While consciousness has been historically a heavily debated topic, awareness had less success in raising the interest of scholars. However, more and more researchers are getting interested in answering questions concerning what awareness is and how it can be artificially generated. The landscape is rapidly evolving, with multiple voices and interpretations of the concept being conceived and techniques being developed. The goal of this paper is to summarize and discuss the ones among these voices connected with projects funded by the EIC Pathfinder Challenge “Awareness Inside” callwithin Horizon Europe, designed specifically for fostering research on natural and synthetic awareness. In this perspective, we dedicate special attention to challenges and promises of applying synthetic awareness in robotics, as the development of mature techniques in this new field is expected to have a special impact on generating more capable and trustworthy embodied systems.

In the context of cybernetic systems and the future of human-AI collaboration, intelligent technologies are increasingly conceived as living machines: systems that regulate themselves through feedback while co-evolving with humans. Unlike traditional automation, these systems do not merely optimize fixed goals—they must continuously adapt in response to user behavior while preserving human agency, intent, and oversight. This paper presents the design and development of a Clinical Decision Support System (CDSS) for stroke neurorehabilitation that operationalizes these principles. While no experimental trials are reported in this manuscript, the system architecture and implementation are described in detail, and validation is planned through a multicenter randomized controlled trial (RCT). By embedding a feedback-driven, clinician-guided control loop into therapy delivery. Integrated within the Rehabilitation Gaming System (RGS), a virtual reality platform for adaptive motor and cognitive training, the CDSS maintains a dynamic model of each patient’s state, derived from behavioral performance, affective self-reports, and adherence patterns. This model drives the adaptive selection of therapeutic activities, forming structured, modifiable care plans. The system enables clinicians to inspect, adjust and refine interventions through a transparent interface, ensuring that algorithmic adaptation remains aligned with expert judgment and patient-specific needs. The CDSS exemplifies a shift from automation to co-regulation by supporting mutual adaptation between human and machine. It represents a concrete step toward the design of cybernetic healthcare systems, those capable of sustaining meaningful, personalized interactions over time, in service of complex and evolving recovery goals.

Collaboration in synthetic systems can inform our understanding of collaboration as a natural phenomenon. A general overview of how collaboration has been studied in the human evolutionary behavioral sciences is presented, and it is argued that further synthesis is needed across the different levels of explanation. At present, two main issues prevent such a synthesis: (a) the current gap between proximate accounts of collaboration from the cognitive sciences and the ultimate levels of explanation from the evolutionary sciences, and (b) methodological limitations which inhibit joint study of collaboration at different levels of description. Synthetic collaborative systems (e.g., robotics and artificial intelligence) can help to address these issues. A unified research program centered on synthetic collaboration provides a way to expand understanding of human and animal collaboration, to design and study human-machine collaboration, as well as to investigate purely synthetic forms of collaboration between intelligent machines. Here, current research is reviewed that employs synthetic methodologies across different fields, the implications of developing synthetic collaborative agents are discussed, and an approach is proposed to study both natural and synthetic collaboration, under the name of collaborative cybernetics. Science requires an almost complete openness to all ideas. On the other hand, it requires the most rigorous and uncompromising skepticism. -Carl Sagan (1995)

The dynamics of the human cortex are highly metastable, driving the spontaneous exploration of network states. This metastability depends on circuit-level edge-of-bifurcation dynamics, which emerge from firing-rate control through multiple mechanisms of excitatory-inhibitory (E-I) homeostasis. However, it is unclear how these contribute to the metastability of cortical networks. We propose that individual mechanisms of E-I homeostasis contribute uniquely to the emergence of resting-state dynamics and test this hypothesis in a large-scale model of the human cortex. We show that empirical connectivity and dynamics can only be reproduced when accounting for multiple mechanisms of E-I homeostasis. More specifically, while the homeostasis of excitation and inhibition enhances metastability, the regulation of intrinsic excitability ensures moderate synchrony, maximizing functional complexity. Furthermore, the modulation bifurcation modulation by the homeostasis of excitation and intrinsic excitability compensates for strong input fluctuations in connector hubs. Importantly, this only occurs in models accounting for local gamma oscillations, suggesting a relationship between E-I balance, gamma rhythms, and metastable dynamics. Altogether, our results show that cortical networks self-organize toward maximal metastability through the multi-factor homeostasis of E-I balance. Therefore, the benefits of combining multiple homeostatic mechanisms transcend the circuit level, supporting the metastable dynamics of large-scale cortical networks.

Although the primary function of excitatory-inhibitory (E-I) homeostasis is the maintenance of mean firing rates, the conjugation of multiple homeostatic mechanisms is thought to be pivotal to ensuring edge-of-bifurcation dynamics in cortical circuits. However, computational studies on E-I homeostasis have focused solely on the plasticity of inhibition, neglecting the impact of different modes of E-I homeostasis on cortical dynamics. Therefore, we investigate how the diverse mechanisms of E-I homeostasis employed by cortical networks shape oscillations and edge-of-bifurcation dynamics. Using the Wilson-Cowan model, we explore how distinct modes of E-I homeostasis maintain stable firing rates in models with varying levels of input and how it affects circuit dynamics. Our results confirm that E-I homeostasis can be leveraged to control edge-of-bifurcation dynamics and that some modes of homeostasis maintain mean firing rates under higher levels of input by modulating the distance to the bifurcation. Additionally, relying on multiple modes of homeostasis ensures stable activity while keeping oscillation frequencies within a physiological range. Our findings tie relevant features of cortical networks, such as E-I balance, the generation of gamma oscillations, and edge-of-bifurcation dynamics, under the framework of firing-rate homeostasis, providing a mechanistic explanation for the heterogeneity in the distance to the bifurcation found across cortical areas. In addition, we reveal the functional benefits of relying upon different homeostatic mechanisms, providing a robust method to regulate network dynamics with minimal perturbation to the generation of gamma rhythms and explaining the correlation between inhibition and gamma frequencies found in cortical networks.

The mammalian hippocampal formation plays a critical role in efficient and flexible navigation. Hippocampal place cells exhibit spatial tuning, characterized by increased firing rates when an animal occupies specific locations in its environment. The mechanisms underlying the encoding of spatial information by hippocampal place cells remain not fully resolved. Evidence suggests that spatial preferences are shaped by multimodal sensory inputs. Yet, existing hippocampal-inspired models typically rely on a single sensory information source. Here, we developed a hippocampus-inspired model that combines motivational and spatial encoding and is based on the fundamental principle of biological autonomy that behavior serves a purpose. That is, in foraging tasks, an agent’s trajectories must be deployed considering the fact that the reward value of environmental stimuli is tied to the agent’s motivational state. In this paper, we introduce a "motivational hippocampal autoencoder" (MoHA) that integrates both interoceptive (motivational) and exteroceptive (visual) information. The MoHA model reproduces hippocampal firing correlates for different motivational states. We show that the representations of MoHA allow a synthetic agent to learn and deploy efficient trajectories in a foraging task, laying the foundation for self-regulated multipurpose reinforcement learning.

The pursuit of artificial consciousness requires conceptual clarity to navigate its theoretical and empirical challenges. This paper introduces a composite, multilevel, and multidimensional model of consciousness as a heuristic framework to guide research in this field. Consciousness is treated as a complex phenomenon, with distinct constituents and dimensions that can be operationalized for study and for evaluating their replication. We argue that this model provides a balanced approach to artificial consciousness research by avoiding binary thinking (e.g., conscious vs. non-conscious) and offering a structured basis for testable hypotheses. To illustrate its utility, we focus on "awareness" as a case study, demonstrating how specific dimensions of consciousness can be pragmatically analyzed and targeted for potential artificial instantiation. By breaking down the conceptual intricacies of consciousness and aligning them with practical research goals, this paper lays the groundwork for a robust strategy to advance the scientific and technical understanding of artificial consciousness.

The mammalian hippocampal formation plays a critical role in efficient and flexible navigation. Hippocampal place cells exhibit spatial tuning, characterized by increased firing rates when an animal occupies specific locations in its environment. However, the mechanisms underlying the encoding of spatial information by hippocampal place cells remain not fully understood. Evidence suggests that spatial preferences are shaped by multimodal sensory inputs. Yet, existing hippocampal models typically rely on a single sensory modality, overlooking the role of interoceptive information in the formation of cognitive maps. In this paper, we introduce the Motivational Hippocampal Autoencoder (MoHA), a biologically inspired model that integrates interoceptive (motivational) and exteroceptive (visual) information to generate motivationally modulated cognitive maps. MoHA captures key hippocampal firing properties across different motivational states and, when embedded in a reinforcement learning agent, generates adaptive internal representations that drive goal-directed foraging behavior. Grounded in the principle of biological autonomy, MoHA enables the agent to dynamically adjust its navigation strategies based on internal drives, ensuring that behavior remains flexible and context-dependent. Our results show the benefits of integrating motivational cognitive maps into artificial agents with a varying set of goals, laying the foundation for self-regulated multi-objective reinforcement learning. ### Competing Interest Statement The authors have declared no competing interest.

From weaving spiders to hibernating mammals and migratory birds, nature presents numerous examples of organisms exhibiting extraordinary autonomous behaviors that ensure their self-maintenance. However, physiological needs often interact and compete. This requires living organisms to handle them as a coordinated system of internal needs rather than as isolated subsystems. We present an artificial agent equipped with a neural mass model replicating fundamental self-regulatory behaviors observed in desert lizards. Our results demonstrate that this agent not only autonomously regulates its internal temperature by navigating to areas with optimal environmental conditions, but also harmonizes this process with other internal needs, such as energy, hydration, security, and mating. This biomimetic agent outperforms a control agent lacking interoceptive awareness in terms of efficiency, fairness, and stability. Additionally, to demonstrate the flexibility of our framework, we develop a ”cautious” agent that prioritizes security over other needs, achieving a Maslow-like hierarchical organization of internal needs. Together, our findings suggest that grounding robot behavior in biological principles of self-regulation provides a robust framework for designing multipurpose, intrinsically motivated agents capable of resolving trade-offs in dynamic environments.

BrainX3 represents an integrated neuroinformatic platform designed to facilitate the effective visualisation, analysis, semantic mapping, and simulation of whole brain dynamics. As the software evolves, emphasis is placed on both studying biological and creating artificial nervous systems; advancing to the needs of researchers. We present the role of BrainX3 in facilitating a comprehensive understanding and exploration of brain data. Through a detailed exploration of its capabilities, we demonstrate how BrainX3 serves as a tool for understanding the brain's functioning. By bridging the gap between empirical data and computational models, BrainX3 empowers users to delve deeper into the complexities of the brain, ultimately advancing our understanding and contributing to the development of living machines and artificial brains. We also highlight recent advancements in the software, with a particular emphasis on the modeling of artificial brains.

As populations worldwide show increasing levels of stress, understanding emerging links among stress, inflammation, cognition, and behavior is vital to human and planetary health. We hypothesize that inflammation is a multiscale driver connecting stressors that affect individuals to large-scale societal dysfunction and, ultimately, to planetary-scale environmental impacts. We propose a “central inflammation map” hypothesis to explain how the brain regulates inflammation and how inflammation impairs cognition, emotion, and action. According to our hypothesis, these interdependent inflammatory and neural processes, and the inter-individual transmission of environmental, infectious, and behavioral stressors—amplified via high-throughput digital global communications—can culminate in a multiscale, runaway, feed-forward process that could detrimentally affect human decision-making and behavior at scale, ultimately impairing the ability to address these same stressors. This perspective could provide non-intuitive explanations for behaviors and relationships among cells, organisms, and communities of organisms, potentially including population-level responses to stressors as diverse as global climate change, conflicts, and the COVID-19 pandemic. To illustrate our hypothesis and elucidate its mechanistic underpinnings, we present a mathematical model applicable to the individual and societal levels to test the links among stress, inflammation, control, and healing, including the implications of transmission, intervention (e.g., via lifestyle modification or medication), and resilience. Future research is needed to validate the model’s assumptions and conclusions against empirical benchmarks and to expand the factors/variables employed. Our model illustrates the need for multilayered, multiscale stress mitigation interventions, including lifestyle measures, precision therapeutics, and human ecosystem design. Our analysis shows the need for a coordinated, interdisciplinary, international research effort to understand the multiscale nature of stress. Doing so would inform the creation of interventions that improve individuals’ lives; enhance communities’ resilience to stress; and mitigate the adverse effects of stress on the world.

State of the art deep reinforcement learning algorithms are sample inefficient due to the large number of episodes they require to achieve asymptotic performance. Episodic Reinforcement Learning (ERL) algorithms, inspired by the mammalian hippocampus, typically use extended memory systems to bootstrap learning from past events to overcome this sample-inefficiency problem. However, such memory augmentations are often used as mere buffers, from which isolated past experiences are drawn to learn from in an offline fashion (e.g., replay). Here, we demonstrate that including a bias in the acquired memory content derived from the order of episodic sampling improves both the sample and memory efficiency of an episodic control algorithm. We test our Sequential Episodic Control (SEC) model in a foraging task to show that storing and using integrated episodes as event sequences leads to faster learning with fewer memory requirements as opposed to a standard ERL benchmark, Model-Free Episodic Control, that buffers isolated events only. We also study the effect of memory constraints and forgetting on the sequential and non-sequential version of the SEC algorithm. Furthermore, we discuss how a hippocampal-like fast memory system could bootstrap slow cortical and subcortical learning subserving habit formation in the mammalian brain.

The hippocampus has been associated with both spatial cognition and episodic memory formation, but integrating these functions into a unified framework remains challenging. Here, we demonstrate that forming discrete memories of visual events in sparse autoencoder neurons can produce spatial tuning similar to hippocampal place cells. We then show that the resulting very high-dimensional code enables neurons to discretize and tile the underlying image space with minimal overlap. Additionally, we extend our results to the auditory domain, showing that neurons similarly tile the frequency space in an experience-dependent manner. Lastly, we show that reinforcement learning agents can effectively perform various visuo-spatial cognitive tasks using these sparse, very high-dimensional representations.

This paper introduces DAC-HRC, a novel cognitive architecture designed to optimize human-robot collaboration (HRC) in industrial settings, particularly within the context of Industry 4.0. The architecture is grounded in the Distributed Adaptive Control theory and the principles of joint intentionality and interdependence, which are key to effective HRC. Joint intentionality refers to the shared goals and mutual understanding between a human and a robot, while interdependence emphasizes the reliance on each other’s capabilities to complete tasks. DAC-HRC is applied to a hybrid recycling plant for the disassembly and recycling of Waste Electrical and Electronic Equipment (WEEE) devices. The architecture incorporates several cognitive modules operating at different timescales and abstraction levels, fostering adaptive collaboration that is personalized to each human user. The effectiveness of DAC-HRC is demonstrated through several pilot studies, showcasing functionalities such as turn-taking interaction, personalized error-handling mechanisms, adaptive safety measures, and gesture-based communication. These features enhance human-robot collaboration in the recycling plant by promoting real-time robot adaptation to human needs and preferences. The DAC-HRC architecture aims to contribute to the development of a new HRC paradigm by paving the way for more seamless and efficient collaboration in Industry 4.0 by relying on socially adept cognitive architectures.