The novel object recognition (NOR) test is one of the most commonly used methods to test memory in rodents. It’s easy to implement, inexpensive and based on rodents’ natural tendency to explore novel stimuli. The NOR task was originally developed for rats, but researchers now mostly use it with mice, which tend to be more anxious—raising important concerns about how the test is interpreted. A study in Behavioural Brain Research disputes the assumption that the NOR test always measures memory.

Using 41 C57BL/6J male and female mice and 18 object pairs, the researchers aimed to separate the effects of memory, object neophobia and innate preference on NOR test outcomes. The objects were of different shapes, shininess and colors. The team found that certain objects were consistently approached or avoided regardless of whether they were novel and that some mice preferred familiar items—likely due to fear of the unfamiliar item rather than memory.

Introducing non-test objects into the home cages before testing increased exploration during initial trials, suggesting reduced neophobia. However, even with this pre-exposure, object preferences varied between experimental cohorts of the same strain, sex and age. This variability shows that subtle, individual differences can confound behavioral outcomes in the NOR task. The team also wanted to understand how properly reported and controlled the experiments recently published were. They then performed a short literature review of 69 NOR studies published in 2024 that revealed that most failed to describe object features, and 20% likely misinterpreted neophobic behavior as memory impairment.

The authors recommend several simple but important improvements to the NOR protocol: balancing how objects are presented using crossover designs, pre-screening objects for their natural appeal to the animals, ensuring pretest habituation and reporting absolute exploration times.

This study shows the complexity behind ‘simple’ behavioral assays and the importance of rigorous experimental design. Without appropriate controls, the NOR test data may be misread as fear or preference instead of memory.

Original reference: Koivisto, H., Maguire, C. & Tanila, H. Behav. Brain Res. 491, 115649 (2025)