Significance: The advances and the miniaturization in functional Near Infrared Spectroscopy (fNIRS) instrumentation offers the potential to move the classical laboratory-based cognitive neuroscience investigations into more naturalistic settings. Wearable and mobile fNIRS devices also provide a novel child-friendly means to image functional brain activity in freely moving toddlers and preschoolers. Measuring brain activity in more ecologically valid settings with fNIRS presents additional challenges, such as the increased impact of physiological interferences. One of the most popular methods to minimize such interferences is to regress out short separation channels from the long separation channels (i.e., superficial signal regression or SSR). Whilst this has been extensively investigated in adults, little is known about the impact of systemic changes on the fNIRS signals recorded in children in either classical or novel naturalistic experiments. Aim: We aim to investigate if extracerebral physiological changes occur in toddlers and preschoolers, and whether SSR can help minimize these interferences. Approach: We collected fNIRS data from 3-to-7 years olds during a conventional computerized static task and in a dynamic naturalistic task in an immersive virtual reality (VR) continuous automatic virtual environment (CAVE). Results: Our results show that superficial signal contamination data is present in both young children as in adults. Importantly, we find that SSR helps in improving the localization of functional brain activity, both in the computerized task and, to a larger extent, in the dynamic VR task. Conclusions: Following from these results, we formulate suggestions to advance the field of developmental neuroimaging with fNIRS, particularly in ecological settings.