The Attentional Set Shifting Task: A Measure of Cognitive Flexibility in Mice

Deficits in cognitive flexibility mediated by the PFC represent a significant disability associated with a variety of neurological and neuropsychiatric disorders, and can have a major impact on outcomes and quality of life of individuals suffering from these disorders. Optimization of the AST for use in mice is vital to investigating the genetic underpinnings in diseases coupled to prefrontal cortical mediated cognitive dysfunction. We have found that the method described above is a reliable measure of a specific cognitive deficit, cognitive flexibility, and closely approximates the results obtained using this task in rats and non-human primates. The method described is the result of an extensive survey of previously described protocols in rats and mice as well as advice and guidance of experts at our institution.

Previous reports using similar tasks in mice have demonstrated the adaptability of such a behavioral measure, albeit with several caveats^2,6,11. Colacicco et al. limited the number of discriminations tested per day to 2, which spread testing over a four day period¹¹. This included repetitions of previously learned stimuli pairs on consecutive testing days. While effective, the aim of developing the protocol presented here is to minimize the duration of the testing period and maximize efficiency. Here, we limit testing to a two day period and eliminate repetitions of previously learned stimuli pairs. Previous researchers have noted that mice require overtraining to form an attentional set and exhibit a small “window of testability” in which they will maintain effortful responding. Our protocol also involves modest overtraining on the intradimensional shift within the initially relevant cue dimension. Particularly when evaluating the normalized data in Figure 4C, it appears that mice tested in our protocol may exhibit a more pronounced deficit during the second intradimensional reversal (relative to the immediately preceding stage). This observation is consistent with the notion of overtraining in order to ensure formation of an attentional set, as mice also exhibited an ID/ED shift. One possibility as to why mice sustain effortful performance in our protocol (Figure 5), which involves an average session length of 2 hr, may be due to utilization of a reverse light cycle. We tested behavior in an isolated and dedicated testing room under red lighting conditions during the first four hours of the dark (wakeful) cycle. However, systematic testing outside the scope of this method manuscript would be required to determine what factors contribute to the duration in which mice will continue to perform the AST, and if deficits in olfactory discrimination are suspected, prescreening may be warranted prior to testing as alternate odor-free methodologies have been reported.

Additionally, during the initial method development, we discovered that young adult mice (8-12 week old) performed with greater variability that slightly older 4-6 month old mice. We speculate that the food restriction protocol affects growing mice differently than more mature mice, and while, in a classic neurodevelopment sense, the mouse brain is considered fully developed by 6 weeks, it is possible that the transition from adolescent to adult maturity and early adulthood synaptic remodeling within the prefrontal cortex confound AST performance in younger mice. However, these fascinating questions remain to be explored.

Other variations of AST protocols in mice include “free” trials, meaning a designated number of trials (typically 4) at the beginning of each stage are not scored, and the animal is allowed to explore both pots until the reward is retrieved. The protocol detailed here does not include any “free” trials. In addition, we set the requirements to meet criterion more stringent (8 consecutively correct trials to criterion instead of the previously described 8/10 trials to meet criterion) and have found that this is an effective indicator of whether the animal has truly learned the rule. Garner et al. showed that a period of overtraining following an IDS (i.e., presentation of the previously relevant stimuli for an additional 50 trials) significantly increased the trials to criterion in the EDS². Here, we include multiple intra-dimensional shifts, in an effort to “overtrain” and ensure the formation of an attentional set on the relevant dimension, in agreement with Bissonette et al. ⁶. Additionally, Garner et al. limited testing sessions to approximately 1 hr (or when the mouse no longer responded)², whereas we have found that C57BL6/J mice adequately and reliably progress through all four stages presented to them each day within approximately a 2 hr period (for control mice). Although, potential differences in motor ability between experimental groups should be considered when interpreting performance. Our procedure restricts the time and labor investment to 17 days for each animal from the start of handling through the completion of testing. It is recommended that an individual experimenter limit the number of mice tested in a single day to no more than 2 (this can be modified based on individual experimenter ability and experience), which allows a single experimenter the ability to complete testing on at least 6 mice per week. If motor function is mildly impaired, the length of each trial can be increased, as correct responding, not speed of responding, is the key dependent variable.

Visual discrimination paradigms similar to those that have been used to assess cognitive performance in non-human primates have recently been adapted for use in rodents, and can be used as an alternative to the odor and medium discrimination tasks described here. This approach may be necessary if significant differences in olfactory discrimination are present between experimental groups of mice. Brigman et al. have shown that mice readily discern compound visual stimuli in an ID and ED shift paradigm¹⁵. While this automated method better represents the test employed to study prefrontal cortex dysfunction in humans and non-human primates, the cost of the operant chambers and touchscreens required to perform this automated test in mice can be prohibitive.

It is recommended that behavior be performed as consistently as possible for each mouse, i.e., habituation, training and testing should be performed at the same time of day for each animal. Furthermore, the behavior room should remain unchanged throughout the experiment, making the use of shared/community space problematic due to uncontrollable traffic flow, odor introduction, etc. If animals are to be transported from a housing room to a remote behavior room (as is the case here), the same route should be taken each time. The behavior room should be in an area where the experimenter and the animals will not be disturbed during the testing procedure. If multiple experimenters are performing the task in the same room, entry and exit from the room should be as limited as possible to avoid any loud noises or disruption which may increase the stress or anxiety of the animal during testing. Additionally, experimenters should be careful in removing and replacing the start gate as well as the pots into and out of the chamber so as to create as little noise and disruption as possible. Experiments performed in rats have demonstrated that AST performance can be disrupted by various stressors.

Scenting of the pots should be performed each week and on the same day each week, at least 2 days prior to the first testing day. It is recommended that the pots are scented in a well-ventilated area outside of the area where the test is to be performed (e.g separate laboratory, hallway, fume hood). This will help to avoid cross contamination of the scents and will avoid confusing the mice with scents that are not part of the test. Additionally, prior to and during testing, strong scents which may distract the animal from the discrimination (i.e., consumption of gum or coffee immediately prior to testing, strong perfume/deodorant, etc.) should be avoided for the aforementioned reason. Should experimenters desire to use alternate digging medium than what is described in this protocol, it is recommended that the medium be of similar density and weight to that described here (excessive weight/density makes it difficult for the mouse to dig adequately). For example, we found that small plastic beads were not suitable because they were difficult for the mice to push aside. Additionally, any media with a very strong scent (e.g., rubber strips and some plastic material) is not suitable as it may interfere with the ability to detect the scents being tested.

Training and testing of an individual mouse should be performed by a single experimenter, however acclimation can be done by any experimenter as long as they have participated in handling. All pots to be used in the four stages performed each day should be within reach so as to limit the amount of time the animal is in the waiting area between stages. This will help maintain the attention of the mouse on each task. Furthermore, once testing has begun it is important not to stop until the completion of the tasks for that day of testing (i.e., all 4 stages for each testing day must be completed in succession).

In conclusion, the protocol described here provides detailed methodological instructions for effective utilization of the AST to measure prefrontal cortical-dependent cognitive function in mice, including description of data analysis and potential caveats. A variety of mouse AST methods are reported in the primary literature, each with its own distinct procedural detail. The unique coupling of both text-based and visual information will be particularly helpful for researchers attempting to incorporate this powerful behavioral assay into their scientific repertoire.