Background

Social interactions are a fundamental aspect of life in many animal species, from honeybees to humans [1,2]. Understanding the neuronal mechanisms that control social behavior is central to neuroscience and also holds translational significance, as aberrant social interactions are a hallmark of many neurodevelopmental and neuropsychiatric disorders [3,4]. Laboratory studies offer unique advantages for tightly controlling environmental conditions as well as recording and manipulating neural circuits relevant to social behavior. Yet a major challenge remains in designing experimental settings that are relevant to the natural history of the studied species. Although multiple rodent social paradigms are available, most still rely on reductionist measures, often at the expense of ecological validity and interpretability [5–8]. Thus, the development of robust and ethologically relevant behavioral assays is essential for advancing mechanistic insight and, ultimately, for informing therapeutic approaches.

Conventional paradigms, such as the widely used three-chamber social interaction (SI) test, have provided valuable insights into sociability and social novelty preference (e.g., [9–11]). They are straightforward to implement, typically lasting 10–30 min, require minimal specialized equipment (e.g., an arena, wire cups, and a recording device), and are simple to analyze using measures such as time spent near a conspecific enclosure or duration of sniffing. However, these assays fail to capture the dynamic, reciprocal, and context-specific nature of social interactions, and the simplistic measures they provide are difficult to interpret in terms of specific alterations in social behavior. For example, increased time spent in close contact in the three-chamber SI test could reflect a stronger drive for affiliative interactions or for aggressive ones. Other paradigms, such as the resident–intruder test (using either same-sex or opposite-sex conspecifics), permit close contact interactions but force animals into close proximity, thereby confounding interpretation by removing the element of agency from social decision-making. Under such conditions, aggression and sexual behaviors are expressed in situations where animals cannot opt out of interactions and cannot display context-specific behaviors, such as fleeing to their own territory, thereby limiting both ecological validity and translational relevance.

To address these limitations, we developed the immersive social interaction assay (ISIA), a behavioral paradigm designed to capture voluntary and dynamic social interactions in mice across prolonged time windows [12]. The ISIA apparatus consists of two home-cage chambers connected by a plastic tube, whose inner diameter can be adjusted with a removable plastic ring. This design allows flexible control over access between chambers: when test mice are implanted with head-bars, inserting a ring into the tube restricts the animal to one chamber while still allowing unrestricted movement of conspecifics without head-bars. Animals can be recorded in the ISIA apparatus over extended durations, with siblings or strangers, in their own home territory, when connected to other territories, and even when linked to behavioral testing chambers.

The ISIA enables freely moving animals to display a wide repertoire of behaviors, including both affiliative and aggressive interactions, while retaining the autonomy to engage in or withdraw from social encounters. This element of choice allows experimenters to probe not only the occurrence of behaviors but also the motivational processes underlying them, as shown in our studies of huddling behavior [12], where mice were willing to forsake their preferred sleeping location, even within their thermoneutral zone, to gain social contact during sleep. The ISIA apparatus also minimizes stress-related confounding factors, since animals can be manipulated without direct handling through chamber connections. Built from modified standard housing cages, the ISIA is cost-effective and readily scalable, from simple two-chamber interactions to complex multi-chamber designs. In addition, with simple overhead USB cameras, the assay can be integrated with modern machine learning pipelines for behavioral identification, tracking, and classification (e.g., DeepLabCut-based key point estimation [13], SLEAP [14], and DAMM [15]), allowing high-throughput and automated analysis of complex social interactions.

Our laboratory first described the ISIA in Sotelo et al. [12]; here, we provide a detailed step-by-step protocol for its construction, setup, and use in an improved design. This protocol is intended to facilitate broader adoption of the ISIA and to advance the ethologically relevant study of social behavior in laboratory animals.