This review summarizes the challenges and recent advances in continual or lifelong learning for artificial systems, focusing on neural networks and their ability to learn and retain knowledge over time. Humans and animals can continuously acquire, refine, and transfer knowledge throughout their lives, a process mediated by neurocognitive mechanisms that support sensorimotor skills and memory. However, machine learning models, especially deep neural networks, struggle with this due to catastrophic forgetting, where new information overwrites previously learned knowledge. This is a major issue for state-of-the-art models that rely on stationary data batches, as they do not account for incrementally available information over time. The review discusses the stability-plasticity dilemma, where systems must balance the need to adapt to new information while preserving existing knowledge. Biological systems, such as the hippocampus and neocortex, play a key role in this process, with the hippocampus handling rapid learning and the neocortex managing long-term storage. The complementary learning systems (CLS) theory suggests that these brain regions work together to learn and retain information effectively. Recent neural network approaches aim to mitigate catastrophic forgetting by regulating synaptic plasticity, allocating additional resources for new information, and using complementary learning systems for memory consolidation. These include methods like learning without forgetting (LwF), elastic weight consolidation (EWC), and dynamic architectures that adapt to new tasks. However, these approaches often require significant computational resources and may not scale well to complex scenarios. The review also highlights the importance of using quantitative metrics to evaluate catastrophic forgetting in large-scale datasets. Additionally, it emphasizes the need for further research to develop robust lifelong learning systems for autonomous agents and robots, particularly in ecological conditions that mimic real-world environments. The integration of multisensory information and the development of more sophisticated learning algorithms are critical for advancing lifelong learning in artificial systems. Overall, the review underscores the interdisciplinary nature of lifelong learning and the potential of neural networks to bridge the gap between biological and artificial learning systems.This review summarizes the challenges and recent advances in continual or lifelong learning for artificial systems, focusing on neural networks and their ability to learn and retain knowledge over time. Humans and animals can continuously acquire, refine, and transfer knowledge throughout their lives, a process mediated by neurocognitive mechanisms that support sensorimotor skills and memory. However, machine learning models, especially deep neural networks, struggle with this due to catastrophic forgetting, where new information overwrites previously learned knowledge. This is a major issue for state-of-the-art models that rely on stationary data batches, as they do not account for incrementally available information over time. The review discusses the stability-plasticity dilemma, where systems must balance the need to adapt to new information while preserving existing knowledge. Biological systems, such as the hippocampus and neocortex, play a key role in this process, with the hippocampus handling rapid learning and the neocortex managing long-term storage. The complementary learning systems (CLS) theory suggests that these brain regions work together to learn and retain information effectively. Recent neural network approaches aim to mitigate catastrophic forgetting by regulating synaptic plasticity, allocating additional resources for new information, and using complementary learning systems for memory consolidation. These include methods like learning without forgetting (LwF), elastic weight consolidation (EWC), and dynamic architectures that adapt to new tasks. However, these approaches often require significant computational resources and may not scale well to complex scenarios. The review also highlights the importance of using quantitative metrics to evaluate catastrophic forgetting in large-scale datasets. Additionally, it emphasizes the need for further research to develop robust lifelong learning systems for autonomous agents and robots, particularly in ecological conditions that mimic real-world environments. The integration of multisensory information and the development of more sophisticated learning algorithms are critical for advancing lifelong learning in artificial systems. Overall, the review underscores the interdisciplinary nature of lifelong learning and the potential of neural networks to bridge the gap between biological and artificial learning systems.