New Content From Current Directions in Psychological Science

Signatures of Reinforcement Learning in Natural Behavior
Catherine A. Hartley, Susan L. Benear, Aaron S. Heller

Across myriad real-world contexts, people encounter the challenge of learning to take actions that bring about desirable outcomes. The theoretical framework of reinforcement learning proposes formal algorithms through which agents learn from experience to make rewarding choices. These formal models capture many aspects of reward-guided human behavior in controlled laboratory contexts. Here, we suggest that the algorithms and the constructs (i.e., states, actions, and rewards) formalized within reinforcement-learning theory can be operationally defined and extended to additionally account for learning in complex natural environments. We discuss several recent examples of empirical studies that provide evidence of signatures of reinforcement learning across diverse human behaviors in everyday environments.

A Framework for Automation in Psychotherapy
Zac E. Imel, Torrey Creed, Brent Kious, Tim Althoff, Dana Atzil-Slonim, Vivek Srikumar

Psychotherapy is a conversational intervention that has relied on humans to manage its implementation. Improvements in conversational artificial intelligence (AI) have accompanied speculation on how technologies might automate components of psychotherapy, most often the replacement of human therapists. However, there is a spectrum of opportunities for human collaboration with autonomous systems in psychotherapy, including evaluation, documentation, training, and assistance. Clarity about what is being automated is necessary to understand the affordances and limitations of specific technologies. In this article we present a framework for categories of autonomous systems in psychotherapy as a guidepost for empirical and ethical inquiry. Categories include scripted or rule-based conversations; collaborative systems in which humans are evaluated by, supervise, or are assisted by AI; and agents that generate interventions. These categories highlight considerations for key stakeholders as psychotherapy moves from unmediated human-to-human conversation to various forms of automation.

Metacognition and Uncertainty Communication in Humans and Large Language Models
Mark Steyvers, Megan A. K. Peters

Metacognition—the capacity to monitor and evaluate one’s own knowledge and performance—is foundational to human decision-making, learning, and communication. As large language models (LLMs) become increasingly embedded in both high-stakes and widespread low-stakes contexts, it is important to assess whether, how, and to what extent they exhibit metacognitive abilities. Here, we provide an overview of the current knowledge of LLMs’ metacognitive capacities, how they might be studied, and how they relate to our knowledge of metacognition in humans. We show that although humans and LLMs can sometimes appear quite aligned in their metacognitive capacities and behaviors, it is clear many differences remain; attending to these differences is important for enhancing the collaboration between humans and artificial intelligence. Last, we discuss how endowing future LLMs with more sensitive and more calibrated metacognition may also help them develop new capacities such as more efficient learning, self-direction, and curiosity.

Empathy for and From Embodied Robots: An Interdisciplinary Review
C. Daryl Cameron, Alan R. Wagner, Martina Orlandi, Eliana Hadjiandreou, India G. Oates, Stephen Anderson

Several years ago, the world was stunned when the cute robot HitchBOT was destroyed. Does empathy for robots—sharing experiences and feeling compassion—make sense for humans? How do people empathize with robots, and what are the ethical and practical implications of doing so? How do people react when robots seem to be empathizing with them? In this review, we detail empirical work on empathy for robots, discuss the ethics of extending empathy toward robots, and consider how to engineer robots that elicit empathy. We then review empirical work on empathy received from robots to explore psychological, philosophical, and engineering implications. In our final section, we suggest how interactions with robots might cultivate human empathy. Can interactions with a robot build human empathy and help it to become more resilient and reliable?

The Structure-Mapping Engine: A Multidecade Interaction Between Psychology and Artificial Intelligence
Dedre Gentner, Kenneth Forbus

This article describes the structure-mapping engine (SME) and its relation to psychological theory and research. SME was created in 1986 as a simulation of structure-mapping theory (SMT) and is still in use, both on its own and as part of larger scale simulations such as CogSketch and Companion that capture analogy’s roles in other cognitive processing. Over the 4 decades since artificial intelligence (AI) first appeared, there has been continual interaction between AI research and human research. We begin by briefly reviewing SMT and the basic construction of SME. After comparing SME with other simulations, we then describe some specific contributions of SME to our understanding of human analogical processing. We close by proposing that these psychological models can become a new technology for AI.

Does Altruism Exist? Implications of Selective Investment Theory for Solving Social Problems
Stephanie L. Brown, R. Michael Brown, David Cavallino

This article provides an overview of the debate within social psychology concerning the possible existence of altruistic motivation. After presenting the social-psychological background, we describe selective investment theory, an evolutionary theory of altruistic motivation, and discuss the underlying neurobiology. We describe evidence of the theory’s generativity within health psychology and consider its implications for solving social problems in the areas of economics, overpopulation, peace negotiations, and environmental protection.

Historical Change in Midlife Development From a Cross-National Perspective
Frank J. Infurna, Yesenia Cruz-Carrillo, Nutifafa E. Y. Dey, Markus Wettstein, Margie E. Lachman, Denis Gerstorf

We summarize empirical evidence documenting that (a) U.S. middle-aged adults have displayed historical trends of elevations in loneliness and depressive symptoms and declining memory and physical health and (b) this pattern is largely confined to the United States and not observed in peer nations. A conceptual model is provided to detail possible explanations for these historical trends. We also discuss future directions to explore whether similar historical trends are transpiring across population subgroups and low- and middle-income nations, and we identify psychosocial resources for promoting resilience. This timely article sheds light on midlife development from a cross-national and historical perspective.

Toward Complementary Intelligence: Integrating Cognitive and Machine AI
Cleotilde Gonzalez, Tailia Malloy

This article calls for complementary human-AI intelligence. Rather than redefining intelligence to fit machine capabilities, we argue for designing AI that complements and extends human cognition. We distinguish between cognitive AI, which is grounded in cognitive science to model human perception, learning, and decision-making, and machine AI, which achieves large-scale performance through data-driven optimization. Building on advances in machine learning alignment and human-AI complementarity, we propose an integrative framework that connects cognitive and machine AI across four routes: embedding integration, aligning human and machine representations; instruction encoding, using machine AI to translate goals into cognitive AI; training agents, using cognitive AI to guide and train machine AI through human-like data; and coevolving agents, enabling cognitive and machine AI to coadapt and improve together over time. These integration routes provide a foundation for complementary intelligence: systems that combine human interpretability with machine scalability and precision to enhance trust, adaptability, and human agency in complex sociotechnical environments.

Dyadic Decisions About Effort: How Caregivers Shape Young Children’s Persistence
Julia A. Leonard, Reut Shachnai

Persistence is essential for learning, but children cannot and should not persist at everything. How do young children decide what is worth their effort? We build a theory of young children’s state persistence as the outcome of a socially guided decision-making process between children and caregivers. Integrating research from metacognition, decision-making, and social learning, we show how caregivers shape two key beliefs that guide children’s effort: What children think they are capable of and whether their effort is worthwhile. Caregivers’ actions, in turn, are guided by their own beliefs about children’s abilities and the value of tasks, creating a dynamic social system of effort calibration. By reframing persistence as a dynamic coconstructed process, we uncover how motivation is built—and where it can break down.

Using Artificial Intelligence to Better Understand Human Intelligence
Gordon Pennycook, Thomas H. Costello, David G. Rand

A consistent pattern emerges from the history of psychology: Technological advances change the way that we understand ourselves. We argue that, in addition to various uses that are already common (e.g., qualitative coding), large language models can be integrated into survey software and act as a virtual research assistant that can generate tailored stimuli on the fly. This creates unprecedented flexibility in developing materials for psychological theory testing. We present an illustrative case study to show how a major lingering debate in the field—that is, whether people really change their mind according to evidence or, instead, rely on motivated reasoning—was pushed forward by using artificial intelligence (AI) to administer personalized experimental treatments. We discuss various potential uses of AI to test hypotheses in psychological science and argue that psychologists should seriously consider using AI to better understand human intelligence.

From a Baby’s Point of View: How Infants’ Face Diets Shape Their Face Perception
Charisse B. Pickron, Laurie Bayet

Individual variations in face-perception expertise become apparent by the second year of life. We propose that infants’ “face diet”—the nature and quantity of their visual interactions with faces—provides a useful lens for understanding how individual differences in face perception arise. In this article, we discuss how the diversity of an infant’s face diet and their interactions with caregivers shape their face-perception and social-learning skills, how a masked face diet may influence infants’ face perception, and how neurodiversity may affect infants’ face diets and learning about faces. These components underscore how face perception develops through both shared and individual pathways, with implications for identifying early-emerging challenges and designing supportive interventions. Future research opportunities include incorporating diverse contexts, improving measurement tools, and examining developmental periods beyond infancy.

Feedback on this article? Email [email protected] or login to comment.