The overall goal of the following experiment is to characterize the response patterns of single neurons. While the experimental subject performs a visual discrimination task involving acquisition, extinction, and reacquisition of condition to responses. First, an experienced subject is trained to emit one of two possible choice responses following the presentation of one of two different visual stimuli.
After the animal reliably performs a discrimination responding to a random one of the two stimuli, it is not followed by reinforcement anymore. Once conditioned responding for that stimulus has deteriorated the responses, again, reinforced continuously leading to a rapid requisition of the response. The results show that individual neurons response patterns gradually change while the animal is engaged in the learning task.
The main advantage of this technique over existing methods is that different learning stages are implemented in a single session within a few hours. This permits tracking individual neurons action potentials over the course of learning. This method can provide insights into the mechanisms underlying adaptive operant behavior at the single cell level.
To begin auto shaping, place a pigeon into a testing chamber that has been set up as described in the accompanying text protocol. Then present a five second visual initialization stimulus on the center key. Immediately upon termination of the initialization stimulus or after a single peck to the response key, whichever comes first, switch off the key illumination and present the food reward by activating the food hopper for two seconds.
Keep the inter trial interval longer than the stimulus presentation. Time to facilitate learning. The initialization stimulus will be reused as the initialization confirmation and choice key in the final task.
Once the animal responds in at least 85%of trials, decrease the inter trial interval stepwise down to 10 seconds and the presentation time down to two seconds. Increase the number of responses required for reinforcement to three. Increase the number of total trials per day so the animal is trained for approximately one hour.
Repeat these training steps for the left and right response keys until response is reliably obtained to the initialization stimulus For all keys. Alternate trials for the left, right, and center key randomly now present the initialization stimulus first at the center and upon getting a response, present the initialization stimulus at either side key. Do not reinforce the center key response.
Alternate which side keys are reinforced randomly from trial to trial. If there is no response to the center key after five seconds, end the trial. This training stage takes about three days following STI initialization.
Introduce two new stimuli, which will later serve as familiar stimuli in the final task, repeat the autos shaping steps with the stimuli to get a response to the familiar stimuli. Training on the new stimuli will usually produce correct responses within four days. Illuminate only the choice key, which is correct for the given familiar stimulus.
This phase of training should take about one week once the subject performs reliably with at least 85%of responses to the correct side key. Introduce free choice trials where both side keys are illuminated during the choice phase. Responding to the correct side is reinforced with food axis for two seconds.
Incorrect responses are followed by house lights going off for two seconds. If no response is given within three seconds, set up the program to terminate the trial and restart the inter trial interval. This phase of training should take about two weeks.
Gradually increase the fraction of free choice trials from 20%to 100%over the course of training. If the subject performs at greater than 90%correct in free choice trials, decrease the reward probability for correct responses to 0.5 while increasing the number of trials per session to 1000. Do not change parameters every day, but choose them flexibly.
Depending on the performance level of the subject. This training phase lasts about four weeks. Pigeons are inherently neophobic and will be reluctant to peck on unfamiliar stimuli.
Therefore, repeat auto shaping for a wide range of diverse visual stimuli. Once training is complete, the final single interval force choice task with novel stimuli can be run under different reinforcement conditions. During the warmup stage, let the subject perform 50 trials with the familiar stimuli.
Only set reward probability for these stimuli to be less than one during all phases to prevent loss of motivation to respond during the acquisition stage. Randomly alternate trials with presentation of the familiar and novel stimuli assign different response keys as correct for the two novel stimuli and reinforce every correct response. Compute the percentage of correct responses as a running average over the last 120 trials.
Acquisition is complete once performance for each novel stimulus exceeds 85%after at least 150 trials have been executed during the extinction stage. Stop reinforcing correct and punishing incorrect responses to a random novel stimulus. Begin the requisition phase when correct.
Responding to the extinction stimulus drops below 60%and the animal has completed at least 150 trials total in this phase. During the reacquisition stage, reinforce correct and punish incorrect responses to the extinction stimulus as was done during the acquisition stage. Terminate the session when performance for this stimulus exceeds 85%and the animal has performed at least 150 trials in this phase.
Implant electrodes. Once animals complete the entire acquisition extinction reacquisition process several times, use a new pair of novel stimuli for each session and advance electrodes at least 125 micrometers or half a revolution of the drive screw before starting. If no action potentials of sufficient signal to noise ratio are observed, abort the session and place the animal in the homepage before trying again the next day.
Using the offline analysis software Ban pass filter all channels from 500 to 5, 000 hertz with steep roll offs, extract spikes with amplitude thresholds and sort them manually using principle component analysis. Custom written MATLAB code for sorting results is available at the MATLAB central file exchange. A well isolated single unit should be a clearly separated cluster.
In principle component space have no sign of multiple units when all recorded wavelengths are overlaid and plotted as a heat map. Have symmetrically distributed peak waveform amplitudes stable recording throughout the session as shown by unchanging peak amplitude no or very few spike events that occur during the refractory period of the proceeding. Spike and a signal-to-noise ratio of at least two.
Inspect the raw channels offline for movement related artifacts and discard any channels if necessary. Test for possible electrical artifacts occurring during key pecking. By examining the time histogram of spike counts relative to each registered key Peck.
Pecking induced artifacts will show up as a peak in the histogram within 50 milliseconds of time, zero as an extra check plot. The waveforms of all spike events registered within 20 milliseconds of a key pack separately and compare them to spike waveforms detected outside of this window. This behavioral data for one animal performing the single interval force choice task shows 100%correct choices from the start of training for novel stimulus one and reaching criterion performance for novel stimulus two within 180 trials in this session, novel stimulus two was randomly chosen to undergo extinction, meaning that all choices following this stimulus are no longer rewarded or punished during extinction.
Performance decreases for the extinction stimulus, but stays high for the other novel stimulus. Criterion level for extinction is reached at trial 370 once correct and incorrect responses are again reinforced and punished during requisition. The 85%performance criterion is reached on Trial 402.
The response pattern of one unit in the needle pale co laterally recorded while an animal was performing. This task show response modulation during the presentation of a novel stimulus. In the acquisition phase, the units respond strongly to novel stimulus.Two.
Early in learning, there is little change in firing during the extinction and reacquisition phases. There is also little response to novel stimulus. One during the session.
Another unit responds during right but not leftward movements suggesting sensory motor coding. After watching this video, you should have a good understanding how to train animals step by step to perform the final single interval forced choice task with dynamic reinforcement contingencies. This technique will pave the way for researchers in the field of learning and memory to explore the reorganization of neural networks during different types of learning at the single cell level.
