Influence of area MT on Smooth Pursuit (view Landscape)

Reading: You will have a copy of W. Newsome, R. Wurtz, M. Dursteler & Mikami, (1985), "Deficits in visual motion processing following ibotenic acid lesions of the middle temporal visual area of the macaque monkey", J. Neurosci 5: 825-840. link to the paper online, when on brown.edu.
Link to my (archive) review of paper.

MT is the medial temporal area of cerebral cortex, a region of direction-selective cells.

In the Newsome experiment there are two regions of retina to consider: the periphery, where the target jumps to as it starts moving, and the fovea, where the SP system will try to hold the target as it moves. It is the region of MT that represents the periphery that is lesioned by 1 micro-liter of ibotenic acid.

Fovea vs Periphery: The 1 μL ibotenic acid injection kills cells, but not fibers of passage, in a localized area. Newsome et al. lesioned cells that monitor visual space along the horizontal meridian.

To begin with, the macaque is foveating, so the target is arranged to appear in the region where the lesion occurred. You will want to appreciate the fovea and the periphery (where the injection occurs) as 2 paths in the model. There will be a diminished response to the target starting to move in the space mapped by lesioned area.

Requirements:
From c:\MatlabR12\work\fold22 you can find read-only simulink model asgn12B05D. Copy and rename the model in your IP folder. Notice provisions are made in the model for peripheral movement detection.

The input on the left is a step + ramp. They combine and are taken to the scope. Next is a 130 msec "saccade visual delay" then a pathway to the saccade generator subcircuit. The saccade generator here does not require EYE feedback from the OMN; it "computes" a saccade pulse duration. The saccade generator combines the pulse with target position ("step" part of the pulse-step control).

The ramp only goes to motion detectors for the periphery and the fovea. The PERIPH detector feeds into the saccade generator input, while the foveal motion detector adds directly to the OMN input.

(1) Control experiments: Run the distribution model to see a response to step-down+ramp target, where the Periph Gain is v. low and the saccade response overshoots the correct response.

Now set the motion detector gains to zero and the ramp slope to zero. Test that the model makes correct saccades to either left or right (+/- 10 deg) saccades. Let your simulation run for 2 seconds, with max step size of 2 msec.

(1b) Turn off the step input and turn on the ramp, to 15-30 deg/sec. How does the eye follow the stimulus? What if you make saccade output always zero, what minimal gains must be set to have the eye to follow the stimulus SPEED?

(2) Now try to reproduce Fig 3 and Fig 9 of Newsome et al., showing that if the peripheral gain is zero the saccade under- or over-reaches the moving target.

(3) What happens if you have a step + ramp, as shown below?

Arrange that you can form a step + ramp target that is reached (by the "uninjured" model) with no saccade at all.

(BONUS: no FTQ) Newsome et al. say that function recovers in a few days, in spite of the lesion. While Newsome did not acknowledge the work of Robinson in highlighting the role of the cerebellum for correcting visuomotor mistakes, investigate (like you did for Gain Mod and Sacc Dysmet asgn's) recovery here. Say the perceived speed detection after lesion is 10% of what it should be. Show a "recovery" mechanism that monitors the error of initial SP and corrects the problem by increasing a downstream gain. Let's say the correction must occur at a site different from the lesion. To begin with, make a "manual" adjustment to show the site of change.

Possible FTQ's: How can you add 50 msec of motor delay to the saccade path? Explain how the saccade pulse duration can be generated in Simulink.