About 150 years ago (October 22, 1850) a German philosopher, Gustav Fechner, thought he saw a solution to problem posed by philosophers for years. For centuries, many philosophers had adopted a dualistic way of looking at the world. There was the physical world and there was the world of the mind--the physical and the mental (or psychic, or spiritual). The body belonged to the physical world and consciousness belonged to the mental world. Illusions of various sorts had shown philosophers that the laws of the physical and the laws of the mental worlds were different. Nevertheless in some way the worlds are intimately related. Our everyday experience argues that this immaterial mind in some way very efficaciously controls the material body. How are these two "realities" related? Descartes, Spinoza, Liebnitz, and Berkeley are just a few of the persons who had given serious thought to this matter.

Fechner saw how it was possible to measure sensations--an integral part of the mental world--in terms of stimulation--the part and parcel of the physical world. He termed the study of the relationship between the psychological and the physical "realities" psychophysics. The methods he developed for measuring the relationship he called the "Psychophysical Methods".

In a way Fechner wanted to do something similar to what Einstein did about 50 years later:

• Einstein: What is the relationship between matter and energy? How is matter converted into energy? That is, how much energy can be created from a given amount of matter? Einstein showed the relationship was E = mc².
• Fechner: What is the relationship between mind and matter. "How" is a physical event (matter) converted into a mind event (sensation). How strong a sensation will be produced by a particular stimulus intensity.The relationship Fechner proposed is "sensation = log (stimulus).

Psychologists still use the procedures developed by Fechner and they are still called psychophysical methods. But what's being measured is not the relationship between mind and matter.

Sensation instead is considered as a hypothetical construct. We measure stimuli and certain types of behaviors that occur after the stimulus has been presented to the individual. To explain the consistency of various behaviors scientists have postulated the concept of sensation--an event that happens inside the animal. Sensation is a hypothetical construct, it is not an element of conscious experience which in the past was called sensation. The term "sensation" is now a technical term and does not have the meaning given to the word when it is used in everyday, nontechnical conversations.

This change in meaning is very important to remember. In testing we talk about the validity of a test--to what extent does the test actually measure what we see it measures. There are several ways that we can go about actually measuring validity. One type of validity is that of face validity. All face validity does is ask, "Does this test convince the subject that it measures what it says it measures?" Some tests have unscored items that increase face validity but if those items were scores would make the test less valid. Many concepts in the basic areas of psychology are tricky for we use as a technical concept an everyday word. What we unconsciously do is try to use the word both ways at the same time. Sometimes it is possible to do this and sometimes it is not possible. Sometimes we can read the word "sensation" in the text and think of our conscious content--at that moment, for that context, the word "sensation" has face validity. Other times, the term "sensations" will be used and it cannot be referenced to your conscious content. In this second usage the term does not have face validity. Nevertheless, in the second case "sensation: may be a good hypothetical construct for it may be doing a find job of integrating and predicting--and for science's criteria, that makes the term useful.

It had been recognized for many many centuries that sensation was not the same as stimulation. Measuring the stimulus does NOT also measure our response to the stimulus (or "does NOT also measure our internal response to the stimulus which we'll call the 'sensation' produced by the stimulus").

• The same water can simultaneously be warm and cold to the same person. (Put your left hand in a basin of cold water and your right in a basin of hot water. After a minute, put both hands in a basin of warm water. Your left hand will feel the warm water as hot and your right hand will feel it as cold.
• The light level in a room depends upon your state of adaptation.
• Change only the intensity of the light and colors will appear or disappear.

Philosophers had stated for many years that you couldn't measure sensation. Two of the reasons were:

• Measurement is substitution. 1+1+1+1 = 4 can mean 1 oz water + 1 oz water + 1 oz water + 1 oz water is the same thing as 4 oz of water. Substitution scheme shows that 4 oz is not merely more water than 1 oz but that it is precisely the same as 4 one-oz dollops of water.
• How can you substitute when what you want to measure doesn't take up any space. A "Red" swatch has more redness than a "Pink" swatch. In what way can we say the "Red" swatch is the same as 4 or 12.7 or however-many "Pink" swatches.

Fechner woke up one morning (October 22, 1850) with a "Eureka! I know how to measure sensation. We'll measure it the same way we measure everything else in physics. We will measure it by the null matching procedure!"

Null matching is the fundamental observational procedure in science. A null-match is achieved when one stimulus is adjusted (made weaker or stronger) until it looks exactly the same as another stimulus. The best agreement (both inter-observer and intra-observer) can be achieved if a given observation can be set up as a null-match or "no difference" judgment. Science begins with simple instruments and the null match. Instruments are designed to give the same results as trained observers' null matching. The useful aspect of the instruments is that an untrained observer can now obtain a value that agrees with a trained scientist. Fechner's "Method of Adjustment" procedure simply standardizes the presentation of the stimuli being judged to reduce observational error or at least to make them balance out. Example: Its 1830. You are a physicist studying light. You want to know how the amount of light from a candle is changed by using different types of wax because its getting harder and harder to buy the wax you have been using. Compare the brightness (a sensation) of a candle made with the replacement wax (the unknown) at 1 foot distance with the brightness of a candle made with the current wax (known). Arrange the candles so the known candle illuminates half of a square of paper and the unknown illuminates the other half. Move the known candle closer or further away until the two half-squares are equally bright. The relative distances of the two light sources when the illuminated squares are equally bright can be used to specify how many times the light energy of the unknown stimulus is to the known stimulus. Example: Titration in Chemistry. How acidic (What is its pH?) is this liquid. Add indicator substance to a liquid. Have an example of the endpoint available. Add acid and base in known amounts until the sample you are titrating has the same color as the endpoint. Since the color (or appearance) of the two test tubes match, the pH of the unknown liquid now is the same as that of the end point. By working backward (how much acid or base did you add) you can specify the pH before the additions. Physicists didn't try to measure amounts by seeing if one thing was twice something else or 4 times something else. Make it the same. If you wanted 2 times, you'd put two standard candles down and do null matching. Over the years instruments were developed which gave the same readings as a trained obsever's null match. The useful aspect of the instruments is that an untrained observer can now obtain the value that agrees with the trained scientist. Nowadays highschool chemistry labs have pH meters. You dip the probe in the liquid and directly read the pH-value. Naturally the pH meters were built to give the same readings that highly skilled chemists achieved with the null matching procedure known as "Titration".

Fechner agreed that there was no way you could ever say that one sensation was four times larger than some other sensation. That there was no way to say that one light is twice as bright an some other light. What Fechner did was to discover that what we can do is to say whether or not a sensation is present and whether or not two sensations are the same or not. These two judgments formed the basis of the field we call classical psychophysics. What Fechner did was to say that you can specify a sensation in terms of the stimulus values needed to produce certain types of sensations--actually certain types of judgments. From these landmark stimuli, we can build the whole structure which will allow us to say that one light is twice as bright as another.

Threshold determinations--liminal values of stimuli: What amounts of stimuli produce certain types of behavior.

Absolute Thresholds (RL). What is the weakest stimulus that will produce a sensation. Defining this stimulus presents several problems because the threshold seemingly varies from moment to moment. Consequently there is no one single value for the RL. Instead there is a range of values.

In addition we have a conceptual problem. If I flash a weak light one time you either see it or you don't see it. Lets say you don't see it. Now let us flash that same light on 10 different trials. You might report seeing the light 8 of the 10 times. Is it above, below, or at threshold? In theory, no matter how weak a stimulus I present, if it is presented enough times, it will elicit a response on some occasion. Consequently, the usual definition of threshold takes into account the variability and defines the RL statistically--as for example, the stimulus which can be reported 50% of the time. But remember, the basic definition will always be the operational definition. In this case the operational definitions are the psychophysical methods developed by Fechner.

Difference Thresholds (DL). What is the smallest difference between two stimuli which can be sensed. Again the actual definition takes into account the statistical nature and variability of the threshold and defines the DL as that stimulus which is just noticeably different from the standard stimulus 50% of the time. What is the DL at 100 gms (3.21 oz; approximately a quarter pound). Here 100 gm is the standard stimulus and using one of the psychophysical methods the DL is 103 gm. That is the stimulus that is just noticeably heavier. The difference between the standard stimulus and the difference limen is the just noticeable difference of the jnd and in this case it is 3 gm.

Scaling. Now we are interested in discovering the way in which sensations (or more operationally defined, the magnitude of our response) increase as the intensity of the stimulus increases. Generally we can say it is a monotonically increasing function (as stimulation becomes larger, our response never gets smaller) but that leaves the field wide open.


Fig. 1. Linear Increase

Fig. 2. Increasing Returns

Fig. 3. A special form of "Decreasing Returns"-- log relationship.

If we could judge when a weight is twice as heavy as some other weight we sould be able to plot the results and find the relationship. As mentioned earlier, philosophers at Fechner's time had generally agreed that this was immpossible. So Fechner said, "Let's use the jnd as our unit of psychological intensity (sensation), and the RL as our zero point. Thus a stimulus 5 jnds above RL produces a sensation one-half as large as a stimulus 10 jnds above threshold. When we do this we generate a sensation-stimulus relationship as follows:

• For weak stimuli small incrases yield large increases in sensation. For intense stimuli large increments are needed for even a small increase in the intensity of the sensation. This is the kind of plot we see in Fig. 3.

Psychophysical Scaling -- Determining the way in which physically specifyable stimuli and sensations evoked by them are relationed.

Psychometric Scaling, Now we are inteerested in scaling the sensations evoked by stimuli that I cannot measure physically in any meaningful way--meaningful here indicates that whatever mesures I can take are not useful in determining factural meaning; the empirically determined relaitonship between two indpeendently defind operational definitions. For example it is possible for me to scale and derive statements which indicate the following:

1. Beauty. Painting A is twice as beautiful as painting B.
2. Pleasantness. This persons voice is 4 units more pleasant than that person's voice.
3. Connotative meaning of words. Egg is 3.7 units more pasive than soldier.
4. Similarity of Form. A rectangle is more similar to an oval than a square is to a circle. True or False?
5. Person B is more racially prejudiced than person A but less so than person C. Furthermore, Person B has become less prejudiced in the last year.

• Judgment and Frame-of-reference psychophysics.
  • This sort of interest asks what variables affect the way in which we make our judgments. For instance this approach says that how big a stimulus is depends upon the class of stimuli this particular item bgelongs to. The statments "a large mouse" and "a large elephant" are both meaningful. But we don't mean that the mouse and elephant are the same size.
• Signal Detection Theory/Sensory Decision Theory.
  • Classical psychophsycis starts from a point of face validity: that there are such things a sensations and then it tires to measure them--at least that is where Fechner started. The classical methods assume that there is something like a threshold which is useful to try to measure.

    Signal detection theory starts from another premise. It says that whenever we are asked to decide wether or not a stimulus occus we are actually a very complex system and the decision that comes out is based on much more than either having or not having a sensation

    It's sort of like this: Suppose we ant to find the RL for a weak light. We put you in the dark and ask you to report when you see this light. Evenn if a light is never presented sometimes you see flashes and you'll wonder if that was just something you imagined or was it really a falsh? What SDT argues is that the criterion you establish to say "That was a flash" depends upon such things as how costly is it to say that there was a flash when there wasn't, how costly is it to say there wasn't one whethn there was; how likely that a flash was presented; etc. Changing any one fo these will change the percentage of times that you will report that a given stimulus intensity produced a sensation. If you really want to understand SDT, then don't phrase it in terms of how "big" of a conscious experience do I ahve to be willing to say yes, there was something there." The better way to give it face validity is to talk in terms of the relative costs of the errors and of the liklihood taht a stimulus was presented as affecting ether or not you had a sensation which you either report or don't report.

An interesting addendum

• How well can bloodhounds identify and track a particular individual. Do canine-based identifications meet the standards for "scientifically validated forensic evidence"? There is a small amount of peer-reviewed research that indicates that "many trained law enforcement dogs cannot perform the tasks that are routinely claimed. These include the ability to enter a scent trail at right angles and automatically determine the direction of quarry movement (MacKenzie, & Schultz, 1987) and the ability to scent-match odors form indivuals to handled objects, under contrlled laboratory conditions (Schoon & deBruin, 1994; Schoon, 1996). Although in some of these studies dogs have proved capable of perfomring such scent-matching tasks at levels greater than chance, their error rates are seldom better than 10 to 20% (Brisbin & Austad, 1993), levels substantially above thoses expected of scientifically validated forensic evidence. The U. S. Supreme Court has stated that 'known or potential rate of error' is a significant factor a court should consider in the admissison of scientific evidence." -- Exerpted from the Letters column. Brisbin, I. L., Austad, S., & Jacobson, S. K. (2000). Canine Detectives: The Nose Knows--or Does It?. Science, 290, 1093.
  • References (in the style used in the journal Science)
    • S. A. MacKenzie, J. A. Schultz, Appl. Anim. Behav. Sci. 17, 353 (1987).
    • G. A. A. Schoon, J. C. deBruin, Forensic Sci. Int. 69, 111 (1994)
    • G. A. A. Schoon, Appl. Anim. Behav. Sci. 49, 257 (1996).
    • I. L. Brisbin Jr, S. N. Austad, Anim. Behav. 46, 191 (1993).

©2002 by Burrton Woodruff. All rights reserved. Modified Friday, June 7, 2002