NROC34: Lec 6: Visual Motion Detection (in flies)

126 views18 pages
29 Mar 2012

For unlimited access to Class Notes, a Class+ subscription is required.

Visual Motion Detection (in flies)
Slide 2:
2 broad categories of things that flies tend to do for motion vision:
ocourse stabilization (when they fly around, have to control where they’re
oobject fixation (interested in an object, need to keep track of it)
compound eye:
ohave very highly structured/complex anatomy
oif you’re interested in certain parameters of visual info, such as light
intensity or color or polarization --> can basically get this with a single
detector; complex eyes = SPATIAL vision = getting info about how these
parameters of visual stimulation vary in space
Slide 3:
spatial info:
odifferences in intensity (or color) at diff locations
oimage formation
Slide 4: Motion
motion changes over time in these patterns of stimulation and this could be due to
objects in visual space moving themselves OR could be due to self-motion
temporal factor
oimage changes
but there are other sources of change in the visual image
How are motion signals computed?
Slide 5: Compound eyes
Facets = ommatidia = small corneal lenses that are aligned with a group of
photoreceptors (rhabdom)
These corneal lenses are pointed in certain directions in space and will capture light
from a certain range of angles around that direction
It makes up a retinotopic mosaic of visual space
Slide 6: Retinotopic Mosaic
Diagram = blow-up of several facets of a compound eye
The range of angles around the direction in which a particular direction in which a
particular ommatidia is facing that will be able to enter the ommatidial lens and
stimulate the associated receptors = acceptance angle for that ommatidia
The angle between the axes of neighboring ommatidia = interommatidial angle
Each ommatidia is looking at a diff region of space and that direction plus the
acceptance angle around that direction represents the receptive field of an
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 18 pages and 3 million more documents.

Already have an account? Log in
Size of the receptive field = determined by size of acceptance angle and the
separation of neighboring receptive fields is determined by the size of the
interommatidial angle
If you want full and efficient coverage of visual space, you’d want those 2 angles to
be the same which is generally the case in compound eyes
oSize of the acceptance equals the separation between so that you don’t have
gaps and no overlap
oHave continuous sampling of visual space
oReceptor field is where the ommatidia is looking = the space that stimulates
the photoreceptor
oInterommatidial angle = angle between the centre of neighboring receptive
fields and so if the size of the receptive field is bigger than the separation
between receptive fields then you have neighboring receptors w/
overlapping receptive fields --> get 2 receptors that are sampling the same
space; if the interommatidial angle is much bigger than the acceptance
angle, then you have neighboring receptors that are looking at 2 places w/
gap in between
Slide 7: Vision = detection of light
Each ommatidium signals the luminance (brightness) in its receptive field = means
being stimulated by photons
Photons have particle-like characteristics such that there’s some randomness in the
arrival of photons to represent particular level of illumination
Intensity of light = rate of photonic stimulation = average # of photons per unit time
oBright light = more photons raining down more quickly than dim light
These photons are pounding into the membrane of the photopigments of
photoreceptor cells = depolarize and activate photoreceptors to generate stimulus
Takes certain amount of stimulation to activate a photoreceptor
Detectability of light is limited by the available light NRG (how many photons are
hitting the detector)
oThere must be some minimum amount hitting photoreceptor over certain
period of time to activate it
oBecause of the fact that it’s a stochastic rainstorm of photons = means
there’s some random fluctuation in any given sample of photons that
photoreceptor sees over a given time window
So, this photon noise is going to limit the resolution of sensitivity of visual sys in
certain ways
Slide 8: Contrast Sensitivity
Have a set of multiple receptors each looking at a diff region of space and making
their measurements of brightness of light and you want to know if there’s a
difference in stimulation in 2 locations
This is what you need if you want to perceive visual patterns = to be able to
discriminate differences in intensity between points in space --> that means spatial
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 18 pages and 3 million more documents.

Already have an account? Log in
oCan think of the same thing as detecting changes in intensity of stimulation
in same location
oSo, can have 2 receptors looking at diff places and say whether there’s a
difference in brightness between the 2 locations (spatial contrast) OR can
have one receptor saying did it get brighter or dimmer since the last time it
looked (temporal contrast)
Changes over time = temporal contrast
Can think of diagram as diff receptors looking at diff regions OR same receptor
over time monitoring a change in brightness in its receptive field
Slide 9: Contrast Sensitivity
It’s easy if the differences are large
Very bright = many photons = just counting up # of photons hitting = get bigger
signal than if there’s low illumination OR overtime
So, if there’s a big difference, then it’s easy task
But if the differences aren’t so big, then you get down to the level where these
stochastic fluctuations by photon noise will become limiting factor
oSo if photoreceptor is getting a signal that is 3 photons +/- 1 photon per
integration time window (in bright spot); dark spot is getting 2 photons +/- 1
photon per integration time window --> harder to be confident that you see
a difference here; difference in activity of these 2 receptors won’t be that
Slide 10: Spatial resolution
Line plot on right represents intensity as a fxn of distance along this stimulus -->
where it’s perfectly bright and then perfectly dark
Can also have these gradual variations --> would be the sinusoidal variation in
With spatial resolution, there are ways to analyze visual patterns that are in some
ways analogous to sound
oCan think of this idea over change in space, how rapidly does the light
intensity (luminance) vary?
oIs it changing quickly (is there a high spatial frequency of variation) or is it
changing slowly (low spatial freq)?
oSpatial freq is a measure of how quickly or slowly as a fxn of distance does
the pattern vary
For a simple pattern, where you have a regularly repeating pattern of light and dark
--> can say here’s a plot of intensity as a fxn of distance
oCan find a period of frequency (x) = cycles/degree
So if you want a visual system to detect this kind of variations in space (spatial
contrast), then the resolution (the shortest distances over which you can resolve
differences) is going to be limited by spacing of your receptors (size of receptive
You need at least two per cycle of change in the pattern b/c have to be able to see
the bright spot contrasted w/ the dark spot
Unlock document

This preview shows pages 1-3 of the document.
Unlock all 18 pages and 3 million more documents.

Already have an account? Log in

Get access

$10 USD/m
Billed $120 USD annually
Homework Help
Class Notes
Textbook Notes
40 Verified Answers
Study Guides
1 Booster Class
$8 USD/m
Billed $96 USD annually
Homework Help
Class Notes
Textbook Notes
30 Verified Answers
Study Guides
1 Booster Class