FAQ

I get many questions during the year, and I thought that I would catalogue them here, since they may address questions or confusions you have had.

--------------------

Could you please explain to me Spotlight Metaphor?

Thank you,

Nichol

Hi Nichol,

I think there should be some discussion of that in the book. You remember that we said that there is too much information for your visual system to process all at once, so it has to select and prioritize, using visual attention. The spotlight metaphor is an analogy that says that visual attention works like a spotlight on a stage. If you think about going to the theater to see a play, there is too much stuff on the stage for you in the audience to keep track of. Luckily the director (and the lighting crew) helps you out by shining a spotlight on the important bits. Your visual attention is supposed to work the same way, according to this theory: your visual attention shines a mental 'spotlight' on important information so that you can selectively process it and ignore everything else.

hope that helps,

dr blaser

-----------------------

Dear professor.

I understand that Fourier analysis has to do with sine waves and their division... but how can I apply it to the situation where you withdrawal the high frequency components of an image and put the image back together?

Thanks, R

Dear R,

Basically, the idea is this: the visual scene that we see around us has important small details but also important large features. Your visual system has to detect both the small stuff and the large stuff. It does this by using different size receptive fields. Large receptive fields detect large, broad features in the scene (like a person standing in front of me), while small receptive fields pick up small details (like the individual hairs in your eyelash).

Now, it turns out that there's a mathematical technique called Fourier Analysis for analyzing images that lets us break an image down into a bunch of abstract patterns, called sinusoidal gratings. Every image can be broken down in this way, into a big set of gratings. Gratings with low 'spatial frequency', that is, with broad stripes, represent the large features in the original image, while gratings with high 'spatial frequency' (lots of thin stripes) represent the fine detail in the original image. So, a Fourier analysis of a picture with lots of fine detail, and not much in the way of broad features like a line drawing cartoon, will result in lots of high frequency gratings. A Fourier analysis of a picture with lots of large, broad features like a sunset on a cloudy day, will result in lots of low frequency gratings. To look at from the other way around, let's say you take a picture, for instance of a city skyline and do a Fourier analysis. Now, throw away all the high spatial frequency gratings and put the image back together again without them. Now you will have an image of the skyline without the detail, in other words a blurry, fuzzy skyline, as if you were looking without your glasses. Or, let's say you do the Fourier analysis of the skyline and throw away the low spatial frequency gratings, and put the image back together again. Now you will have a picture with detail, but without broader features, in other words something that looks like a line drawing cartoon of the skyline, just edges and corners and dots and lines.

The good thing about gratings is that we can use them to construct a very good test of visual acuity. If you measure how well a person can see these different gratings, you are measuring the Contrast Sensitivity Function for that person. (And notice that once we know how well a person can see gratings of various spacial frequencies, we know how well they can see any image whatsoever, since all images can be broken down (or built up) out of these gratings).

Another connection is this: since the visual system uses different size receptive fields to analyze pictures in order to find the small details (high spatial frequencies) and the large features (low spatial frequencies), your visual system is in essence doing a Fourier analysis of the visual scene.

Hope that helps,

dr blaser

------------

Good Afternoon Professor,

I am having trouble finding the distinction between physical signals and psychological experiences. Also what do you want us to know about action potential?

Thanks, C

Dear C,

The most important thing is to recognize that there are perceptions, which are psychological, like experiencing the color red or the taste of sweet. On the other hand, there are the physical 'stimuli' that produced these perceptions, in this case, long wavelength photons and sugar molecules.

Action potential is just another word for the 'firings' or neural 'spikes' that are the language of neural activity. These are the bursts of electricity that neurons produce. This is what we were raising or lowering (from baseline) by stimulating various parts of the receptive fields of retinal ganglion cells and simple cells in the PVC. The rate of these action potentials can range from basically 0 to 1000 spikes per second.

Hope that helps,

dr blaser

-----------

Hello,

I was wondering if you could explain what the difference is between the 'likelihood principle' and the 'fundamental problem of seeing depth.'

Thanks, V.

Dear V,

Sure, I can help out with that.

The likelihood principle says that, in general, the visual system goes with the 'most likely' interpretation of a particular image. The example that I gave in class was that if you see an image of a piece of wire that looks to be straight, your visual system assumes that it is looking at a piece of wire that is actually straight in 3D, even though it could actually be a bent wire hanger that you just happen to be looking at from the right angle so that it looks straight. Since that is a one-in-a-million chance, you go with the more likely interpretation.

The fundamental problem of seeing depth is just the fact that the 'information' that you have to figure out depth, the raw material so to speak, is just the two images on each of your retinas, *which are completely flat*. It just shows what a marvelous detective your depth system is - it can figure out the 3D shape and the distances and positions of objects just based on two flat pictures.

see you in class,

dr blaser

-------------------------

Good Afternoon Professor,

I am having trouble with four specific questions on the study guide.

Any help that you can give me would be greatly appreciated, seeing how I have been struggling with these last four for two days.

Hi Carla,

Let me see if I can help out a bit.

1) Why are we better at judging faces right-side up, rather than upside down?

Remember that we saw that there are actually individual neurons in visual cortex that are 'tuned' for faces? There's a special topic on my website about that too. These neurons are tuned for rightside-up faces - after all, that's the type of face that we encounter in everyday life, so that's the kind of face you'd want to be good at seeing. Having such specialized neurons sounds quite a bit like Template matching, and not so much like Structural description ('recognition-by-components'), so that's why it is relevant to our discussion about object recognition.

2) The role of geons play in recognition by components structural description theories?

Geons are the 'components' - they are like the lego pieces that all objects are built out of, according to structural description (recognition-by-components') theories.

3) which parts of the visual system are trichromatic? Which use opponent mechanisms?

The retina - more specifically the cones - are trichromatic (you have three cone types, S, M, and L, tuned for short, medium, and long wavelength light, respectively). After that - some ganglion cells, LGN, and visual cortext color areas - almost everything is recoded in terms of opponencies: red vs. green and blue vs. yellow.

4) What are color afteraffects?

A color aftereffect is when your color vision gets 'biased'. You have opponent mechanisms, for instance, red vs. green. If you stare at something red, or hang out in a room with a red light bulb, or put on red sunglasses - this red system goes crazy but eventually gets tired-out, fatigued. Then, since it is weak, and green is relatively strong, everything will look kinda greenish. And vice versa - look at a lot of green, then for a little while everything will look reddish. Same for blue vs. yellow.

Good luck with the rest of your studying,

dr blaser

---------------

Hello Professor,

I have a couple more questions.

1) Why is color constancy a good thing?

The first thing to realize here is that - even though you aren't aware of it - the 'color' of the illumination (the light falling onto a scene, like your living room, or a beach or a classroom) differs radically from place to place and sometimes time to time. The light falling in a classroom contains a lot of short wavelength light, while the light falling in your living room contains a lot of long wavelength light. What that means is that if you did not have color constancy, then everything in class would look quite bluish, and everything in your living room would look quite reddish. So, for example, if you took a piece of paper and looked at it in sunlight, it would look white - ok, no problem - but then you bring that piece of paper into class and - wtf? - it looks blue, then you bring it home and suddenly it looks red, as if it were swapped with another color of paper. Color is a very important property of objects (red berries might be poisonous and blue healthy), so you don't want colors to be changing all over the place just because the light is changing. You want red berries to always look red and blue to always look blue no matter where you are eating them. Color constancy is what allows you to keep your color experience constant in spite of these changes in illumination. Basically, your visual system just looks around in an particular scene and tries to figure out the average color falling on the scene. So if everything looks reddish - like in your living room - then your visual system turns down the 'red' knob in your head, so to speak, and cranks up green to compensate. Just as if you were making color adjustments on a faulty TV.

2) Is stereopsis the most precise depth cue?

Yes!

Hope that helps,

dr blaser

-------------------------------

1) How do reichardt detectors recognize direction tuning and speed tuning?

Direction is determined by two things. The first is the placement of the little receptive fields. If they are next to one another, then the detector is for either right or left motion. If they are one above the other, then it is for up-down motion, etc. Then you have to look where the delay is. If the delay is on the right, then the detector is tuned for leftward motion, if it is on top, then it is tuned for downward motion, etc. Speed tuning is determined by the length of the delay (shorter delay means the detector is tuned for faster motion) and the spacing of the two little receptive fields (far apart means faster tuning).

2) learned taste aversions?

3) Specific Hungers Theory?

Those two definitions are in the book.

4)How do senses accomplish transduction?

This depends on the sense. In vision the photoreceptors convert photons into neural electrical activity. In the ear, the hair cells convert mechanical vibrations into neural electrical activity. In the nose and tongue, special receptors convert chemical molecules into neural electrical activity.

---------------------------

Hi Dr. Blaser:

Two questions I have that weren't answers in class yesterday. Would it be possible for you to clarify what you're looking for regarding these two items on the study guide? thanks!

1. Diagrams of how a lens creates an image of an extended object (not just a 'point source'):

Hi Ben, for this one check out my webpage. I think it's the lecture topic called 'light and lenses'. There's a diagram there of a waving man and his image through the convex lens. That's what I had in mind.

2. The visual spectrum and its general place in the broader electromagnetic spectrum:

Here I just want you to know that the ROYGBIV is just a small section of a broader spectrum that contains other wavelengths, like gamma rays, x-rays, microwaves, etc... There's a figure in the book on page 28 showing this, with the wavelength measurements and a picture of the roygbiv rainbow.

Hope that helps!

dr blaser

----------------------------

Dr. Blaser

Can you help me recognize an illuminant and reflectance spectral power distribution and what it means?

Ok, to answer this, first remember what each of these terms mean. Luminance is the amount of light *coming off* of something (like a lightbulb, or off of a surface like the sun reflecting off of your shirt), Illuminance is the amount of light *falling onto* something (so the amount of light from the sun that hits your shirt in the first place), and Reflectance is the percentage of light that a surface reflects (so, a white shirt reflects about 90% of the light that hits it, while a black shirt only 10%).

Now, for the spectrum part. We say that luminance is the amount of light coming off of something. Ok, but what about the wavelengths? Does the light coming out of your lightbulb have lots of long wavelengths (like a red disco light) or lots of short wavelengths (like a blue disco light), or a whole big mix of all different wavelengths (like white light, or sunlight for instance)? This is what the spectrum shows. The x-axis on the graph is 'amount of light' and the y-axis shows the wavelength of the light.

The idea is basically the same for Illuminance spectrum and for Reflectance spectrum, but there you are graphing what kind of wavelengths are falling onto a surface, and what kind of wavelengths the surface likes to reflect.

Does that make sense?

Here's an image from the book showing the Reflectance spectrum for raw and cooked meat. See that the raw reflects more long wavelength than the cooked? That's why it looks redder! :)

FAQ

1:09 PM

Dear professor.

Could you help me understand Weber’s law and JND’s?

Thanks, C

Dear C,

Firstly, you should know that the 'Just Noticeable Difference' (say in discriminating that one weight is heavier than another, or that one light is brighter than another) is a constant percentage of whatever you are starting with. So, maybe you need to have something be 10% heavier for you to notice the weight increase, or 1% brighter.

The other connection to make is that this is how neural receptive fields work: their response is based on the percentage difference of the light falling in the plus areas versus the minus areas. So, there might be 50% more light falling in the plus areas, which tends to drive the response upward, or 30% more light in the minus areas which tends to drive it downward.

hope that helps,

dr blaser

--------------

Hi professor Blaser,

I'm very confused about this question regarding the increase in db's.

Standard JND

1 db 1.1

10 db 11

20 db 22

50 db 55

Based on these data, what db loudness is ‘one JND louder’ than at 100 db sound?

Answer=110 db's.

Are the standards 1, 10,20, and 50 determined by a formula, and if so how? Or are the randomly chosen?

I understand multiplying the JND (.1) by each standard =10.1=1+10=11, 20.1=2+20=22, etc. but doesn't each JDN have to correspond to the next standard?

Also, I am lost at how the answer 110 db's was determined?

I know it is probably not very difficult but I just don't get it and am frustrated.

Thank you,

Hi J,

Ok - good questions. Yes, the standards are just arbitrary. Sometimes we'll use various standards in the laboratory for experiments, to get a view of how sensitive an observer is. Sometimes the standards just come from real world applications. Let's say that you work for Dell and are deciding whether the engineering department should put better screens in the laptops that would make them brighter, say from 10 watts to 12 watts, but raise the cost of the units by $50. Should you do it? Well - will the users even notice the increase? In other words, would a 20% increase in wattage result in a JND increase? You would have to do an experiment to figure that out. Or rely on data from some experiments that have already been run.

In the example you are talking about, that whole table showing 1->1.1 and 10->11 and 20->22, etc. was just supposed to tip you off that the JND's come from a 10% increase in db level. Once you know that, you can predict how much louder you need to make the 100 to increase it by one JND: 10%. So, a 10% increase from 100 is 110.

Does that make sense?

Hope that helps,

dr blaser

--------------

Prof. Blaser~

I just wanted to see what you could tell me about the differences between absolute and relative depth perception cues. I cannot seem to figure out how to differ the two.

Also, when I was looking over my notes, I noticed that I didn’t seem to write down what the definition of color constancy is. The same with metamers. What are these things, and what do I need to know about them?

Thanks a lot professor, S.

Dear S,

Glad to see you started your studying early! Absolute depth cues let you know the distance of some object from you, in actual units of distance, like feet. Accommodation and Vergence are definitely absolute. You can easily calculate from the power of your lens where it's focussed. That answer will be in meters. Same for the geometry of Vergence.

Contrast that however with a relative cue, like occlusion. How could you possibly figure out from, say, the knowledge that "the car is in front of the building" how far away the car is? Relative cues just give you information on the depth order of things. This is in front of that, which is behind that, etc.

Color constancy: perceptually, this is the fact that color appearances of objects appear quite stable, even if the light falling on them changes. If you put on red sunglasses, eventually everything goes back to looking normal. A room doesn't appear to have radically different colors if you change your light bulbs (even though there may be, in fact, a radical change in the illuminant from, for instance, incandescent to florescent). The visual system does this by figuring out the 'average' color in the scene (which is probably due to the light source, or sunglasses, it reasons) and factors this out. If everything in the room is looking kinda reddish, the visual system turns down its red knob (or turns up green, is another way to think of it) to compensate.

Metamers are two physically different things that appear identical perceptually. Monochromatic yellow and Red+Green yellow both look yellow to you, but they are very different physically. A bunch of tiny 1mm black and white lines looks like a big grey square from across the room; those too are metamers.

----------------

Dr. Blaser,

I have a few questions on color vision, i am sure i will have more later.

1. What is purple's wavelength mixture?

Just use the color wheel for this; you'll see that it takes a mixture of red and blue light. Purple is a non-spectral color, for humans with normal color vision. That means that we can't get the perception of the color purple from a single wavelength. It takes two, and those two turn out to be wavelengths that, on their own, appear red and blue.

2. Sunlight is spectral because it is a mixture of colored lights, so then what is "white sunlight?" (in the isaac newton, prism, part you reference white sunlight, isn't that sort oxymoron?

No - wait: sunlight is not a spectral color. Spectral colors are in the spectrum, white is not in the spectrum. Spectral colors are those that can be generated, perceptually, from looking at a single wavelength. There is no single wavelength that appears white. It takes a combination of wavelengths, for instance, red and blue and green, all beamed into the eye at the same time, will create the experience of 'white'.

3. Subtractive mixtures? i understand that when you mix a lot of paint colors together, you get brown, i guess my question is can you apply subtractive mixtures to the color wheel?

To do subtractive color mixtures, you'd need a subtractive color wheel; the one we have only works for additive. You can think about it this way: in additive color mixtures, you start with some kind of lights(s) and you add more to it. So, maybe you start with red and blue. That looks purple. Then you do an additive color mixture and add in some green. Okay, now that looks white. All of that is captured in the color wheel. On the other hand, in subtractive color mixtures, you are subtracting wavelengths, so you are removing lights from the mixture. We can talk more about this later....

4. in the trichromacy of cone color vision section, you specify a qualification that instead of x=r+g+b, we have x+b=r+g, so couldn't you just say x=r+g-b?

Yes, that's fine. The original equation was just put that way, notice, so that it is completely additive. Yours, which is perfectly fine, requires subtracting blue. But for our purposes, just focus on the simple case: x = r+g+b. Keeping in mind that we can change the levels of r and g and b (that is, say, turn on red and green full blast and turn off blue to create x=yellow, for instance)

5. Dichromacy- you say that the result of dichromacy is "lights of all possible wavelength compositions can be perceptually matched by a mixture of just two primary lights," is this saying that not all color experiences are possible, because not all colors can be matched?

Your color experiences are completely dependent on what kind of color system you have. Monochromats have no color experiences (all wavelenghts are 'metamers'). Dichromats have a lot, but have huge parts of the spectrum which are metamers. Thrichromats have fewer metamers, but a lot more than people with 4 cone types. If you have 100 cone types, you have yet far fewer metamers, but you still have some. etc.

6. are both types of mosaic color blindness in women just referring to the genetic acquisition of colorblindness? i ask this only because colorblindness in women is so rare.

Colorblindness in women is rare because women have a 'backup' copy of their genetic material on their other Y chromosome. So, if their color blind on one, they use the info on the other, and then they'll have normal vision. Men have no such backup, and will be out of luck...

7. i am a little confused by simultaneous color contrast. (this may be a totally separate question but, color opponent cells, lets say +R-G, due to the idea of simultaneous color contrast would not have two responses if both red and green were shown in the cells receptive field?)

8. Color constancy is such that if the illumination spectrum decreased, the reflectance would also decrease however due to color constancy, the color of an apple would not change, just maybe the hue would change?

Be careful here. Reflectance is a fixed property of a surface. If the surface doesn't change, like a ripening banana or repainting the table, reflectance is fixed. It doesn't have anything to do with the lighting conditions. If a mirror reflects 99% of the light that falls on it, well, that's that. That's true on earth, on mars, in daylight or at dusk. Now, if the illuminant spectrum changes (doesn't have to 'decrease', could just change - like more warm reddish light from regular bulbs versus cooler blue-biased light from fluorescent bulbs), then - I think this is what you meant before - different relative strengths of wavelengths will be reflected from an object. Color constancy notices this change, and corrects for it - cancels it out - so that your color experience can be stable. It is not perfect, and sometimes gets fooled, or can't compensate enough.

With all this stuff, take it step by step. One thing that can be confusing, that maybe is a sticking point for some understanding is this:

There is the physical spectrum, which is just a bunch of different wavelengths of photons (radio waves, ultraviolet, etc.). This is purely physical, and has a life absolutely independently of humans and vision and eyes, and colors etc.

Now,

different creatures are 'tuned into' different parts of this spectrum. We happen to have light sensitive things called retinas that respond to the part of this physical spectrum around 400-600 nanometers (I forget the exact numbers, that's not important).

Okay, fine. But, notice even at this stage, we are not saying anything about colors - just the fact that we have some photopigment stuff that gets a chemical reaction from photons with wavelengths in this range.

Now, if you have three different cone types, then your brain can compare and contrast the relative activity of each of these. By virtue of this calculation, the brain can make an educated guess at the wavelengths of light that are hitting the retina at any given moment. Was it 450? 525? A little 400 and a little 425 and a lot of 550? etc. This sort of calculation - critically - requires at least two cone types. If you only have one, you are out of luck at figuring out anything about the wavelength. You know that you are getting something in the 400 to 600 range, perhaps, but you have no idea what. Importantly too though, is that even with three cone types, you can't discriminate all the wavelengths and combinations, you still have some metamers: different wavelengths and combinations that can't be distinguished from one another. Other creatures may have more cone types to reduce the number of metamers.

Great. Now, if you can start to figure out different wavelengths, then the mind is now in a position to assign different experiences - perceptions - to these different wavelengths. After all, that's the way the mind let's you know what's 'out in the world': it gives you perceptual representations of physical stimuli. These experiences that get assigned to the different wavelengths are the *colors*. Now, you have a spectrum of color experiences, like when you look at a rainbow, or the output of a prism. But don't be confused, this rainbow-spectrum is purely psychological. A dichromat, like a color-blind person, or a dog, sees a different spectrum, a different rainbow, with fewer colors. A monochromat doesn't see the rainbow or 'spectrum' at all, just a gray stripe, maybe. Someone with 4 cone types thinks we're losers, as they see an even richer rainbow, with other unimaginable colors stuck in between the ones we share.

The point is that there is the physical spectrum, and then there is the perceptual experience of the spectrum. The former has nothing to do with observers or eyes or minds. The latter, in contrast, is completely in the eye of the beholder.

hope that helps,

dr blaser

---------------

Hi Dr. Blaser,

I just have two quick questions. Any answers you can send my way would be greatly appreciated. Thanks again.

1) The first is about Monochromats. I know that they only have one type of photoreceptor, but does it have to be rods? Could it also just cones? If it can also be one type of cone does is matter if it is a S, M, or L cone?

Great questions!

As you suspected, a monochromat is a monochromat no matter which photoreceptor they have. It may be that they just have rods or just S cones or just M cones or just L cones by themselves. Usually, it's the case that they have only rods, but it doesn't have to be that way.

2) My other question is about geons and the role they play in recognition when talking about the structural description theory. I'm assuming that they are "building blocks" of different shapes and what not. If that's what they are, then I understand the role they would play. However, if that's not what they are or how the function then I guess I'm a bit lost.

You are dead-on with your definition of Geons! Think of them as more sophisticated Legos - you stick them together in various ways to build objects. Stick a curvy cylinder on top of a flat 'brick' and you get a briefcase, etc.

Good luck studying - it sounds like you have a good grasp on the material so far!

dr blaser

-------------

Hi professor Blaser, How are Ganglion cell receptive fields measured?

Hi C,

Ganglion cell receptive fields are measured by sticking an electrode into a ganglion cell, and then measuring its electrical activity ('firing rate' or 'spikes per second'), while you shine lights on various photoreceptors on the retina. Here's some more details about that: when your light hits some group of photoreceptors that are wired up to the ganglion cell (in other words, in the cell's receptive field), you'll likely get a change in the ganglion cell's activity relative to baseline: either an increase in firing rate or a decrease in firing rate. If you get an increase, that means your light landed in the ON (or '+') area of the ganglion cell's receptive field; if you get a decrease, that means your light fell on the cell's OFF ('-') area.

2. What is the ‘language’ that Ganglion cells use to communicate with the brain?

Isomerization converts photons into the 'language' of the brain, and the 'language' of the brain is electrical activity. In basically all neural structures (except for photoreceptors that have graded potentials), this electrical language takes the form of shorts bursts (firing/spike/impuse) of electricity, called an 'action potential'.

Hope that helps,

dr blaser

---------------

Hello Professor Blaser,

I'm unsure about what is meant when you ask about the graded potential and action potential. If you can help me that would be great. Thanks, M

Hi M,

Action potential means the neuron sends out a little burst of electricity, always the same amount of electricity, just sometimes with the bursts coming faster or sometimes coming out more slowly. A good analogy is a water-gun fight: one person has a little plastic gun, with a pump trigger, so pump the trigger, squirt out some water, wait for the trigger to pull back, squirt again, always squirting the same amount of water: that's like an action potential. The other guy has a shampoo bottle full of water, and he's just constantly squeezing it, sometimes firmly, sometimes softly. That's a graded potential - electricity always coming out, just sometimes more electricity (higher voltage), sometimes less (lower voltage).

good luck studying!

dr blaser