A Behavioral Assay for Investigating the Role of Spatial Memory During Instinctive Defense in Mice

Instinctive defensive behaviors are largely considered to be hardwired stimulus-responses, such as the C-start movement in fish and amphibians, which moves the animal away from a threat source¹. However, defensive behaviors can be more adaptive if they flexibly take into accountinformation learned about the current environment. One example of such flexibility is the switch from escape to freezing shown by rodents when faced with threat, depending on prior knowledge about the presence or absence of a shelter in the environment^2,3. Other examples of flexibility in innate behaviors include adapting the flight initiation threshold or escape speed depending on the distance between a prey and its shelter⁴, distance to threat^5,6, and previous experience^11,12, as well selecting different defensive strategies depending on the sensory properties of the aversive stimulus⁷, or even suppressing defensive behavior in face of competing motivations such as hunger^8,9,10. The dependency of action selection on acquired knowledge about the spatial features of the environment makes instinctive defensive behaviors in mice a powerful model for studying goal selection, spatial memory and navigation. Here, we describe the adaptation of a commonly used behavioral task, the Barnes maze (BM) spatial memory assay¹³, to determine the influence of the spatial environment on defensive action selection in mice, and their navigation strategies when escaping towards a shelter.

The standard Barnes maze used to study learning and spatial memory in mice consists of a circular ~90 cm diameter platform with 20 equally spaced holes, of which 19 are closed, and one leads to an underground shelterwhich mice seek in order to avoid the open-field environment of the platform. Often, weak aversive stimuli (buzzer, bright light, fan) are used continuously throughout the assay to make the environment aversive and thus promote entry into the shelter¹⁴. In the most commonly used assay^15,16, the animal has one habituation trial where it is guided to the shelter manually by the experimenter immediately after being positioned on the platform. This is followed by a 4-day acquisition period, where each day the mouse is allowed to navigate freely in the maze for 3 min, after which it is again manually guided to the shelter if it does not reach it during the exploration period. The last stage of the assay is one probe trial on the 5^th day (although a long-term seven-day memory trial is also commonly performed) when the animal explores the maze with all holes closed. Learning and long-term memory are quantified by the time taken to find the shelter and pokes in the wrong holes during the acquisition period, and by the time spent near the closed shelter hole in the probe trial. Typical results show a decrease in the number of errors and latency to reach the shelter during acquisition, and an above-chance proportion of time spent in the quadrant containing the closed target hole in the probe trial¹⁵.

While several variants of the BM assay have been previously described^17,18,19, the paradigm we describe here has three fundamental changes from the standard assay. First, the animal is left to explore the maze and find the shelter on its own, and testing is done in the same session, shortly after the shelter has been found, during periods where the animal is engaged in exploratory behavior. Although this setting does not test long-term memory of the shelter location, it is designed to be a naturalistic scenario that mimics the exploration of novel territory, under the threat of predation. In addition, it allows testing of how animals adapt to acute changes in the environment, such as sudden changes in landscape. Second, a key aspect of our assay is that the experimenter never forces the animal in or out of the shelter, which can disorient the animal and exclude path integration²⁰ as a viable navigation strategy²¹. Path integration is a navigation strategy that uses self-motion cues, such as proprioceptive and vestibular cues, arising from the integration of motor outflow, to update the current position of the animal and navigate towards a goal, which is not possible if the animal is passively moved by the experimenter. Third, we use innately aversive visual²² and auditory²³ stimuli to elicit escape to shelter, which is easily distinguished from ongoing foraging behavior and allows the assessment and quantification of specific navigation strategies used during defense from imminent threats. We propose that this assay will be useful for dissecting the role of spatial memory in the selection and implementation of defensive behaviors, and more generally to the broader study of goal-directed navigation and short-term spatial memory. The protocols described here were introduced by Vale et al. in 2017, to which we refer the readers for more in depth discussion of the rationale for experiments and results.