The Social Contagion of Risk Perceptions of Emerging Technologies through Evolutionary Game in Networks

Abstract

Emerging technologies are leading the fourth industrial revolution, bringing profound changes to modern life. However, the contagion of public risk perceptions pertaining to these technologies may result in new social stability risks according to the social amplification of risk framework (SARF). As such, understanding the formation and contagion mechanisms behind public risk perceptions of emerging technologies is critical to promoting their increased receptivity. Based on an evolutionary game theoretical approach to complex networks, this paper constructs a social contagion model of the risk perception of emerging technologies and uses simulations to analyze the influence of factors such as individual cognition and social environment. Results indicate the following: (1) the rate of risk perception contagion gradually increases with the level of individual risk perception and trust in communicators; (2) the risk perception contagion rate gradually decreases as the level of risk governance participation increases; and (3) comparing different social networks, kinship networks have a greater effect than social media networks on risk perception contagion.

1. Introduction

The digital era brought forth by the fourth industrial revolution has allowed rapidly developing new technologies to penetrate all aspects of life. Recently, artificial intelligence (AI) technology, epitomized by chatGPT, has caused an uproar around the world. Emerging technologies can improve the quality of public services and the efficiency of public decision-making [1], but they carry significant ethical, moral, and economic risks. The social dissemination of risk-related information may amplify levels of individual risk perceptions, which can lead to concern, panic, and resistance to emerging technologies. Such social barriers may even affect social stability and socioeconomic development. Therefore, effective risk communication is critical for improved public awareness of science. Risk communication is a two-way interactive process that requires the continuous dissemination of factual, relevant information and the timely evaluation of public risk perception levels. As risk communicators must constantly adjust their strategies based on public feedback, exploring the contagion and formation of public risk perceptions under varying environments may provide broader, more comprehensive perspectives on how to govern emerging technologies.

The formation of public risk perceptions begins with the processing of risk information [2], which is received by the public through varying external networks, processed into subjective risk perceptions, and transmitted to other individuals. Therefore, an individual’s risk perceptions of emerging technologies are continuously reshaped through communication and exchange with others. However, current research on risk perception is dominated by psychometric methods, which focus on static elements of risk perceptions and are mostly applied to attitude analyses of high-risk events [3], such as natural disasters and emerging technologies. In these studies, the public simply adopts certain behaviors and reactions based on static risk perceptions [4,5]. In the current technological revolution led by AI, public concern over and awareness of the risks of new technologies have gradually increased, and individual cognition, expert opinions, and social networks have become important factors influencing public risk perception. At the same time, in the age of social and digital media, the media of risk information dissemination have become increasingly diverse and no longer limited to networks of relatives and friends. How the transformation of processed risk information into individual risk perceptions is connected to the formation and reconstruction of public risk perceptions in social networks must be further addressed.

This paper focuses on two scientific questions. First, how does risk perception diffuse in social media versus kinship networks? Second, what are the mechanisms and effects of various factors on risk perception contagion? We use theories of risk perception to analyze its contagion as it relates to emerging technologies, construct an evolutionary game model of risk perception contagion based on complex networks, and analyze the effects of different factors on the contagion behavior of public risk perception. The innovations of this paper are as follows: Firstly, relevance to emerging technologies. With the rapid development and widespread adoption of emerging technologies such as artificial intelligence and gene editing, these technologies are increasingly becoming a part of people’s lives and work environments. Unlike traditional technologies, emerging technologies present various ethical and risk issues, causing not only convenience but also some degree of anxiety. Therefore, researching the risk perception of emerging technologies and their dissemination among the public has practical significance and extends the scope of risk perception studies. Secondly, focus on risk perception dynamics. Previous research on risk perception has primarily focused on cognitive differences between experts and the public, as well as risk communication to improve public understanding of science. This study, however, employs theoretical analysis to explore how risk perception functions within real-world social systems, specifically focusing on the process of risk perception spread among public individuals. Thirdly, the application of evolutionary game theory on complex networks. This paper innovatively applies evolutionary game theory within complex networks to the field of social sciences. It uses different network structures to represent various social environments: small-world networks to depict relationships among family and friends, and scale-free networks to represent relationships between experts and the public. The amplification effect of risk perception is mapped onto the strategic interactions between individuals, simulating the amplification and dissemination of public risk perception regarding emerging technologies. This approach extends the research scope and applicable methods in this field.

2. Literature Review

2.1. Emerging Technologies and Their Risk Perception

Risk is traditionally defined as an uncertain negative outcome [6], so social scientists often consider the concept of uncertainty referring to the probability according to which the uncertain outcome may occur. A broad definition of risk must delineate the negative consequences of a risky event, the severity of loss, and its uncertainty [7]. Technological risk is defined as the potential physical, social, and economic consequences, i.e., the likelihood of loss, for citizens throughout the life cycle of technological development [8]. The characteristics of emerging technologies include radical novelty, relatively fast growth, coherence, prominent impact, and uncertainty and ambiguity [9], where the characteristics of high impact, uncertainty, and ambiguity lead to the inevitable perception of emerging technologies as posing potential risks to society. In risk society theory [10], technological risk has become the primary type of risk in modern society [11], leading to a deepening awareness of new technologies and their potential consequences [12].

The study of risk perception originated in psychology, where it referred to individual, subjective risk assessments [13]. In the social sciences, risk perception is more often applied to study people’s intuitive perceptions and attitudes toward risky events such as natural disasters and emerging technologies, and their subsequent adoption of certain behaviors and reactions [3,4,5]. Technological risk perception refers to the processing of potential physical signals and risk information associated with a technology into subjective judgments about its severity, success probability, and acceptability [14,15,16,17]. It follows that the formation of an individual’s perception of technological risk includes both subjective judgement formed directly through physical senses and signals and the social construction of risk information and consequences communicated across various modalities. Technological risk perception and awareness are therefore not equivalent to objective measures of technological risk. Rather, they constitute an adaptive social, cognitive, and emotional process of evaluating uncertainty in the development of innovative technologies [18]. Thus, for the same technology, subjective public risk perceptions are often different from expert judgments based on the analysis of objective measures [19,20]. Early scholars used a psychometric paradigm to study public attitudes toward technological risk perception [21], transforming individual risk perceptions collected through questionnaires into risk cognitive maps based on the dimensions of familiarity and apprehensiveness. The uncertainty and ambiguity surrounding emerging technologies often place their risk perception in the highest section of the risk cognitive map, as seen with technologies like DNA technology and nuclear energy [17]. However, the actual risk level of these technologies may not be as high. This means that for some emerging technologies, public risk perception can significantly exceed the risk calculations made by experts. The formation of public risk perception toward emerging technologies is largely influenced by individual subjective understanding and the collective effect of risk perception diffusion among other individuals in society. Although emerging technologies, natural disasters, and other emergencies are all characterized by risk, there are certain differences in their research paradigms. Risk perception studies of natural disasters are mostly used to analyze the effects of risk perception on preventative and protective behaviors [22], often investigating only the immediate time periods before and after a disaster event. In contrast, risk perception studies of technology tend to analyze public acceptance and the social risks associated with technological resistance. The window of interest is typically the development and application stages of a new technology, and public risk perception is often a result of long-term, repeated interactions of risk information exchange embedded in a social environment.

In studies of technology acceptance, its degree is often evaluated in terms of both perceived risk and perceived benefit [23]. An individual’s risk perception of technologies is the most fundamental method for evaluating potential risk and is therefore often used to judge the probability of negative consequences and concerns [24]. Public judgment of risk is usually expressed as an attitude of support or opposition to a particular technology [25], and thus public opposition depends primarily on a technology’s riskiness. Studies of public opposition toward technologies such as nuclear technology [26], nanotechnology [27], and gene technology [28,29] have shown that, once faced with a potentially grave negative outcome, individuals tend to associate these technologies with high risk and low benefit, creating conflict with experts’ perceptions of technological risk, which influences scientific and technological decisions [30].

2.2. Factors Influencing Risk Perception

Public risk perception is an important explanatory variable for public behavior that can determine attitudes toward risky technologies [31]. However, risk perception, as a socially constructed form of awareness, is also influenced by other factors. In their studies of the risk perception of high-risk technologies, scholars have found that individual characteristics, the properties of the technology itself, and the social environment can have differing impacts on public risk perception. Furthermore, it has been emphasized that neither technical nor quantitative approaches that start with risk characteristics of the technology alone are sufficient to reflect the complex patterns of individual risk perceptions [32,33]. The social amplification of risk framework (SARF), based on risk perception theory, depicts a structure in which risk is amplified or weakened by social factors through the transmission of information, where these social factors are reflected at the individual and social levels [34,35,36]. After receiving risk signals, individuals make judgments based on their knowledge and experience, amplifying or attenuating risk based on their psychological distance from it [37,38]. This “unrealistic” risk perception interacts with a series of psychological, social, institutional, or cultural processes to further deviate from objective risk perception at the social response level. Individual risk perceptions are internalized over time through sociocultural learning and mediated through media coverage, peer influence, and other communication processes [39,40,41], with differences in perceptions influenced by both individual and external factors, such as social networks and the media. In addition, research in risk communication has emphasized that the process of forming and evolving public risk perceptions is influenced by the information’s sources, codes, communication agents, and communication channels [42]. Others have emphasized how trust in information sources and channels shapes individuals’ perceptions and information processing, which results in dynamic, differing transformations of risk information into risk perception [43].

2.3. Social Contagion of Risk Perception

Traditional psychometric approaches focus on intrinsic, individualistic factors in the formation of individual risk perceptions. Although social amplification of risk studies has highlighted the role of information in the construction of individual perceptions [44], information is not homogeneous, and its ascribed meaning is an amalgamated construct based on levels of understanding, subjective perception, and social influence. In other words, communicated risk information has already been subjectively processed by individuals. Risk perception, as a mechanism of individual cognition, has been extensively studied in the process of collecting, processing, and forming cognitive thought. However, these theories cannot explain how risk perceptions vary across or within groups, nor how they interact among individuals. The social contagion theory of risk perception [45] emphasizes that individually relevant social ties, and their resulting networks and self-organizing systems, influence individual perceptions, and it finds that the stronger the relationships between actors in a network, the more they tend to share similar risk perceptions, attitudes, and behaviors. In a study of environmentally controversial technological risks, a questionnaire measured perceived risks and benefits, and groups were clustered by attitude classifications based on their levels of risk perception [46]. It was found that, through the influence of connected individuals in the network, respondents tended to cluster into “groups or communities of like-minded individuals” [47].

2.4. Developing the Analytic Framework

Research on emerging technologies, factors of risk perception, and the contagion of risk perception provides the theoretical foundation for this paper. The risk perception of emerging technologies by individuals in society is influenced not only by their own characteristics but also by the impact of family, peers, or experts. Specifically, each individual has the right to choose whether to share their perception of technological risks and will decide whether to believe in the potential risks based on their understanding. This results in a social amplification effect of risks, which can be conceptualized as a strategic behavior reflecting dynamic evolutionary processes in populations, exhibiting stable or abrupt changes with limited rationality and some inertia. Additionally, social individuals and their connections form a complex network with different structural characteristics, where nodes represent individuals with various traits and connections represent information dissemination pathways. Therefore, the diffusion of risk perception regarding emerging technologies among the public occurs through strategic interactions and communication within the social system (i.e., the complex network), influenced by the network’s structure. This paper employs evolutionary game theory within complex networks to explore the diffusion process of risk perception in social systems, aiming to uncover the “black box” of risk perception diffusion. This section provides a detailed review of the existing literature and explains the diffusion of risk perception addressed in this paper. (Figure 1).

3. Modeling the Contagion of Risk Perception of Emerging Technologies

3.1. Basic Concepts and Hypotheses

We propose five factors, originating from both internal, individual factors and external social environments, that influence the contagion of risk perception of emerging technologies. Internal factors focus on the level of individual risk perception, psychological distance between individuals and risks, and degree of individuals’ participation in risk governance. External factors focus on the social environment, i.e., the social network structure, in which individuals live, and the degree of trust in risk communicators they encounter.

The level of individual risk perception refers to an individual’s subjective risk perception regarding a certain technology or product (e.g., driverless technology or nuclear power), understanding that knowledge and experience significantly contribute to the construction of one’s risk perception. The degree of closeness of individuals’ feelings, attitudes, and behaviors, i.e., the psychological distance [48], affects subjective judgments, just as familiarity is an important dimension in the traditional risk perception measurement framework. The degree to which individuals can participate in a technology’s governance process may represent their sense of control over it. The stronger the sense of control, the lower the level of risk perception. The social environment in which individuals live refers to the channels through which they diffuse their risk perceptions, i.e., their social network, in which they both receive information and disseminate their subjectively processed information. In this process, the structure and relationships of the network determine the path of risk information adoption and dissemination, which in turn affects the breadth and depth of risk perception contagion. The degree of information adoption is influenced by the degree of trust in the connected science communicators; the stronger the competence or professionalism of the communicators and the higher their social influence, the more likely transmitted information will be adopted by individuals and influence their risk perception levels.

In summary, the extent to which individuals diffuse risk perception and information on emerging technologies is influenced by their own factors; psychological distance to the technology; participation in governance activities; and outside experts, family, and friends. An individual’s risk perception is then diffused through one’s social networks, including kinship networks and mass media.

This paper sets out the following hypotheses:

Individuals act as influential agents in the public social system, and an individual can engage in risk contagion and non-contagion after one’s own risk perception of emerging technologies, whose initial level is constant and then changes with learning exchanges.

Due to information asymmetry and other conditions, individuals in a network are considered to possess bounded rationality in the game process and therefore cannot make the ultimate interest-maximizing decision in a single game. Thus, risk perception must spread in a lengthy process of learning and communication through a repeated game over time.

An individual’s risk perception is influenced by one’s initial perceived level of risk, the psychological distance between individuals and risks, the degree of individuals’ participation in risk governance, the degree of trust in the risk communicators they come into contact with, and the overarching potential negative outcomes, whose culmination amplifies or attenuates the network contagion of risk perceptions. At the same time, an individual’s social environment is reflected in one’s personal connections and relationships, which ultimately influence the choice to diffuse risk perception or not, as well as to determine the target of contagion.

3.2. Model Construction

3.2.1. Evolutionary Game Model Construction

The contagion of risk perception among individuals ultimately occurs in either their social network’s kinship networks or media networks. Based on evolutionary game theory, the payoff matrix of the two-party game is constructed as shown in

Table 1. The payoff matrix of the evolutionary game.

		Individuals 2 (Group2)
		Contagion	No Contagion
Individuals 1 (Group1)	Contagion	$R_{11}$$; V_{11}$	$R_{12}$$; V_{12}$
	No contagion	$R_{21}$$; V_{21}$	$R_{22}$$; V_{22}$

.

Payoffs of the public two-sided game are shown in Table 1. According to the previous research [49], $R_{ij}$ and $V_{ij}$ are the payoffs obtained by individuals after making the decisions (contagion or non-contagion of risk perception), where there are the equations as follows:

$$R_{11}=\left[\right(1 - \theta )\cdot {\alpha }_1+{\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_1]\cdot \eta$$

(1)

$$V_{11}=\left[\right(1 - \theta )\cdot {\alpha }_1+{\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_1]\cdot \eta$$

(2)

$$R_{12}=\left(1 - \theta \right)\cdot {\alpha }_1-\xi \cdot {\alpha }_2$$

(3)

$$V_{12}=\left(1 - \theta \right)\cdot {\alpha }_2+{\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_1$$

(4)

$$R_{21}=\left(1 - \theta \right)\cdot {\alpha }_1+{\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_2$$

(5)

$$V_{21}=(1-\theta )\cdot {\alpha }_2-\xi \cdot {\alpha }_1$$

(6)

$$R_{22}=(1-\theta )\cdot {\alpha }_1$$

(7)

$$V_{22}=(1-\theta )\cdot {\alpha }_2$$

(8)

(1) ${\alpha }_i(i=1,2)$ represents the individual’s risk perception of emerging technologies; $\eta \in (1,+\infty )$ is the superimposed utility of risk, reflecting the amplification of risk perception throughout the process of communication and exchange between individuals; and $\xi \in \left(0, 1\right)$ represents the negative feedback of risk perception contagion.

(2) $\beta \in \left(0, 1\right)$ is the psychological distance between the individual and the risk. Because the degree of trust in the communicator has an amplifying effect on the level of individual risk perception, we let ${\displaystyle \frac{1}{\gamma }}$ represent the degree of trust in the communicator contacted, where $\gamma \in \left(0, 1\right)$. Finally, $\theta \in \left(0, 1\right)$ is the degree of individual participation in risk governance.

After one round of communication, individuals will establish their payoffs based on the above factors associated with the risk perception of emerging technologies and decide whether to diffuse their own risk perception. In this process, individuals will also be influenced by their social environment. In this paper, the small-world and scale-free network structures from complex network theory are used to represent the kinship–friend network and authoritative expert (or high-influence) networks, respectively [50,51].

3.2.2. Complex Network Model Construction

We constructed a social network representing the public’s social environment, where $G(V,E)$ is the set of all nodes (representing individual members of the public), and $E$ is the set of all edges. A connecting edge between two nodes indicates a certain relationship between individuals, and thus a certain probability to produce risk perception contagion behavior.

The small-world network is a complex network proposed by Watts and Strogatz. Given a coupling network of $N$ nodes, each node is connected to $K/2$ of its left and right neighboring nodes, and each edge is reconnected randomly with probability $P$ without repetition or self-loops. The node degree of the network follows a Poisson distribution, and the network has a high clustering coefficient and a short average path length [52].

The scale-free network proposed by Barabási and Albert starts from a connected network of $m_0$ nodes. Newly added nodes are connected to $m$ existing nodes according to $P=k_i/\sum k_j$, forming a network with scale-free properties. Such networks have a degree distribution that approximately follows a power law, a low clustering coefficient, and a long average path length [53].

Individuals are placed into different structures of social networks, and after the initial game, they randomly choose the neighboring subjects with which they are associated to compare their payoffs. With the game payoff matrix shown in Table 1, if the individual’s payoff is $p{r}_i<p{r}_j$, then subject $i$ will adjust its game strategy and choose the strategy of node $j$ according to probability $W$ [54].

$$W_{i\to j}={\displaystyle \frac{1}{1+\mathit{exp}[(p{r}_j-p{r}_i)/k]}}$$

(9)

In Equations (1)–(9), $k$ is the degree of noise, representing uncontrollable factors that interfere with a subject’s imitation or learning. A larger $k$ indicates that more uncontrollable factors exist in the network. We chose a neutral noise intensity of $k=1$ for simulation purposes. After imitating the other node’s strategy, the individual selects the contagion target node with a certain probability ${\gamma }_{is}$ for the continued contagion of risk perception. All individuals in the social network communicate, learn, and select contagion target nodes according to the above rules. With increasing iterations, whether individuals choose to diffuse their risk perception will stabilize. At the game’s end, the ratio of nodes that chose contagion to the total number of nodes represents the contagion of risk perception of emerging technologies in the social system.

3.3. Equilibrium Analysis

From a macro perspective, assume that the social system is divided into two groups. In the initial stage of the game, suppose that a certain proportion $x\in (0,1)$ of subjects in group 1 choose to diffuse risk perception, and hence a proportion $1-x$ chooses not to diffuse. Suppose also that the proportion of individuals in group 2 who choose not to diffuse risk perception is $y\in (0,1)$, and the proportion who choose not to diffuse is $1-y$. $U_1^x$ is the expected payoff of individuals who choose to diffuse risk perception, $U_1^{1-x}$ is that of individuals who choose not to diffuse, and $U_{2A}$ is the average payoff of group 1, where

$$U_1^x=y\cdot R_{11}+(1-y)\cdot R_{12}$$

(10)

$$U_1^{1-x}=y\cdot R_{21}+(1-y)\cdot R_{22}$$

(11)

$$U_{1A}=x\cdot U_1^x+(1-x)\cdot U_1^{1-x}$$

(12)

Then, assuming that $U_2^y$ and $U_2^{1-y}$ represent the expected payoffs of those in group 2 who choose to diffuse and not to diffuse, respectively, $U_{2A}$ is the expected payoff, as shown in Table 1, where

$$U_2^y=x\cdot V_{11}+(1-x)\cdot V_{21}$$

(13)

$$U_2^{1-y}=x\cdot V_{12}+(1-x)\cdot V_{22}$$

(14)

$$U_{2A}=y\cdot U_2^y+(1-y)\cdot U_2^{1-y}$$

(15)

The replicator dynamic equation of the public individuals—G1 is

$$f(x,y)={\displaystyle \frac{dx}{dt}}=x(1-x)(U_1^x-U_1^{1-x})\hspace{0.33em}=x(1-x)\left\{y\right[(\eta -1)(1-\theta ){\alpha }_1+(\eta -1)\cdot {\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_2+\xi \cdot {\alpha }_2]-\xi \cdot {\alpha }_2\}$$

(16)

The replicator dynamic equation of the public individuals—G2 is

$$g(x,y)={\displaystyle \frac{dy}{dt}}=y(1-y)(U_2^y-U_2^{1-y}) \hspace{0.33em}=y(1-y)\left\{x\right[(\eta -1)(1-\theta ){\alpha }_2+(\eta -1)\cdot {\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_2+\xi \cdot {\alpha }_1]-\xi \cdot {\alpha }_1\}$$

(17)

In the above replicator dynamic system, the equilibrium points, $(x,y)\in \left\{\right(x,y\left)\left|0\le x\le 1,0\le y\le 1\right.\right\}$, include $(0,0)$ $(0,1)$ $(1,0)$ $(1,1)$ $(x^{\ast },y^{\ast })$.

This is where

$$x^{\ast }={\displaystyle \frac{\xi \cdot {\alpha }_1}{(\eta -1)\cdot A+\xi \cdot {\alpha }_1}}$$

(18)

$$y^{\ast }={\displaystyle \frac{\xi \cdot {\alpha }_2}{(\eta -1)\cdot A+\xi \cdot {\alpha }_2}}$$

(19)

The Jacobian matrix, J, of the replicator dynamic system is

$$J=\left[\begin{array}{cc}\left(1-2x\right)\cdot \left\{y\left[\left(\eta -1\right)\cdot A+\xi \cdot {\alpha }_2\right]-\xi \cdot {\alpha }_2\right\}& x\left(1-x\right)\left[\left(\eta -1\right)\cdot A+\xi \cdot {\alpha }_2\right]\\ y\left(1-y\right)\left[\left(\eta -1\right)\cdot B+\xi \cdot {\alpha }_1\right]& \left(1-2y\right)\cdot \left\{x\left[\left(\eta -1\right)\cdot B+\xi \cdot {\alpha }_1\right]-\xi \cdot {\alpha }_1\right\}\end{array}\right]$$

(20)

where

$$A=(1-\theta )\cdot {\alpha }_1+{\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_2$$

(21)

$$B=(1-\theta )\cdot {\alpha }_2+{\displaystyle \frac{\beta }{\gamma }}\cdot {\alpha }_1$$

(22)

Namely,

$$\begin{gathered} DetJ=\left(1-2x\right)\left(1-2y\right)\left\{y\left[\left(\eta -1\right)A+\xi {\alpha }_2\right]-\xi {\alpha }_2\right\} \\ \left\{x\left[\left(\eta -1\right)B+\xi {\alpha }_1\right]-\xi {\alpha }_1\right\}-xy\left(1-x\right)\left(1-y\right) \\ \left[\right(\eta -1)A+\xi {\alpha }_2]\left[\right(\eta -1)B+\xi {\alpha }_1] \end{gathered}$$

(23)

$$TrJ=\left(1-2x\right)\left\{y\left[\left(\eta -1\right)A+\xi {\alpha }_2\right]-\xi {\alpha }_2\right\}\hspace{0.33em}+(1-2y)\left\{x\right[(\eta -1)B+\xi {\alpha }_1]-\xi {\alpha }_1\}$$

(24)

According to the determinant (Det) and the trace (Tr) of J, six situations are given in

Table 2. Analysis of equilibrium points.

(x, y)	Det J	Tr J	Stability
$(0,0)$	+	−	ESS
$(0,1)$	+	+	Unstable point
$(1,0)$	+	+	Unstable point
$(1,1)$	+	−	ESS
$(x^{\ast },y^{\ast })$	−	+	Saddle point

Note: * represents the saddle points of system.

.

4. Simulation Experiments and Results

4.1. Method and Parameter Setup

According to preliminary questionnaire responses and expert interviews, the emergency management research team from Tsinghua University, and the scoring of relevant academic experts, we determined the simulation-related parameters to be as shown in

Table 3. Simulation parameters.

Factor	Influence Level	Parameter Range	Classification Notes
$\mathrm{Psychological} \mathrm{distance} \mathrm{to} \mathrm{risk} \left(\beta \right)$	Low	$\beta \in [0,0.4]$	Based on averaged and normalized results of preliminary questionnaire.
	Moderate	$\beta \in (0.4,0.7]$
	High	$\beta \in (0.7,1]$
$\mathrm{Risk} \mathrm{governance} \mathrm{participation} \left(\theta \right)$	Low	$\theta \in [0,0.3]$	Based on types of participation available to individuals to govern emerging technologies, as determined by an expert panel.
	Moderate	$\theta \in (0.3,0.7]$
	High	$\theta \in (0.7,1]$
$\mathrm{Trust} \mathrm{in} \mathrm{the} \mathrm{risk} \mathrm{communicators} \left({\displaystyle \frac{1}{\gamma }}\right)$	Low	$\gamma \in [0,0.2]$	Based on averaged and normalized results of preliminary questionnaire.
	Moderate	$\gamma \in (0.2,0.6]$
	High	$\gamma \in (0.6,1]$

.

Following an evolutionary game analysis of complex networks, the simulation procedure is as follows:

Step 1: Generate scale-free and small-world networks with a certain number of nodes.

Step 2: Initialize the simulation parameters and set their values as shown in

Table 4. Values of fixed parameters in the payment matrix.

Network Structure	Network Size	Number of Iterations	$\mathit{\beta }$	$\mathit{\theta }$	$\mathit{\gamma }$	$\mathit{\alpha }$	$\mathit{\eta }$	$\mathit{\xi }$
Scale free	200 nodes	30	0.2	0.2	0.2	1	1.5	0.5
			0.5	0.5	0.5
			0.8	0.8	0.8
Small world	200 nodes	30	0.2	0.2	0.2	1	1.5	0.5
			0.5	0.8	0.5
			0.8	0.5	0.8

.

Step 3: Randomly assign game behaviors to nodes at a 1:1 ratio, assigning 1 to diffuse individual risk perceptions of emerging technology, and 0 to remain silent (i.e., no contagion).

Step 4: Execute the simulation algorithm. Subjects in the social network select neighboring nodes for payoff comparison and play multiple games to determine individual behavior based on their payoffs and Equation (9).

Step 5: Repeat steps 2–4, terminating the simulation after reaching a set number of iterations. To prevent network structure instability from interfering with the results, we repeated the simulation an average of 30 times each. We further simulated contagion in both small-world and scale-free networks under various conditions of influential factors, as depicted in contagion graphs.

4.2. Discussion of Results

As the network nodes represent realistic individuals, their risk perception transmission behavior is sensitive to the psychological distance, degree of participation in risk management, and degree of trust in the risk communicator. Therefore, to ensure consistency, we fixed the values of other parameters to explore the effects of the psychological distance between individuals and risk, the degree of individuals’ participation in risk governance, and the degree of trust in the risk communicators on the contagion of risk perception of emerging technologies. Parameter settings are shown in Table 4.

Using MATLAB 2018(b) software, scale-free and small-world networks with 200 nodes each were initially generated, as shown in Figure 2.

4.2.1. Psychological Distance between Individuals and Emerging Technological Risks

Following the steps in Section 4.1, the psychological distance between the simulated individual and risk takes a value between 0.05 and 0.95, and the contagion of risk perception in the social system is as shown in Figure 3 and Figure 4.

As shown by the figures, the contagion of risk perception gradually decreases as the psychological distance between individuals and risk increases. At $\beta$ = 0.2, individuals in the social system mostly choose to diffuse their risk perceptions, and as the number of iterations increases, risk perceptions eventually spread throughout the network. Conversely, at $\beta$ = 0.8, individuals in the social system gradually stop diffusing their risk perceptions, in accordance with theoretical predictions. Individuals closer to the emerging technology may care more about it, or they interact with it more frequently in daily life. Thus, they are sensitive to potential risk and tend to increase their risk perception level and spread it to other individuals after receiving the risk signal. As the psychological distance gradually increases, individuals’ sensitivity to technological risks decreases, such that when they receive information about others’ risk perceptions, they fail to empathize with them. Eventually, psychologically distant individuals in the network tend to filter out the risk information and halt contagion. Comparing contagion between social media networks and kinship–friend networks, we found that their trends in risk perception contagion were not significantly different. However, the level of risk perception contagion among kinship networks was higher than that of social media networks under the same psychological distance, suggesting that word-of-mouth risk perception contagion is more infectious and may spur more individuals to diffuse their own risk perceptions.

Conclusion 1 The contagion efficiency of the risk perception of emerging technologies gradually decreases with the increasing psychological distance of individuals from emerging technological risks.

4.2.2. Trust between Individuals and Their Risk Communicators

The degree of trust $\gamma$ of simulated individuals toward their risk communicator takes a value between 0.05 and 0.95. The resulting risk perception contagion across social networks is shown in Figure 5 and Figure 6.

From Figure 5 and Figure 6, it can be seen that as $\gamma$ decreases, the trust level ${\displaystyle \frac{1}{\gamma }}$ of individuals toward risk communicators increases, and it gradually increases the contagion of risk perceptions across the network. At the same time, trust in risk communicators has a greater influence than psychological distance on contagion behavior; e.g., when $\gamma$ = 0.8, risk contagion behavior in the social system is already in a rising-stable state because individuals’ risk perceptions are sensitive to their trust in the information source, and the technical knowledge and influence of the communicator determines the trust level of the receiving individuals. When individuals receive information about technology risk from professionals, they tend to believe and agree with their risk perceptions and spread them to other individuals, and when they receive risk information from people unrelated to technology, their trust level is relatively low, as is their probability of choosing to spread risk perceptions. When trust in the communicator reaches a certain level, contagion behavior stabilizes, indicating little difference between the contagion of risk perceptions by authoritative experts and experts, for example. Comparing the two network structures, with trust in risk communicators held equal between them, small-world networks were able to reach equilibrium after fewer game iterations than scale-free networks. This suggests that diffusing risk perceptions among relatives and peers is more efficient, converging after fewer information exchanges. At the same time, under identical conditions, the level of trust in communicators has a stronger influence on individual risk perceptions than their psychological distance. In other words, individuals tend to spread the risk judgements of professionals or trusted communicators regardless of their own psychological distance to technological risks, while they do not readily trust and spread the risk judgment of non-professionals, even for technological risks of greater negative consequence.

Conclusion 2 The contagion efficiency of risk perception of emerging technologies gradually increases with the level of trust in risk communicators.

4.2.3. Individual Participation in Risk Governance of Emerging Technologies

Figure 7 and Figure 8 show the contagion of risk perceptions across the social system at individual involvement in risk governance values between 0.05 and 0.95.

As the figures show, when the level of participation in risk governance is 0.2, risk perceptions fully diffuse throughout the social network. When it increases to 0.8, the contagion of risk perceptions decreases to 0.2 or even less. This relationship can be explained as follows. When public participation in the governance of emerging technologies risks is low, it manifests itself as a wait-and-see state regarding the development and risks of emerging technologies. When an individual cannot participate in the governance of emerging technologies, the perceived sense of control over the technology decreases, allowing the uncertainty and ambiguity inherent to emerging technologies to lead to panic. As public participation in technology governance increases, individuals in the network become more aware of and familiar with the risks posed by emerging technologies, diminishing their fear of the technologies and improving their sense of control. This improved sense of control or diminished fear has two effects: it reduces levels of individual risk perceptions, and individuals now hold a stauncher, less persuadable judgement of technological risk, leading to the reduction in or even suppression of risk perception contagion. Comparing the two network structures, regardless of the level of participation in governance, risk perception contagion among close friends and family was more sensitive to the attenuating effect of participation in governance, while this was less impactful in social media networks. Thus, participation in risk governance among close friends and family seems to have a stronger effect on risk perception contagion, likely because risk judgements among them are more persuasive, and more likely to form consistent attitudes and behaviors toward technology.

Conclusion 3 As public participation in emerging technology risk governance deepens, the contagion of risk perception in the social system gradually decreases.

It is evident that among the analyzed factors of influence, the positive influence of a small-world network structure on the contagion of risk perception of emerging technologies is stronger than that of a scale-free network structure, i.e., risk perception is more efficiently transmitted among kin and friends. This is because small-world networks are characterized by short average path lengths, high clustering coefficients, and more direct connectivity across nodes. Therefore, information, knowledge, and personal experience can spread more efficiently, allowing connected individuals to quickly imitate the contagion of risk perception and rapidly spread this behavior across shorter paths. Conversely, scale-free networks have higher clustering coefficients, where each node is individually connected to influential nodes; although they can be directly influenced by big-V nodes (influencers), it takes more iterations for risk perception to reach either complete contagion or complete lack of it. In short, contagion efficiency of risk perception in scale-free networks is low.

Conclusion 4 The positive effect of small-world networks on the contagion of risk perceptions is higher than that of scale-free networks, i.e., risk perceptions are more likely to spread among kin and friends.

5. Conclusions and Suggestions for Countermeasures

As emerging technologies continue to develop, their potential moral and ethical risks have gradually received greater attention from the public. The resulting risk perceptions of emerging technologies have spread throughout social networks, bringing with them potentially detrimental anti-technology sentiments and other socially destabilizing risk factors. We drew from risk perception theory to analyze the contagion process of risk perception of emerging technologies among individuals, and to simulate the influence of various factors in a complex network evolutionary game model. Results show that individuals’ risk perception and degree of trust in risk communicators have a positive stimulating effect on the contagion of risk perception of emerging technologies, involvement in risk governance has a negative inhibiting effect, and kinship networks better facilitate the contagion of risk perceptions than social media networks.

There are several implications of these results on the risk governance of emerging technologies. The management of this risk requires a degree of government intervention and adaptive governance, which should be guided by defined strategic and developmental needs. First, regarding the technologies themselves, the uncertainty of emerging technologies means there may be unforeseen consequences to their proliferation and application. Thus, relevant stakeholders should be prepared to quickly react to, analyze, and deal with crises of emerging technology, and to anticipate potential problems in a targeted manner. Second, the government should provide public guidance and organizational education on the development of emerging technologies and encourage public participation in risk management discussions. Third, public risk perceptions should be incorporated into policy processes and their evaluation, integrating a wide span of public opinion. For emerging technologies that are more controversial and uncertain, decision-makers may seek to engage in grassroots media efforts to answer common questions and encourage public participation in risk, with the goal of improving risk communication. Finally, risk communication mechanisms and channels should be actively constructed to reconcile differing perceptions between experts and the public on risky technologies and improve public awareness of science.

This paper’s research on the contagion of risk perception of emerging technologies was limited to simulation analysis. Further research may benefit from combining real-world data with simulation models to validate and expand upon our conclusions.