Despite significant advancements in deep learning, artificial neural networks (ANNs) still suffer from catastrophic forgetting in continual learning, where training on new tasks causes them to easily forget previously learned information. In contrast, the human brain retains diverse information through declarative and nondeclarative memory systems ([Bear et al., 2020, Figure 24.1, p. 838]), storing it in either short-term or long-term memory. A key factor that protects humans from drastic forgetting is thought to be the reactivation of neural activity patterns representing previous experiences--referred to as memory replay (Wilson and McNaughton [1994], Rasch and Born [2007], Oudiette and Paller [2013], Van de Ven et al. [2016]). To address catastrophic forgetting in ANNs, previous works have attempted to mimic the brain's memory replay mechanism. Notably, studies such as Van de Ven et al. [2020], Millichamp and Chen [2021], Ran et al. [2024] have demonstrated that brain-inspired mechanisms can help retain performance during continual learning in AI. Motivated by these findings, we aim to draw inspiration from the brain to develop mechanisms for long-term memory in AI. Specifically, we focus on analyzing the impact of brain-inspired components on AI performance and providing insights to guide future research directions. 1

During periods of quiescence, such as sleep, neural activity in many brain circuits resembles that observed during periods of task engagement. However, the precise conditions under which task-optimized networks can autonomously reactivate the same network states responsible for online behavior is poorly understood. In this study, we develop a mathematical framework that outlines sufficient conditions for the emergence of neural reactivation in circuits that encode features of smoothly varying stimuli. We demonstrate mathematically that noisy recurrent networks optimized to track environmental state variables using change-based sensory information naturally develop denoising dynamics, which, in the absence of input, cause the network to revisit state configurations observed during periods of online activity. We validate our findings using numerical experiments on two canonical neuroscience tasks: spatial position estimation based on self-motion cues, and head direction estimation based on angular velocity cues. Overall, our work provides theoretical support for modeling offline reactivation as an emergent consequence of task optimization in noisy neural circuits.

The extremely limited working memory span, typically around four items, contrasts sharply with our everyday experience of processing much larger streams of sensory information concurrently. This disparity suggests that working memory can organize information into compact representations such as chunks, yet the underlying neural mechanisms remain largely unknown. Here, we propose a recurrent neural network model for chunking within the framework of the synaptic theory of working memory. We showed that by selectively suppressing groups of stimuli, the network can maintain and retrieve the stimuli in chunks, hence exceeding the basic capacity. Moreover, we show that our model can dynamically construct hierarchical representations within working memory through hierarchical chunking. A consequence of this proposed mechanism is a new limit on the number of items that can be stored and subsequently retrieved from working memory, depending only on the basic working memory capacity when chunking is not invoked. Predictions from our model were confirmed by analyzing single-unit responses in epileptic patients and memory experiments with verbal material. Our work provides a novel conceptual and analytical framework for understanding the on-the-fly organization of information in the brain that is crucial for cognition.

People are better at recalling new names and faces if they are played recorded prompts for them while they enjoy good quality sleep, a study has found. Northwestern University experts explored how quality of slumber affects'targeted reactivation' -- a process used to enhance memory consolidation during sleep. In tests involving subjects trying to learn 80 new peoples' names, recorded prompts played during deep sleep improved subsequent recall by 1.5 names on average. However, the benefits of this memory reactivation process were only seen when the subjects had good quality, undisturbed sleep, the researchers noted. It is possible reactivation may even be detrimental to recall if used with interrupted sleep, the team added -- potentially offering a way to weaken unwanted memories.

South Korea's infectious disease experts said Thursday that dead virus fragments were the likely cause of over 260 people here testing positive again for the novel coronavirus days and even weeks after marking full recoveries. Oh Myoung-don, who leads the central clinical committee for emerging disease control, said the committee members found little reason to believe that those cases could be COVID-19 reinfections or reactivations, which would have made global efforts to contain the virus much more daunting. "The tests detected the ribonucleic acid of the dead virus," said Oh, a Seoul National University hospital doctor, at a press conference Thursday held at the National Medical Center. He went on to explain that in PCR tests, or polymerase chain reaction tests, used for COVID-19 diagnosis, genetic materials of the virus amplify during testing, whether it is from a live virus or just from fragments of dead virus cells that can take months to clear from recovered patients. The PCR tests cannot distinguish whether the virus is alive or dead, he added, and this can lead to false positives.

During sleep and awake rest, the hippocampus replays sequences of place cells that have been activated during prior experiences. These have been interpreted as a memory consolidation process, but recent results suggest a possible interpretation in terms of reinforcement learning. The Dyna reinforcement learning algorithms use off-line replays to improve learning. Under limited replay budget, a prioritized sweeping approach, which requires a model of the transitions to the predecessors, can be used to improve performance. We investigate whether such algorithms can explain the experimentally observed replays. We propose a neural network version of prioritized sweeping Q-learning, for which we developed a growing multiple expert algorithm, able to cope with multiple predecessors. The resulting architecture is able to improve the learning of simulated agents confronted to a navigation task. We predict that, in animals, learning the world model should occur during rest periods, and that the corresponding replays should be shuffled.

One week ago, we reported that Microsoft's first foray into Twitter chat "artificial intelligence" did not quite work as expected: once unleashed into the wild, the chat robot named "Tay" proceeded to have a spectacular implosion, and in the span of just a few hours upon interacting with the broader Twitter population, was transformed from a polite teenage girl impersonator into an all out sociopath, unleash ingtweets covering everything from racist outbursts, N-words, conspiracy theories, genocide, incest, Obama-slurs, and even outright Nazism. "The AI chatbot Tay is a machine learning project, designed for human engagement. It is as much a social and cultural experiment, as it is technical. Unfortunately, within the first 24 hours of coming online, we became aware of a coordinated effort by some users to abuse Tay's commenting skills to have Tay respond in inappropriate ways. As a result, we have taken Tay offline and are making adjustments."