Which is better: use the same animal to evaluate social interaction at different stages of life (30, 60 and 90 days) or a different animal for each stage?
Social scientists have to deal with this issue on a regular basis. The issue is whether you want to observe changes longitudinally over time (watch the same mice over time, and see if their behaviors change), or whether you think you can observe like aged-groups of different mice at different ages (e.g., observe a group of 30 day old mice; a group of 60 day old mice, etc), and infer changes in behavior from those age specific observations. Now, you could do both, and see how your observations correspond. Also, since you have mice and can control their environments and genetic composition, you eliminate (perhaps) some of the need to observe the same group over time. However, one could argue that a group of mice develop a dynamic over time that affects their behavior at different ages. So, you perhaps need some more specific hypotheses/research questions before deciding how to do this. There is a large enough literature on, for example, the aggressive behavior in mice over the past 60 years that you can get some ideas from reading that literature.
I am not familiar with mice social interaction, but you might be interested in simulating human social interaction and different determinants of social interaction, including, temporal. Book Social Interaction, Globalization and Computer-Aided Analysi...