Anyone wondering why I haven’t posted about the results of the OKR data? I have to keep some secrets … and while we’re on that subject, make sure to attend my presentation (hint hint!). 🙂
Last week was completely full of experiments, and also included troubleshooting of a camera. More on that below.
We have four different treatments, and four experimental conditions with varying LTD functionality. Don’t remember what LTD was? The short explanation is that LTD is the opposite of making connections between neurons. The cerebellum comes pre-made with a lot of unnecessary connections, so LTD just has to break those connections that contribute to errors in performance.
In order to control for order effects, we randomize the order that each subject receives each treatment. We also wait at least three days between experiments (in reality, this is usually about a week) so that the subjects have ample time to forget all the training they had during the experiment (whether 1 Hz Gain Up or 1 Hz Gain Down, see last week’s post).
Here’s what my experiment randomization table looks like. I’ve labeled four treatments (A-D).
As you can see, all subjects (numbered 1-26 with some gaps) receive OKR training. Then each treatment is done at least four times a week with the exception of A in week 4 (which is a small mistake, but now cannot be corrected). Treatments are in a random order, and all are the same. This way, we can be sure that learning is from the treatment itself rather than due to order effects.
We have a camera in the Raymond Lab that is designed to look at eye movements by taking a video of the eye, and then match those eye movements to the signal made by movement of the magnet. This is called calibration — it calibrates the magnet signal to the actual eye movements, and produces a scaling factor, which tells us how many degrees of eye movement corresponds to how many volts in the signal we receive.
Because we are trying to create a new experimental rig downstairs, I have been helping troubleshoot the new camera for this rig. The camera has to be positioned very precisely relative to the eye during experiments, and because I have done many experiments, I have developed a knack for this. I also know the math behind how the camera calculates eye position based on the two images it receives (yes, it’s actually two cameras). If you want to know what it is, you can look at this paper.
Next week will be the last week of experiments!