L4 - Spatial Vision

Question 1

What is the organisation of V1?

Answer 1

• There are 6 layers between different types of neurons, some layers are input layers - to process and send it out to next processing stage, and then output layers
• All neurons stacked on top of each other all have the same optimal orientation = also called orientation columns
• The different orientation columns, neurons change their optimal orientation by 10 degrees, so all orientations are covered within 0.5 mm
• Most neurons are binocular, but can prefer input from a particular eye, so those that responds to right eye are grouped together and same for left = ocular dominance bands

Question 2

What is a hypercolumn?

Answer 2

• Hypercolumn is a collection of neurons within V1 that all respond to the same part of visual space, and have same receptive fields but within the visual space, they respond to different orientations. All respond to same orientations grouped together, but different sizes (extract different information)

Question 3

What are sinewaves/sinusoidal gratings?

Answer 3

• Black and white stripes: luminance changes, not a hard edge between bars
• Can change to look at different properties of neurons in V1
• Black to white is one cycle
• Change orientation to test neurons in different orientations
• Change spatial frequency - make the bars fatter/thinner to change luminance profile = number of cycles per image
• Change contrast: looking at light intensity differences, changes luminance profile as peaks and troughs become closer together, does not change cycles
• Change phase: where does the cycle start e.g black/white or white/black or somewhere in middle

Question 4

What is visual angle?

Answer 4

• When estimating size of stimulus, does not matter what size is in the env, only have access to what is on the retina
• Size of retinal image is important
• If you use object of the same size, and distance of stimulus from observer, size of projection changes e.g closer to you = projection is larger
• Can use the angle as the number to describe the size of the retinal image
• How many cycles fit into one degree so spatial frequency is that (looking at fat bars or thin bars)

Question 5

Why do we need sinewave gratings?

Answer 5

• Each image can be broken down into sinewaves components - fourier analysis (smallest component you need to build up any image that can exist in the world)
• Visual system conducts the equivalent of a local fourier analysis

Question 6

What is the fourier analysis?

Answer 6

• When you add three different sinewave gratings, you get a far more complex one that changes in luminance, orientations and contrast. Large combinations can make any pictures
• Can decompose any image or sound and decompose them into their individual aspects

Question 7

How does early vision act as a local fourier analysis?

Answer 7

• Hypercolumns contain neurons tuned to different orientations and spatial frequencies
• All of these neurons analyse the same patch of visual space
• Together they extract spatial frequencies and orientations contained in their local patch - conduct a local fourier analysis
• If you look in the middle of the mixed grating, you have a hypercolumn with different orientations and spatial scales and frequency in the neurons, which of these would respond to the grating?
• Each individual neuron responds to the separate component, they decompose the picture into their gratings to respond to

Question 8

What is an adaptation?

Answer 8

• Both a method and a process in the visual system
• Method: stare at the same stimulus for a long time
• Process: As a consequence of long exposure, neurons that are tuned to the stimulus property decrease their sensitivity = results in the tilt aftereffect

Question 9

How to explain the tilt aftereffect?

Answer 9

• When you look at a grating with perfectly vertical lines, have one neuron that has optimally tuned so responds in its maximum response.
• Other neurons respond but just not optimally
• When you look at all the neurons together, they create an activation profile across visual cortex, some are maximally activated and others are less so
• When presented tilted right grating and ask ppt to stare
• Those neurons that are tuned are less sensitive, other neurons respond maximally, and others do not, as you keep staring, those that are initially tuned, become least sensitive and other neurons will become less less sensitive
• Then when you show the vertical again, maximal response is diminished, so the most optimal one is a mixed one from the left side because of both stimulus

Question 10

The relationship between contrast sensitivity and spatial scale

Answer 10

• We have a small area in mid-spatial frequency = we are most sensitive, and lower in sensitivity in high and low spatial frequency = called contrast sensitivity function = allows a window of visibility and a resolution limit
• Sensitivity of our visual system differs for different spatial scales

Question 11

What produces the Contrast Sensitivity Function?

Answer 11

• Population of neurons tuned to similar spatial frequency forms a spatial frequency channel
• Neurons can respond to different sizes so form different channels depending on the size they react to, all the neurons together create an overall channel

Question 12

What is the independence of spatial frequency channels?

Answer 12

• They are relatively independent
• When pasteing 2 different photos: einstein and monroe, monroe is communicated with low spatial frequencies and einstein is high spatial frequency neurons
• Further away you are sat = less you can receive the high spatial frequency neurons, these channels make it look like monroe, but the low spatial frequency scale is dominant which is why we see einstein when we are close.