The size of the stimulus change required for detection is called the threshold. But this is not an adequate definition: There is no specific stimulus change above which a stimulus is always detected as changed and below which it is never detected as changed. Instead, an observer occasionally reports detecting very small changes, sometimes misses large changes, and reports intermediate sized changes on some intermediate percent of the times that they are presented. This is true not only of people but of all communication channels: phone lines, TV signals, and electronic instruments as well as sensory systems like vision and hearing.

This gradual (rather than abrupt) increase in delectability happens because every information system has "noise" (~meaningless activity) in it, and the senses are information systems. It's called noise because it does what "real" noise does to hearing: you have trouble understanding what people say when there's a lot of noise in the room. Noise always exists in all information channels, physical as well as biological. One example for a physical system most of you know well is the "snow" you may get on a weak TV channel. Something like that snow is always present in vision and in every other sense channel. Anything that disturbs your ability to detect a signal contributes noise, so a shift in attention is another source of "noise".

Because of the variation introduced by noise, threshold is a statistical concept. So a more accurate definition of threshold takes into account its statistical nature. That definition is: The change in a stimulus that is detectably different 50% of the time . If the change is from no stimulation at all, that threshold stimulus is called the absolute threshold. Otherwise the threshold is a difference threshold, because it measures the threshold for detecting the difference between stimuli: a stimulus before and after it is changed. A common way to measure threshold asks observers to report differences the perceive between an unchanging standard stimulus and each of several other stimuli, called comparison stimuli, which differ by small amounts from that standard.

To measure threshold, observers do many trials, often hundreds, comparing the standard with each of the different comparison stimuli. These data provide the percent of trials on which each comparison stimulus is detected as different from the standard. As the difference between the standard stimulus and the comparison stimulus increases, the change in per cent of comparison stimuli detected as different also increases. For mathematical reasons, the point half way between never reporting and always reporting the difference is the least affected by random error in measurement, so it is (usually) used to define the threshold.

Figure 2 illustrates a few trials in the determination of difference threshold for sound intensity. It shows an observer listening to pairs of tones. The first is always the standard of 40.0 units. The second is the comparison which varies from 40.4 to 41.6 on the seven trials shown. The observer says "yes" when she hears the second tone as louder and "no" when she does not. Note that on these trials she reports 40.8 as louder but 41.0 as not louder, illustrating the statistical nature of detecting small changes.

Modern techniques for measuring many psychological processes are based on methods developed for the measurement of simple psychological dimensions like the one outlined above. These measurement techniques can measure complex psychological processes as well as simple psychological dimensions. They have been used to measure complex perceptual processes like how physicians read X-rays or how men recognize (or fail to recognize) positive and negative emotional facial expressions in women. Most psychological measurement (except mental and personality testing) is based on the concept of threshold. How this is done is described later in this exercise.

Q2. [This question may look tricky; it really isn't. Just ask yourself: "Which of these specifically fit the definition (for scientific purposes) and explanation of threshold presented above? Which just "sort of" fit (the way you might use 'threshold' in ordinary speech"?)]

[ Mark EACH item True (T) or False (F) ] Threshold is
 T F     A. the difference between the standard stimulus and the comparison stimulus detected on 50 % of trials
 T F     B. the point half way between "never" and "always" detecting a difference.
 T F     C. the least difference between two stimuli that can be detected
 T F     D. the difference between two stimuli that is consistently detected
 T F     E. defined statistically
 T F     F. the basis of measuring many complex psychological processes

Hint for A-E; Hint for F

Here is a specific example of how threshold works. Brad likes to listen to heavy metal rock, which Dad hates, so Brad has to play it softly enough so Dad can't hear it in his study. Sometimes Brad tries increasing the sound intensity (physical stimulus) so it will sound louder (psychological change). If he increases it only a little bit, Dad rarely notices. As he increases the intensity more, Dad notices it more often. Figure 3 gives a table illustrating this relation.

Q3. [Use the data in the table to answer this question.] When Brad increased the loudness by about 1.5 dB (units of sound intensity),
  A. Dad noticed it about half the time
  B. the intensity was half way to the intensity Dad could hear in his study
  C. Dad noticed the music 1.5 times more often
  D. the sound intensity reached Dad's threshold for hearing the music in his study
  E. A and D are both correct
  F. A, B, C, and D are all correct
Hint

Figure 4 below illustrates the difference threshold for salt. The X (horizontal) axis represents increasing concentration (the measure of intensity for the chemical taste stimuli) above the standard 0.1% salt solution (about a small pinch of salt in a teaspoon of water). The Y (vertical) axis represents the percent of time the observer detects the difference at each concentration. These percentages are based on the results of hundreds of trials.

To find the value of the threshold from the graph, find the point where the curve matches 50 % detection on the Y axis. Then drop a perpendicular line to the X axis. This line points to the value of the stimulus intensity that would produce 50% detection as different from the standard.