What It Feels Like to Be an Agent: Prerequisites, Core Mechanisms, and Cognitive Foundations

The Topic is temporary closed. We look forward to open it again in due time.

--


An agent is fundamentally defined by its capacity for autonomous, goal-directed behaviour within a dynamic environment. Despite the ubiquitous development of “agentic” systems across various domains such as cognitive science, philosophy, and human-robot interaction, a systematic and rigorous taxonomy of what an “agent” is remains fragmented. The term is applied liberally and inconsistently, describing everything from a thermostat regulating temperature, to a reinforcement-learning policy optimising a reward, to a large language model orchestrating tools, to a human deliberating over a moral choice. Each field imports its own assumptions about autonomy, intentionality, and goal-directedness, yet these are rarely made explicit or reconciled across disciplines, leaving little common ground for comparison or cumulative progress.

This fragmentation has real consequences. Without shared criteria, it is difficult to determine which properties are truly necessary for agency, which are merely sufficient, and which reflect nothing more than an observer’s tendency to attribute intentions to complex behaviour. Establishing a principled, interdisciplinary account of what a system must have, and must do, to qualify as an agent is therefore both a foundational scientific challenge and a timely practical one.

This Research Topic seeks to establish a potentially unified framework detailing the minimal structural prerequisites, core mechanisms, and cognitive foundations required for a system to function as a true agent. It aims to bring together interdisciplinary contributions addressing not only what an agent is, but also conditions that promote the perception of a system as a true agent. Ultimately, this Research Topic aspires to contribute to the development of an integrative view about what constitutionally and functionally defines an agent.

Areas of interest welcome contributions from the following, but not limited, domains:

- Artificial intelligence, human-robot interaction, and autonomous systems: how to move beyond basic reactive systems, such as standard chatbots or simple robotic devices, to fully autonomous, goal-directed AI and robotic agents.

- Cognitive Science and Computational Neuroscience: how to determine the simplest biological and/or computational systems that qualify as agents

- Comparative Cognition: comparing agent capabilities across humans, animals, and machines.

This Research Topic welcomes Original Research articles, Systematic Reviews, Meta-Analyses, Mini Reviews, Longitudinal Studies, Cross-Cultural Research, Theoretical Contributions and Perspectives addressing the cognitive and philosophical foundations of what it feels like to be an agent.