Dissecting novel object exploration in a fully automated homecage-based novel object recognition test.

The novel object recognition (NOR) test is widely used to assess memory in rodents, offering strong ethological validity, cross-species relevance, and specificity for hippocampal-parahippocampal function. However, standard implementations are often confounded by uncontrolled factors. Here, we present a fully automated, homecage-based NOR test for evaluating long-term object memory in mice. Our empirically informed computational model demonstrates the robustness of this approach despite uncertainties in defining exploratory behavior. Mice reliably preferred novel over familiar objects after both 24-h and 7-day delays, with recognition emerging already at a distance. Results were replicated across two facilities. Notably, recognition after 24 h depended on prior interactions with the replaced object, but not after 7 days. We also show that external factors can bias exploration, which can be mitigated using relative discrimination measures. This automated paradigm enhances standardization, reproducibility, and our understanding of the factors influencing object exploratory behaviors and object memory.