Reference no: EM132223708

Neural Data Analysis Lab: Encoding

Project

The primary goal of this lab is to introduce you to fundamental methods of analyzing spike trains of single neurons to characterize their encoding properties: raster plots, peri-stimulus time histograms, and tuning curves. The dataset for this lab is called "Lab5_CenterOutTrain" and is found under "Modules/Data for labs" in Canvas. Put it in your current directory and load it.

This behavioral data was collected using a manipulandum, which is an exoskeleton which fits over the arm and constrains movement to a 2-D plane. Think of the manipulandum as a joystick controlled with the whole arm. The behavioral task was the center out paradigm pioneered by Georgopoulos and colleagues (1982). The subject first holds the cursor over the center target for 500 ms. Then a peripheral target appears at one of eight locations arranged in circle around the center target. In our task there is an instructed delay, which means that after the peripheral target appears the subject must wait 1000-1500 ms for a go cue. After the go cue, the subject moves to and holds on the peripheral target for 500 ms and the trial is completed.

Obviously, MI neurons should respond during a time window centered around the go cue, since this when voluntary movement begins. However, MI neurons also respond during the instructed delay. This is somewhat surprising, because the subject is holding still during this time. The usual interpretation is that the subject is imagining or preparing for movement to the upcoming target. This means that MI is involved in planning as well as execution of movement.

The "Background" section of this lab will walk you through creating a raster plot and peri-stimulus time histogram for the example data found in the variable "spike." For your project, you will analyze the data in the struct "unit." This contains the spike times for 143 neurons. You also have information about 158 trials, including the instruction times, the go cue times, and direction of the target. The targets are labeled 1 through 8. Target 1 is at zero degrees, which we defined as moving to the right of the target. Target 2 is at 45 degrees, Target 3 is 90 degrees (moving straight up) and so number counterclockwise to Target 8 which is 315 degrees. The "Background" section will also walk you through how to fit a tuning curve to raw data. If we treat the direction to the peripheral target as the stimulus, we can arrange the neuronal responses in a tuning curve. These can be described with a cosine curve, where the phase of the fitted cosine corresponds to the preferred direction of the neuron.
Be sure to fulfill the "Lab Report Requirements".

Lab Report Requirements

1. Plot a raster plot and PSTH for all eight directions for a two second period centered on the go cue for one of the neurons (report which one). To get them all on the same graph, you can use the function subplot. For example, subplot(3,3,1) targets the upper right graph in a 3x3 grid, subplot(3,3,2) targets the upper middle graph and so on. Plot them such that the location of the target corresponds to the location of the subplots (see I. Project).

2. Compute the mean firing rate for each of your 143 neurons in each of the 8 directions for a 2-second epoch centered on the go cue time. Then fit a cosine tuning curve for the mean firing rate as a function of the target angle.

3. Plot the tuning curve (the mean firing rate in each direction and fit cosine tuning curve) for your best neuron. Report which neuron it was, and how you determined it was the "best."

4. Plot a histogram of the preferred directions of all 143 neurons. Picking out the preferred direction from the fit parameters is a little tricky, since the value maybe off a multiple of 2*pi. Make sure to compensate for that.

5. Discuss how good you think the cosine-tuning curve is as a model of your neuron's responses. What percentage of neurons do you think are well fit by a cosine-tuning curve?

Attachment:- Lab Neural Data Analysis.rar