Renovating the Barnes maze for mouse models of dementia with STARR FIELD: A 4-day protocol for learning rate, retention, and cognitive flexibility.

Land-based spatial mazes are a low-stress method to evaluate learning in rodent models of dementia. By using innate exploratory and hiding behavior, the Barnes maze requires fewer trials, allowing examination of early learning rate and retention, as well as executive and motivational features that can be characteristic of non-amnestic dementias. However, unwanted odor cues may disrupt interpretation of acquisition rate during typical learning trials. We designed and tested our Barnes FIELD protocol (Find the Invisible Exit to Locate the Domicile) to improve reproducibility, allow evaluation of learning trials, and limit experimenter influence. The protocol uses 3D-printed escape shuttles and docking tunnels, allowing mice to exit the maze to the home cage. We show evidence that our shuttles mitigate the possibility of undesired cues. We demonstrate the feasibility of our protocol across several models of cognitive impairment and aging, and develop an additional stage, the STARR (Spatial Training And Rapid Reversal) maze, to better challenge behavioral flexibility. By examining commonly used outcome measures we identify important considerations for interpretation. These insights are used to evaluate several models of cognitive change, including deficits in an Alzheimer's disease mouse model and behavioral flexibility in a model of brainstem dysfunction. This work provides comprehensive instructions to build, perform, and analyze a robust spatial maze that expands the range of behavioral and motivational outcomes that can be identified and screened. Our findings will aid interpretation of traditional protocols, enhance rigor and reproducibility, and provide an updated method to screen for cognitive changes in mice.