There are two types of light meters, reflected meters (drawing on left) aimed at the subject from the camera or from the camera location, which reads light reflected from the subject. And incident meters (drawing on right) aimed at the camera from the subject position, which reads the actual direct light incident upon the subject - at the subject, but totally independent of the subject. Since cameras are not located at the subject, they can only use reflected meters.
Reflected meters (are the necessary type in cameras) are aimed at the subject from the camera, and read the light reflected from the subject. A white dress will reflect a lot of light (bright), which the meter sees and will darken to be middle gray. A black dress will not reflect much light (dark), which the meter sees and will brighten to be middle gray. Every reading will be placed at about middle gray (so white is usually underexposed, or black is usually overexposed). Fortunately, many average scenes do average out to a middle gray (a mix of many shades, trees, sky, etc) and so often do come out near right, but there are many exceptions. We just learn these things, that we must realize what we are metering, and make necessary correction (Exposure Compensation).
When we walk up to a scene with major areas of untypical light or dark color, the alarms should immediately sound. We need to realize that we will need to take corrective action (because the reflective meter is going to be fooled). This is no big deal, it's just photography, and it's easy to learn how to handle it.
Incident meters (are necessarily hand held at subjects position) instead read the actual light source (meter is aimed at camera, Not aimed at the subject). The middle of that range will be placed at middle gray. So this leaves room for white dresses to appear white (brighter), and black dresses to appear black (darker). Incident meters are usually close to correct, because incident measures the actual light, which is independent of the variable reflection from the subjects colors. Meters like Sekonic often offer both metering methods, for continuous light and for flash.
Incident meters tell you settings for how bright the light is. Reflected meters tell you setting about how much of it is reflecting from the subject's colors. Reflected meters are extremely popular, but we need to learn a bit of skill about using them. Incident meters are extremely desirable for manual studio flash. The article on next page compares a reflected meter in a camera, with a handheld Sekonic L-308S incident meter.
The simplest trivial test is shown next, anyone can easily check this. Spot Metering DOES NOT mean correct exposure of the scene, or even of the spot. It only means the spot will come out middle gray. Results depend on which spot you choose, and how you will compensate it. No compensation was done here, and each Spot case here varies exposure a couple of stops. Reflective meters only seek a middle gray result, including Spot metering. Spot metering simply produces a middle gray spot, on whatever Spot you choose. If not a wise choice, then it is up to you to know how to compensate it correctly. My concern is that beginners don't seem to realize this, but Spot metering is an advanced subject, very far from point&shoot.
Example of Spot Metering: D800 camera A mode, f/5.6, ISO 100, hazy bright sun, near noon. Black and gray and white cards.
My point was to show that Spot metering reproduced the black card as middle gray at 1/60 second, or the gray card as middle gray at 1/400 second, or the white card as middle gray at 1/1250 second. Yes, the 1/60 second certainly helps the background shadows, but it blows out the cards - the card results were overexposed, and simply not the black or gray they should be (the black card comes out middle gray). Or of course, spot metering on the white card makes it come out middle gray. Spot metering merely makes the selected spot be middle gray. That could be correct or not, which is your choice and your call and your responsibility. Note of course, Spot metering only sees the spot, and totally ignores all the rest of the scene.
So the point is, if using Spot metering, you also better know how to compensate so the spot comes out its correct brightness. There is much thinking involved in spot metering. If spot metering on a human face, it is common knowledge that a proper starting point is to open up one more stop. You need to know too. The human brain is an important light meter accessory. :)
As said on previous page, black reflects little light, and is seen as dark, and the camera meter overexposes it to get middle gray. White reflects a lot of light, and is seen as bright, so the meter underexposes it to get middle gray. Simply is how reflected meters work. It is all they can do. It is very helpful to realize this. The Center Weighted metering saw a large center area of mixed colors, and averaged that result to middle gray, which is generally about correct in typical cases. Matrix metering saw that too, but also saw the sky area at the edge, which was pretty bright, so in this case, it reduced the exposure (1/3 stop). But IMO, my subject was in the center, and I'm a fan of Center Weighted metering.
We spot meter to ensure that spot comes out middle gray intensity (specifically, for the purpose to be independent of all other areas). We might Spot meter on a face to isolate it from the background, but hopefully then we know we must compensate about +1 EV to make the face come out right. Odds are that Center metering (which includes more spots, more area) normally has a better chance. I'm just trying to say, Spot metering is no magic plan, it is very special, and instead it requires considerably more thinking and knowledge about what we are trying to do. We should know how we plan to compensate it. The concept of Spot Metering has no meaning for Incident metering (which the camera cannot do anyway). If we meter the incident light itself, all spots will be reproduced with reasonable accuracy.
Spot metering has unique meaning for TTL flash. The flash system does NOT do Spot metering (flash has its own central system). But selecting Spot does switch TTL BL mode to be TTL mode - meaning no balancing with background. This is because ambient Spot is about the spot, and has no knowledge or concern about the background. And again, only ambient light does spot metering, the flash system does not, but TTL mode does switch off reducing the flash for the ambient. Spot metering is often used for flash indoors to force TTL (more below).
General camera reflective metering is greatly affected by the reflectivity of the subject's reflected colors.
Exposure methods that are independent of variable subject color reflection:
Gray cards: We can directly meter the reflection off of a 18% gray card (from close, including nothing else in that meter's view), and then use the same exposure settings for a scene in the same light (assuming scene and card are in the same light, and at the subjects position - same distance from the light - inverse square law always applies). Then the camera will expose the subject in the same way. This simulates using Incident metering, in that the light incident on the gray card will be independent of the reflectivity of the actual subject, and will instead depend on the reflectivity of the card (chosen for the purpose). Specifically, this means that any tones in the subject which average to match the cards 18% reflectance will also come out middle gray.
Note that the term middle gray card (analog) and midpoint of the histogram (digital) are not the same thing - very different concepts - not really related. Gray cards were a film concept, and histograms are a digital concept. Kodak has not made an 18% gray card in over twenty years. Kodak sold that part of the business to Silver Pixel Press around 1995, who sold it to Tiffen in 2000. Both sold Kodak cards, and Tiffen still makes Kodak Wratten filters, but apparently now they sell rights to "Kodak" gray cards to others. Anyway, generally, the card idea is that if we think this subject's average overall tone matches an 18% card, the result will come out about right. And it often about does, at least for many typical wide range scenes, but Kodak told us if we meter the reflection off of their 18% gray card, and then use that exposure for a scene in the same light, we should increase that exposure by 1/2 stop (1/3 stop today). If we meter on an 18% card, and then open Kodak's 1/2 stop, which is the same as metering on a 12.5% card, all else equal (our light meters mostly all use 12.5%). This method provides a known reflectivity to be metered, metering an actual middle gray subject, which the meter will make it come out middle gray, therefore striving for a more accurate result - specifically independent of the subject. Be sure to angle the card slightly to prevent any shadows or overly bright glare/flare. In the Spot metering example pictures above, the Center Weighted metering chose 1/3 stop open from the 18% card Spot metering value (which is not quite 1/2 stop, but the camera was set to do 1/3 stops).
One advantage is that this gray card method is independent of the subject's own reflectivity values (like incident metering is independent). Any error this way depends on the difference in the scenes actual average, and the 18% card's average. It can also be done several other ways. If in bright sun, metering from the clear deep blue sky (low, and away from sun) works about the same as an 18% card. A green grass lawn is close too. In the old days, Kodak told us metering on the yellow box the film came in (included black ink) was about the same. Don't expect great precision, things will vary every time you move your camera - we are speaking ballpark. And this logic will always still depend on your actual subject also being average.
Or we can use the open palm of our hand as a gray card. Metering from our outstretched palm is not "average", and will be nearly one stop brighter than 18% (is more like spot metering from the face). Then we know that we will want to set the camera for about one stop more exposure. No matter what we meter, a reflected meter will place the result at middle gray - regardless of where it "ought" to be placed. Then we compensate for how THAT middle gray differs from the actual scene's actual gray average. The metered object must of course be in the same light (and same distance from light source) as the intended subject.
And there are other metering tricks possible (called technique). Say the subject is sitting in the shade under a tree with very bright background, or is indoors with their back to a bright window. In those cases, our camera meter mostly sees the bright sunshine background behind the subject. It does not understand that the background is not our subject. It exposes for the bright sunlight, and the picture of the subject comes out very dark. Spot metering can help then, or faster, we can just aim the camera down lower, maybe as far down as to focus on their feet, specifically down enough to exclude the bright background, and then meter there. The light level is often about the same lower there, and the subject distance is essentially the same there (our brain is a great tool to evaluate situations), but this view specifically omits the bright problem outside. Half press the shutter, and hold to lock this reading (locks metering and focus), then raise camera to reframe, and the final shutter click will use that locked meter setting for this subject (Nikon AE-L Exposure Lock options can interfere or modify this method - menu C1 on some). But this new result will be greatly better than ignoring it, normally it is very satisfactory. If you instead also want to maintain the exposure of the bright outside, then you meter on that, and will need to use fill flash, to equalize things.
Old timers all understand this. Novices imagine their fancy modern camera ought to understand it too, but this blind faith will be disappointed. Instead, it all works as mentioned above. The meter is a helpful guide indeed, but just watch, and always do what you see you need to do. Sometimes it can be too big a job for a dumb computer, but the human brain can always help.
Ansel Adams thought his Zone System's middle zone 5 ought to be 18% reflectivity, and he promoted the 18% notion in photography (in the 1930s). I cannot imagine that Ansel could have ever seen a digital image or a histogram back then, so don't confuse the two systems, analog film and digital. However, our meters are set to 12.5%, (thought to be more realistic, on average), and Kodak says we will need to open 1/2 stop if metering on their 18% card (this 1/2 stop more is equivalent of 12%). So any and all scenes are metered to come out about 12.5% middle tone average - regardless if they are typical 12.5% scenes or not (the reflectance of the scenes colors can vary away from "typical"). If photographing that card itself (as previous page), then no matter which card you meter, black, white, or gray, these cases do cause very different exposure settings, with results always near the same middle tone. None are necessarily at 128 on the histogram, which is simply the wrong idea (the histogram data is gamma encoded anyway - and the 18% card was an analog film concept).
It should be obvious that any plan to calibrate a light meter by metering an 18% gray card is not the best plan. This was popular in analog film days, and it is not far wrong (perhaps a bit practical, but it is NOT theory). Kodak always told us if metering on their 18% gray card, we should open 1/2 stop more exposure (1/3 stop today). But a digital histogram is something else (a related link, explaining why 18% gray is Not the midpoint of the histogram we see, is at Histograms are Gamma Encoded). Metering the 18% card has no calibration significance (any scene will come out middle tone, a white card and the black card too... what significance is the 18%?) The correct way the 18% card is used is that we can meter from it, and then use that reading for the exposure of some other general scene in the same light. Kodak always said to open 1/2 stop more. Most wide range scenes (beach scene, mountain scene, sky and water and trees and human faces) do often average about middle tone, which is what lets reflective meters be useful, since that is what they do too.
The camera meter is a "reflected" light meter. A reflected meter is aimed at the subject, and reads the light reflected from the subject (which is very variable, depending on what it is) The reflected meter tries to make all scenes it sees to average out to be near middle tone. That is all it can do. This is more easily seen with a blank paper subject, but all scenes come out averaging a middle tone. The meter is calibrated to make that result tone be about 12.5%. Post processing can change it. This value is NOT related to the center of the histogram. This value is not related to an 18% gray card - 18% is not a factor. The meter's "calibration" value and result causes this one same resulting middle gray tone - from any subject.
Techie details: Sekonic specifications (L-358 for example) specifically says their reflected K=12.5. Wikipedia explains more and says "ISO 2720:1974 recommends a range for K of 10.6 to 13.4 with luminance in cd/m². Two values for K are in common use: 12.5 (Sekonic, Canon, Nikon) and 14 (Kenko and Pentax). The difference between the two values is approximately 1/6 EV." (Minolta meters are Kenko now)
A quick summary: There are normally many colors in our subjects scene, but we can only set one camera exposure for any one picture. So correct exposure generally means something like this:
The mode topics below are about metering ambient light. Nikon TTL flash metering will be a different concept.
Spot metering - Speaking of metering ambient: Beginners beware. Spot Metering (ambient) is a techie tool that requires greater experience, definitely NOT point&shoot (for ambient). Only a small spot in the frame is considered, at the current focus point. Spot metering isolates this small area from the rest of the frame. The idea is NOT about "correct" exposure, the only idea is that this small spot area will be made to come out middle gray (correct or not), and all the rest of the frame is ignored - the rest comes out however it does, when this spot is made middle gray. Only that small spot matters, that small area will become middle tone.
Speaking of metering flash: Spot metering is NOT Spot Metering for TTL flash - it is still regular old flash metering. Spot mode does switch TTL BL flash mode to be TTL mode (i.e., Spot is never concerned about ambient background as is TTL BL). For ambient, yes, it is also Spot metering, not concerned about any of the rest of the frame - it only sees this tiny spot. Spot metering of ambient has no concept of a background to match, so Nikon TTL BL flash mode will change to become TTL mode if Spot Metering. Note that's about the only change for indoor flash, since the indoor ambient level is typically far below the flash level (ambient metering ignored, Spot or not), so then all Spot really does for indoor flash pictures is to force actual TTL mode (vs default TTL BL mode).
Spot metering is NOT a general purpose ambient metering mode for beginners - Except it does become more general with metered flash in dim ambient areas. It may be the only way to switch TTL BL to be TTL mode. Novices mistakenly confuse Spot metering with a general purpose metering mode, and assume it means that this spot will come out correctly exposed. But reflected meters don't "come out correctly exposed", and Spot metering is a very special case (only about that one spot), and it just means that spot will come out middle gray, which is NOT about being "correct" - unless hopefully, we choose a spot which ought to come out middle gray (which puts the full responsibility back on the photographer). Frankly, for beginners not understanding the details yet, bracketing multiple exposures seems more helpful than spot metering - helpful both to produce immediate results, and to aid learning about the compensation that various scene types require. Pretty soon, you will just already know. Yes, spot metering can isolate a face from the influence of a black or a white background, but should that face come out middle gray? The skill we need to know is how to compensate it, either way. That face will likely need about one stop additional exposure to make it as bright as it should be, so this is two tricks to learn. The original idea of spot metering was to meter the highlights, and to meter the shadows, to be able to determine the range between them, as a way to determine a middle compromise. Center and Matrix metering try to automate that averaging for us. Accuracy is normally better when a larger overall area is averaged to be middle gray.
Center Weighted metering - The entire frame is metered in some degree, but significantly greater importance (weighting) is given to a large center area (default diameter is roughly half frame height, but adjustable). There is no sharp dividing line at the circle, the boundary slowly tapers, typically a Gaussian Bell Curve. Usually the center area is weighted about 75%, i.e., tones in the center count as 3x more important (toward the picture's average) than at the outside edges. So the outside edges are only weighted about 25% (not as important as the center), but the edges are not ignored (i.e., white edges will meter higher than black edges.) All of this area is averaged to one average tone, and exposed so that this average tone is made to come out about middle tone. It is very suitable for many typical or average scenes, especially portraits, because subjects are often in the center of the scene. Note that while we might move the focus sensor towards one edge of the frame, Center metering remains in the center.
Matrix metering - The focus point area has substantial weighting, but a wide area of the frame is also watched for exceptions, with the camera firmware making other judgments to factor in areas with exceptions. These rules are unknown to us, it does what it does, a mystery to us, but often pretty close. Don't assume too much magic however (and Nikon TTL flash meters in the center of the frame).
For those new to these terms, the on-line Nikon Glossary says this (Words are by Nikon, bold added by me):
Spot Metering: A metering method in which meter sensitivity is concentrated within a small circle in the center of the viewfinder. Recommended for very precise metering; requires extensive knowledge of lighting for really effective use.
Center Weighted Metering: Meter sensitivity is biased toward the center of the viewfinder. Recommended when the subject is in the center of the picture.
Matrix Metering: An advanced metering mode in which the camera's computer sets exposure based on its analysis of the scene's components. It is generally regarded as the most accurate metering for most lighting situations, including those with complex lighting. Matrix Meter or 3D RGB Color Matrix Meter gathers information from the red, green, and blue sensors and factors in distance information provided by the lens as it evaluates proper exposure calculation. This meter instantly analyzes a scene's overall brightness, contrast, and other lighting characteristics, comparing what is sees against an onboard database of over 30,000 images for unsurpassed exposure accuracy, even in the most challenging photographic situations. By the time the 3D Matrix meter has made its considerations of colors by hue and saturation, tonal ranges by brightest and darkest, areas of similar tonality that are connected or separated, distance to the subject, and compared that to its database generated from photographic images, it's got a very good idea of what the exposure should be.
If you are a beginner, Matrix is where you should start. As your skills grow, and they will, you will acquire a better understand of when it might be beneficial to use other light metering options.
What is the database of over 30,000 images? Over the years, Nikon has studied the color, area of coverage, focus distance, contrast, size and shape of shadows and highlights and exposure characteristics of over 30,000 actual photographic images and incorporated this data as a reference source for the expert exposure system that is the 3D Color Matrix Meter.
Yeah, yeah, sure. :) Maybe Matrix can help beginners, but no one knows what Matrix does, or will do, other than we know it watches the full frame. It often seems bad to me in that if there is some bright spot somewhere at the edge, it will cut back on the exposure of the full frame (like next below, and like in the first pictures above). Just one notion.
I am a fan of Center Weighted metering myself (usually where my subject is). The Nikon default is the center area diameter metered is half of the frame height, but there are menu options to resize it. 75% center weighted means that while the outer edges are seen, this center counts for 3/4 of the importance of the exposure. Not at all a sharp dividing line, its view is gradually feathered out to the edges, but the effect from the edges is minor. The center subject is the important thing.
The camera metering modes of Center, Matrix, and Spot refer only to the ambient light. The TTL flash system has its own metering method and area (which is a center area, similar, but not exactly same as Center Weighted). However, there can be a big effect on flash. Spot metering mode does switch the flash metering mode from TTL BL (balanced) metering to be TTL metering. Here is a quick example. It has a bright North window at left (intended to be hard for Matrix), and a black card at the center focus spot. Nothing easy about it. Nikon D800, camera A mode, SB-800 hot shoe direct flash, no compensation. All pictures are f/5.6 ISO 400.
Showing that Matrix can affect the ambient differently than Center (depending on what the metering area sees), but Matrix or Spot does not affect the metering zone of the flash. Spot metering certainly affects the ambient (the Spot tries to become middle gray, which is NOT the same as becoming "correct"), and Spot Metering selection does switch TTL BL flash mode to be TTL flash mode, and TTL mode is NOT balanced with ambient, but Spot does not affect the metering zone of the flash.
The flash pictures were all were 1/60 second (minimum shutter speed with flash in camera A mode), which is less ambient exposure than even the 1/40 second metered above, which underexposes indoor ambient, a couple of stops here. A better try would be to use camera Manual mode, to be able to set any shutter speed desired to obtain any desired ambient effect. But here, we let automation run, this is camera A mode, which does 1/60 second in indoor flash situations. Basically indoor levels have reversed things - the reduced ambient is now the lesser fill level for the TTL flash (outdoors in fully metered bright sun would be the opposite). Larger differences certainly are possible sometimes, but this is an (easy) middle gray subject, and these flash pictures are mostly alike, the point being: TTL flash metering is not affected by Matrix or Spot metering of the ambient. Two complications are that TTL BL flash level is reduced to accommodate bright levels of ambient, in Center or Matrix metering modes. In Spot mode, TTL BL is changed to be TTL mode.
In the middle row with TTL BL flash, Matrix metering of brighter window ambient reduced the (balanced) flash level slightly. However, notice that the window dark shadow at right behind chair is still visible (there are two exposures here, ambient and flash). If this had been a full proper metered ambient exposure, we would see a large difference in TTL BL and TTL flash mode.
Spot metering would have still affected the ambient if faster shutter speed had not forced underexposure. What Spot metering does for flash is to change TTL BL metering mode to be TTL mode (no influence to reduce TTL flash by the ambient). Said again, Spot metering only affects the ambient metering, except it changes TTL BL flash metering to instead be TTL metering, but which is Not Spot. There are only slight hints in this try, hard to see here with the other influences, so claims are based more on experience. TTL BL indoors does often tend to need more +EV Flash Compensation than TTL (maybe +1 EV), but both are dependent on reflective metering. Balancing flash with ambient is not much factor indoors, normally ambient is well down at minimum shutter speed with flash (unless ISO is high). This picture is an exception, aimed into a bright window light, more in between indoor and outdoor (regarding bright ambient). Otherwise, my routine starting point for indoor TTL BL is +1 EV Flash Compensation. But for TTL mode indoors, then 0 EV (starting point). In bright sun outdoors, it is TTL that needs about -2 EV compensation, when TTL BL does not.
The last row with TTL flash are all rather equal flash exposure (not affected by Matrix or Spot metering). When comparing TTL BL and TTL, larger differences are possible, but Spot is more alike because both are TTL (which does not Spot Meter).
This one test situation should not be considered typical of all comparison situations. We can easily contrive other TTL vs TTL BL situations to show different results, either way between those two. TTL is a computer program looking at light from reflected subject colors and trying to decide what it is, and what TTL exposure should be. TTL BL is also that, but with more program trying to decide how much to reduce the flash due to how the ambient reflected colors look. I hope to not be discouraging, it is not difficult at all. Just pick one system (either one you have), and learn what it does and how to use it, and all will be good.
You can of course also do simple tests like this. You should, so you can see it, and will know what to believe. At least in photography, we ought to be able to see what we claim to believe. But do try a different thing or two to see if notions still hold up.
Note that Spot metering with flash indoors can normally be said to have Only the effect of selecting TTL flash mode, because:
There is more about TTL and TTL BL flash metering in Fundamentals Part 4.
These rose and card test pictures (on previous page) used Matrix metering (because many people do), but I normally use Center metering, to emphasize the central subject more than the edges. Center metering is very useful for portraits, or even this rose (Center more nearly matches what TTL flash metering will do). Center metering still works for scenes like landscapes that may not have any central subject (they do still have an average tone). We ought not to imagine any great understanding is present in any metering system. It is just a computer chip, which sees some light, and can measure it, but with absolutely no idea what it represents, or how bright it ought to look. The meter's only goal is to make whatever it is come out middle tone - all it knows, all it can do.
My human brain can almost visualize and understand Center metering, to help predict the result, and my own opinion is to shun the automation of Matrix metering and TTL BL flash, simply because I know it never has a clue about what the subject is, and I'm never sure what the computer is going to do. Possibly the automation can help sometimes, but as a personal philosophy, I simply try to avoid those unknown complications. :) But regardless of your method, you will always need to watch and stand ready to help. YOU are the photographer. No one else is going to do it. Your human brain is the only intelligent tool present that can judge "correct" or not. Your human brain is the only intelligence present that can recognize a black cat in a coal mine from the polar bear on the snow. Any reflected meter is going to make both come out to be middle gray. Realizing this makes all the difference.
This is NOTHING new, reflective light meters have always worked this way, for many decades. Don't skip that previous Kodak link: Accurate Exposure With Your Meter.
(Oops! Kodak is gone now, but this article is here and here.
In contrast to Reflected meters, Incident meters measure the incident light directly, independent of, and without reference to the subject - so then it is not about how the subject's colors reflect light. Incident is about the light itself, and typically more accurate (without that influence) - But incident meters must meter the light actually incident at the subject's exact position (Inverse Square Law). Direct sunshine may not matter where you measure it (it is typically the same everywhere), but any incident light more local than that cannot be metered from the camera position.
So there are often more factors than merely aiming the camera (thinking about what we see does always help). It was shown how reflective light meters work (meters in cameras are reflective meters). Reflected meters only have this one way they can work. They are a dumb silicon chip, without any human brain. The meter has absolutely zero recognition about what the subject is, and has zero knowledge how it ought to look. The meter does not know if the subject is black and ought to be dark, or if it is white and ought to be bright. The human brain probably recognizes it instantly (in its surroundings), but the meter has absolutely no clue - no brain. The meter simply sees a blob of light. It can measure that light accurately, but it has absolutely no clue what it means. Its best try is going to make that light come out middle tone, which is a good compromise, considering - not too dark and not too bright, perhaps usable. The reflective light meter's every answer is "middle gray". However, every subject may not be. It is good that we realize this, to work with it, instead of against it.
The reflective meter can only attempt to reproduce whatever it may be, to come out as a middle tone average. And this middle tone will be halfway ballpark (not too dark, not too bright, but close enough that we can recognize the subject), but if that result is correct or not depends entirely on the subject that it sees. Most typical scenes do have wide range (portraits, beaches, mountains, jungles), and do average out about middle tone, which is how the system works. But not every scene is a typical scene.
The photographer does need to keep a close watch on things. Bottom line, we can moan and complain about what the meter does, or we can simply just watch and learn to easily fix it. This latter way works much better. It is called experience.
We certainly need to be aware of this, so a similar experiment is one of those things which you should try at home. To become a believer, you should do similar tests as these, yourself (previous page), to see it, to believe it, to be able to use it. Just do it, so you understand it. If you are metering a light scene (meaning, colors which reflect much of the light efficiently), it is going to come out underexposed. If you are metering a dark scene (meaning, colors which do not reflect so much), it is going to come out overexposed. Because, both will come out averaging middle gray, which may or may not be a "correct" result for that scene. Our human brain can recognize this and plan ahead. The least we can do is to look at the result and compensate. What we ought not do is to act surprised. :)
The meter is a big help, and it is still pretty easy. The idea is, you have eyes, you have a brain, and if you use them, then with just a little experience, you quickly come to know that a light scene with enhanced reflection (a white dress, a beige wall, the snow covered ground, etc) will need additional exposure, more than TTL thinks. You quickly come to know that a dark scene with relatively little reflection (black tuxedo, dark colors, dark open background, etc) will need less exposure than TTL thinks. Pretty soon, you will just already know about how much more. A little experience goes a long way, just think a little about what you see in front of your camera. Digital is easy, it allows us to see the first exposure result, and to react.