Perspectives on Psychological Science最新文献

We illustrate how standard psychometric inventories originally designed for assessing noncognitive human traits can be repurposed as diagnostic tools to evaluate analogous traits in large language models (LLMs). We start from the assumption that LLMs, inadvertently yet inevitably, acquire psychological traits (metaphorically speaking) from the vast text corpora on which they are trained. Such corpora contain sediments of the personalities, values, beliefs, and biases of the countless human authors of these texts, which LLMs learn through a complex training process. The traits that LLMs acquire in such a way can potentially influence their behavior, that is, their outputs in downstream tasks and applications in which they are employed, which in turn may have real-world consequences for individuals and social groups. By eliciting LLMs' responses to language-based psychometric inventories, we can bring their traits to light. Psychometric profiling enables researchers to study and compare LLMs in terms of noncognitive characteristics, thereby providing a window into the personalities, values, beliefs, and biases these models exhibit (or mimic). We discuss the history of similar ideas and outline possible psychometric approaches for LLMs. We demonstrate one promising approach, zero-shot classification, for several LLMs and psychometric inventories. We conclude by highlighting open challenges and future avenues of research for AI Psychometrics.

Research on clinical versus statistical prediction has demonstrated that algorithms make more accurate predictions than humans in many domains. Geopolitical forecasting is an algorithm-unfriendly domain, with hard-to-quantify data and elusive reference classes that make predictive model-building difficult. Furthermore, the stakes can be high, with missed forecasts leading to mass-casualty consequences. For these reasons, geopolitical forecasting is typically done by humans, even though algorithms play important roles. They are essential as aggregators of crowd wisdom, as frameworks to partition human forecasting variance, and as inputs to hybrid forecasting models. Algorithms are extremely important in this domain. We doubt that humans will relinquish control to algorithms anytime soon-nor do we think they should. However, the accuracy of forecasts will greatly improve if humans are aided by algorithms.

One of the early goals of artificial intelligence (AI) was to create algorithms that exhibited behavior indistinguishable from human behavior (i.e., human-like behavior). Today, AI has diverged, often aiming to excel in tasks inspired by human capabilities and outperform humans, rather than replicating human cogntion and action. In this paper, I explore the overarching question of whether computational algorithms have achieved this initial goal of AI. I focus on dynamic decision-making, approaching the question from the perspective of computational cognitive science. I present a general cognitive algorithm that intends to emulate human decision-making in dynamic environments, as defined in instance-based learning theory (IBLT). I use the cognitive steps proposed in IBLT to organize and discuss current evidence that supports some of the human-likeness of the decision-making mechanisms. I also highlight the significant gaps in research that are required to improve current models and to create higher fidelity in computational algorithms to represent human decision processes. I conclude with concrete steps toward advancing the construction of algorithms that exhibit human-like behavior with the ultimate goal of supporting human dynamic decision-making.

Meta-analytic evidence indicates that the quality of the attachment relationship that infants establish with their primary caregiver has enduring significance for socioemotional and cognitive outcomes. However, the mechanisms by which early attachment experiences contribute to subsequent development remain underspecified. According to attachment theory, early attachment experiences become embodied in the form of cognitive-affective representations, referred to as internal working models (IWMs), that guide future behavior. Little is known, however, about the cognitive architecture of IWMs in infancy. In this article, we discuss significant advances made in the field of infant cognitive development and propose that leveraging insights from this research has the potential to fundamentally shape our understanding of the cognitive architecture of attachment representations in infancy. We also propose that the integration of attachment research into cognitive research can shed light on the role of early experiences, individual differences, and stability and change in infant cognition, as well as open new routes of investigation in cognitive studies, which will further our understanding of human knowledge. We provide recommendations for future research throughout the article and conclude by using our collaborative research as an example.

Fear is an emotion triggered by the perception of danger and motivates safety behaviors. Within the context of the COVID-19 pandemic, there were ample danger cues (e.g., images of patients on ventilators) and a high need for people to use appropriate safety behaviors (e.g., social distancing). Given this central role of fear within the context of a pandemic, it is important to review some of the emerging findings and lessons learned during the COVID-19 pandemic and their implications for managing fear. We highlight factors that determine fear (i.e., proximity, predictability, and controllability) and review several adaptive and maladaptive consequences of fear of COVID-19 (e.g., following governmental health policies and panic buying). Finally, we provide directions for future research and make policy recommendations that can promote adequate health behaviors and limit the negative consequences of fear during pandemics.

Humans evolved to be hyper-cooperative, particularly when among people who are well known to them, when relationships involve reciprocal helping opportunities, and when the costs to the helper are substantially less than the benefits to the recipient. Because humans' cooperative nature evolved over many millennia when they lived exclusively in small groups, factors that cause cooperation to break down tend to be those associated with life in large, impersonal, modern societies: when people are not identifiable, when interactions are one-off, when self-interest is not tied to the interests of others, and when people are concerned that others might free ride. From this perspective, it becomes clear that policies for managing pandemics will be most effective when they highlight superordinate goals and connect people or institutions to one another over multiple identifiable interactions. When forging such connections is not possible, policies should mimic critical components of ancestral conditions by providing reputational markers for cooperators and reducing the systemic damage caused by free riding. In this article, we review policies implemented during the pandemic, highlighting spontaneous community efforts that leveraged these aspects of people's evolved psychology, and consider implications for future decision makers.

Faced with the challenges of motivating people to vaccinate, many countries have introduced policy-level interventions to encourage vaccination against COVID-19. For example, mandates were widely imposed requiring individuals to vaccinate to work and attend school, and vaccination passports required individuals to show proof of vaccination to travel and access public spaces and events. Furthermore, some countries also began offering financial incentives for getting vaccinated. One major criticism of these policies was the possibility that they would produce reactance and thus undermine voluntary vaccination. This article therefore reviews relevant empirical evidence to examine whether this is indeed the case. Specifically, we devote separate sections to reviewing and discussing the impacts of three major policies that were implemented during the COVID-19 pandemic: vaccination mandates, vaccination passports, and the provision of financial incentives. A careful analysis of the evidence provides little support that these policies backfire but instead can effectively promote vaccination at the population level. The policies are not without limitations, however, such as their inability to mobilize those that are strongly hesitant to vaccines. Finally, we discuss how policy-level interventions should be designed and implemented to address future epidemics and pandemics.

The COVID-19 pandemic challenged the public health system to respond to an emerging, difficult-to-understand pathogen through demanding behaviors, including staying at home, masking for long periods, and vaccinating multiple times. We discuss key challenges of the pandemic health communication efforts deployed in the United States from 2020 to 2022 and identify research priorities. One priority is communicating about uncertainty in ways that prepare the public for disagreement and likely changes in recommendations as scientific understanding advances: How can changes in understanding and recommendations foster a sense that "science works as intended" rather than "the experts are clueless" and prevent creating a void to be filled by misinformation? A second priority concerns creating a culturally fluent framework for asking people to engage in difficult and novel actions: How can health messages foster the perception that difficulties of behavior change signal that the change is important rather than that the change "is not for people like me?" A third priority entails a shift from communication strategies that focus on knowledge and attitudes to interventions that focus on norms, policy, communication about policy, and channel factors that impair behavior change: How can we move beyond educating and correcting misinformation to achieving desired actions?