The article describes the experimental procedures for the commonly used linear track virtual reality (VR) paradigm in mice as well as determining the feasibility of running complex VR tasks by testing a Y-shaped signal discrimination task.
Virtual reality (VR) combined with head-fixation is increasingly being utilized in behavioral neuroscience studies as it allows complex behavioral assays to be performed in head-fixed mice. This enables precise behavioral recordings while incorporating various neurophysiological techniques that require head-fixation to minimize movement-related signal noise during neural recordings. However, despite the growing use of VR, there is little published data on the detailed methodology of how to implement it. In this study, a training protocol is developed whereby male and female C57B16/J mice are trained to run down a virtual linear corridor, the length of which is increased from 1-3 m over multiple training sessions. Building upon this foundation, this study investigated the feasibility of mice performing complex behaviors within VR using a Y-maze paradigm. The task required navigating to the arm with black walls from the choice point in the Y-maze. After reaching a criterion of two consecutive days equal to or greater than 70% correct, the mice progressed to increasingly difficult sensory discrimination. The findings provide important details on the methodologies useful for the successful training of mice in VR and demonstrate that mice exhibit learning capabilities in navigating the Y-maze. The methodology presented not only offers insights into training duration in VR-based assays but also underscores the potential for probing intricate behaviors in mice, opening avenues for more comprehensive neuroscience investigations.
Virtual reality tasks have emerged as a powerful method of behavioral assessment in mice due to head-fixation, which allows for mechanical stability that would be compromised in freely behaving mice1. This method enables reduced movement artifacts in electrophysiological recordings2,3 and optical imaging4,5,6,7. It also facilitates repeatable behaviors8 and precise eye-tracking9. In the experimental setup, the mouse is fixed in place and situated atop an air-supported spherical treadmill. This apparatus allows for the intricate exploration of visually guided behavior within the VR environment. As the mouse moves on the treadmill, its locomotion synchronizes seamlessly with its navigation within the virtual landscape, which is visually depicted on the screen surrounding the mouse.
The aim of this study is two-fold: to address key challenges within experimental behavioral neuroscience and to contribute to the advancement of methodologies in this field. Firstly, despite the increased use of VR in academic research10,11,12, there remains a notable absence of comprehensive methodologies and training protocols, hindering the adoption of this technology by new investigators. The primary goal was to fill this gap by delineating a detailed training regimen for the linear track paradigm, as depicted in prior studies13,14,15. A commercially available system is used to describe these operational procedures. As a disclaimer, these procedural guidelines have components specific to this system; however, for a discussion of the generalizability of this protocol, see the discussion. The objective was to outline the behavioral procedures, the typical timeline for performing these procedures, and the success rate for training mice to run on a simple linear track.
Second, there remains a lack of documentation on the implementation of complex maze tasks within this paradigm in mice. Complex virtual assays have been developed in rats11. However, mice have reduced visual acuity in comparison16 and often perform worse in complex tasks17. While some investigations have focused on specific tasks such as evidence accumulation or spatial novelty18, the focus here was in elucidating the training methodologies required for mice to engage in decision-making paradigms within VR environments. To address this challenge, a signal discrimination task was devised where the mice were tasked solely with learning to associate the color/luminance (black versus white) of the rewarded arm with the reward, achieved by selecting the black arm at the choice point of the Y-maze, with the correct arm randomized on each trial. This task was designed to require interaction with the virtual cues and provide insight into the perceptual discrimination abilities of the mice.
In summary, this study addresses critical gaps in the field of experimental behavioral neuroscience by providing comprehensive training protocols for using VR paradigms in mice and elucidating methodologies for complex decision-making tasks within this framework. By leveraging insights from previous research and innovative experimental designs, this study aims to streamline research practices and advance the understanding of neural mechanisms underlying behavior. The following sections will delve deeper into the experimental procedures and results and discuss the findings.
All procedures involving animals were conducted in strict adherence to the protocols established by the NIEHS Animal Care and Use Committee, ensuring compliance with ethical standards and welfare guidelines. C57BL/6Tac mice, approximately 8 weeks old, were utilized for the study.
1. Surgery for head-bar implantation
2. Fluid restriction
NOTE: Water restriction induces a state of thirst in mice, heightening their motivation for liquid rewards. However, meticulous implementation is necessary to ensure the preservation of mouse well-being20.
3. System setup
4. Behavioral Tasks
NOTE: In accordance with established methodologies in behavioral neuroscience, the formulated tasks employ a reward-based associative learning technique. By employing immediate rewards to reinforce specific behaviors, animals are trained effectively to execute repetitive tasks, facilitated by the teleportation capability of VR. Within a virtual behavioral framework, teleportation functionality affords mice the ability to engage in tasks without the stress associated with physical manipulation, concurrently reducing the setup duration necessary for analogous real-world tasks. During the training sessions, use dim red overhead lighting within the experimental setting. This precaution is recommended due to the diminished visual perceptive sensitivity in mice to red light, which mitigates potential interference with their perception of the virtual reality (VR) screens, as opposed to the use of white light22.
This pilot study aimed to outline methodologies for the efficient training of mice in two distinct tasks: a simple corridor and a complex decision-making task (the Y-maze visual discrimination task). These data served as the basis for establishing temporal guidelines for behavioral training in VR.
The procedural steps start by outlining the surgical implantation of the head-bar in Figure 1. This implant serves to stabilize the mouse's skull during behavioral assessments, thereby enhancing the precision of neural recordings, particularly when employed in conjunction with electrophysiology or imaging techniques.
Figure 2 and Figure 3 illustrate the hardware components and setup of the experimental system. Figure 2 details the water delivery system, which utilized a Petri dish fountain method. This involved affixing a 60 mm x 15 mm Petri dish concave-side down onto the cage floor, securing a smaller 35 mm x 10 mm Petri dish concave-side down at the center of the larger dish, and placing another 60 mm x 15 mm Petri dish concave-side up on top of the smaller dish to serve as a water reservoir. The height of the upper dish was carefully adjusted to prevent contamination by bedding material while ensuring mice had easy access to water.
Figure 3 presents the system hardware and mouse positioning guidelines. Figure 3A depicts the VR setup, which featured a six-screen array with a spherical treadmill positioned centrally. Figure 3B shows the optimal placement of the mouse on the treadmill, with the head aligned in a natural position and all four paws in contact with the surface. Figure 3C compares correct and incorrect mouse placement relative to the head-bar, emphasizing that the midsagittal plane of the mouse should be centered, rather than aligning with the head-bar itself.
Figure 4 presents reward acquisition curves on a line graph, illustrating the expected learning periods for 1 m, 2 m, and 3 m narrow corridors in VR based on predefined parameters for progression. It depicts the average velocities of mice across respective track lengths, demonstrating a gradual increase in speed as evidence of task learning and improvement commensurate with increased difficulty. A bar graph is also shown illustrating the average number of days required for mice to reach the criterion for the linear tracks, as well as a bar graph displaying mean velocities for each track length. Following this, the progressive stages of the linear track task learned by the mice are illustrated as well. These tasks were designed to replicate methodologies established in academic literature while ensuring a learning curve feasible for mice, facilitating their advancement through the levels.
Finally, Figure 5 provides data pertaining to the Y-Maze task. The figure illustrates the progressive nature of the task, beginning with a straightforward discrimination between solid black and white arms. This initial stage serves as a foundational step, establishing the mice's ability to distinguish between contrasting visual cues. Subsequent levels of the task introduce increasing complexity by incorporating additional percentages of the contrasting color to each arm, thereby challenging the mice's discrimination abilities further. The gradual augmentation of task difficulty is exemplified by the transition from solid black and white arms to arms composed of 90% of one color and 10% of the other. Notably, the data presented in Figure 5 indicates that while discrimination accuracy improves with each level of progression, some mice consistently demonstrate a threshold of visual discrimination capability, reaching a maximum of 80%/20% white/black discrimination. This observation underscores the limitations inherent in the mice's visual discrimination abilities within the context of the Y-Maze task, providing valuable insights into the task's feasibility and the cognitive capacities of the subjects. Subsequently, the progressive stages of the Y-maze track task, which were designed to align with established methodologies in the literature, are detailed. These stages ensured a feasible learning curve for the mice, supporting their gradual advancement through the levels.
Figure 1: Surgical instructions for head-bar implantation. (A) The incision site is marked on the mouse's cranium. (B) The screws should be implanted 1 mm to the left of the interfrontal suture slightly below bregma and 3 mm to the right of the interfrontal suture slightly above lambda. (C) The head-bar should be placed along the interfrontal suture. (D) Apply dental cement over the head-bar implant. (E) Actual visualization of the head-bar after the application of dental cement. Please click here to view a larger version of this figure.
Figure 2: Water delivery system using a petri dish fountain method. A 60 mm x 15 mm Petri dish was fixed concave-side down on the cage floor. A smaller 35 mm x 10 mm petri dish was centered on the larger dish, with another 60 mm x 15 mm Petri dish placed concave-side up on top to serve as a reservoir. This setup ensured that the water remained uncontaminated by bedding and accessible to the mice. Please click here to view a larger version of this figure.
Figure 3: System hardware and positioning of the mouse guidelines. (A) This displays the VR setup utilized. A six-screen setup was utilized, with the spherical treadmill placed in the middle. (B) Side view of optimal mouse placement on the spherical treadmill. The mouse head is in a natural position, while all four paws are on the spherical treadmill. (C) Top view of correct versus incorrect placement of the mouse in regard to the head-bar. For correct placement, the midsagittal plane of the mouse should be centered rather than the head-bar itself. Please click here to view a larger version of this figure.
Figure 4: Linear track data. (A) The presented data depict the daily rewards collected within each 30 min trial period. Mice progressed to longer track lengths once they achieved an average of 2 rewards per min over 2 consecutive days, totaling 60 rewards (threshold). (B) As mice acquired proficiency in the task, their velocities exhibited a gradual increase, indicative of the efficacy of reward reinforcement. The graph illustrates the average daily velocity of each mouse on the track in cm/s, portraying a linear progression in learned behavior. (C) This bar graph illustrates the duration taken by each mouse to acquire proficiency on individual track lengths, with the respective means and standard error depicted for each track length. (D) This bar graph demonstrates the mean and standard error of the average daily velocities achieved by each mouse across various track lengths. The nearly linear progression suggests a learned enhancement in running speed. (E) This illustrates the progression of the linear track task, which requires 2 consecutive trial days of 60 rewards before advancement to a longer version of the maze. Please click here to view a larger version of this figure.
Figure 5: Y-Maze data. (A) This shows the distribution of rewards acquired at different stages of the Y-maze progression. This analysis focused exclusively on a subset of four mice that completed all phases of the linear track, thereby ensuring an equitable representation of both male and female participants. (B) This visual representation illustrates the stages of the Y-Maze task, wherein mice advance upon achieving two consecutive days of 70% correct choices. Please click here to view a larger version of this figure.
This study employed a comprehensive approach to investigate the behavioral responses of mice in VR environments, focusing on the implementation of surgical procedures, fluid restriction protocols, system setup, and behavioral tasks. These findings contribute to the field by providing procedural details, time frames for training, and success rates. This will enable more effective adoption of VR procedures in mice and facilitate planning and implementation for labs interested in using this procedure in their research.
The surgical implantation of head-bars was essential for facilitating head-fixed behavioral experiments in VR environments. By carefully following established protocols and providing appropriate post-operative care, the successful integration of head-bars was ensured while minimizing adverse effects on the animals’ health and behavior. Additionally, fluid restriction protocols were implemented to regulate water intake and maintain hydration and thirst levels among the mice. The gradual acclimation process and periodic access to water were crucial for ensuring the animals’ welfare while facilitating the execution of behavioral tasks.
The setup of the VR behavioral system involved the integration of hardware and software components to create immersive virtual environments for the mice. The utilization of fully immersive virtual displays, liquid reward systems, styrofoam balls as spherical treadmills, and head holders enabled precise control over experimental conditions and data acquisition. Behavioral tasks, including the linear track and Y-maze paradigms, were carefully designed to investigate key aspects of mouse behavior, such as locomotion, decision-making, and reward processing.
Despite best efforts to optimize experimental procedures, several challenges were encountered during the study. Variability in individual mouse responses and technical issues related to hardware and software integration posed challenges to data collection and analysis. Additionally, the reliance on fluid restriction protocols necessitated careful monitoring of the animal’s hydration status and adjustment of experimental procedures accordingly. At times, mice would struggle when placed on the ball, not drink from the reward spout, or freeze and fail to run on the ball. Although some of these challenges may be temporary, it is crucial to monitor the mice to ensure they are not experiencing impediments in their progress. Mice that fail to show advancement compared to their peers should be withdrawn from the study. One similar experiment had 4 of 55 mice removed due to their inability to learn the paradigm25. Mice exhibiting consistent immobility on the ball for 5 consecutive days were excluded from the study following thorough assessments of their weight, ability to access the reward spout for drinking, and positioning on the ball to ensure no underlying issues were present. In these cases, it is up to the discretion of the researcher to decide what strategy to take to resume the study efficiently.
These training protocols were designed to progressively challenge the mice while ensuring their proficiency in executing behavioral tasks. Criteria for progression from the linear track to the Y-maze paradigm were based on the mice’s ability to meet predetermined performance thresholds, such as achieving consecutive days of successful trials and reward acquisition. The implementation of rigorous training protocols allowed us to assess the mice’s behavioral capabilities and adaptability to increasingly complex tasks. These carefully structured protocols provide a robust framework for researchers in the field of behavioral neuroscience, offering a systematic approach to evaluating and training animals for diverse experimental paradigms. By outlining clear criteria for progression, researchers can efficiently gauge the learning curve of experimental subjects and curate training paradigms accordingly. Furthermore, this methodological approach fosters reproducibility and standardization across experiments, facilitating comparative analyses and advancing the understanding of cognitive processes and learning mechanisms in animal models.
When designing a VR paradigm for mice, it is crucial to recognize the range of approaches available concerning task complexity and training progression. This protocol offers a broad framework for constructing an experimental design, yet it remains up to the investigator to tailor specific aspects such as reward delivery, bias control, stimulus type, task progression, and system parameters according to the needs of the study. For instance, some studies opt for a more streamlined approach, focusing on immediate task engagement. An example is Krumin et al. which implemented a single, consistent T-maze task rather than employing a progressive learning regimen between different tasks. In contrast, other studies offer diverse trial design components, such as stimulus reinforcement strategies and auditory cues. The study utilized auditory feedback as a punishment for incorrect trials and provided only water as a reward for correct trials26. Conversely, Zhao et al. employed a 10% sucrose solution as a reward for correct trials and did not incorporate any form of punishment for incorrect trials27. Instead, they focused on mitigating incorrect responses through methods such as anti-bias training, which involved increasing the probability of switching the cue direction from the animal’s previous choice and adjusting the daily water allowance to enhance motivation. Differences in experimental design, such as the presence of spatial cues throughout the task, can lead to varying interpretations of neural coding, as evidenced by Zhao et al. finding posterior parietal cortex cell selectivity explained by trajectories and spatial preferences, in contrast to Harvey et al.’s observed choice-dependent activation sequences27,28. It is important to note that the specific hardware used included six LCD monitors, an extendable lick spout, and an air-cushioned styrofoam ball treadmill. There are a number of differences across virtual reality systems across labs, including the use of projectors29 versus computer monitors, non-spherical treadmills30, and fixed10 versus extendable lick spouts.
In conclusion, this study provides valuable insights into the behavioral responses of mice in VR environments and demonstrates the feasibility of employing immersive technology to investigate complex behaviors. Future research endeavors may focus on refining experimental protocols, exploring neural mechanisms underlying decision-making processes, and translating findings to clinical applications. By continuing to advance the understanding of mouse behavior, scientists can further elucidate the neural circuits and cognitive processes underlying complex behaviors in both health and disease.
The authors have nothing to disclose.
This research was funded by the National Institutes of Environmental Health Sciences (ZIC-ES103330). Special thanks to K. Krepinksy of Phenosys for his help on the hardware and software properties of the system, to T. Viney of the University of Oxford for his assistance with behavioral paradigms, and finally to G. Vargish of the NIH for his guidance on his piloting procedures and surgical methods.
2.4 mm Screws (00-96 X 3/32) | Protech International | 8L0X3905202F | For Added Headbar Stability |
Bupivocaine | Hospira | NDC:0409-1162-19 | Local Anesthetic |
Buprenorphine | Wedgewood Pharmaceuticals | SKU: BUPREN-INJ010VC | Analgesia |
Buzzers | Wahl | 1565q | For Shaving Surgical Region |
Drill and microinjection robot | Neurostar | 17129-IDA | Stereotaxis |
GLUture | Zoetis | 32046 | Surgical Adhesive |
Head-bar Implant | Luigs-Neumann | 130060 | Mouse Head Implant |
Heating Pad (Lectro-Kennel) | K&H Manufacturing | 100212933 | Post-operative |
Hemostats | World Precision Instruments | 501291 | Surgical Tool |
Hydrogen Peroxide | Swam | L0003648FB | Cleaning Agent |
Isoflurane | Dechra | B230008 | Surgical Inhalation Anesthetic |
Isoflurane/O2 Delivery device w Nosecomb attachments | Eagle Eye Anesthesia Inc. | Model 50 Anesthesia | Surgical Device |
Metabond | Parkell | CB-S380 | Adhesive Cement |
Microscissors | Fine Science Tools | 15000-08 | Surgical Tool |
Oxygen | Praxair | UN1072 | Surgical Oxygen |
Povidone-Iodine Swabstick | Dynarex | g172095-05 | Surgical Tool |
Saline | Hospira | NDC:0409-1966-02 | Hydration Agent |
Sterile Cotton Tipped Applicator (Q-tips) | Puritan | 25-806 2WC | Surgical Tool |
Sucrose | Fisher Chemical | CAS 57-50-1 | Primary Reinforcer/Motivator/Reward |
Tweezers | World Precision Instruments | 504505 | Surgical Tool |
Virtual Reality System | PhenoSys | JetBall-TFT | The JetBall, an air cushioned spherical treadmill allows an animal to navigate effortlessly in a virtual world projected on 6 surrounding monitors. |
White petrolatum lubricant eye ointment ointment | AACE Pharmaceuticals | NDC:71406-124-35 | Eyelube |
.