The field of abstract reasoning is moving towards a more integrated approach, combining visual and linguistic reasoning to achieve human-like intelligence. Recent research has highlighted the importance of visual abstraction in solving complex puzzles, and has proposed novel methods for incorporating visual priors into existing models. The use of vision-centric perspectives and image-to-image translation problems has shown promising results, with some models achieving competitive performance with leading language models. Furthermore, the development of datasets and benchmarks, such as ARCTraj, has enabled the study of human-like reasoning and has the potential to advance explainability and generalizable intelligence. Noteworthy papers include: ARCTraj, which introduces a dataset and methodological framework for modeling human reasoning through complex visual tasks. ARC Is a Vision Problem, which formulates ARC within a vision paradigm and achieves substantial improvements over existing methods. Think Visually, Reason Textually, which introduces a synergistic approach combining vision and language for abstract reasoning. How Modality Shapes Perception and Reasoning, which provides a principled account of how modality imposes perceptual bottlenecks and shapes model perception.