, both the statistical properties of word’s initial and final character positional probabilities are processed as segmentation cue. As lexical familiarity increases, the extent to such segmentation roles decreases, which initially begins with final character, and then occurs with initial character. Later, both the roles of initial and final character positional probabilities disappear with the establishment of a more-integral representation of novel words.
References | Related Articles | Metrics
The transition of latent classes of children’s learning engagement in primary school against the background of the “double reduction” policy
YANG Jingyuan, YU Xiao, ZHANG Jingyi, LU Lifei, YANG Zhihui
2024, 56 (3):  295-310.  doi: 10.3724/SP.J.1041.2024.00295
Abstract ( 2154 )   PDF (1153KB) ( 3739 )   Peer Review Comments
Learning engagement, an important indicator of the learning process, has garnered extensive attention. Developmental contextualism and the integrative model of engagement posit that the interaction between individuals and environmental factors results in heterogeneous learning engagement development among individuals. Previous studies have demonstrated learning engagement heterogeneity among primary school students. However, in the context of the “double reduction” policy, the dynamic development of children’s learning engagement remains unclear. Moreover, positive parenting style, teacher-student relationships, and peer relationships, as important environmental factors, may predict children’s learning engagement transitions. Thus, this study adopts a people-centered research method to address these issues from a longitudinal perspective.
This study recruited participants from three ordinary public primary schools in Shandong Province, China. Participants at T1 (June 2021, before the implementation of the “double reduction” policy) were 378 children (164 boys; mean age: 9.97 ± 0.91 years old). Participants at T2 (December 2021, six months after the implementation of the policy) were 357 primary school students (155 boys; mean age: 10.50 ± 0.94 years old). Participants at T3 (June 2022, a year after the implementation of the policy) were 347 primary school students (147 boys; mean age: 10.97 ± 0.91 years old). Students completed the Children’s Learning Engagement Scale (at T1, T2, and T3), Short-form Egna Minnen av Barndoms Uppfostran (at T1 and T2), Student Teacher Relationship Scale (at T1 and T2) and Children’s Peer Relationship Scales (at T1 and T2) during the three measurements. Latent profile analysis and latent transition analysis were employed in this study to explore children’s potential learning engagement subtypes and examine transitions between different subtypes across the three waves. Multiple logistic regressions were also used to investigate the impact of various environmental factors (i.e., positive parenting style, student-teacher relationships, and peer relationships) on the latent transitions of different learning engagement subtypes.
All data were analyzed by SPSS 26.0 and Mplus 8.0. The results revealed four distinct subgroups of learning engagement among primary school students: the “Low Engaged”, “Moderately Engaged”, “High Absorption with Vigorous Disengagement”, and “Highly Engaged” groups. In addition, due to the “double reduction” policy, students in the “Moderately Engaged” and “Highly Engaged” groups displayed relative stability, while those in the “Highly Disengaged” group tended to transition toward the “Moderately Engaged” group. Regarding the “High Absorption with Vigorous Disengagement” group, the findings indicated a higher likelihood of transitioning to the “Moderately Engaged” group from T1 to T2; however, from T2 to T3, these students were more likely to remain in their original subgroup. Moreover, the study identified the varying roles of different environmental factors in children’s learning engagement subgroups. Specifically, under the “double reduction” policy, positive parenting style and teacher-student relationships exhibited robust effects on children’s learning engagement transitions. The predictive effects of teacher-student relationships varied across different learning engagement subtypes among primary school students. Additionally, the study found that peer relationships had a positive influence on the transition of children within the “Moderately Engaged” group following the implementation of the “double reduction” policy.
This study provides the first evidence of heterogeneity and dynamic changes in learning engagement among Chinese primary school students, which indicates that following the implementation of the “double reduction” policy, family-school-collaborative education has made initial progress. These findings not only enhance our understanding of the dynamic development of learning engagement among primary school students but also provide empirical evidence regarding the effectiveness of the “double reduction” policy implementation.
References | Related Articles | Metrics
The influence mechanism of team reflexivity training on team ambidexterity development
LI Cirong, LI Chunxuan
2024, 56 (3):  311-325.  doi: 10.3724/SP.J.1041.2024.00311
Abstract ( 665 )   PDF (531KB) ( 925 )   Peer Review Comments
Performing explorative and exploitative behaviors simultaneously is a key means for teams to quickly adapt to environmental and task changes. How to improve team ambidexterity is an important topic of concern in theoretical and management practice. Although scholars have conducted preliminary analyses on the antecedents of team ambidexterity, conclusions at the between-team level can only be used to identify ambidextrous teams but not to answer the question of how to cultivate team ambidexterity. Based on the "differentiation-integration" framework, this study argues that the realization of team ambidexterity requires team members to obtain and process different types of information. Open collective reflexivity activities provide a rich source of information for teams. However, reflexivity activities are highly complex and resource-consuming, and companies need to use reflexivity interventions (e.g., team reflexivity training) to guide teams to engage in reflexivity activities on their own initiative. The team information processing model states that teams enhance team effectiveness and adaptability through two paths of information sharing and integration. Based on the above deduction, this study suggests that meta-knowledge sharing and perspective picking are the key cognitive mechanisms through which team reflexivity training positively influences team ambidexterity development.
We test our theoretical propositions in an experimental study and a quasiexperimental study. In Study 1, we conducted a course experiment with students and seven wave measurement waves over 4 months, resulting in 630 observations from 90 teams. We invited 360 undergraduates majoring in economics or management from a university in southern China. We randomly and equally assigned 360 college students into 90 teams and then divided the teams into experimental and control groups. We gave the experimental group team reflexivity training and assisted them with reflexivity activities in subsequent sessions, while the control group was given team building training to avoid a placebo effect. We measured team ambidexterity at all seven measurement waves and team reflexivity after and before intervention using established scales and items. Conditional latent growth modeling was applied to test the slope difference of the team ambidexterity trend between the experimental and control groups. To investigate the theoretical hypotheses in Study 2, we further conducted a quasiexperimental study, which took one year and involved three measurement waves; the study resulted in 222 observations from 74 teams. We invited a total of 656 employees from R&D teams in 26 companies engaged in high-tech industries related to information technology, precision instruments, and biopharmaceuticals in a southern Chinese province in this study. Seventy-four R&D teams were randomly and equally divided into experimental and control groups. We gave the experimental group team reflexivity training in the first month and required them to conduct a formal reflexivity activity at a regular time each week (or two weeks) thereafter. We measured team ambidexterity in the first and second measurement waves and meta-knowledge sharing and perspective taking in the second and third measurement waves. To account for the mediating effect of meta-knowledge sharing and perspective taking between team reflexivity training and team ambidexterity development, latent change score modeling was applied.
The statistical analyses supported our hypotheses. The results of Study 1 showed that teams that did not participate in team reflexivity training showed a nonsignificant downward trend in team duality; in contrast, teams that participated in reflexivity training showed a significant upward trend in team ambidexterity. Based on this, for Study 2, we further analyzed the mediating role of meta-knowledge sharing and perspective taking and improved the external validity of the Study 1 finding with a quasiexperimental research design. It was found that teams' meta-knowledge sharing and perspective taking improved after participating in reflexivity training, which led to an increase in team ambidexterity.
By increasing our understanding of how to improve team ambidexterity and the key information cognitive mechanisms of it, our study contributes to the literature in three ways. First, this study provides rich empirical evidence for ambidexterity research by confirming the role of team reflexivity training in sustainably enhancing team ambidexterity. The findings support the consistent view of team reflexivity training research that it is effective in enhancing team adaptability as a management intervention. At the same time, this study bridges the gap regarding how to help teams build the capacity to perform ambidextrous behaviors, responding to the call for research on "exploring how to guide paradoxical coping into a beneficial developmental process”. Second, based on the "differentiation-integration" framework and the team information processing model, this study infers and confirms that team meta-knowledge sharing and perspective taking are important cognitive processes that influence the development of team ambidexterity through team reflexivity training. The findings are not only consistent with the view that "information exchange and adoption among team members is necessary for team ambidexterity" but also expand ambidexterity research from a cognitive perspective. Meanwhile, the findings enrich the narrow research on the team information processing model in enhancing team adaptability and flexibility and reaffirm the fundamental role of efficient information processing in determining team effectiveness. Third, this study introduces the element of time in the empirical study of team ambidexterity for the first time, deepening the understanding of the nature of ambidexterity dynamics. The results found that team ambidexterity was unable to show positive trends over time, which is consistent with the expected negative self-reinforcing effect. This suggests that our team members are not willing to consistently adopt complex behavioral patterns such as ambidexterity for work but instead prefer specific activities due to behavioral inertia.
Our findings also offer empirical evidence that companies need to provide reflexivity courses for their teams to help members acquire and develop good work rethinking habits. At the same time, supervisors can activate and optimize the team information processing process by developing corresponding systems (e.g., a set time and frequency), providing necessary support (e.g., venue and accompanying guidance), and building a good team climate to continuously improve team ambidexterity.
References | Related Articles | Metrics
Positive effects of leader perceived overqualification on team creativity
WANG Yating, CHEN Zhijun, LI Rui, ZHOU Mingjian
2024, 56 (3):  326-338.  doi: 10.3724/SP.J.1041.2024.00326
Abstract ( 1030 )   PDF (482KB) ( 1393 )   Peer Review Comments
With the spread of higher education and the global economic downturn, the overqualification phenomenon is increasingly becoming common and popular. Prior research has mainly focused on the negative effects of perceived overqualification. However, some scholars are currently urging a deeper exploration of the positive implications of perceived overqualification. Although most studies have focused on employee perceived overqualification and its impact on work attitudes, behaviours and personal well-being, information is limited on the phenomenon of leader perceived overqualification and its effects. For organisations, understanding the effects of leader perceived overqualification on teams is crucial for effective talent management. Therefore, our study draws on self-regulation theory and the process-based theory of team creative synthesis to propose and test a mediated moderation model that explores when and why leader perceived overqualification influences team creativity.
To test the proposed hypotheses, we conducted a multi-wave and multi-source field study. We collected data from five hospitals in North China, and the final sample consists of 106 head nurses and their 847 nurses. At time 1, head nurses were asked to report their demographics and perceived overqualification. At time 2 (two months later), head nurses were asked to report their perceptions of team capability and psychological entitlement. Additionally, nurses were asked to evaluate leader encouragement of creativity and abusive supervision. At time 3 (two months later), nurses rated their team creative process engagement. Lastly, head nurses were asked to assess team creativity.
Results provided support for our theoretical model and revealed the following findings. (1) The interaction between leader perceived overqualification and leader perceived capability significantly predicted leader encouragement of creativity, such that the positive relationship between leader perceived overqualification and leader encouragement of creativity was stronger when team capability was higher rather than lower. (2) Team creative process engagement mediated the relationship between leader encouragement of creativity and team creativity. (3) Leader encouragement of creativity and team creative process engagement mediated the interactive effect of leader perceived overqualification and team capability on team creativity, such that the indirect effect was stronger when team capability was higher.
The preceding results provide several important theoretical contributions. Firstly, this research enriches the outcomes of perceived overqualification by investigating the positive impact of leader perceived overqualification on team creativity. Secondly, this research identifies leader perceived team capability as an important boundary condition for the positive effects of leader perceived overqualification. Thirdly, by exploring the chain mediating roles of leader encouragement of creativity and team creative process engagement, this study opens the ‘black box’ of the effect of leader perceived overqualification on team creativity and expands the understanding of the positive implications of perceived overqualification. Lastly, by examining the relationship between leader perceived overqualification and team creativity, this study enriches the antecedents of team creativity from the leader characteristic perspective.
References | Related Articles | Metrics
Cognitive diagnostic assessment based on signal detection theory: Modeling and application
GUO Lei, QIN Haijiang
2024, 56 (3):  339-351.  doi: 10.3724/SP.J.1041.2024.00339
Abstract ( 664 )   PDF (1965KB) ( 862 )   Peer Review Comments
Cognitive diagnostic assessment (CDA) is aimed at diagnose which skills or attributes examinees have or do not have as the name expressed. This technique provides more useful feedback to examinees than a simple overall score got from classical test theory or item response theory. In CDA, multiple-choice (MC) is one of popular item types, which have the superiority on high test reliability, being easy to review, and scoring quickly and objectively. Traditionally, several cognitive diagnostic models (CDMs) have been developed to analyze the MC data by including the potential diagnostic information contained in the distractors.
However, the response to MC items can be viewed as the process of extracting signals (correct options) from noises (distractors). Examinees are supposed to have perceptions of the plausibility of each options, and they make the decision based on the most plausible option. Meanwhile, there are two different states when examinee response to items: knows or does not know each item. Thus, the signal detection theory can be integrated into CDM to deal with MC data in CDA. The cognitive diagnostic model based on signal detection theory (SDT-CDM) is proposed in this paper and has several advantages over traditional CDMs. Firstly, it does not require the coding of q-vector for each option. Secondly, it provides discrimination and difficulty parameters that traditional CDMs cannot provide. Thirdly, it can directly express the relative differences between each options by plausibility parameters, providing a more comprehensive characterization of item quality.
The results of two simulation studies showed that (1) the marginal maximum likelihood estimation approach via Expectation Maximization (MMLE/EM) algorithm could effectively estimate the model parameters of the SDT-CDM. (2) the SDT-CDM had high classification accuracy and parameter estimation precision, and could provide option-level information for item quality diagnosis. (3) independent variables such as the number of attributes, item quality, and sample size affected the performance of the SDT-CDM, but the overall results were promising. (4) compared with the nominal response diagnostic model (NRDM), the SDT-CDM was more accurate in classifying examinees under all data conditions.
Further, an empirical study on the TIMSS 2011 mathematics assessment were conducted using both the SDT-CDM and the NRDM to inspect the ecological validity for the new model. The results showed that the SDT-CDM had better fitting and a smaller number of model parameters than the NRDM. The difficulty parameters of the SDT-CDM were significantly correlated with those of the two- (three-) parameter logical models. And the same was true of the discrimination parameters for the SDT-CDM. However, the correlation between the discrimination parameters of the NRDM and those of the two- (three-) parameter logical models was low and not significant. Besides, the classification accuracy and classification consistency of the SDT-CDM were higher than those of the NRDM. All the results indicated that the SDT-CDM was worth promoting.
References | Related Articles | Metrics
Modeling the dependence between response and response time: A bifactor model approach
GUO Xiaojun, BAI Xiaoyun, LUO Zhaosheng
2024, 56 (3):  352-362.  doi: 10.3724/SP.J.1041.2024.00352
Abstract ( 482 )   PDF (945KB) ( 670 )   Peer Review Comments
In the realms of psychological and educational testing, the computerization of tests is becoming more prevalent, facilitating the acquisition of process data from test-takers. In the domain of process data, response time and response represent the two most commonly utilized variables. Responses provide critical insights into the answers provided by test-takers, while response time, as an essential source of information, is increasingly garnering attention from researchers. The proposal of hierarchical model (HM) has provided a fundamental modeling framework for the joint analysis of response time and response, and it is becoming increasingly popular in current research practices. However, relying solely on the association between item and subject parameters is insufficient to adequately explain the correlation between response time and response. Consequently, researchers have proposed various enhanced models to address these limitations, although some challenges persist.
The bifactor model explains common variance through a general or global factor, while a local or specific factor explains the common variance of additional partial items. In psychological and educational testing, it is possible to capture not only the test-takers’ response times on test items but also their responses. From the perspective of the bifactor model, response times and responses to test items measure different local factors. Specifically, a test's response time measures the test-taker's speed trait, while the response to the test measures their ability trait. Test-takers are also influenced by a combination of time and accuracy when responding to the test, known as general latent traits or global factors, or speed-accuracy trade-off ability. This test structure aligns well with the bifactor model and provides a new perspective on analyzing the relationship between test-taking response time and response dependence. Based on this, this study proposes a bifactor hierarchical model (Bi-HM) to explore the dependency between response time and response.
In the simulation study, it was found that the MPLUS program utilizing MLR (Maximum Likelihood Robust), could accurately estimate the parameters of the Bi-HM and was not influenced by the level of item parameter correlation. Conversely, when disregarding the relationship between response time and response in the HM, notable bias in the parameter estimates occured. In the empirical data analysis, the Bi-HM demonstrated significantly superior model fit indices compared to the HM. Moreover, the Bi-HM effectively captured the dependency between response and response time at both the participant and item levels. This dependency is closely associated with item difficulty and time intensity factors.
Based on the findings mentioned above, it is evident that the Bi-HM, which adopts a bifactor model perspective, excels in parameter estimation and data fitting, demonstrating excellent scalability.
References | Related Articles | Metrics
Theory and History of Psychology
New research paradigms and agenda of human factors science in the intelligence era
XU Wei, GAO Zaifeng, GE Liezhong
2024, 56 (3):  363-382.  doi: 10.3724/SP.J.1041.2024.00363
Abstract ( 909 )   PDF (809KB) ( 1340 )   Peer Review Comments
This paper first proposes the innovative concept of “human factors science” to characterize engineering psychology, human factors engineering, ergonomics, human-computer interaction, and other similar fields. Although the perspectives in these fields differ, they share a common goal: optimizing the human-machine relationship by applying a “human-centered design” approach. AI technology has brought in new characteristics, and our recent research reveals that the human-machine relationship presents a trans-era evolution from "human-machine interaction" to "human-AI teaming." These changes have raised questions and challenges for human factors science, compelling us to re-examine current research paradigms and agendas.
In this context, this paper reviews and discusses the implications of the following three conceptual models and frameworks that we recently proposed to enrich the research paradigms for human factors science. (1) human-AI joint cognitive systems: this model differs from the traditional human-computer interaction paradigm and regards an intelligent system as a cognitive agent with a certain level of cognitive capabilities. Thus, a human-AI system can be characterized as a joint cognitive system in which two cognitive agents (human and intelligent agents) work as teammates for collaboration. (2) human-AI joint cognitive ecosystems: an intelligent ecosystem with multiple human-AI systems can be represented as a human-AI joint cognitive ecosystem. The overall system performance of the intelligent ecosystem depends on optimal collaboration and design across the multiple human-AI systems. (3) intelligent sociotechnical systems (iSTS): human-AI systems are designed, developed, and deployed in an iSTS environment. From a macro perspective, iSTS focuses on the interdependency between the technical and social subsystems. The successful design, development, and deployment of a human-AI system within an iSTS environment depends on the synergistic optimization between the two subsystems.
This paper further enhances these frameworks from the research paradigm perspective. We propose three new research paradigms for human factors science in the intelligence ear: human-AI joint cognitive systems, human-AI joint cognitive ecosystems, and intelligent sociotechnical systems, enabling comprehensive human factors solutions for AI-based intelligent systems. Further analyses show that the three new research paradigms will benefit future research in human factors science. Furthermore, this paper looks forward to the future research agenda of human factors science from three aspects: “human-AI interaction”, “intelligent human-machine interface”, and “human-AI teaming”. We believe the proposed research paradigms and the future research agenda will mutually promote each other, further advancing human factors science in the intelligence era.
References | Related Articles | Metrics