From Exploration to Structured Navigation: Learning Dynamics of Freely Moving Marmosets during Foraging

Spatial memory plays a crucial role in guiding navigation and optimizing foraging efficiency. While extensive research has explored navigation strategies in rodents, how these strategies emerge and evolve in freely moving non-human primates remains poorly understood. Here, we conducted a longitudinal study to investigate the dynamics of spatial learning and navigation strategies in freely moving common marmosets ( Callithrix jacchus) performing a foraging task in a semi-naturalistic environment. Marmosets were trained to collect rewards from eight spatially distributed dispensers. Using behavioral analyses combined with machine learning methods, including unsupervised clustering and probabilistic modeling, we examined how their navigation strategies evolved over time. Our results reveal a progressive improvement in performance, accompanied by an increasing structuring of movement patterns. Initially, marmosets exhibited exploratory behaviors, with high transition variability between food dispensers. As training progressed, their navigation became more structured, with the emergence of recurrent transition patterns and a shift towards route-based navigation. However, despite this increasing regularity, their navigation retained a degree of flexibility, allowing them to dynamically adapt their routes rather than following a fixed sequence. These findings suggest that marmosets develop structured yet flexible navigation strategies through experience. Their behavior reflects a balance between efficiency and adaptability and supports the emergence of a flexible representation of space. Our study provides novel insights into spatial memory development in primates and highlights the value of semi-naturalistic paradigms for studying navigation strategies and spatial learning in freely moving animals.