New Variations for Strategy Set-shifting in the Rat

Set-shifting, a form of behavioral flexibility, requires an attentional shift from one stimulus dimension to another. We extended an established rodent set-shifting task¹ by requiring attention to different stimuli according to context. The task was combined with specific lesions to identify neuron subtypes underlying a successful shift.

Behavior flexibility is critical for survival in this changing world. One of the behavior paradigms testing this ability is set-shifting task which is being used for both primates and rodent. This task requires attentional shift from one stimulus dimension to another for changing behavioral strategy.

In a typical version of the rat set-shifting task animals have to attend to a previously irrelevant stimulus dimension. However, during real life situation, we often change our action strategy by paying attention not only to the previously irrelevant stimulus dimension, but also to the other stimuli, such as completely novel and historically relevant cue. In this article, we introduce new variation of the rat set-shifting task.

We need three different conditions for the set shift. In all three conditions, the same change in behavioral strategy is required which include two phases, an initial response strategy and a subsequent switch to a visual cue strategy. During the response strategy, a reward is given when the correct lever is pressed.

The lever remains the same throughout this phase. Next, a change to the visual cue strategy occurs when a new rule to obtain a reward is applied, as animals are required to select the lever on which a visual cue is illuminated. In Condition one, no light is given in an initial response and animals have to attend to a novel light cue.

In Condition two, a light cue indicates the correct side in the initial response strategy, but this cue is not necessarily used for making a choice. In this case, the animals attend to a historically relevant cue. In Condition three, the visual cue randomly appears above either lever, regardless if it's correct or not, so it has to be ignored.

This case requires that the animals have to attend to the previously irrelevant cue in the next visual cue strategy. On arrival of animals, groups of two or three rats are housed together. After a week, they are each moved into individual cages and food restricted.

Five days before experiments, they are gently handled daily for five minutes. Each operant chamber is composed of a house light, a pure tone generator, two identical visual cue indicators, two levers, and a food tray. On the first day of experiments, animals are put into an operant chamber for 20 minutes of habituation.

Once animals have been assigned to a particular chamber, all the behavioral tasks are performed there. After habituation, 10-15 pellets are given when returned to their home cage, allowing the animals to become familiar with a reward. The following day, magazine training begins.

Here, a total of 20 pellets are given to each animal at a rate of one pellet per minute. The next stage is a continuous reinforcement schedule. Here animals can obtain one pellet by pressing a lever once.

This continues until they receive 60 rewards or 40 minutes pass in each session. When animals have completed a session for at least two consecutive days, they proceed to the next phase. Here animals are trained on the lever press trial.

The trial commences with a three second tone. Two seconds after the tone ceases, either the left or right lever is presented. The animals are required to press the lever within ten seconds to obtain a reward.

They perform 80 trials per session with an inter-trial interval of 20-30 seconds. When they make fewer than ten omissions out of 80 trials, they proceed to the last training session. The Side Bias Test determines the animal's preference for either lever.

A successful trial occurs when there are at least two lever presses on each of the left and right levers. First, animals need to choose one lever to receive a reward. This selected side is counted.

In the next attempt, they are required to choose the opposite side to their first choice to obtain a reward. If they respond to the same side, no reward is given and the trial is continued until animals press the other side. This test has seven trials and enables us to define each animal's side preference.

From this point on, the testing session begins. A daily session has 80 trials. In this session, animals have to respond to one side of the levers that is the opposite to their preference.

Similar to the performed lever press trial training, two seconds after the tone delivery, two levers are presented, and animals are allowed to make a choice. If it's correct, they obtain a reward. If they respond to the other side, the trial is counted as an incorrect trial and no reward is given.

When animals do not respond within ten seconds, the trial is counted as an omission. This initial learning of the response strategy is continued for four days. In this phase, there are three different conditions which differ in how a visual cue is presented.

In Condition one, no light cue is given. In Condition two, a light cue is illuminated on the correct side, but animals do not have to use this cue because they learn this strategy based on lever location. In Condition three, a light cue is randomly presented either on the left or right side, which is an irrelevant light cue.

After animals have learned a response strategy based on the lever location, the behavioral rule changes to a visual cue strategy. In Condition one, animals must attend to a completely novel stimulus. In Condition two, they must attend to a previously relevant cue.

In Condition three, animals must attend to a previous irrelevant cue. Importantly, all the conditions require the same change in behavioral strategy. In this visual cue learning, the number and type of errors are analyzed in detail.

First, taking all the trials in which a light cue is illuminated on the opposite side of the previously correct lever, if animals respond to the previously correct lever, this error is classified as either a perseverative or regressive error. To separate those errors, the number of errors in a ten trial moving window is counted by advancing the window one trial at a time. Perseverative errors are scored until animals make fewer than eight errors out of the ten trial window.

After this point, subsequent errors are regarded as regressive errors. In contrast, never-reinforced errors are scored when animals make a response to the previously incorrect lever in which a light cue did not exist. Using these new variations in behavioral experimental design our recent study compared intact rats against rats with selective lesions of the striatal cholinergic interneurons.

The lesion animals were intact in acquisition of initial response strategy. As well, they did not show a significant decline in the percentage of correct responses across three conditions and in any treatment after the set-shift. However, careful analysis of error types indicated differences.

When cholinergic interneurons of the ventral striatum were ablated, animals scored more perseverative errors when they had to shift their attention to a novel stimulus. On the other hand, animals with a loss of cholinergic interneurons of the dorsal medial striatum were more perseverative to the old strategy when they had to attend to the previously irrelevant stimulus. Furthermore, this condition and associated impairment of decreased number of never-reinforced errors was observed.

In contrast, neither lesion affected a behavioral shift in which the light cue remained relevant. These results indicate that the ventral cholinergic system appears to be needed when attention to a novel stimulus is required, whereas the dorsal medial cholinergic system may be important when attention to a previously irrelevant cue is needed. Here we introduced a new variation of the set-shifting task for testing behavioral flexibility in the rat.

We believe application of these new paradigms make it possible to further understand neural mechanism underlying a flexible control of actions upon a change in behavioral rules.