Large language models (LLMs) have shown remarkable capabilities, but still struggle with processing extensive contexts, limiting their ability to maintain coherence and accuracy over long sequences. In contrast, the human brain excels at organising and retrieving episodic experiences across vast temporal scales, spanning a lifetime. In this work, we introduce EM-LLM, a novel approach that integrates key aspects of human episodic memory and event cognition into LLMs with no fine-tuning, enabling them to handle practically infinite context lengths while maintaining computational efficiency. EM-LLM organises sequences of tokens into coherent episodic events using a combination of Bayesian surprise and graph-theoretic boundary refinement in an online fashion. When needed, these events are retrieved through a two-stage memory process, combining similarity-based and temporally contiguous retrieval for efficient, human-inspired access to relevant information. Experiments on the LongBench and $\infty$-Bench benchmarks demonstrate EM-LLM's superior performance, consistently outperforming the state-of-the-art retrieval model InfLLM across various baseline LLMs. In addition, EM-LLM outperforms its popular counterpart, RAG, in a wide range of tasks, while requiring similar resources. Notably, EM-LLM's performance even surpasses full-context models in most tasks, while successfully performing retrieval across 10 million tokens -- a scale computationally infeasible for such models. Finally, our analysis reveals strong correlations between EM-LLM's event segmentation and human-perceived events, suggesting parallels between this artificial system and its biological counterpart, thereby offering a novel computational framework for exploring human memory mechanisms.

We present Langevin Predictive Coding (LPC), a novel algorithm for deep generative model learning that builds upon the predictive coding framework of computational neuroscience. By injecting Gaussian noise into the predictive coding inference procedure and incorporating an encoder network initialization, we reframe the approach as an amortized Langevin sampling method for optimizing a tight variational lower bound. To increase robustness to sampling step size, we present a lightweight preconditioning technique inspired by Riemannian Langevin methods and adaptive SGD. We compare LPC against VAEs by training generative models on benchmark datasets; our experiments demonstrate superior sample quality and faster convergence for LPC in a fraction of SGD training iterations, while matching or exceeding VAE performance across key metrics like FID, diversity and coverage.

Backpropagation has rapidly become the workhorse credit assignment algorithm for modern deep learning methods. Recently, modified forms of predictive coding (PC), an algorithm with origins in computational neuroscience, have been shown to result in approximately or exactly equal parameter updates to those under backpropagation. Due to this connection, it has been suggested that PC can act as an alternative to backpropagation with desirable properties that may facilitate implementation in neuromorphic systems. Here, we explore these claims using the different contemporary PC variants proposed in the literature. We obtain time complexity bounds for these PC variants, which we show are lower bounded by backpropagation. We also present key properties of these variants that have implications for neurobiological plausibility and their interpretations, particularly from the perspective of standard PC as a variational Bayes algorithm for latent probabilistic models. Our findings shed new light on the connection between the two learning frameworks and suggest that in its current forms, PC may have more limited potential as a direct replacement of backpropagation than previously envisioned.

Human perception and experience of time are strongly influenced by ongoing stimulation, memory of past experiences, and required task context. When paying attention to time, time experience seems to expand; when distracted, it seems to contract. When considering time based on memory, the experience may be different than what is in the moment, exemplified by sayings like "time flies when you're having fun." Experience of time also depends on the content of perceptual experience-rapidly changing or complex perceptual scenes seem longer in duration than less dynamic ones. The complexity of interactions among attention, memory, and perceptual stimulation is a likely reason that an overarching theory of time perception has been difficult to achieve. Here, we introduce a model of perceptual processing and episodic memory that makes use of hierarchical predictive coding, short-term plasticity, spatiotemporal attention, and episodic memory formation and recall, and apply this model to the problem of human time perception. In an experiment with approximately 13,000 human participants, we investigated the effects of memory, cognitive load, and stimulus content on duration reports of dynamic natural scenes up to about 1 minute long. Using our model to generate duration estimates, we compared human and model performance. Model-based estimates replicated key qualitative biases, including differences by cognitive load (attention), scene type (stimulation), and whether the judgment was made based on current or remembered experience (memory). Our work provides a comprehensive model of human time perception and a foundation for exploring the computational basis of episodic memory within a hierarchical predictive coding framework.

Human experience of time exhibits systematic, context-dependent deviations from clock time; for example, time is experienced differently at work than on holiday. Here we test the proposal that differences from clock time in subjective experience of time arise because time estimates are constructed by accumulating the same quantity that guides perception: salient events. Healthy human participants watched naturalistic, silent videos of up to 24 seconds in duration and estimated their duration while fMRI was acquired. We were able to reconstruct trial-by-trial biases in participants' duration reports, which reflect subjective experience of duration, purely from salient events in their visual cortex BOLD activity. By contrast, salient events in neither of two control regions-auditory and somatosensory cortex-were predictive of duration biases. These results held despite being able to (trivially) predict clock time from all three brain areas. Our results reveal that the information arising during perceptual processing of a dynamic environment provides a sufficient basis for reconstructing human subjective time duration.

Memorising vocabulary is an important aspect of formal foreign language learning. Advances in cognitive psychology have led to the development of adaptive learning systems that make vocabulary learning more efficient. These computer-based systems measure learning performance in real time to create optimal study strategies for individual learners. While such adaptive learning systems have been successfully applied to written word learning, they have thus far seen little application in spoken word learning. Here we present a system for adaptive, speech-based word learning. We show that it is possible to improve the efficiency of speech-based learning systems by applying a modified adaptive model that was originally developed for typing-based word learning. This finding contributes to a better understanding of the memory processes involved in speech-based word learning. Furthermore, our work provides a basis for the development of language learning applications that use real-time pronunciation assessment software to score the accuracy of the learner’s pronunciations. Speech-based learning applications are educationally relevant because they focus on what may be the most important aspect of language learning: to practice speech.

We propose a novel neural network architecture, named the Global Workspace Network (GWN), which addresses the challenge of dynamic and unspecified uncertainties in multimodal data fusion. Our GWN is a model of attention across modalities and evolving through time, and is inspired by the well-established Global Workspace Theory from the field of cognitive science. The GWN achieved average F1 score of 0.92 for discrimination between pain patients and healthy participants and average F1 score = 0.75 for further classification of three pain levels for a patient, both based on the multimodal EmoPain dataset captured from people with chronic pain and healthy people performing different types of exercise movements in unconstrained settings. In these tasks, the GWN significantly outperforms the typical fusion approach of merging by concatenation. We further provide extensive analysis of the behaviour of the GWN and its ability to address uncertainties (hidden noise) in multimodal data.

Simulation of population dynamics is a central research theme in computational biology, which contributes to understanding the interactions between predators and preys. Conventional mathematical tools of this theme, however, are incapable of accounting for several important attributes of such systems, such as the intelligent and adaptive behavior exhibited by individual agents. This unrealistic setting is often insufficient to simulate properties of population dynamics found in the real-world. In this work, we leverage multi-agent deep reinforcement learning, and we propose a new model of large-scale predator-prey ecosystems. Using different variants of our proposed environment, we show that multi-agent simulations can exhibit key real-world dynamical properties. To obtain this behavior, we firstly define a mating mechanism such that existing agents reproduce new individuals bound by the conditions of the environment. Furthermore, we incorporate a real-time evolutionary algorithm and show that reinforcement learning enhances the evolution of the agents' physical properties such as speed, attack and resilience against attacks.

Active inference is a Bayesian framework for understanding biological intelligence. The underlying theory brings together perception and action under one single imperative: minimizing free energy. However, despite its theoretical utility in explaining intelligence, computational implementations have been restricted to low-dimensional and idealized situations. In this paper, we present a neural architecture for building deep active inference agents operating in complex, continuous state-spaces using multiple forms of Monte-Carlo (MC) sampling. For this, we introduce a number of techniques, novel to active inference. These include: i) selecting free-energy-optimal policies via MC tree search, ii) approximating this optimal policy distribution via a feed-forward `habitual' network, iii) predicting future parameter belief updates using MC dropouts and, finally, iv) optimizing state transition precision (a high-end form of attention). Our approach enables agents to learn environmental dynamics efficiently, while maintaining task performance, in relation to reward-based counterparts. We illustrate this in a new toy environment, based on the dSprites data-set, and demonstrate that active inference agents automatically create disentangled representations that are apt for modeling state transitions. In a more complex Animal-AI environment, our agents (using the same neural architecture) are able to simulate future state transitions and actions (i.e., plan), to evince reward-directed navigation - despite temporary suspension of visual input. These results show that deep active inference - equipped with MC methods - provides a flexible framework to develop biologically-inspired intelligent agents, with applications in both machine learning and cognitive science.

A significant obstacle in using phenomenological models of spiking neurons for large-scale simulations is the approximation of the optimal parameters for a type of neuron, given the available experimental data. Here we show a method for optimizing the parameters of such models, based on a combination of different frequency-current and voltage-current relations of a neuron as well as known physiological properties. We also present a python toolbox which uses NeMo spiking neural network simulator and provides a fast GPU-based implementation of our method. As a benchmark, our toolbox was used to fit Izhikevich equations to neurological data obtained from a cat's thalamic relay cell. Our resulting model was able to predict the firing patterns of known membrane potential traces of this neuron, although they were not explicitly defined during training. A further comparison between this neuron model and a previous approach, when both models are used in the simulation of a generic thalamic nucleus, revealed that the distribution of neuronal avalanches is significantly different and conforms better to power law-like distributions, thus increasing the likelihood of a critical regime and the biological plausibility of the simulation.

Low-frequency oscillations have been the target of extensive research both in cortical structures and in the basal ganglia, due to numerous reports of associations with brain disorders and the normal functioning of the brain. Whereas a number of computational models of the basal ganglia investigate these phenomena, these models tend to focus on intrinsic oscillatory mechanisms, neglecting evidence that points to the cortex as the origin of this oscillatory behaviour. In this work we constructed a neural model of the basal ganglia circuitry and used it to investigate the relationship between frequency of oscillatory cortical input, dopamine and the effectiveness of the basal ganglia as an action selection device. Our simulations show the impact of the phase offset between different cortical inputs. This was found to be highly dependent on the frequency band and to have a strong influence on basal ganglia effectiveness. In addition, the level of dopamine in the system was found to modulate this effect, also depending on the input's frequency band.

This paper describes the NeuroBot system, which uses a global workspace architecture, implemented in spiking neurons, to control an avatar within the Unreal Tournament 2004 (UT2004) computer game. This system is designed to display humanlike behavior within UT2004, which provides a good environment for comparing human and embodied AI behavior without the cost and difficulty of full humanoid robots. Using a biologically inspired approach, the architecture is loosely based on theories about the high-level control circuits in the brain, and it is the first neural implementation of a global workspace that has been embodied in a complex dynamic real-time environment. NeuroBot's humanlike behavior was tested by competing in the 2011 BotPrize competition, in which human judges play UT2004 and rate the humanness of other avatars that are controlled by a human or a bot. NeuroBot came a close second, achieving a humanness rating of 36%, while the most human human reached 67%. We also developed a humanness metric that combines a number of statistical measures of an avatar's behavior into a single number. In our experiments with this metric, NeuroBot was rated as 33% human, and the most human human achieved 73%.