While Large Language Models (LLMs) have achieved remarkable success across various applications, they also raise concerns regarding self-cognition. In this paper, we perform a pioneering study to explore self-cognition in LLMs. Specifically, we first construct a pool of self-cognition instruction prompts to evaluate where an LLM exhibits self-cognition and four well-designed principles to quantify LLMs' self-cognition. Our study reveals that 4 of the 48 models on Chatbot Arena--specifically Command R, Claude3-Opus, Llama-3-70b-Instruct, and Reka-core--demonstrate some level of detectable self-cognition. We observe a positive correlation between model size, training data quality, and self-cognition level. Additionally, we also explore the utility and trustworthiness of LLM in the self-cognition state, revealing that the self-cognition state enhances some specific tasks such as creative writing and exaggeration. We believe that our work can serve as an inspiration for further research to study the self-cognition in LLMs.
Cognitive processes are realized across an extraordinary range of natural, artificial, and hybrid systems, yet there is no unified framework for comparing their forms, limits, and unrealized possibilities. Here, we propose a cognition space approach that replaces narrow, substrate-dependent definitions with a comparative representation based on organizational and informational dimensions. Within this framework, cognition is treated as a graded capacity to sense, process, and act upon information, allowing systems as diverse as cells, brains, artificial agents, and human-AI collectives to be analyzed within a common conceptual landscape. We introduce and examine three cognition spaces -- basal aneural, neural, and human-AI hybrid -- and show that their occupation is highly uneven, with clusters of realized systems separated by large unoccupied regions. We argue that these voids are not accidental but reflect evolutionary contingencies, physical constraints, and design limitations. By focusing on the structure of cognition spaces rather than on categorical definitions, this approach clarifies the diversity of existing cognitive systems and highlights hybrid cognition as a promising f
What fundamental research questions are essential for advancing toward brain-inspired AI or AGI capable of performing any intellectual task a human can? We believe the key question today is the relationship between cognition and computation (RCC). For example, the widely discussed question "Will artificial intelligence replace the human mind?" is, in essence and in scientific terms, an issue concerning RCC. We have chosen to classify RCC into four categories: 1. The relationship between the primitives of cognition and the primitives of computation. 2. The relationship between the anatomical structure of neural representation of cognition and the computational architecture of artificial intelligence. 3. The relationship between emergents in cognition and emergents in computation. 4. The relationship between the mathematical foundations of cognition and computation. The cumulative empirical evidence and theoretical analyses led us to formulate the "Global-first" principle, which highlights the contrast between "Global-first" cognition and local-first computation in RCC, offering a specific and well-defined starting point for understanding RCC.
LLMs are widely regarded as alien intelligences, systems whose cognitive operations are fundamentally unlike our own. Apparent similarities to human cognition are therefore often seen as the result of anthropomorphic projection. We argue that this framing is mistaken. LLMs clearly differ from humans in important respects, including their physical substrate, learning history, and the environments with which they interact. These differences make it all the more striking that contemporary LLM-based systems converge with human cognition on a number of principles of cognitive organization with longstanding support in cognitive science. We identify structural correspondences across five dimensions: inferential organization, computational architecture, representational structure, prediction-driven learning, and reinforcement-learning-like mechanisms supporting goal-directed action. These correspondences support a broader model of intelligent cognition in which core principles long used to explain human intelligence also characterize contemporary LLM-based systems.
Recent advancements in video generation have witnessed significant progress, especially with the rapid advancement of diffusion models. Despite this, their deficiencies in physical cognition have gradually received widespread attention - generated content often violates the fundamental laws of physics, falling into the dilemma of ''visual realism but physical absurdity". Researchers began to increasingly recognize the importance of physical fidelity in video generation and attempted to integrate heuristic physical cognition such as motion representations and physical knowledge into generative systems to simulate real-world dynamic scenarios. Considering the lack of a systematic overview in this field, this survey aims to provide a comprehensive summary of architecture designs and their applications to fill this gap. Specifically, we discuss and organize the evolutionary process of physical cognition in video generation from a cognitive science perspective, while proposing a three-tier taxonomy: 1) basic schema perception for generation, 2) passive cognition of physical knowledge for generation, and 3) active cognition for world simulation, encompassing state-of-the-art methods, cla
Various effects in human cognition, often considered `non-classical', have been argued to be most naturally modelled by quantum-like models of decision making. We extend this approach to describe models of cognition and decision-making in general probabilistic process theories, which include both classical probabilistic models and quantum instrument models as special cases. We show how many aspects of quantum-like cognition can be described diagrammatically in process theories, before using our approach to assess the arguments for quantum-like models. While standard Bayesian classical models are insufficient, we prove that any sequential decision data can in fact be given a more general form of classical instrument model, and see that even simple deterministic models can exhibit all cognitive effects. Restricting attention to instruments induced by measurements, such as classical Bayesian and quantum POVM models, rules out such a result, but is challenged by the fact that such instruments cannot account for certain effects. Finally, we argue that to strictly rule out classical instrument models one should make use of parallel composition in the modelling of joint decisions, and fin
Cognition is a core part of and a common topic among philosophy of mind, psychology, neuroscience, AI, and cognitive science. Through a mechanistic lens, I propose a framework of defining, modeling, and analyzing cognition mechanisms. Firstly, appropriate terms are introduced and used in explanations related to the framework and within the definition of a mechanism. I implicitly contend that this terminology essentially characterizes a conceptual world required for discussions in this paper. Secondly, a mathematical model of a mechanism based on directed graphs is proposed. Thirdly, the definition of a base necessary for a mechanism to be classified as a cognition mechanism is proposed. I argue that the cognition base has the features of the cognition self of humans. Fourthly, three ways to mechanistically look at mechanisms is defined and specific instances of them are suggested. Fifthly, standards for visualization and presentation of mechanisms, cognition mechanisms, and the instances to mechanistically look at them are suggested and used to analyze cognition mechanisms through appropriate examples. Finally, the features of this paper are discussed and prospects of further devel
AI governance still tends to assume that the relevant object is a bounded model or a bounded agent. That assumption is getting weaker. Real systems increasingly distribute cognition across models, tools, humans, context stores, retrieval layers, runtime policies, authorization boundaries, and delegated institutional roles. In such systems, the central governance problem is no longer only what the system did, but whether later institutions can still read what the system was. This paper introduces the khipu problem for distributed AI: the record can survive while the reading practice needed to interpret it decays. Logs, traces, model versions, tool calls, outputs, and approval artifacts may remain available while the institutional capacity to read them as parts of one coherent cognitive episode disappears. We argue that this failure is better understood as loss of interpretive continuity than as ordinary lack of observability. The result is a distinct governance failure. Institutions must classify, trust, audit, and constrain systems whose relevant identity is distributed across components and whose legibility depends on surrounding interpretive scaffolding. The problem is not merely
Many of the most consequential dynamics in human cognition occur \emph{before} events become explicit: before decisions are finalized, emotions are labeled, or meanings stabilize into narrative form. These pre-event states are characterized by ambiguity, contextual tension, and competing latent interpretations. Rogue Variable Theory (RVT) formalizes such states as \emph{Rogue Variables}: structured, pre-event cognitive configurations that influence outcomes while remaining unresolved or incompatible with a system's current representational manifold. We present a quantum-consistent information-theoretic implementation of RVT based on a time-indexed \emph{Mirrored Personal Graph} (MPG) embedded into a fixed graph Hilbert space, a normalized \emph{Quantum MPG State} (QMS) constructed from node and edge metrics under context, Hamiltonian dynamics derived from graph couplings, and an error-weighted `rogue operator'' whose principal eigenvectors identify rogue factor directions and candidate Rogue Variable segments. We further introduce a \emph{Rosetta Stone Layer} (RSL) that maps user-specific latent factor coordinates into a shared reference Hilbert space to enable cross-user compariso
This article proposes a formal rapprochement between cognitive load theory and embodied cognition by reconceptualizing psychological representations as dynamic multiscale attractors within a temporal-hierarchical prediction architecture. The apparent conflict between the two theories dissolves when viewed through a complex systems lens. Cognitive load theory describes compressed representations operating at medium timescales, while embodied cognition describes fast sensorimotor loops. These two theories describe complementary, timescale-separated processes that operate simultaneously without contradiction. Drawing on dynamical systems theory, hierarchical predictive processing, and a six-node open-systems architecture, the article proposes that learning is best understood as attractor sculpting across coupled temporal layers, from millisecond sensorimotor loops through seconds-to-minutes working memory compression to the slow, years-long reshaping of knowledge structures. Three theoretical reconciliations are developed: time-scale separation, spatially extended hierarchies, and developmental trajectories from novice to expert configurations. From these understandings, five novel, t
Recent studies on audio models show brain-tuning - fine-tuning models to better predict corresponding fMRI activity - improves brain alignment and increases performance on downstream semantic and audio tasks. We extend this approach to a multimodal audio-video model to enhance social cognition, targeting the Superior Temporal Sulcus (STS), a key region for social processing, while subjects watch Friends. We find significant increases in brain alignment to the STS and an adjacent ROI, as well as improvements to a social cognition task related to the training data - sarcasm detection in sitcoms. In summary, our study extends brain-tuning to the multi-modal domain, demonstrating improvements to a downstream task after tuning to a relevant functional region.
4E views of cognition seek to replace many of the long-held assumptions of tra- ditional cognitive science. One of the most radical shifts is the rejection of the sandwich model of cognition [8], which holds that mental processes are located be- tween action and perception. Subversion of such a long-held assumption requires an accessible theoretical alternative with firm experimental support. One unifying thread among the emerging 4E camps is their shared insistence that sensorimotor contingencies (SMCs) are such an alternative.
Complex adaptive agents consistently achieve their goals by solving problems that seem to require an understanding of causal information, information pertaining to the causal relationships that exist among elements of combined agent-environment systems. Causal cognition studies and describes the main characteristics of causal learning and reasoning in human and non-human animals, offering a conceptual framework to discuss cognitive performances based on the level of apparent causal understanding of a task. Despite the use of formal intervention-based models of causality, including causal Bayesian networks, psychological and behavioural research on causal cognition does not yet offer a computational account that operationalises how agents acquire a causal understanding of the world. Machine and reinforcement learning research on causality, especially involving disentanglement as a candidate process to build causal representations, represent on the one hand a concrete attempt at designing causal artificial agents that can shed light on the inner workings of natural causal cognition. In this work, we connect these two areas of research to build a unifying framework for causal cognitio
The interplay between cognition and gaming, notably through educational games enhancing cognitive skills, has garnered significant attention in recent years. This research introduces the CogSimulator, a novel algorithm for simulating user cognition in small-group settings with minimal data, as the educational game Wordle exemplifies. The CogSimulator employs Wasserstein-1 distance and coordinates search optimization for hyperparameter tuning, enabling precise few-shot predictions in new game scenarios. Comparative experiments with the Wordle dataset illustrate that our model surpasses most conventional machine learning models in mean Wasserstein-1 distance, mean squared error, and mean accuracy, showcasing its efficacy in cognitive enhancement through tailored game design.
Recent advancements have explored agentic zero-shot 3D understanding by reformulating it as video keyframe understanding with Multimodal Large Language Models (MLLMs). However, existing methods face an intrinsic bottleneck due to the finite observation perspectives inherent in videos and the implicit perception of 3D scenes. In this paper, we propose a collaborative multi-agent framework that assigns a Planning Agent to handle high-level viewpoint planning and supplement novel perspectives, and a Perception Agent to explicitly summarize the 3D scene into a structured holistic cognitive map. Specifically, Planning Agent first analyzes this cognitive map to determine query-relevant viewpoints and supplements missing critical perspectives to ensure comprehensive observation. Subsequently, Perception Agent documents object-level attributes from these views by assigning consistent instance identifiers across viewpoints, thereby integrating fragmented observations into the holistic cognitive map. In parallel, it provides feedback to filter out mismatched candidate objects and guide subsequent viewpoint planning. Through this closed-loop iterative process, two agents collaboratively figur
A model of the evolution of cognition is used to derive a Requirement Equation (RE), which defines what computations the fittest possible brain must make, or must choose actions as if it had made those computations. The terms in the RE depend on factors outside an animals brain, which can be modelled without making assumptions about how the brain works, from knowledge of the animals habitat and biology. In simple domains where the choices of actions have small information content, it may not be necessary to build internal models of reality; short cut computations may be just as good at choosing actions. In complex domains such as 3D spatial cognition, which underpins many complex choices of action, the RE implies that brains build Bayesian internal models of the animals surroundings; and that the models are constrained to be true to external reality.
A pervasive research protocol of cognitive neuroscience is to train subjects to perform deliberately designed experiments and record brain activity simultaneously, aiming to understand the brain mechanism underlying cognition. However, how the results of this protocol can be applied in technology is seldom discussed. Here, I review the studies on time processing of the brain as examples of this protocol, as well as two main application areas of neuroscience (neuroengineering and brain-inspired artificial intelligence). Time processing is an indispensable dimension of cognition; time is also an indispensable dimension of any real-world signal to be processed in technology. So one may expect that the studies of time processing in cognition profoundly influence brain-related technology. Surprisingly, I found that the results from cognitive studies on timing processing are hardly helpful in solving practical problems. This awkward situation may be due to the lack of generalizability of the results of cognitive studies, which are under well-controlled laboratory conditions, to real-life situations. This lack of generalizability may be rooted in the fundamental unknowability of the world
Cognition is not passive data accumulation but the active resolution of uncertainty through symmetry breaking. This paper argues that both cognitive evolution and development unfold via sequential symmetry-breaking transitions that disrupt innate regularities across space, time, self, and representation. First, spatial symmetry is broken through bilateral body plans and neural codes like grid and place cells, which privilege egocentric orientation and localized encoding. Second, reinforcement learning introduces temporal asymmetry by favoring future rewards, establishing a directional flow of inference. Third, goal-directed simulation breaks spatiotemporal symmetry between internal self-models and the external world, enabling embodied inference and solving the combinatorial search problem. Fourth, social cognition via mentalizing and imitation breaks the symmetry between minds, allowing agents to infer others' beliefs. Finally, language imposes a linear, recursive structure onto unordered thought, breaking expressive symmetry through syntax and grammar. These asymmetries are unified by the Context-Content Uncertainty Principle (CCUP), which frames cognition as a cyclical entropy-mi
A large body of compelling evidence has been accumulated demonstrating that embodiment - the agent's physical setup, including its shape, materials, sensors and actuators - is constitutive for any form of cognition and as a consequence, models of cognition need to be embodied. In contrast to methods from empirical sciences to study cognition, robots can be freely manipulated and virtually all key variables of their embodiment and control programs can be systematically varied. As such, they provide an extremely powerful tool of investigation. We present a robotic bottom-up or developmental approach, focusing on three stages: (a) low-level behaviors like walking and reflexes, (b) learning regularities in sensorimotor spaces, and (c) human-like cognition. We also show that robotic based research is not only a productive path to deepening our understanding of cognition, but that robots can strongly benefit from human-like cognition in order to become more autonomous, robust, resilient, and safe.
The article develops a generative model of the human translating mind, grounded in empirical translation process data. It posits that three embedded processing layers unfold concurrently in the human mind, and their traces are detectable in behavioral data: sequences of routinized/automated processes are observable in fluent translation production, cognitive/reflective thoughts lead to longer keystroke pauses, while affective/emotional states may be identified through characteristic typing and gazing patterns. Utilizing data from the CRITT Translation Process Research Database (TPR-DB), the article illustrates how the temporal structure of keystroke and gaze data can be related to the three assumed hidden mental processing strata. The article relates this embedded generative model to various theoretical frameworks, dual-process theories and Robinson's (2023) ideosomatic theory of translation, opening exciting new theoretical horizons for Cognitive Translation Studies, grounded in empirical data and evaluation.