Reaction Time
Reaction Time
Minimum RT and comparison of senses
Relation to other aspects of behavior
When a human subject follows instructions to make a specific response as soon as he can after the presentation of a specific signal, the latency of the response is called reaction time (RT). Average values of between 150 and 250 milliseconds (msec.) are typically found, for example, where the subject must press a telegraph key when a light is flashed; however, under some conditions, RTs even shorter than one hundred msec. and even longer than one second may occur. The term “RT,” is also applied to the continuous lag of responses to the stream of ongoing events.
Beginnings of the RT experiment are found in two sources. One is Helmholtz’ study of the speed of neural transmission. In an unsuccessful attempt, reported in 1850, to adapt the nerve muscle preparation technique to the intact human organism, he measured the latency of voluntary hand responses to light electric shocks applied to different areas of the skin. It may be supposed that Helmholtz did not appreciate what could be learned through this approach as he did not proceed with it. However, it was not long before many others were actively experimenting and theorizing on the problem.
The other source is the astronomer Friedrich Bessel’s analysis in 1822 of differences among observers in their estimates of the instant, within a series of clock ticks, in which a star passed a cross wire, the “eye and ear” method. With the interest thus aroused in the variability of human behavior, and with the concurrent development of accurate timing apparatus designed to eliminate this factor from astronomical observations, there eventually resulted the measurement of true RTs by two astronomers. One was O. M. Mitchel, an American, who used light and sound signals in 1858; the other was A. Hirsch, a Swiss, who also used electric shock in 1861–1864.
Three early programs of research on RT are of special importance. The first is that of the Dutch physiologist Frans Donders, who in 1868 introduced the experiment in which response is made dependent upon a choice between signals, i.e., the disjunctive RT. By subtracting the RT to a simpler situation from that to a more complex situation, he hoped to find the time required for the additional mental act. The second is that of Siegmund Exner, who first used the term “reaction time” in 1873. His emphasis was on the importance of preparation and on the subject’s feeling of involun-tariness of the response. The third is that of Wundt, started in 1880, which was notable for the attempt to measure the RT for the identification of a stimulus. At least the germ of much of the later experimentation on RT and of the accompanying conceptionaUzations is found in the work of these pioneers. (See Boring [1929] 1950, chapter 8; and Woodworth 1938, chapter 14 for a more complete historical coverage.)
Two characteristics of RT have especially attracted the attention of investigators. First, RT is shortened by “strength” conditions, but not below an irreducible minimum. Second, RT is increased by “complexity” conditions. Much of the research in this field has aimed at developing functional relations between RT and variables concerned with strength or complexity.
Strength variables
Considered here are variations which, in some respect, might be expected differentially to activate, facilitate, or energize some part of the chain of processes between signal and response.
Intensity of signal
It regularly has been observed that more intense signals have a shorter RT. Moreover, two signals of different frequency but of the same subjective intensity will produce extremely similar distributions of RT. If the signal is the cessation of a stimulus, a shorter RT is again found for a strong stimulus. RT to a sudden change of stimulus intensity is shorter as the amount of change is greater.
Discriminability
Discriminability may also be shown to affect RT, and with a wider variety of examples, by the disjunctive, or choice, procedure. The more different the signals, the shorter the RT. This has been demonstrated for frequency of tones, length of lines, gradation of colors, number of objects, and distance between points of tactile stimulation.
Motivation
Motivation may, in several ways, be shown to have an effect. For example, informing the subject of his RT after each trial decreases RT. A more marked reduction is found by applying an electric shock for high RTs.
Preparation
In most studies of RT a warning signal is given, perhaps a second or two before the signal for response. If the foreperiod (or warning interval) is increased to seven or eight seconds, RT will be lengthened. Too short a warning interval will increase RT, but a greater increase will result from highly variable or unexpected fore-periods (Klemmer 1956). Degree of muscular tension was found to vary inversely with RT under such manipulations.
Practice
Practice has been found effective in reducing RT, especially for complex conditions, where considerable reduction may take place over many daily sessions. Warm-up within a session has also been noted, lowest values not ordinarily being attained before the tenth or fifteenth trial.
Minimum RT and comparison of senses
A typical subject has a mean auditory RT of about 140 msec, under usual testing conditions and mean visual RT of about 180 msec. The difference is similar to that required to set the respective sensory nerves into action by appropriate stimuli to the sense organs, and the argument has been made that the difference in RT for the two modalities lies in this differential for receptor stimulation. The argument would be more convincing if their irreducible minimums of RT were composed of known processes. However, in the case of auditory stimulation, only about 55 msec, is accounted for by measurable “peripheral” events and in visual stimulation about 85 msec. (Woodworth [1938] 1960, p. 19). It is risky to assume that, except for the matter of stimulation, the same mechanisms are involved in auditory and visual RT. As a matter of fact, in the study where the lowest RT values were found for the two modalities, the two mean RTs were both close to 110 msec. (Hovland & Brad-shaw 1935). In this study the view was presented that the typical difference might be explained by the poorer figure-ground segregation under usual visual testing conditions (or, in other terms, a lower signal-noise ratio), a factor that was eliminated by making the visual signal a bright light in completely dark surroundings. In view of the difference in time for the stimulus process apparently demonstrated for the two modalities, this then would imply, if the subtractive is accepted, that the unknown processes require less time for visual than for auditory signals as the irreducible minimums are approached.
RT for touch is sometimes as short as that for sound, sometimes perhaps 25 msec, longer. There are special difficulties in finding RT for other types of stimulation: taste, smell, pain, etc. One problem is the isolation of specific sense qualities: a sharp point will be felt as pressure as well as pain. Another is to ascertain the instant when the stimulus acts on the receptor, as in the case of olfaction. The RTs reported for these modalities generally range between 300 msec, and one second.
Complexity conditions
Complexity is introduced by any circumstance that changes the situation from that of the simple RT, where the same signal appears on each trial, requiring the same response, and in which there is no problem of the relation between signal and response.
Multiple options
The RT where the choice of reactions depends upon which of a set of signals appears has been approached from the point of view of information theory. The same increase in RT has been found between two and four options —one and two bits—as between four and eight options—two and three bits (Hyman 1953). However, the effect of number of options is much reduced by extensive practice and by elimination of coding requirements, for example, using a tactual signal to the finger. Also, if options permitting simultaneous response of several fingers are allowed, RT does not increase greatly with an enormous increase in the number of possible patterns.
Indirectness
Any introduction of indirectness or incompatibility between signal and response will increase RT. If the lights on a display no longer directly indicate the required response, RT will rise (Garvey & Knowles 1954).
Prior involvement
If there is an unpredictable succession of two signals, each requiring a response, RT to the second signal is elevated over normal. The shorter the interval between signals, the greater will be the elevation. The explanation of a “psychological refractory period” holds that the processing of a new response cannot start until the end of the processing of the first response (Welford 1959). Alternatively, it has been suggested that the preparation for the second response is interfered with by the first event (Poulton 1950).
Explanations of RT
It may be well to start with the major problem raised by the relation between intensity of signal and RT. Since speed of neural transmission is independent of intensity of stimulus, the relation between intensity and RT cannot be explained in terms of delays in transmission along neurons. Four modes of explanation have been proposed (Chocholle 1963).
Two of the modes specify kinds of neurological mechanisms; the other two imply the properties they must have. The first explanation is that there is actually a faster over-all neural transmission for stronger signals despite the constant transmission speed along single neurons. One possibility cited is that of more rapid traversal of synapses. This appears inadequate to account for the extent of differences found. Another possibility, that a stronger stimulus obtains access to more direct pathways in the central nervous system (CNS), has not yet been directly tested. The second mode of explanation is that there are sensing mechanisms in the CNS which are activated more quickly by a stronger stimulus because of a higher rate of impulses along ascending neurons or because of the involvement of more fibers. This, too, has yet to be verified.
The third mode of explanation is that RT is short to the extent to which part processes between signal and response can be prepared (Poulton 1950). Perhaps full preparation of the effectors (motor set) cannot be attained or maintained if involvement is required of the mechanisms of detection (sensory set).
The final mode of explanation is in terms of a stochastic decision process which distinguishes between a noisy signal and noise alone (McGill 1963). A strong signal might reach a “criterion count” sooner than a weaker one.
The last two modes of explanation lend themselves to aspects of RT other than its relation to intensity. With fewer alternatives, there can be greater preparation for those that do occur—especially for those most likely—as well as a “clearing of the ways” for the response. Also relevant for this expectancy formulation is the relation between RT and variability of the length of the foreperiod.
The kind of decision model described explains the relation between RT and number of alternatives by the time necessary to distinguish among the “noisy” alternatives. The information-theory approach appears to give a fixed amount of time to each successive binary decision, although the means by which this is accomplished have not been stipulated [seeInformation theory; see also Bricker 1955].
Relation to other aspects of behavior
Reaction time is related to several other areas of investigation.
Association time
In the association experiment, the subject makes a verbal response to a stimulus. Unlike in the case of RT, he employs some already available association: he reads a word, names an object, gives the opposite of a word, or simply replies to one word with another (Woodworth [1938] 1960, chapter 3).
Operant conditioning
In operant conditioning, reinforcement of an operant is first made contingent upon the occurrence of a signal. Next, reinforcement is made to depend upon the rapid appearance of the operant after the signal. Latencies may become very short, in the range of values found for RT (Skinner 1946). Thus, experiments with animal subjects can be made to parallel those with human subjects. The relation of latency to signal intensity has already been studied. [SeeLearning, articles oninstrumental learningandreinforcement.]
Tracking
In tracking procedures the subject keeps a pointer on a moving target or keeps a target from being moved off center. If the position of the target varies continuously it is difficult—although possible in theory—to determine the RT to the appearance of an error. However, the step-tracking task amounts to no more than a variant of the RT procedure.
Instrumentation
Several procedural aspects are relevant to a discussion of RT.
Signals
The greatest problem in instrumentation is to make the signal appear substantially instantaneously. Neon lamp and tachistoscopic techniques are superior in this respect to incandescent lamps. Although tones may be introduced instantaneously, a click will be heard at their onset, often contrary to the aims of the experiment.
Responses
Manipulation of the telegraph key and the light-pressure snap switch are both satisfactory responses. The response may be either a press or a release. Finger responses are usual, but any movement of the hand, arm, leg, etc. may be employed. A microphonic switch is the preferred method of recording a vocal response. In the step-tracking procedure, a pointer or lever is generally moved laterally.
Recording
An immediately accessible RT after each trial may be obtained by means of a continuously running synchronous clock. The direct-current clutch is engaged with the signal and disengaged with the response. Although such clocks are sufficiently accurate for most purposes and are widely employed, electronic counters are superior in avoiding mechanical problems and in not producing clicks. Moreover, permanent registration may be made by means of print-out devices. In the step-tracking procedure, graphic techniques, often utilizing signals produced by transducers, are used to find an index of the start of the response. It is also possible to use such instruments in conjunction with key pressing and other ungraded responses that provide an early indication of the onset of the response. Muscle action potentials provide an even earlier indication (Bartlett 1963). The delay arising from the mass of the recording element may be avoided by use of a camera technique in conjunction with a cathode-ray tube or by use of magnetic tape recording.
Special conditions
For a given condition, where the RT for a three-year-old is 500 msec., an exponential decrease to 150 msec, might be expected to about age thirty, with a shallow increase starting thereafter. For as yet unexplained reasons, slightly lower RTs are found for males than for females.
Predictive use has been made in a number of ways. In testing or training automobile drivers a measure is made of time taken to apply the brake on the appearance of a danger signal. Some negative correlations have been revealed between RT and measures of athletic performance.
A variety of special conditions have been tested; following are a few examples. Prolonged vigilance has been found to increase RT, but no consistent effect has been found to result from simple loss of sleep. Benzedrine can overcome the effect of prolonged vigilance but otherwise has no discernible effect. Alcohol has been found to increase both the mean and the variability of RT.
Robert M. Gottsdanker
[Other relevant material may be found inAttention; Hearing; Nervous system; Senses; Vision.]
BIBLIOGRAPHY
Bartlett, Neil R. 1963 A Comparison of Manual Reaction Times as Measured by Three Sensitive Indices. Psychological Record 13:51–56.
Boring, Edwin G. (1929) 1950 A History of Experimental Psychology. 2d ed. New York: Appleton.
Bricker, Peter D. 1955 Information Measurement and Reaction Time: A Review. Pages 350–359 in Henry Quastler (editor), Information Theory in Psychology. Glencoe, III.: Free Press.
Chocholle, RenÉ 1945 Variations des temps de réaction auditifs en fonction de l’intensité à diverses fréquences. Année psychologique 41/42:65–124.
Chocholle, RenÉ 1963 Les temps de réaction. Volume 2, pages 63–112 in Paul Fraisse and Jean Piaget (editors), Traiteé de psychologie expérimental. Paris: Presses Universitaires de France.
Garvey, W. D.; and Knowles, W. B. 1954 Response Time Patterns Associated With Various Display-Control Relationships. Journal of Experimental Psychology 47:315–322.
Hovland, Carl I.; and Bradshaw, Dorothy A. 1935 Visual Reaction Time as a Function of Stimulus Background Contrast. Psychologische Forschung 21: 50–55.
Hyman, Ray 1953 Stimulus Information as a Determinant of Reaction Time. Journal of Experimental Psychology 45:188–196.
Klemmer, Edmund T. 1956 Time Uncertainty in Simple Reaction Time. Journal of Experimental Psychology 51:179–184.
McGill, William J. 1963 Stochastic Latency Mechanisms. Volume 1, pages 309–360 in R. Duncan Luce, Robert R. Bush, and Eugene Galanter (editors), Handbook of Mathematical Psychology. New York: Wiley.
Poulton, E. C. 1950 Perceptual Anticipation and Reaction Time. Quarterly Journal of Experimental Psychology 2:99–118.
Skinner, B. F. 1946 Differential Reinforcement With Respect to Time. American Psychologist 1:274–275.
Teichner, Warren H. 1954 Recent Studies of Simple Reaction Time. Psychological Bulletin 51:128–149.
Welford, A. T. 1959 Evidence of a Single-channel Decision Mechanism Limiting Performance in a Serial Reaction Task. Quarterly Journal of Experimental Psychology 11:193–210.
Woodworth, Robert S. (1938) 1960 Experimental Psychology. Rev. ed. by Robert S. Woodworth and Harold Schlosberg. New York: Holt.