Direction Selectivity in motion-sensitive neurons of the visual pathway
and miscellaneous topics/demos in perception of motion

Motion is a vector defined by direction and speed.

2006: We will review 2 paper from Fried, Munch & Werblin:
"Mechanisms and circuitry underlying directional selectivity in the retina", Nature 420: 411 (2002)
link to Nature 2002 paper and

Notes on 2002 Nature paper: Voltage clamping is used to separate excitatory and inhibitory inputs to a DS cell: During clamping the amount of current needed to "clamp" the voltage of the cell is measured while the cell is stimulated. To begin with, the key result is Fig 1(e), where inhibition is seen off-center from the physical position of the DS g-cell being recorded from (the "spatial profile" they call it). Time is now right-to-left. Compare to Fig 1(b). In Fig 1(f) the same result is obtained with flashing spots, not moving bars. They estimate the inhibitory offset at 150 microns of distance on the retina: the cell responsible for the "inhibitory lobe, the starburst amacrine, has a dendritic field that extends the distance required.

Fig 2 shows simultaneous recordings from DS cell and starburst amacrines. Fig 2(d, e) has DS g-cell processes in green, starburst amacrines in cyan.

"Directional selectivity is formed at multiple levels by laterally offset inhibition in rabbit retina", Neuron 46: 117 (2005)
link to Neuron 2005 paper

Notes on 2005 Neuron paper: Are any amacrines cells already direc. selec. before reacting with DS G-cells?

linked from:
http://www.hms.harvard.edu/bss/neuro/bornlab/nb204/papers

Previously we had considered, on the topic of directional selectivity, a paper by Werblin's post-doc advisor, Horace Barlow, and W. Levick, reviewed below.

Notes on Barlow & Levick (1965)

Horace Barlow & Bill Levick, "The mechanism of directionally selective units in the rabbit's retina," J. Physiol. 179: 477 (1965), Barlow and Levick recorded from rabbit retinal ganglion cells and showed that a population of the cells are selective for the direction of movement a small target takes, moving through a cell's "receptive field" of a 10 or so square degrees of visual angle. In the Discussion section of their paper Barlow and Levick put forth two models, or hypotheses, for explaining direction selectivity, and associate parts of the models with more distal cells in the retina, such as amacrine cells.

Definition of receptive field (RF): The region of space in which visual stimuli can influence firing of a single neuron, such as ganglion cell of rabbit retina. Testing with small ON and OFF stationary stimuli is one way to map spatial extent of RF.

Basic Observation: Response in spikes/sec to one direction of movement, but not the opposite direction.
Fig 2: No significant trend in latencies across RF: How can asymmetric latencies explain DS?
Figs 3, 4: DS occurs throughout the RF
Fig 5: the slit experiment: the subunit of DS appears to be a fraction of the total RF, and is "duplicated" a dozen times across the field.
Fig 6: Collision: Two slit timing experiment: response the the second slit depends on which slit went first.




What is a critical test, to decide between the two models? Full field sustained stimulus: excitatory should respond in sustained way, inhibitory model in transient way. Inhibitory model provides the correct prediction.

Another approach: What is delta T? the order of msec...one test: response to v. slow movement in null direction: response breaks through. Another test: response to 2 slits timed for null and preferred apparent motion:

Maturana and Frenk, in pigeons: Collision test of preferred and null simultaneously moving:

after Barlow: Nigel Daw tried a pharmacology test in rabbit retina, antagonizing inhibition with picrotoxin delivered through an artery headed to retina.

%-------------OTHER TOPICS IN PERCEPTION OF MOTION, below ----------------- -------------

Demo: Pulfrich effect

Both eyes are yoked to be fixating the same point. The filtered eye is reporting what it saw about 25 msec earlier than the unfiltered eye (increased retinal delay). The diagrams below illustrate why you have the perception of nearness for left-right motion and left eye filter. In the CNS are 3x3 cells, each receiving a different left-right pair of inputs.

An explanation in terms of binocular units in the visual cortex is given below:

Juggling in strobe light
What can you learn about perception of motion-apparent motion-from the time between strobe flashes at the "cut off" frequency below which juggling breaks up? That interval is about 100 msec. Presumably the brain has the ability to correctly predict the path of movement if the sample rate is greater than 10 Hz.

Illusion of apparent motion and color. jump to link. fixate on cross in middle. How long does it take to fatigue the "magenta" pathway?

Optic flow
Computing time to collision by knowing only theta, theta dot and x0. Solve for time to collision = Tc.
Certainly Tc is y0/v, but we need to compute what they are in the terms given.

with a derivation of the solution shown in the detail below.

Cortical area MT ("medial temporal"): most of the neurons there responds best to motion, of particular speed, direction, with some limitation of receptive field. See asgn area MT.