All procedures in this study were approved by the Institutional Animal Care and Use Committee (IACUC) at Bryn Mawr College. Obtain IACUC or otherwise applicable regulatory approval before ordering laboratory animals and beginning experimentation.

1. Animal preparation

1. Acquire male and female adult Sprague Dawley rats.
   NOTE: The rats can be delivered before 65 days of age, but do not begin procedures until after this point to ensure that both males and females are fully mature.
2. Pair-house same-sex rats for as long as possible, as long-term isolation is a stressor²⁴. For food restriction, singly house rats just prior to the operant strategy shifting protocol.
3. After 1 week of acclimation, gently begin to handle rats for 3–5 min per day. Collect the body weight of each rat. Additionally, if interested in assessing how gonadal hormones may affect the results, collect vaginal lavage for female rats (described in section 2).
4. Restrict (from food) animals that will be run in the operant strategy shifting paradigm at least 3 days before the training begins so that they successfully learn the task. Ensure that water is always freely available. Written records should be maintained for each animal to document daily food and fluid consumption, hydration status, and any behavioral and clinical changes used as criteria for temporary or permanent removal of an animal from a protocol (Morton 2000; NRC 2003b).
   1. If employing a stress procedure for more than 3 days before training, adjust the food restriction to match the number of days of stress (e.g., 5 days of restraint plus food restriction²⁵).
   2. Each day, deliver 80% of the normal daily food intake (i.e., 4 g of food per 100 g of body weight)²⁶. Use the daily weight collection for the rat to calculate how much food to give each day.
   3. Continue the food restriction through the training and testing days. However, do not place food in the home cage until after the rat has completed training or testing for the day, or else they will not be motivated to perform the tasks for a food pellet reward. Ensure that the timing of food delivery to rats upon completion of the task is fairly unpredictable since this helps to avoid reduced motivation to perform in the operant chamber (in favor of simply waiting for food in the home cage afterwards).
      NOTE: Animals undergoing the restraint stress paradigm do not exhibit significantly greater weight loss than control, unstressed subjects. However, various stress procedures may themselves induce weight loss, resulting in rats receiving less food than unstressed counterparts during body weight-based food restriction. This may present an additional, confounding stressor. If this appears to be the case, alternatively use a fixed amount of food given to each subject, regardless of weight²⁷.

2. Vaginal lavage

NOTE: Gonadal hormones (i.e., estrogen and progesterone) are known to affect the stress response and cognition^28,29,30. These hormones fluctuate over the estrous cycle of female rats³¹. If interested in tracking the estrous cycle of freely cycling female rodents to correlate with stress or cognitive flexibility data, collect vaginal lavage as described below. Representative data considering estrous cycle stage are not provided.

1. To obtain vaginal lavage samples from females, gather warm water in a clean beaker, a glass eyedropper, a “lavage” slide (microscope slide with acrylic paint circles to hold the lavage sample), and one empty beaker.
2. Fill the eyedropper with a small amount of warm water (~0.5 mL), then insert the tip into the vagina of the female rat (by lifting by its tail). Expel the sterile water 2x–3x and expel the collected fluid onto a microscopic slide. Do not overflow the lavage slide circle.
3. Expel any excess liquid into the empty beaker. Label the lavage slide with rat numbers and put the samples from each rat in that order so it is clear which sample belongs to each rat.
4. Thoroughly rinse the eyedropper by pipetting clean warm water and dispensing it into the “excess” beaker several times before filling the eyedropper to sample the next rat.
5. Carefully carry the lavage slide to a brightfield microscope to image the lavage sample and classify the day within the estrous cycle as described in Becker et al³¹.
   NOTE: Ideally, lavaging should be done for a few weeks to properly track a female’s cycle and should be performed at a very similar time each day to control for circadian rhythms. Preferably, this procedure should be performed before stress and operant strategy shifting procedures. The use of cotton swabs and sterile saline can also be used as an alternative to this eye-dropper technique. Data for female rats can be analyzed post-hoc according to estrous cycle day (consider days of cycle when stress is performed and/or day of cycle when testing occurs).

3. Equipment and software

1. Use operant chambers for behavioral training and testing.  
   1. Ensure that the chambers contain at least two retractable levers with two stimulus lights above, a house light, and a dispenser for reinforcement for these tasks.
   2. Check that the levers are on the either side of the central reinforcement delivery area with one stimulus light above each lever.
   3. Use the house light to illuminate the chamber without interfering with detection of the light stimulus (it is best if the house light is on the back wall of the chamber, opposite to the levers and stimulus lights).
2. Use dustless food pellets (here, 45 mg pellets are used: 18.7% protein, 5.6% fat, and 4.7% fiber) for reinforcement in food-restricted rats. Do not use pellets high in sucrose or fat (unless there is interest in how stress affects palatable food intake).
3. Control the presentation of stimuli, lever operation, and data collection from a computer with software capable of operating the chamber (Table of Materials).
   NOTE: For information related to coding of programs using this software, contact the authors. MED-PC scripts are included as supplemental files. This software collects information about the animal’s responses for each trial (which lever is pressed, whether it is correct/incorrect/no response, and latency to make the choice). From this information, users can calculate various measures in the behavioral paradigm, as described in the behavioral analysis section.
4. Perform training/testing at the same time each day to control for circadian rhythms in stress hormones³² (and other relevant measures).
5. Fill the bottom tray of each operant box with fresh bedding to collect feces/waste. Following each session, dump each tray, clean trays and chamber interior with alcohol wipes or an IACUC-approved disinfectant, and replace with fresh bedding before placing a new animal in the chamber.

4. Stress procedures

1. Decide whether the stress procedure should be performed before, during, and/or after training on the operant strategy shifting paradigm (e.g., 5 days of restraint stress prior to 3 days of operant training vs. 3 days of operant training followed by a single restraint and testing).
2. Execute the stress procedure at the same time daily with respect to operant training. (e.g., 30 min of restraint stress starting at 9 A.M., followed by placement in the operant chamber).
3. Perform the stress procedures in a separate room from both the colony room and strategy shifting paradigm rooms (to ensure there are no confounding factors associated with witness stress)³³. Briefly, place the rat in a Broome-style transparent restraint tube and seal the opening, taking care not to pinch the limbs or tail. Between each subject, use an IACUC-approved disinfectant to clean the Broome-style tube. Improper sanitization can leave pheromones behind that can adversely affect the testing protocol.
   NOTE: Estimate how long the first group of rats will spend in the operant chambers. This will vary depending on training vs. test day; however, after running several cohorts, an average time to complete each task to estimate future tasks can be calculated.
4. Depending on how many operant chambers are available, stagger the stress procedure for subjects. For example, four rats undergo restraint stress and are placed in four operant chambers. One hour later, four more animals undergo stress procedures to be followed by the operant chamber.

5. Training

NOTE: This paradigm is modified from the operant set-shifting procedure developed by Floresco et al. such that it can be completed in 3 days¹⁹.  Training procedures for rats require 3 days (1 day to learn each task as described below). It is rare that a rat does not learn these tasks. If a rat fails to learn each task, it should be excluded from the final study. See Figure 1A for a visual depiction of the training paradigm described below.

1. Before placing the rat in the chamber, ensure that there are enough food pellets in the dispenser and that the operant boxes are properly functioning. To accomplish this, load and initiate a training or test day program in an empty chamber, manually testing that the correct lever appropriately delivers one reward per lever press.
2. Training the rat to press each lever
   1. Before placing the rat in the box for the first day of training, manually set one food pellet reward on the correct lever, as designated upon loading the training procedure within each chamber.
   2. Train the rat using a fixed ratio (FR-1) schedule, such that each correct lever press is rewarded with one reinforcement. Counterbalance the correct lever per day across subjects and/or experimental conditions (shaping only one lever at a time) by designating the correct lever upon loading the training procedure on the computer operating the chambers.
   3. Allow the rat to press the lever until it reaches the criterion by pressing the correct lever 50x, usually completing the task between 30–45 min.
   4. The following day force the rat to perform this task on the opposite lever using the same program as the first day of training, but designate the opposite lever as the correct one. There is no need to “shape” the lever with a food pellet on this day of training. Typically, this criterion is quickly acquired after rats have learned to press the first lever.
3. Training the rat to respond to the light cue
   1. On the third day of training, illuminate the light above both levers for 15 s trials, during which the rat may press one of lever to potentially receive a food pellet reward. During the light discrimination task, this program will randomly select which lever is correct on a trial-by-trial basis.
   2. If the rat presses the correct lever, ensure that the lights remain illuminated for 3 s and the reward is delivered, followed by a 5 s period, during which the lights are shut off preceding the next trial. If the rat presses the incorrect lever, ensure that no reward is delivered and that lights are shut off for 10 s preceding the next trial.
   3. Following this last day of training, calculate “side bias” to determine if the rat has a preference for the left or right lever by dividing the number of presses of one lever divided by the total number of lever presses. On the test day, the rat will start on its least preferred side to ensure that it is learning the specific response-reward contingency, rather than responding to a preferred lever.

6. Testing

NOTE: See Figure 1B for a visual depiction of the testing paradigm described below.

1. On day 4 (test day), place the rat in the operant chamber following stress procedures and test them in side discrimination, side reversal, and light discrimination tasks serially. Ensure that the light discrimination task only illuminates the light above the “correct” lever. In each task, rats must consecutively achieve eight correct trials to complete each discrimination without pressing the unrewarded, incorrect lever. An incorrect lever press will reset this chain of trials.
   1. Test rats using the side discrimination task. Using the side discrimination program, reward the rat for pressing the lever on its least preferred side as determined from the third day of training, regardless of the light cue. The task ends upon pressing the correct lever 8x consecutively (excluding omissions).
   2. Perform the side reversal test by running rats using the side discrimination program again, but this time designating the lever opposite to the correct one from the side discrimination task as correct. Ensure that the rat is rewarded for pressing this lever, regardless of the light cue. The task ends upon pressing the correct lever 8x consecutively (excluding omissions).
   3. Perform the light discrimination task, which rewards the rat for pressing the lever with the light illuminated above. Each operant testing is complete upon pressing the correct lever 8x consecutively (excluding omissions).
      NOTE: Based on previous studies, these tasks encode a minimum of 30 trials, regardless of consecutive presses, to ensure that rats have sufficient time to learn the rules of each task¹⁸. Thus, if the rat consecutively achieves eight correct trials before 30 trials have occurred, the task will remain engaged until 30 trials are completed.

7. Behavioral analysis

NOTE: The data acquired for each animal on the test day are automatically recorded and saved by the computer, as long as a MED-PC script for each task been initiated and allowed to complete (see supplementary materials for MED-PC scripts).

1. Open the data for each test day task (side discrimination, side reversal, and light discrimination) using the computer program. The main measures recorded by the program are trials to criterion, errors in criterion, and time to criterion. These measures are described in detail below.
   NOTE: The authors have generated a MATLAB script that allows for automation of the analysis process as well as analysis of perseverative vs. regressive errors (contact authors for code information to streamline data analysis).
   1. Use trials to criterion (which refers to the total number of trials [not including omissions] necessary for the rat to consecutively complete eight correct trials, including those eight trials) as the main indicator of accuracy. This data is located in the first column in array B in a data file generated by the MED-PC script for any of the tasks on test day.
   2. Examine the total errors made during each task. This data is located in the third column of array B in a data file generated by the MED-PC script for any of the tasks on test day. These errors are also categorized into perseverative or regressive errors. Perseverative errors are committed when the rat continues to follow the earlier rule from the previous task. Regressive errors are committed after it has disengaged from the previous rule but continues to try to acquire the new rule (for more details on how these types of errors are calculated, refer to the published method¹⁸).
   3. If the rat did not respond to a light cue within 15 s, the trial is categorized as an omission, and it will not count towards the total number of trials to criterion. Calculate this by first adding together the number of correct responses (located in the second column of array B in data file) and number of errors (located in the third column of array B in data file). Next, subtract this number from the total number of trials to criterion (this is the last number in the first column of array B in a data file, different from the trials to criterion).
   4. Use start and finish times recorded by the program (located at the top of a data file generated by the MED-PC script for any of the tasks on test day) to calculate time to criterion. Latency to the first lever press can also be calculated from the data file by subtracting the variable K (elapsed time in seconds from the first lever press) from the time to criterion.
   5. Average the data for each behavioral measure for rats within the same treatment group. Perform appropriate statistical analyses (depending on how many variables are being examined).

8. Brain substrates

1. Determine an interested brain area and/or aspect of cognitive flexibility. For example, if stress increases perseverative errors in the side reversal task, the orbitofrontal cortex (OFC) may be of particular interest, as previous lesion studies have indicated this brain region plays a role in many forms of reversal learning (i.e., spatial reversal tested in the side reversal task)^34,35,36. In this example, sacrifice rats after the strategy shifting paradigm is completed and examine c-fos (measure of neural activation³⁷) in the OFC using described immunohistochemical methods²⁵ and described briefly here.
   1. First, extract brains from animals and cut into 40 µm slices.
   2. Wash the tissue in phosphate-buffered saline (PBS) 4x for 5 min each, then incubate in 0.3% hydrogen peroxide for 10 min to quench endogenous peroxidases.
   3. Wash tissue in PBS 2x for 5 min each, then incubate in mouse anti-c-fos primary antibody (1:500), 3% normal donkey serum (NDS), and 0.3% Triton X overnight.
   4. The next day, wash tissue in PBS 3x for 5 min each, then incubate in biotin-SP-conjugated donkey anti-mouse sary antibody (1:500) for 2 h.
   5. Wash tissue in PBS 3x for 5 min each, then incubate in avidin-streptavidin AB complex for 1 h.
   6. Wash tissue in PBS 3x for 5 min each, then incubate in DAB solution for up to 10 min as tissue undergoes an oxidation chromogenic reaction.
   7. Wash tissue in PBS 3x for 5 min each, then mount the brain slices on glass microscope slides.
   8. Coverslip the tissue using toluene based mounting medium and image using a brightfield microscope.
      NOTE: Here, as reflected in the representative results, rats are sacrificed 30 min after the strategy shifting paradigm ends, roughly 60–90 min after the reversal task has been completed (depending on each rat’s performance in the light task). This should represent optimal timing for c-fos expression³⁸, reflecting performance in the reversal task.
2. Alternatively, cannulate a specific brain area for drug injection or viral injection prior to the execution of stress or the operant strategy shifting paradigm.
   NOTE: Researchers may want to examine how manipulating neural substrates alters the effects of stress on cognitive flexibility. For example, researchers can block a particular neurotransmitter receptor in the prefrontal cortex prior to testing.