Science Communication as a Collective Intelligence Endeavor: A Manifesto and Examples for Implementation.

Effective science communication is challenging when scientific messages are informed by a continually updating evidence base and must often compete against misinformation. We argue that we need a new program of science communication as collective intelligence-a collaborative approach, supported by technology. This would have four key advantages over the typical model where scientists communicate as individuals: scientific messages would be informed by (a) a wider base of aggregated knowledge, (b) contributions from a diverse scientific community, (c) participatory input from stakeholders, and (d) better responsiveness to ongoing changes in the state of knowledge.

The effects of recursive communication dynamics on belief updating.

Many social interactions are characterized by dynamic interplay, such that individuals exert reciprocal influence over each other's behaviours and beliefs. The present study investigated how the dynamics of reciprocal influence affect individual beliefs in a social context, over and above the information communicated in an interaction. To this end, we developed a simple social decision-making paradigm in which two people are asked to make perceptual judgments while receiving information about each other's decisions. In a Static condition, information about the partner only conveyed their initial, independent judgment. However, in a Dynamic condition, each individual saw the evolving belief of their partner as they learnt about and responded to the individual's own judgment. The results indicated that in both conditions, the majority of confidence adjustments were characterized by an abrupt change followed by smaller adjustments around an equilibrium, and that participants' confidence was used to arbitrate conflict (although deviating from Bayesian norm). Crucially, recursive interaction had systematic effects on belief change relative to the static baseline, magnifying confidence change when partners agreed and reducing confidence change when they disagreed. These findings indicate that during dynamic interactions-often a characteristic of real-life and online social contexts-information is collectively transformed rather than acted upon by individuals in isolation. Consequently, the output of social events is not only influenced by what the dyad knows but also by predictable recursive and self-reinforcing dynamics.

Adversarial Dynamics in Centralized Versus Decentralized Intelligent Systems.

Artificial intelligence (AI) is often used to predict human behavior, thus potentially posing limitations to individuals' and collectives' freedom to act. AI's most controversial and contested applications range from targeted advertisements to crime prevention, including the suppression of civil disorder. Scholars and civil society watchdogs are discussing the oppressive dangers of AI being used by centralized institutions, like governments or private corporations. Some suggest that AI gives asymmetrical power to governments, compared to their citizens. On the other hand, civil protests often rely on distributed networks of activists without centralized leadership or planning. Civil protests create an adversarial tension between centralized and decentralized intelligence, opening the question of how distributed human networks can collectively adapt and outperform a hostile centralized AI trying to anticipate and control their activities. This paper leverages multi-agent reinforcement learning to simulate dynamics within a human-machine hybrid society. We ask how decentralized intelligent agents can collectively adapt when competing with a centralized predictive algorithm, wherein prediction involves suppressing coordination. In particular, we investigate an adversarial game between a collective of individual learners and a central predictive algorithm, each trained through deep Q-learning. We compare different predictive architectures and showcase conditions in which the adversarial nature of this dynamic pushes each intelligence to increase its behavioral complexity to outperform its counterpart. We further show that a shared predictive algorithm drives decentralized agents to align their behavior. This work sheds light on the totalitarian danger posed by AI and provides evidence that decentrally organized humans can overcome its risks by developing increasingly complex coordination strategies.

Machine Impostors Can Avoid Human Detection and Interrupt the Formation of Stable Conventions by Imitating Past Interactions: A Minimal Turing Test.

Interactions between humans and bots are increasingly common online, prompting some legislators to pass laws that require bots to disclose their identity. The Turing test is a classic thought experiment testing humans' ability to distinguish a bot impostor from a real human from exchanging text messages. In the current study, we propose a minimal Turing test that avoids natural language, thus allowing us to study the foundations of human communication. In particular, we investigate the relative roles of conventions and reciprocal interaction in determining successful communication. Participants in our task could communicate only by moving an abstract shape in a 2D space. We asked participants to categorize their online social interaction as being with a human partner or a bot impostor. The main hypotheses were that access to the interaction history of a pair would make a bot impostor more deceptive and interrupt the formation of novel conventions between the human participants. Copying their previous interactions prevents humans from successfully communicating through repeating what already worked before. By comparing bots that imitate behavior from the same or a different dyad, we find that impostors are harder to detect when they copy the participants' own partners, leading to less conventional interactions. We also show that reciprocity is beneficial for communicative success when the bot impostor prevents conventionality. We conclude that machine impostors can avoid detection and interrupt the formation of stable conventions by imitating past interactions, and that both reciprocity and conventionality are adaptive strategies under the right circumstances. Our results provide new insights into the emergence of communication and suggest that online bots mining personal information, for example, on social media, might become indistinguishable from humans more easily.

Using multisession tDCS stimulation as an early intervention on memory bias processing in subthreshold depression.

Transcranial direct current stimulation (tDCS) as an intervention tool has gained promising results in major depression disorder. However, studies related to subthreshold depression's (SD) cognitive deficits and neuromodulation approaches for the treatment of SD are still rare. We adopted Beck's cognitive model of depression and tested the tDCS stimulation effects on attentional and memory deficits on SD. First, this was a single-blinded, randomized, sham-controlled clinical trial to determine a 13-day tDCS modulation effect on 49 SD (27: Stimulation; 22: Sham) and 17 healthy controls. Second, the intervention effects of the consecutive and single-session tDCS were compared. Furthermore, the attentional and memory biases were explored in SD. Anodal tDCS was administrated over left dorsolateral prefrontal cortex for 13 consecutive days. Attentional and memory bias were assessed through a modified Sternberg task and a dot-probe task on the 1st, 2nd, and 15th day while their EEG was being recorded. After the 13-day tDCS stimulation (not after single-session stimulation), we found reduced memory bias (Stimulation vs. Sham, p = .02, r2  = .09) and decreased mid-frontal alpha power (p < .01, r2  = .13). In contrast, tDCS did not affect any attentional related behavioral or neural indexes (all ps > .15). Finally, reduced depressive symptoms (e.g., BDI score) were found for both groups. The criteria of SD varied across studies; the efficacy of this protocol should be tested in elderly patients. Our study suggests memory bias of SD can be modulated by the multisession tDCS and alpha power could serve as a neural index for intervention.

Beyond collective intelligence: Collective adaptation.

We develop a conceptual framework for studying collective adaptation in complex socio-cognitive systems, driven by dynamic interactions of social integration strategies, social environments and problem structures. Going beyond searching for 'intelligent' collectives, we integrate research from different disciplines and outline modelling approaches that can be used to begin answering questions such as why collectives sometimes fail to reach seemingly obvious solutions, how they change their strategies and network structures in response to different problems and how we can anticipate and perhaps change future harmful societal trajectories. We discuss the importance of considering path dependence, lack of optimization and collective myopia to understand the sometimes counterintuitive outcomes of collective adaptation. We call for a transdisciplinary, quantitative and societally useful social science that can help us to understand our rapidly changing and ever more complex societies, avoid collective disasters and reach the full potential of our ability to organize in adaptive collectives.

A variational-autoencoder approach to solve the hidden profile task in hybrid human-machine teams.

Algorithmic agents, popularly known as bots, have been accused of spreading misinformation online and supporting fringe views. Collectives are vulnerable to hidden-profile environments, where task-relevant information is unevenly distributed across individuals. To do well in this task, information aggregation must equally weigh minority and majority views against simple but inefficient majority-based decisions. In an experimental design, human volunteers working in teams of 10 were asked to solve a hidden-profile prediction task. We trained a variational auto-encoder (VAE) to learn people's hidden information distribution by observing how people's judgments correlated over time. A bot was designed to sample responses from the VAE latent embedding to selectively support opinions proportionally to their under-representation in the team. We show that the presence of a single bot (representing 10% of team members) can significantly increase the polarization between minority and majority opinions by making minority opinions less prone to social influence. Although the effects on hybrid team performance were small, the bot presence significantly influenced opinion dynamics and individual accuracy. These findings show that self-supervized machine learning techniques can be used to design algorithms that can sway opinion dynamics and group outcomes.

The role of decision confidence in advice-taking and trust formation.

In a world where ideas flow freely across multiple platforms, people must often rely on others' advice and opinions without an objective standard to judge whether this information is accurate. The present study explores the hypothesis that an individual's internal decision confidence can be used as a signal to learn the accuracy of others' advice, even in the absence of feedback. According to this "agreement-in-confidence" hypothesis, people can learn about an advisor's accuracy across multiple interactions according to whether the advice offered agrees with their own initial opinions, weighted by the confidence with which these initial opinions are held. We test this hypothesis using a judge-advisor system paradigm to precisely manipulate the profiles of virtual advisors in a perceptual decision-making task. We find that when advisors' and participants' judgments are independent, people can correctly learn advisors' features, like their accuracy and calibration, whether or not objective feedback is available. However, when their judgments (and thus errors) are correlated-as is the case in many real social contexts-predictable distortions in trust can be observed between feedback and feedback-free scenarios. Using agent-based simulations, we explore implications of these individual-level heuristics for network-level patterns of trust and belief formation. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Confidence, advice seeking and changes of mind in decision making.

Humans and other animals rely on social learning strategies to guide their behaviour, especially when the task is difficult and individual learning might be costly or ineffective. Recent models of individual and group decision-making suggest that subjective confidence judgments are a prime candidate in guiding the way people seek and integrate information from social sources. The present study investigates the way people choose and use advice as a function of the confidence in their decisions, using a perceptual decision task to carefully control the quality of participants' decisions and the advice provided. The results show that reported confidence guides the search for new information in accordance with probabilistic normative models. Moreover, large inter-individual differences were found, which strongly correlated with more traditional measures of metacognition. However, the extent to which participants used the advice they received deviated from what would be expected under a Bayesian update of confidence, and instead was characterised by heuristic-like strategies of categorically ignoring vs. accepting advice provided, again with substantial individual differences apparent in the relative dominance of these strategies.

Modularity and composite diversity affect the collective gathering of information online.

Many modern interactions happen in a digital space, where automated recommendations and homophily can shape the composition of groups interacting together and the knowledge that groups are able to tap into when operating online. Digital interactions are also characterized by different scales, from small interest groups to large online communities. Here, we manipulate the composition of groups based on a large multi-trait profiling space (including demographic, professional, psychological and relational variables) to explore the causal link between group composition and performance as a function of group size. We asked volunteers to search news online under time pressure and measured individual and group performance in forecasting real geo-political events. Our manipulation affected the correlation of forecasts made by people after online searches. Group composition interacted with group size so that composite diversity benefited individual and group performance proportionally to group size. Aggregating opinions of modular crowds composed of small independent groups achieved better forecasts than aggregating a similar number of forecasts from non-modular ones. Finally, we show differences existing among groups in terms of disagreement, speed of convergence to consensus forecasts and within-group variability in performance. The present work sheds light on the mechanisms underlying effective online information gathering in digital environments.

The network limits of infectious disease control via occupation-based targeting.

Policymakers commonly employ non-pharmaceutical interventions to reduce the scale and severity of pandemics. Of non-pharmaceutical interventions, physical distancing policies-designed to reduce person-to-person pathogenic spread - have risen to recent prominence. In particular, stay-at-home policies of the sort widely implemented around the globe in response to the COVID-19 pandemic have proven to be markedly effective at slowing pandemic growth. However, such blunt policy instruments, while effective, produce numerous unintended consequences, including potentially dramatic reductions in economic productivity. In this study, we develop methods to investigate the potential to simultaneously contain pandemic spread while also minimizing economic disruptions. We do so by incorporating both occupational and contact network information contained within an urban environment, information that is commonly excluded from typical pandemic control policy design. The results of our methods suggest that large gains in both economic productivity and pandemic control might be had by the incorporation and consideration of simple-to-measure characteristics of the occupational contact network. We find evidence that more sophisticated, and more privacy invasive, measures of this network do not drastically increase performance.

The perceptual and social components of metacognition.

When deciding whether or not to bring an umbrella to work, your confidence will be influenced by the sky outside the window (direct evidence) as well as by, for example, whether or not people walking in the street have their own umbrella (indirect or contingent evidence). These 2 distinct aspects of decision confidence have not yet been assessed independently within the same framework. Here we study the relative contributions of stimulus-specific and social-contingent information on confidence formation. Dyads of participants made visual perceptual decisions, first individually and then together by sharing their wagers in their decisions. We independently manipulated the sensory evidence and the social consensus available to participants and found that both type of evidence contributed to wagers. Consistent with previous work, the amount people were prepared to wager covaried with the strength of sensory evidence. However, social agreements and disagreement affected wagers in opposite directions and asymmetrically. These different contributions of sensory and social evidence to wager were linearly additive. Moreover, average metacognitive sensitivity-namely the association between wagers and accuracy-between interacting dyad members positively correlated with dyadic performance and dyadic benefit above average individual performance. Our results provide a general framework that accounts for how both social context and direct sensory evidence contribute to decision confidence. (PsycINFO Database Record

Brain Mechanisms Underlying the Brief Maintenance of Seen and Unseen Sensory Information.

Recent evidence of unconscious working memory challenges the notion that only visible stimuli can be actively maintained over time. In the present study, we investigated the neural dynamics underlying the maintenance of variably visible stimuli using magnetoencephalography. Subjects had to detect and mentally maintain the orientation of a masked grating. We show that the stimulus is fully encoded in early brain activity independently of visibility reports. However, the presence and orientation of the target are actively maintained throughout the brief retention period, even when the stimulus is reported as unseen. Source and decoding analyses revealed that perceptual maintenance recruits a hierarchical network spanning the early visual, temporal, parietal, and frontal cortices. Importantly, the representations coded in the late processing stages of this network specifically predicted visibility reports. These unexpected results challenge several theories of consciousness and suggest that invisible information can be briefly maintained within the higher processing stages of visual perception.