Tutorial: 2D Classification

In the third lesson, part two, we looked at the stretch reflex dependency on stretch velocity. It could have been any other stimulus-response (or input-output) relation, like the M-H-relation or alike.

In the sixth lesson we conditioned the stretch reflex with an electrical stimulation. We used a very simple stretch protocol: Only one single combination of stretch amplitude, velocity, etc.

Both of these setups apply a one-dimensional classification scheme - we use either the main or the sub classification. Why not combine the two, to see if conditioning acts different on different stretches? - For this we'll use use two-dimensional classification.

We'll start out from the settings derived in the previous lesson, ttCondt0.mat.

Conditioned Stimulus-Response

We'll investigate the stretch reflex of the Soleus muscle and measure the reflex in the EMG from that muscle. EMGs from other muscles will be acquired along with position and force / torque - the way we've done before.

As a result the following will not be changed relative to the above configuration file:

• Acquisition
• Analog Input
• Trigger

Other settings discussed below can be found in the ttIoCon0.mat.

The two setup that we are combining contained 8 and 16 classes respectively. In a 2D scheme that would make up a total of 8*16=128 classes. Let's say you want at least 10 stretches in each class and an interval between stretches of five seconds then one experiment would last one hour and 47 tiresome minutes. I'll suggest reducing the number of classes in both dimensions to achieve a shorter duration of the experiment.

Let's assume that the most important stretch velocities corresponds to rise times at or shorter than 120 ms. Also let's assume the effect of conditioning is most profound when the electrical stimulation is applied around 50 ms before stretch onset - say from 80 to 20 ms before. We'll still monitor the unconditioned stretch reflex for all velocities.

Classification

For Mr. Kick to control both the stretch velocity and the time interval between conditioning stimulation and the stretch we set both Sub and Main Classification to Stimulus Type. Choose Random for Stimulus Select. Let Main classification Act On Analog Output (stretch) and Sub classification Act On Event Timing.

The range of stretch velocities given above requires five main classes. The range of time intervals require seven sub classes plus one without stimulation - eigth in all. A total of 40 classes is defined. With a five sec interval between stretches and ten stretches in each class one experiment will still last for more than 33 minutes.

In the Y-Analysis section we set on-line monitoring to measure the peak amplitude (Maximum) of the Soleus EMG from 40 to 140 ms after stretch onset.

Event Timing

Event timer #0 controls data acquisition as well as analog output (stretch). event timer #1 controls the electrical stimulatior.

To monitor the unconditioned stretch reflex we'll turn event timer #1 off in one class (0.0).

In all other classes electrical stimulation comes first. We'll set a zero delay for event timer #1, and delay event timer #0 by the following amounts: 80, 70, 60, 50, 40, 30, 20 ms (class 0.1 through 0.7).

Analog Output

For the range of velocities chosen we'll set the five main classes Rise Time to the following values: 20, 40, 60, 80, 120 ms.