Two Mechanisms of Vision: Ambient Vision and Focal Vision (1 Viewer)

[Omid]

Human vision involves two parallel processes; one ambient, determining space at large around the body, the other focal which examines detail in small areas of space. More generally, in all mammals, there are two brain pathways for processing information from the eyes, an evolutionarily ancient one and a more modern one. The ancient pathway, which is also present in vertebrates such as fish and frogs whose brains do not have a cortex, runs from the eyes to the optic tectum in the midbrain. The other pathway, that evolved in the mammalian line, runs to the primary visual cortex.

Ambient vision runs automatically beneath the level of consciousness. This visual system

1. is color blind
2. has wide field of view (mediated by the peripheral photo-receptor cells of the retina)
3. has low spatial resolution
4. works day and night
5. uses a "body centered" frame of reference
6. detects metric (absolute) size and direction of objects that constitute space at large
7. has extremely fast reaction time (direct access to certain muscles)

The focal visual system

1. has color discrimination
2. has narrow field of view (about 1-2 degrees) mediated by cells in the fovea region of the retina
3. has high spatial resolution
4. works only during daytime
5. uses an "object centered" frame of reference (sees parts of an object relative to each other)
6. detects non-metric (relative) size only; is mostly blind with respect to direction of objects with respect to the observer
7. has slow reaction time

Furthermore, focal vision requires attention while ambient vision does not. Ambient vision works automatically beneath the level of consciousness (in the same manner that our heart and digestion system work).

I understand that many readers of this forum will find this theory extremely surprising and difficult to imagine. I felt the same way when I first came across this theory about two years ago when I read a seminal paper published in 1968 by the late Prof. Colwyn B. Trevarthen. Dr. Trevarthen was a New Zealander- British scientist working on visual perception at Harvard University when he first formulated this theory. He passed away last summer at the age of 93.

I gave a presentation on this topic at a major European optical manufacturing firm a few months ago. I will attach my presentation slides for your enjoyment. If I sense that Bird Forum members are interested, I will be happy to continue the discussion and explore the implications of this theory on our visual experience when using magnifying optical devices such as binoculars.

Have a nice weekend!
-Omid

 

[Maljunulo]

Fascinating

 

[Cavendish]

Really interesting - thanks for sharing. When using binoculars I would imagine both systems are engaged. But perhaps some users are more attuned to ambient vision and therefore will always subconsciously prefer a wider FOV in their binoculars than others?

 

[tenex]

I understand that many readers of this forum will find this theory extremely surprising and difficult to imagine.

Not at all, 1968 is almost sixty years ago, and this makes immediate sense. (It reminds me of Kahneman's "fast" vs "slow thinking".) It's easy to see how the research didn't extend to binoculars. But your slideshow doesn't actually propose anything further about them yet. Surely your talk didn't end there?

When using binoculars I would imagine both systems are engaged. But perhaps some users are more attuned to ambient vision and therefore will always subconsciously prefer a wider FOV in their binoculars than others?

Variations in AFOV are fairly minor, it's always quite small, so I'd imagine the principal effect of any binocular is to cut out ambient vision (almost?) entirely. So the first sort of result one could expect is having panning make some people sick; perhaps there are others? (I believe ambient vision is related to the vestibular system, not mentioned above.)

 

[Omid]

I am pleased to see that this topic has been found interesting. Thank you, members who responded favorably! Let's continue:

In humans, the ambient visual system (which predates the focal system in terms of its evolutionary history) controls our eye movements. It monitors a wide field of view continuously and detects areas that have salience, then it tries to orient the eyes towards those points such that their image falls on the high-resolution central part of the reina (the fovea). After about 200 milliseconds of fixation, the eyes are directed to a new point, and they stay focused on that point (or direction) for another 200 milliseconds before moving to a new point (the is an average value, fixation time can vary from about 150 to 300 millisecond).

The focal visual system recognizes objects. Both systems together construct a mental 3D map of the outside world through these rapid fixations on certain salient points in the scene. The ambient visual system decides what those salient points are and directs the eyes towards them through a series of rapid eye movements called saccads. Once a target point is fixated, the ambient visual system stabilizes the eyes on the fixation point and tries to avoid slippage of the object image from the fovea.

The world we "see" is constructed based on a series of disjointed, narrow FoV, retinal snapshots weaved together into multiple 3D mental maps. A mixture of two of these mental maps constitute what we consciously experience {see footnote}. Several other mental maps exist and are used to control and guide our "actions". Our hands and our oculomotor muscles do not see the same visual world that our "conscious ego" sees!!

Here, we only focus on the consciously perceived world. To construct the 3D visual world that we consciously experience, the ambient visual system takes into consideration the direction of each fixated object with respect to the body and its distance (which it magically infers using multiple factors, some visual and some tactile). Multiple high-resolution images captured by the focal visual system are then scaled (magnified, minified or even rotated) before being "pasted" in the correct spatial position in the low-resolution 3D mental map that ambient visual system has already built.

The diagram below shows a simplified representation of visual pathways in the human brain. The focal vision pathway goes from the eyes to the visual cortex at back of our head. The ambient vision is subserved through the tiny brain nucleus called "Superior Colliculus". This pathway is analogues to what birds and reptiles have. These species do not have a visual cortex. So, it seems that evolution did not replace this primitive visual system in higher animals but added a second system on top of it!! That's why we have two visual systems looking through the same eyes



Next topics:

How does magnification affect focal vision?
How does magnification affect ambient vision? (sneak preview: large apparent field of view in binoculars disrupts and confuses our ambient visual system leading to anxiety, fear and a sensation of motion sickness as tenex correctly noted in post #4).

Thank you again for your attention
-Omid

Footnote: A living creature needs to create and maintain an egocentric worldview and an allocentric worldview at the same time. His actions and survival depend on continuously switching between these two views (or maps). This conflict also arises frequently in visual arts and cinema. See the book "The Power of The Center" by Rodulf Arnheim.

 

[Holger Merlitz]

Human vision involves two parallel processes; one ambient, determining space at large around the body, the other focal which examines detail in small areas of space. More generally, in all mammals, there are two brain pathways for processing information from the eyes, an evolutionarily ancient one and a more modern one. The ancient pathway, which is also present in vertebrates such as fish and frogs whose brains do not have a cortex, runs from the eyes to the optic tectum in the midbrain. The other pathway, that evolved in the mammalian line, runs to the primary visual cortex.

Ambient vision runs automatically beneath the level of consciousness. This visual system

1. is color blind
2. has wide field of view (mediated by the peripheral photo-receptor cells of the retina)
3. has low spatial resolution
4. works day and night
5. uses a "body centered" frame of reference
6. detects metric (absolute) size and direction of objects that constitute space at large
7. has extremely fast reaction time (direct access to certain muscles)

The focal visual system

1. has color discrimination
2. has narrow field of view (about 1-2 degrees) mediated by cells in the fovea region of the retina
3. has high spatial resolution
4. works only during daytime
5. uses an "object centered" frame of reference (sees parts of an object relative to each other)
6. detects non-metric (relative) size only; is mostly blind with respect to direction of objects with respect to the observer
7. has slow reaction time

Furthermore, focal vision requires attention while ambient vision does not. Ambient vision works automatically beneath the level of consciousness (in the same manner that our heart and digestion system work).

I understand that many readers of this forum will find this theory extremely surprising and difficult to imagine. I felt the same way when I first came across this theory about two years ago when I read a seminal paper published in 1968 by the late Prof. Colwyn B. Trevarthen. Dr. Trevarthen was a New Zealander- British scientist working on visual perception at Harvard University when he first formulated this theory. He passed away last summer at the age of 93.

I gave a presentation on this topic at a major European optical manufacturing firm a few months ago. I will attach my presentation slides for your enjoyment. If I sense that Bird Forum members are interested, I will be happy to continue the discussion and explore the implications of this theory on our visual experience when using magnifying optical devices such as binoculars.

Have a nice weekend!
-Omid

Dear Omid,

Thanks a lot for these observations! I believe that these two modes of vision have also given rise to different, rather complementary approaches to the accessment of binocular performance: There is the approach by the Zeiss guys, Köhler and Leinhos, who arrived at the (in)famous twilight factor in certain limit cases. Their performance measurements were based on the recognition of details on Landolt rings, which is fully foveal performance. On the other hand, there is the approach by Berek of Leitz, who determined threshold contrasts for target detection. Here, it is not about details, but about whether or not a target can be seen at all under difficult conditions, and this is, I believe, dominated by ambient vision.

In my book I show how these two approaches can lead to rather different results, and this is no surprise. Unfortunately, Köhler & Leinhos did not appreciate Bereks approach and rather criticized is as being wrong, and since Berek passed away shortly after the publication of his work, there was noone left to defend him, so that his approach got forgotten, until I have revived it in a publication of 2015.


Cheers,
Holger

 

[tenex]

Some care is required to give a brief summary of complex material, and issues of interpretation can arise, so I have a few specific questions already.

In humans, the ambient visual system (which predates the focal system in terms of its evolutionary history) controls our eye movements. It monitors a wide field of view continuously and detects areas that have salience, then it tries to orient the eyes towards those points such that their image falls on the high-resolution central part of the reina (the fovea). After about 200 milliseconds of fixation, the eyes are directed to a new point, and they stay focused on that point (or direction) for another 200 milliseconds before moving to a new point (the is an average value, fixation time can vary from about 150 to 300 millisecond).

The focal visual system recognizes objects. Both systems together construct a mental 3D map of the outside world through these rapid fixations on certain salient points in the scene. The ambient visual system decides what those salient points are and directs the eyes towards them through a series of rapid eye movements called saccads. Once a target point is fixated, the ambient visual system stabilizes the eyes on the fixation point and tries to avoid slippage of the object image from the fovea.

Actually, saccades. It's nice to see evolution credited this time around; it makes so much sense of things like this. I'm curious what the ambient system's criteria for "areas of salience" might be, if it's only the focal system that "recognizes objects"... merely differences of contrast etc? (Or motion of course.) And why do you claim that the ambient system is in subconscious "control" of all eye movement, including the orientation and use of focal vision, the length of its fixation and the size of its saccades (typically much longer and smaller respectively than for ambient vision)? Rather, focal vision is (primarily?) intentional. It seems from a bit of searching that the ambient system is dominant in the first couple of seconds orienting to a scene, before focal vision takes over.

To construct the 3D visual world that we consciously experience, the ambient visual system takes into consideration the direction of each fixated object with respect to the body and its distance (which it magically infers using multiple factors, some visual and some tactile).

Here again you seem to be attributing too much agency to the ambient visual system itself. Any process incorporating tactile input[?] is more than a visual system.

How does magnification affect focal vision?
How does magnification affect ambient vision? (sneak preview: large apparent field of view in binoculars disrupts and confuses our ambient visual system leading to anxiety, fear and a sensation of motion sickness as tenex correctly noted in post #4).

Surely you mean small AFOV does? Although there must still be more room than I imagined for the ambient system to operate, since researchers seem to consider focal/central vision to be around 5-6°, while binoculars have ~60° AFOV, which of course is still a good deal less than normal vision. And I recall the effect of magnification on focal vision being previously explored in parts of your thread New Horizons II.

I came upon a thought-provoking detail: researchers have reported more and longer fixations on objects in natural scenes vs urban ones! But overall there seem to be contradictory results on many issues regarding central vs peripheral, focal vs ambient vision, which seems surprising and worrisome. (Studies generally involve two-dimensional images rather than direct observation of scenes, for obvious practical reasons.)

 

[Omid]

Dear Omid,

Thanks a lot for these observations! I believe that these two modes of vision have also given rise to different, rather complementary approaches to the accessment of binocular performance: There is the approach by the Zeiss guys, Köhler and Leinhos, who arrived at the (in)famous twilight factor in certain limit cases. Their performance measurements were based on the recognition of details on Landolt rings, which is fully foveal performance. On the other hand, there is the approach by Berek of Leitz, who determined threshold contrasts for target detection. Here, it is not about details, but about whether or not a target can be seen at all under difficult conditions, and this is, I believe, dominated by ambient vision.

In my book I show how these two approaches can lead to rather different results, and this is no surprise. Unfortunately, Köhler & Leinhos did not appreciate Bereks approach and rather criticized is as being wrong, and since Berek passed away shortly after the publication of his work, there was noone left to defend him, so that his approach got forgotten, until I have revived it in a publication of 2015.


Cheers,
Holger

Hello Holger,

Thank you very much for taking time and reading my post(s) in this thread. I remember reading about the two assessment methods you mentioned in your excellently written paper published in the Journal of Optical Society of America (2015). Is there any new material added in your book? If not, then pleases allow me a few days to read the paper again and refresh my memory. Then we can discuss if we can interpret those two approaches in terms of ambient vs focal vision.

In the meantime, I invite you to read the attached short paper by the late American psychologist Dr. Herschel W. Leibowitz. Items 2 and 3 on his list are particularly relevant to our discussion. I also highly recommend the attached seminal paper by Dr. Margaret Livingston (professor of neuroscience at Harvard) and her late supervisor Dr. David Hubel (winner of Nobel prize in 1981). Table 1 in Livingstone and Hubel's paper summarizes the distinct characteristics of the two visual systems.

Sincerely,
-Omid

 

[Holger Merlitz]

Hello Holger,

Thank you very much for taking time and reading my post(s) in this thread. I remember reading about the two assessment methods you mentioned in your excellently written paper published in the Journal of Optical Society of America (2015). Is there any new material added in your book? If not, then pleases allow me a few days to read the paper again and refresh my memory. Then we can discuss if we can interpret those two approaches in terms of ambient vs focal vision.

In the meantime, I invite you to read the attached short paper by the late American psychologist Dr. Herschel W. Leibowitz. Items 2 and 3 on his list are particularly relevant to our discussion. I also highly recommend the attached seminal paper by Dr. Margaret Livingston (professor of neuroscience at Harvard) and her late supervisor Dr. David Hubel (winner of Nobel prize in 1981). Table 1 in Livingstone and Hubel's paper summarizes the distinct characteristics of the two visual systems.

Sincerely,
-Omid

Hello Omid,

Thanks a lot for these papers, which I will study when I have a little free time! If you have my publication of 2015, you will have all the information you need. The book simplifies everything a bit, reduces the usage of math, and it may here and there add some explanations and speculations, but nothing entirely new. The comparison with the Blackwell data set is only shown in the paper.

Cheers,
Holger

 

[Omid]

Hi Holger,

I read your paper again and I was impressed with your keen knowledge of classic works on visual acuity and attention to details. For example, you correctly observed the very small range of contrast that exists in viewing natural scenes (a natural black object has a reflectance of about 2-5% and a white object about 90%, so maximum contrast ratio in nature is about 9 to 1. Practical situations where one tries to spot a bird or wild animal in nature admit much lower contrast ratios, may be 1.5 to 1 or even less.

I don't know how exactly Kohler and Leinhos arrived at their formula, but it gives the following results as you have summarized: for daylight performance only magnification matters; for nighttime performance only aperture matters, and in the no man's land (twilight), they mix the two factors and come up with the infamous "twilight factor".

In the main part of the paper, starting from Berek’s human vision model, you eventually arrived at the threshold contrast curves shown in Fig 0.3. Again, I don't know how Berek developed his formulas, but the curves in your Fig. 03 closely resemble similar curves obtained by other searchers (see plots below). Essentially, what these curves show is that human pattern vision needs both "sufficient contrast" and "sufficient brightness" to produce best acuity.

If I were to write my own version of your paper, I would start with a set of reliable contrast sensitivity curves from the literature and would proceed by making arguments as follows:

Dependence of Visual Acuity on Illuminance:
At low light levels, contrast sensitivity of the eye is approximately 8% and maximum resolution is approximately 6 cycles per degree. As ambient light levels increase, the eye's contrast sensitivity improves, and it can now detect contrasts as low as 0.5% for spatial frequencies in the range of 5 to 10 cycles per degree. Note that for photopic (daylight) viewing, sensitivity to contrast diminishes both for high and low spatial frequencies. (This last little detail is not anticipated by either K-L or B theories discussed in your paper. But then again, those researchers were mainly concerned with smallest possible objects and also with low light conditions. So, it's OK. I don't blame them.)

What does a binocular do?

• It reduces the apparent spatial frequency of the target by a factor of M. This means, moving towards left side of the graphs shown below. Under low light conditions, moving towards left on each graph always leads to better contrast sensitivity. Under daylight conditions, magnification is not always beneficial: if an object already has features in the range of 5-10 cycles/degree, enlarging it further does not help!
• It could reduce illuminance too. A telescope cannot increase illuminance of a reflective extended object, nor can it increase its contrast. But, if the instrument exit pupil is smaller than eye pupil, it could reduce perceived target illuminance. The amount of this loss can be measured approximately by comparing exit pupil area with eye pupil size (to be more precise, we need to take Stiles-Crawford Effect into account as well). So, depending on exit pupil size, we may need to move down on the graphs below if we are using such low-aperture binoculars.

Now, comparing binoculars for their "scientific target detection capability" becomes simple: start from a given point on the graphs (pick your desired object size & illuminance), then go left proportional to M and a bit down (if exit pupil is less than 7mm) and then see which binoculars gives better contrast sensitivity! I suspect we will get conclusions very similar to what you already pointed out in your paper: 8X56 binoculars are king!!

Cheers,
-Omid

PS1. Looking at self-luminous point sources is a whole different story. A telescope can (and often does) increase the luminance of a point source as its aperture is increased.

PS2. We didn't talk about "focusing binoculars" while trying to detect a small object under low contrast conditions. Inability of an observer to focus his binoculars under low contrast conditions is not considered in any of the above scientific theories (That's why they are called scientific!)

PS2. The fact that human vision has two mechanisms is not considered either! To detect something (e.g. a bird), it must be "seen" first by our ambient visual system in the peripheral parts of the field of view. Its salience must be established and then the two "foveae" are directed towards it via coordinated saccadic movements of the eyes. The ambient visual system has very different contrast and spatial frequency sensitivity characteristics. In fact, contrast is not its primary stimulus, motion is. That's why staying still is a great way to prevent detection by other animals. A realistic binocular performance measure should determine how various binoculars increase the salience of a peripheral target (if they do so at all).

PS4. When the object illuminance is in the scotopic range (e.g. nighttime), there is no focal vision and therefore no pattern or form recognition capability in the human visual system. The curves in Fig. 5, 6 and 7 should not go to the scotopic zone. They should stop at mid mesopic zone.

 

[Maljunulo]

This conversation, while fascinating, has reached the point where (for me) it may as well be written in Akkadian, on small clay tablets.

 

[Canip]

My view: at last a fascinating conversation with a „niveau“*** that may keep some of us from quitting the BF over the usual nerve-wrecking
„SFL is better than HDX“ - „no it isn’t“
exchanges in some threads here.

Sure, I do have an elitist attitude, but as I said some time ago, I tend to be proud of it

*** „niveau“ may look like arrogance, but only from below

 

[Holger Merlitz]

Hi Holger,

I read your paper again and I was impressed with your keen knowledge of classic works on visual acuity and attention to details. For example, you correctly observed the very small range of contrast that exists in viewing natural scenes (a natural black object has a reflectance of about 2-5% and a white object about 90%, so maximum contrast ratio in nature is about 9 to 1. Practical situations where one tries to spot a bird or wild animal in nature admit much lower contrast ratios, may be 1.5 to 1 or even less.

I don't know how exactly Kohler and Leinhos arrived at their formula, but it gives the following results as you have summarized: for daylight performance only magnification matters; for nighttime performance only aperture matters, and in the no man's land (twilight), they mix the two factors and come up with the infamous "twilight factor".

In the main part of the paper, starting from Berek’s human vision model, you eventually arrived at the threshold contrast curves shown in Fig 0.3. Again, I don't know how Berek developed his formulas, but the curves in your Fig. 03 closely resemble similar curves obtained by other searchers (see plots below). Essentially, what these curves show is that human pattern vision needs both "sufficient contrast" and "sufficient brightness" to produce best acuity.

If I were to write my own version of your paper, I would start with a set of reliable contrast sensitivity curves from the literature and would proceed by making arguments as follows:

Dependence of Visual Acuity on Illuminance:
At low light levels, contrast sensitivity of the eye is approximately 8% and maximum resolution is approximately 6 cycles per degree. As ambient light levels increase, the eye's contrast sensitivity improves, and it can now detect contrasts as low as 0.5% for spatial frequencies in the range of 5 to 10 cycles per degree. Note that for photopic (daylight) viewing, sensitivity to contrast diminishes both for high and low spatial frequencies. (This last little detail is not anticipated by either K-L or B theories discussed in your paper. But then again, those researchers were mainly concerned with smallest possible objects and also with low light conditions. So, it's OK. I don't blame them.)

What does a binocular do?

• It reduces the apparent spatial frequency of the target by a factor of M. This means, moving towards left side of the graphs shown below. Under low light conditions, moving towards left on each graph always leads to better contrast sensitivity. Under daylight conditions, magnification is not always beneficial: if an object already has features in the range of 5-10 cycles/degree, enlarging it further does not help!
• It could reduce illuminance too. A telescope cannot increase illuminance of a reflective extended object, nor can it increase its contrast. But, if the instrument exit pupil is smaller than eye pupil, it could reduce perceived target illuminance. The amount of this loss can be measured approximately by comparing exit pupil area with eye pupil size (to be more precise, we need to take Stiles-Crawford Effect into account as well). So, depending on exit pupil size, we may need to move down on the graphs below if we are using such low-aperture binoculars.

View attachment 1632340

Now, comparing binoculars for their "scientific target detection capability" becomes simple: start from a given point on the graphs (pick your desired object size & illuminance), then go left proportional to M and a bit down (if exit pupil is less than 7mm) and then see which binoculars gives better contrast sensitivity! I suspect we will get conclusions very similar to what you already pointed out in your paper: 8X56 binoculars are king!!

Cheers,
-Omid

PS1. Looking at self-luminous point sources is a whole different story. A telescope can (and often does) increase the luminance of a point source as its aperture is increased.

PS2. We didn't talk about "focusing binoculars" while trying to detect a small object under low contrast conditions. Inability of an observer to focus his binoculars under low contrast conditions is not considered in any of the above scientific theories (That's why they are called scientific!)

PS2. The fact that human vision has two mechanisms is not considered either! To detect something (e.g. a bird), it must be "seen" first by our ambient visual system in the peripheral parts of the field of view. Its salience must be established and then the two "foveae" are directed towards it via coordinated saccadic movements of the eyes. The ambient visual system has very different contrast and spatial frequency sensitivity characteristics. In fact, contrast is not its primary stimulus, motion is. That's why staying still is a great way to prevent detection by other animals. A realistic binocular performance measure should determine how various binoculars increase the salience of a peripheral target (if they do so at all).

PS4. When the object illuminance is in the scotopic range (e.g. nighttime), there is no focal vision and therefore no pattern or form recognition capability in the human visual system. The curves in Fig. 5, 6 and 7 should not go to the scotopic zone. They should stop at mid mesopic zone.

Hi Omid,

These are useful ideas and they could indeed be published. The binocular, when reducing the spacial frequencies of motifs, is effectively making use of Ricco's law, which in a nutshell says that the larger the object, the lower the contrast at which it its barely detectable. Berek essentially started with the two limiting regimes, Ricco's and the Weber-Fechner law, and then he mathematically interpolated between them. This way he derived his 'universal' formula for treshold contrasts, and then, in a second paper, he combined all that with the imaging properties of binoculars to derive their performance parameters. I can send you all that stuff via Email if you like to follow it up.

Cheers,
Holger

 

[tenex]

This conversation, while fascinating, has reached the point where (for me) it may as well be written in Akkadian, on small clay tablets.

This is presumably due to absent context. A graph for example is supposed to help one understand a situation, but what are "trolands", why inverse-modulation-threshold, etc etc? Perhaps reading the attached papers will help. It would be nice to have a title and link also for Holger's 2015 paper, which is otherwise not immediately visible on his site. And at some point to get back to ambient vision.

 

[Omid]

Hi Omid,

These are useful ideas and they could indeed be published. The binocular, when reducing the spacial frequencies of motifs, is effectively making use of Ricco's law, which in a nutshell says that the larger the object, the lower the contrast at which it its barely detectable. Berek essentially started with the two limiting regimes, Ricco's and the Weber-Fechner law, and then he mathematically interpolated between them. This way he derived his 'universal' formula for treshold contrasts, and then, in a second paper, he combined all that with the imaging properties of binoculars to derive their performance parameters. I can send you all that stuff via Email if you like to follow it up.

Cheers,
Holger

Hi Holger,

When I said, "If I were to write my own version of your paper" I meant it as a figure of speech. Obviously, you are far more qualified to extend/update your own work

I will return to the thread's original topic shortly.

Best,
-Omid

 

[Holger Merlitz]

This is presumably due to absent context. A graph for example is supposed to help one understand a situation, but what are "trolands", why inverse-modulation-threshold, etc etc? Perhaps reading the attached papers will help. It would be nice to have a title and link also for Holger's 2015 paper, which is otherwise not immediately visible on his site. And at some point to get back to ambient vision.

Sorry, I haven't cited it properly: Performance of binoculars: Berek's model of target detection

A preprint version can be downloaded here on my webpage.

Cheers,
Holger

 

[Omid]

Visual Salience

Visual salience is defined as the distinct subjective perceptual quality that makes some items in the world stand out from their neighbors and immediately capture our attention. A living creature uses his sense of vision for survival, therefore objects or events in the surrounding environment that are most relevant to him will be attributed a higher degree of salience.

Motion is a primary component of salience
Evolutionary older species such as amphibians, do not have foveas like we do. Therefore, they do not have Focal Vision. Their entire visual system is based on Ambient Vision.

A dead fly placed nearby does not attract a frog's attention. The frog does not seem to see or, at any rate, is not concerned with the detail of stationary parts of the world around him. He will starve to death surrounded by food if it is not moving. His choice of food is determined only by size and movement. He will leap to capture any object the size of an insect or worm, providing it moves like one. He can be fooled easily not only by a bit of dangled meat but by any moving small object. His sex life is conducted by sound and touch (See the seminal 1959 paper by Lettvin and his colleagues at MIT).

Motion is, in fact, the primary stimulus for all biological vision
Physiological research such as those conducted by Lettvin and his colleagues as well as extensive studies by the eminent psychologist James J. Gibson have revealed that primary stimulus for biological vision is not the light intensity pattern on the retina but rather the change in this intensity pattern caused by motion. In mathematical terms, if the light intensity pattern on the retina is denoted U(x,y), this two-dimensional function (or image) is not the stimulus for vision; stimulus for vision is actually dU(x,y)/dt caused by eye motion, object motion or both.

Other factors that contribute to salience
Other than finding food, a living creature is concerned with two other challenges: avoiding predators and finding a proper mate. A frog's sex life is conducted by sound and touch. His choice of paths in escaping enemies does not seem to be governed by anything more devious than leaping to where it is darker. Since he is equally at home in water and on land, why should it matter where he lights after jumping or what particular direction he takes? He does remember a moving thing providing it stays within his field of vision, and he is not distracted.

For the human visual system, anything which has a "face" has very high salience. This means, our peripheral visual system is subconsciously searching for faces in the wide visual field around us. The most prominent features of a face are eyes. So, eyes have very high salience, especially if they include a dark disk surrounded by white matter. White sclera in the eyes is a uniquely human trait. Other primates have dark eyes that do not stand out as human eyes do. Even more salient than a face with two white eyes, is a face with two white eyes which are directly looking at you! And even more salient than that, is a beautiful female face looking directly at you while smiling!

In addition to objects that move and have attractive (or threatening) eyes, our peripheral visual system is particularly attentive to objects of unusually high contrast or unusually different color. Color red is unique in this respect and serves as a very powerful attractor of attention. Prof. Semir Zeki of University College, London, even claims that there are unique aspects in perceiving the color red in area V4 of the human visual cortex.

What happens once a salient object (or direction in FoV) is detected
Once an object of high salience in our peripheral visual field is detected, our Ambient visual system will often execute a saccadic eye movement so that said object is brought to the center of our visual field for closer examination by our focal vision. Furthermore, the blurry moving retinal image that results from the fast saccadic movement of the eyes is deleted from our conscious perception. We can never see our own eye saccades in the mirror!

Once an object is fixated, our visual system may maintain the fixation or jump to a new fixation point based on a competition between a mental force for attention and a mental tendency for exploration. It is interesting that even during a fixation, the eyes are never perfectly stable. They make random movements and drift around the point of fixation before being brought back by a small saccade. The small saccades that maintain the fixation point on the fovea are distinct from large saccades that direct our fovea to a new fixation point.

 

[Omid]

Enlarging visual angles to create "a magnified object" vs enlarging visual angles to create "a magnified space"

The most basic way to describe binoculars effect on vision is by considering their magnification factor M. This means, binoculars increase visual angles subtended by "an object" by a factor of M.

If we assume M=10 and the object is a small bird sitting alone on a tree branch 10 meters away, the visual experience of seeing said bird via a10X telescope could be described in several ways:

• It appears that the bird remains at the same location, but its size is magnified (it looks like a large bird).
• It appears that the bird has come closer (as if the bird is sitting on a branch only 1m away from the observer).
• It appears that the observer has moved forward (as if he is looking at the bird from a point of view (PoV) 9m ahead of his actual location).

Which one of these scenarios do you think is more representative of our visual perception?

Now another experiment:

Imagine you have a polaroid SLR camera which takes nice color pictures and prints them out instantaneously. You go to Venice, Italy and stay at the center of Piazza San Marco with your camera. Then you take a series of pictures each in a direction slightly to the left of the previous picture. You carefully adjust your rotation angle so that your pictures show a continuous 360-degree field of view once stitched together in a panoramic display.

You bring all the pictures home and paste them around the inner wall of a circular tent of suitable radius which you have set up in your backyard. If the radius of the tent is correct, your pictures will cover the entire 360 degree of the inner wall, and you'll get a panoramic view of Piazza San Marco if you stand at the center of the tent and turn your head around.

During your next year, you decide to do something more impressive: you use a 10X telephoto lens on the same camera and take a sufficient number of photos to cover 360 degrees. You bring all the magnified images home and paste them around the inner walls of a suitably sized tent to create a 360-degree "magnified display" of Piazza San Marco. You stand at the center of the tent as before and look around.

What do you think you will see? Will you see a "magnified panorama" of Piazza San Marco?!! Can you see the details of paintings on the facade of the St Mark's Basilica now?

 

[Maljunulo]

I’m not sure how I would distinguish between having the bird move closer to me, and my moving closer to the bird.

That’s almost Einstinian.

 

[tenex]

• It appears that the bird remains at the same location, but its size is magnified (it looks like a large bird).
• It appears that the bird has come closer (as if the bird is sitting on a branch only 1m away from the observer).
• It appears that the observer has moved forward (as if he is looking at the bird from a point of view (PoV) 9m ahead of his actual location).

Which one of these scenarios do you think is more representative of our visual perception?

What does "more representative" mean? Presumably by "location" you mean distance, but one could choose any of these descriptions at will, as that's all they are, and people often do. One simply sees a magnified image of the bird (and everything else in the FOV) with the wider context excluded.

You bring all the pictures home and paste them around the inner wall of a circular tent of suitable radius which you have set up in your backyard. If the radius of the tent is correct, your pictures will cover the entire 360 degree of the inner wall, and you'll get a panoramic view of Piazza San Marco if you stand at the center of the tent and turn your head around.

During your next year, you decide to do something more impressive: you use a 10X telephoto lens on the same camera and take a sufficient number of photos to cover 360 degrees. You bring all the magnified images home and paste them around the inner walls of a suitably sized tent to create a 360-degree "magnified display" of Piazza San Marco. You stand at the center of the tent as before and look around.

What do you think you will see? Will you see a "magnified panorama" of Piazza San Marco?!! Can you see the details of paintings on the facade of the St Mark's Basilica now?

If the prints cover 360°, they (and the unmentioned print size) will determine the radius. If the 10x prints are the same size, they will likewise determine a radius 10x larger. But why should we care? The level of discussion needs to be raised a bit again. (If that's what this is meant to be.)