Quantifying the impact of persuasiveness, cautiousness and prior beliefs in (mis)information sharing on online social networks using Drift Diffusion Models

TL;DR

It is found that the natural shape of these social online networks provides a fertile ground for any news to rapidly become viral, yet it is found that, for the case of X, limiting the number of followers of the most connected users proves to be an appropriate and feasible containment strategy.

Abstract

Misleading newsletters can shape individuals' perceptions, and pose a threat to societies; as we witnessed by lowering the severity of follow-up stay-at-home orders and burdening a significant challenge to the fight against COVID-19. In this research, we study (mis)information spreading, reanalyzing behavioral data on online sharing, and analyzing decision-making mechanisms using the Drift Diffusion Model (DDM). We find that subjects display an increased instinctive inclination towards sharing misleading news, but rational thinking significantly curbs this reaction, especially for more cautious and older individuals. On top of network structures with similar characteristics as X, Mastodon, and Facebook, we use an agent-based model to expand this individual knowledge to a large scale where individuals are exposed to (mis)information through friends and share (or not) content with probabilities driven by DDM. We found that the natural shape of these social online networks provides a fertile ground for any news to rapidly become viral. Yet we have found that, for the case of X, limiting the number of followers of the most connected users proves to be an appropriate and feasible containment strategy.

Figures (33)

• Figure 1: Information spreading in a society considering the neurocognitive mechanisms driving each individual response. On the right side, there is the social network on top of which the spreading process occurs, $i.e.$, information diffuses in the social media platform where users interact. Every time an $S$ individual shares content (dark pink), its $NS$ neighbors will see it, and a single decision-making process is activated (left figure) for each individual to decide whether to share (pink) or not the content (green). The piece of news has not yet reached Grey nodes.
• Figure 1: Sensitivity analysis of the free parameters of DDM as a function of RTs. Each curve corresponds to a different age range; rows correspond to misleading (bottom) and reliable (top) content. We average among headlines ($12$ per row).
• Figure 2: Probability Distribution Function of the Response Times for the case of Misleading Information. The estimation of the free parameters values (a, $\nu$, z and $t_0$ in the legend box) are obtained with the Python-based toolbox HDDM wiecki2013hddm version=$0.6.0$. For simplicity, we display headline number $3$ (the remaining headlines are displayed in the Supplementary Information). Dots correspond to empirical data, lines are obtained using Eq. \ref{['RT-PDF']} with the corresponding values of the free parameters, and crosses correspond to stochastic simulated data. Each panel corresponds to a different age bin: (a) $16-24$ years old, (b) $25-31$ years old, (c) $32-37$ years old, (d) $38-47$ years old, and (e) $48-88$ years old; and the corresponding number of participants for each case are displayed in the titles. The left side of each curve (violet) corresponds to data collected with not sharing answers, while the right side (blue) corresponds to sharing answers.
• Figure 2: Expected fraction of individuals towards sharing. Each box plot shows the distribution of each age range averaging over $12$ headlines for the case of misleading (blue) and reliable (green) headlines.
• Figure 3: Probability Distribution Function of the Response Times for the case of Reliable Information. Details on the description are equal to Fig. \ref{['PDF-misleading']} but considering the case of reliable headlines. By comparing both scenarios, we can observe that for reliable news responses tend to have longer response times and more they are more shared, as we can also observe in Fig. \ref{['statistics-share']}.