PerceptualandMotor Skills, 1991, 72, 663-673.
O Perceptual and Motor Skills 1991
REACTION TIME AND MOVEMENT TIME MEASURED IN A KEYPRESS AND A KEYRELEASE CONDITION "' ROALD A. BJORKLUND National Institute of Occupational Health Summary.-Reaction time (RT) and movement time (MT) are measured in two conditions, a key-press and a key-release condition. The latter makes a greater demand on perceptual feedback than the former. RT increases in the key-release condition and a fraction of M T denoted key-press time (KT) decreases in the key-press condition. It is argued that KT of a response button in the simple reaction-time experiment may serve as a proper measure of the motor component. The foreperiod prior to onset of the reacting stimulus affects RT, KT, and MT, suggesting that preparatory set or expectancy influences both the perceptual and motor components of simple human performance. The results indicate that the relation of RTand M T depends upon methodological conditions.
The relation between the perceptual and motor components of simple human performance has been investigated in a great number of reaction-time experiments (Henry, 1961; Lotter, 1960; Smith, 1961; Keele, Pokorny, Corcos, & Ivry, 1985; Fowler, Taylor, & Porlier, 1987). The perceptual or premotor component is measured as reaction time (RT) defined as the elapsed time from delivery of an imperative stimulus until the subsequent initiation of a predetermined response. The motor component is frequently measured as movement time (MT) defined as the elapsed time from initiation of the response until a prescribed action is completed. Two models have been proposed to describe the relation between RT and MT: the RT - MT independence and the RT - MT dependence (Phillips & Glencross, 1985). The researchers holding the first model argue that RT and MT reflect independent processing stages or processing structures (Henry, 1952, 1961; Fitts & Peterson, 1964; Frowein & Sanders, 1978; Sanders, 1980, 1983), and the researchers representing the other model argue that RT and MT are linked together more strongly and so are less independent (Howell, 1953; Kerr, 1966; Klapp, Wyatt, & Lingo, 1974; Fowler, Taylor, & Porlier, 1987). Occasionally, the studies that investigate the RT - MT relation have considerable limitations. The initiation of the response is not well defined or specified, and MT includes a wide range of responses, for instance, gross motor skills, vocal behaviour, and discrete finger-movements. Groves (1973) investigated the independence between RT and MT in a gross motor skill, -
'N.Halleraker made the apparatus, S. Andresen wrote the computer program, E. Melsom
produced the figures, and M. Gulliksen, E. Dahlin, B. Nyjordet, and G . Pettesen assisted in the experiments. Their contributions are appreciated. 'Address correspondence to R. A. Bjorklund, Department of Physiology, National Institute of Occupational Health, P O Box 8149 Dep, 0033 Oslo 1, Norway.
the racing start in swimming. Mean R T and M T were quantified for each subject by counting frames of film taken of the performance for five trials. However, the racing start in swimming from initiation of body movement upon the flash of the starter's pistol until the swimmer's feet left the starting block measured by frames on film, gives an uncertainty in MTexceeding 2 0 msec. and contributes to a great variance. I n choice reaction-time experiments, M T is defined as the elapsed time from release of a ready 'home' button until key press of one of 2 to 10 target buttons (Ledwith, 1970; Frowein, 1981). The subjects in the two-choice reaction-time experiment of Guiard and Requin (1973) had to move a stylus 30 cm in a 40° upward angle, either to left or right in two conditions, guided from the ready to the target position by parallel metal rods or not. M T performed in the condition without guidance increased significantly compared to the guided condition. The results indicated that rapid M T between separate positions depends upon tactile feedback from the metal rods during the motor phase of performance. The above studies have two critical limitations. First, movement from a ready to a target button requires perceptual feedback necessary to correct and adjust the movement. Thus, M T is not a 'pure' motor component, but adjustment of the appropriate spatiotemporal movements requires sensorimotor coordination. Secondly, M T measured as a gross motor skill is less dependent upon perceptual feedback, but this response is not defined precisely. Variability in R T and M T can be obtained strictly as a function of a response's susceptibility to a physical measurement (Pachella, 1974). The amount of time it takes to initiate a response (RT) is not always free of the effects of response characteristics, such as response length or complexity. To the extent that the operational procedures become confounded with the experimental conditions, the further interpretation of the obtained measures is limited. The study of the R T - M T relation needs more appropriate methods. The objective of the present study is to compare the RTs and MTs in a key-press and in a key-release condition. Both conditions tend to give exact measures. I t is assumed that the key-press condition requires only minor sensory feedback. The key-release condition requires more sensorimotor coordination but is less dependent upon visual feedback than the traditional choicereaction time tasks.
Subjects Twenty male students from the University of Oslo served as paid subjects. The mean age of the subjects was 24.0 yr., with range of 20 to 28 years.
RT AND MT AT KEY PRESS AND KEY RELEASE
665
Apparatus An 80386 24 Mhz IBM-compatible personal computer with 110-card containing the Intel 8255A programmable peripheral interface programmed in Turbo Pascal v5.5 (Borland, 1989) controlled the variables i n the study. Signals from a programmable crystal oscillator (Statek PXO-1000) confirmed that exactness in the time measures was 2 1 msec. Stimulus.-A large (diameter 20 mm), highly luminous (103 cd/mz), red light-emitting diode (LED) (Sharp LT9520D) acted as reacting stimulus. Fig. 1 shows the arrangement of the stimulus and the two response buttons on a green response unit with 5 cd/m2 constant luminosity. The front plate of the response unit subtended an angle of llSOto the horizontal plane. Viewing distance was 35 cm, and the light-emitting diode subtended a visual angle of 3.3O.
FIG.1. Stimulus and response configuration in the experiments, S = stimulus, 1 = response button in Exp. 1, equivalent with ready button in Exp. 2, 2 =target button in Exp. 2 (distances in cm)
Responses.-The response in Exp. 1 (key-press condition) was key press of button 1. I n Exp. 2 (key-release condition), the response consisted of three components: (a) release of the ready-button (Button ' 1') and movement towards the target button (Button '2'), (b) key press of the target button, and (c) movement from the target button back to the ready button. Both buttons (Marquardt 6420.0101) had similar characteristics, 18-mm x 18-mm surface area, 1.6-mm total vertical movement distance, and key force changing between 0.0 and 0.87 Newton during the travel distance, measured as 0.65 Newton 0.02 mm before circuit was connected in the downward movement phase. Foreperiod.-In both experiments there were 200 trials with foreperiods changing randomly in a predetermined order (variable foreperiod design) within a range of 1 log unit in equal logarithmic steps: 400-7111265-2249-4000 rnsec., denoted FP, to FP,. The five foreperiods were repeated eight times with equal probability in a randomized order, giving a sequence of 40 values. Five duplications of the sequence generated 200 foreperiods which were stored in the computer and used in the same order for all subjects in both experiments. Measures.-In Exp. 1, the measurement of reaction time (RT) started from the onset of the stimulus and terminated after a 0.9 mm downward movement of the response button (denoted RT,) simultaneously, as the stimulus disappeared, the measurement of the key-press time (KT,) started. KT, continued during the last 0.7 mm phase of the downward key travel, during the down position and the first 0.8 mrn upwards movement, whereupon KT, ceased and the next foreperiod began. The 0.1-mm difference in distance before release and depression in the upward and downward key travel was made possible by the spring suspension in the buttons. I n Exp. 2, the subjects held the 'ready' button down before presentation of the stimulus, otherwise the computer went into a resting state. RT was measured as the time from onset of the stimulus until 0 . 8 mm release of the ready button, denoted RT,. Simultaneously, movement time from Button 1 to Button 2 started (MT, -,). MT,_, ceased after a 0.9-mm depression of the target button, at the same time as measurement of the key-press time of Button 2 started (KT,). The last phase of the response from Button 2 to Button 1 (MT,_,) began when target button travelled upwards and terminated when the ready button moved down, concurrently with the start of the next foreperiod. Values from all measurements were stored in the computer for later statistical analysis by the SPSS/PC + 3.0 package (Norusis, 1988) in a design with repeated measures, with the means of each subject at the five foreperiods as cells. Procedure Each subject received extensive training in both conditions. They start-
667
RT AND MT AT KEY PRESS AND KEY RELEASE
ed with Exp. 1 and continued on Exp. 2 after 15 min. rest. It was assumed that the design counteracted negative within-subjects transfer of response movements between the two experiments. The subject was seated with the forearms placed on a table in a shielded room (3.1 x 3.2 x 3.0 m). The response unit was located upon the table. In Exp. 1, the index finger of the writing hand rested just upon the response key before onset of each stimulus. In Exp. 2, the same key (now acting as a ready button) was held depressed between trials. Before both experiments the subject was instructed to react as fast as possible immediately upon observing the stimulus. RESULTS Table 1 summarizes the main findings of Exp. 1 and Exp. 2. Mean RT, in the key-press condition (Exp. 1) was 244 msec. and mean RT, was 267 msec. in the release condition (Exp. 2). The Pearson correlation between the two means was significant (Table 2). TABLE 1 MEANS,MEDIANS, STANDARD ERRORS OF MEANS,STANDARD DEVIATIONS, AHD RANGES MEASUREMENTS (MSEC.) OBTAINED FORMEASURES UNDERTWOCONDITIONS Measures (key press) RT, (key release) (key press) KT, MT,. , (key release) (key release) KT, MT,. , (key release) RT,
M
Mdn
SEM
SD
244 267 110 126 73 101
237 24 1 113 115 64 98
8.5 10.1 6.3 6.9 4.6 8.6
37.8 45.1 28.3 31.1 20.6 35.6
OF
Range 198-327 221-367 60-162 64-182 46-123 48-182
Fig. 2 depicts the correlations of RT, and RT, for each of the 20 subjects. Linear regression analysis yielded a significant positive slope between the two measures (RT, = RT, x .7132 + 53.8, r = .851, p < .001). TABLE 2 PEARSON CORRELATION COEFFICIENTSAMONGM ~ S U R E(SEE S TULE 1) Measures
RT,
RT2
.85lt ,268 .031 ,222 -.277
KT, KT, MT, - 1
RT2
KT,
KT,
MT,.,
.lo3 -.042 ,267 -.361
,491 ,394 .43 1
,338 .398
.63St
MT,., Note.-One-tailed significance: ' p < .01, t p < ,001.
The effect of foreperiod on RT, and RT, is shown in Fig. 3. In both experiments, RT increased at the long foreperiods. The variance analysis with repeated measures showed significant within-subjects difference of 23
R. A. BJBRKLUND
FIG.2. Mean RT, against mean RT, for 20 subjects
rnsec. on RT between the two conditions (Table 3). The variance analysis also gave a significant effect of foreperiod in both experiments. From Fig. 3 it can be seen that the two lines representing RT, and RT, parallel each other as a function of foreperiod.
220 400
1000 FOREPERIOD (ms)
4000
FIG.3. Mean RT, (key press) and mean RT, (key release) at five foreperiods
RT AND MT AT KEY PRESS AND KEY RELEASE
669
The values in Table 1 show a considerable range in the measures obtained for the four motor variables, and Table 2 are the correlation matrices for the corresponding mean values. No significant correlations appeared between the perceptual and motor phase in the two experimental conditions. Table 2 shows a positive significant correlation between the two movement components (MT,_, and MT,_ ,) in Exp. 2 ( r = .64, p < .01) and between the sum of the three motor phases in the key-release condition and KT, in key-press condition (r = .53, p < .01), indicating that measure of the simple key press in Exp. 1 to some extent predicts the response pattern in Exp. 2 . Fig. 4 depicts means from the four motor measures related to the length of foreperiod.
400
1000 FOREPERIOD (ms)
4000
FIG. 4. The motor components at five foreperiods (KT, =key press time in Exp. 1, MT,., =movement time from ready button to tar et button in Exp. 2, KT, =key press rime of target button in Exp. 2, MT,. , =movement time &m target button to ready button in Exp. 2)
The KT, results in Fig. 4 were not affected by foreperiod, although the other three motor components all increased with the longer foreperiods. Table 3 summarizes a within-subject analysis based upon the four motor measures. KT, held constant across the five foreperiods, but KT, increased at the longer foreperiods. Movement time from the ready button to the target button, MT,.,, was longer than the reverse travel M T 2 _ ,(p
O Perceptual and Motor Skills 1991
REACTION TIME AND MOVEMENT TIME MEASURED IN A KEYPRESS AND A KEYRELEASE CONDITION "' ROALD A. BJORKLUND National Institute of Occupational Health Summary.-Reaction time (RT) and movement time (MT) are measured in two conditions, a key-press and a key-release condition. The latter makes a greater demand on perceptual feedback than the former. RT increases in the key-release condition and a fraction of M T denoted key-press time (KT) decreases in the key-press condition. It is argued that KT of a response button in the simple reaction-time experiment may serve as a proper measure of the motor component. The foreperiod prior to onset of the reacting stimulus affects RT, KT, and MT, suggesting that preparatory set or expectancy influences both the perceptual and motor components of simple human performance. The results indicate that the relation of RTand M T depends upon methodological conditions.
The relation between the perceptual and motor components of simple human performance has been investigated in a great number of reaction-time experiments (Henry, 1961; Lotter, 1960; Smith, 1961; Keele, Pokorny, Corcos, & Ivry, 1985; Fowler, Taylor, & Porlier, 1987). The perceptual or premotor component is measured as reaction time (RT) defined as the elapsed time from delivery of an imperative stimulus until the subsequent initiation of a predetermined response. The motor component is frequently measured as movement time (MT) defined as the elapsed time from initiation of the response until a prescribed action is completed. Two models have been proposed to describe the relation between RT and MT: the RT - MT independence and the RT - MT dependence (Phillips & Glencross, 1985). The researchers holding the first model argue that RT and MT reflect independent processing stages or processing structures (Henry, 1952, 1961; Fitts & Peterson, 1964; Frowein & Sanders, 1978; Sanders, 1980, 1983), and the researchers representing the other model argue that RT and MT are linked together more strongly and so are less independent (Howell, 1953; Kerr, 1966; Klapp, Wyatt, & Lingo, 1974; Fowler, Taylor, & Porlier, 1987). Occasionally, the studies that investigate the RT - MT relation have considerable limitations. The initiation of the response is not well defined or specified, and MT includes a wide range of responses, for instance, gross motor skills, vocal behaviour, and discrete finger-movements. Groves (1973) investigated the independence between RT and MT in a gross motor skill, -
'N.Halleraker made the apparatus, S. Andresen wrote the computer program, E. Melsom
produced the figures, and M. Gulliksen, E. Dahlin, B. Nyjordet, and G . Pettesen assisted in the experiments. Their contributions are appreciated. 'Address correspondence to R. A. Bjorklund, Department of Physiology, National Institute of Occupational Health, P O Box 8149 Dep, 0033 Oslo 1, Norway.
the racing start in swimming. Mean R T and M T were quantified for each subject by counting frames of film taken of the performance for five trials. However, the racing start in swimming from initiation of body movement upon the flash of the starter's pistol until the swimmer's feet left the starting block measured by frames on film, gives an uncertainty in MTexceeding 2 0 msec. and contributes to a great variance. I n choice reaction-time experiments, M T is defined as the elapsed time from release of a ready 'home' button until key press of one of 2 to 10 target buttons (Ledwith, 1970; Frowein, 1981). The subjects in the two-choice reaction-time experiment of Guiard and Requin (1973) had to move a stylus 30 cm in a 40° upward angle, either to left or right in two conditions, guided from the ready to the target position by parallel metal rods or not. M T performed in the condition without guidance increased significantly compared to the guided condition. The results indicated that rapid M T between separate positions depends upon tactile feedback from the metal rods during the motor phase of performance. The above studies have two critical limitations. First, movement from a ready to a target button requires perceptual feedback necessary to correct and adjust the movement. Thus, M T is not a 'pure' motor component, but adjustment of the appropriate spatiotemporal movements requires sensorimotor coordination. Secondly, M T measured as a gross motor skill is less dependent upon perceptual feedback, but this response is not defined precisely. Variability in R T and M T can be obtained strictly as a function of a response's susceptibility to a physical measurement (Pachella, 1974). The amount of time it takes to initiate a response (RT) is not always free of the effects of response characteristics, such as response length or complexity. To the extent that the operational procedures become confounded with the experimental conditions, the further interpretation of the obtained measures is limited. The study of the R T - M T relation needs more appropriate methods. The objective of the present study is to compare the RTs and MTs in a key-press and in a key-release condition. Both conditions tend to give exact measures. I t is assumed that the key-press condition requires only minor sensory feedback. The key-release condition requires more sensorimotor coordination but is less dependent upon visual feedback than the traditional choicereaction time tasks.
Subjects Twenty male students from the University of Oslo served as paid subjects. The mean age of the subjects was 24.0 yr., with range of 20 to 28 years.
RT AND MT AT KEY PRESS AND KEY RELEASE
665
Apparatus An 80386 24 Mhz IBM-compatible personal computer with 110-card containing the Intel 8255A programmable peripheral interface programmed in Turbo Pascal v5.5 (Borland, 1989) controlled the variables i n the study. Signals from a programmable crystal oscillator (Statek PXO-1000) confirmed that exactness in the time measures was 2 1 msec. Stimulus.-A large (diameter 20 mm), highly luminous (103 cd/mz), red light-emitting diode (LED) (Sharp LT9520D) acted as reacting stimulus. Fig. 1 shows the arrangement of the stimulus and the two response buttons on a green response unit with 5 cd/m2 constant luminosity. The front plate of the response unit subtended an angle of llSOto the horizontal plane. Viewing distance was 35 cm, and the light-emitting diode subtended a visual angle of 3.3O.
FIG.1. Stimulus and response configuration in the experiments, S = stimulus, 1 = response button in Exp. 1, equivalent with ready button in Exp. 2, 2 =target button in Exp. 2 (distances in cm)
Responses.-The response in Exp. 1 (key-press condition) was key press of button 1. I n Exp. 2 (key-release condition), the response consisted of three components: (a) release of the ready-button (Button ' 1') and movement towards the target button (Button '2'), (b) key press of the target button, and (c) movement from the target button back to the ready button. Both buttons (Marquardt 6420.0101) had similar characteristics, 18-mm x 18-mm surface area, 1.6-mm total vertical movement distance, and key force changing between 0.0 and 0.87 Newton during the travel distance, measured as 0.65 Newton 0.02 mm before circuit was connected in the downward movement phase. Foreperiod.-In both experiments there were 200 trials with foreperiods changing randomly in a predetermined order (variable foreperiod design) within a range of 1 log unit in equal logarithmic steps: 400-7111265-2249-4000 rnsec., denoted FP, to FP,. The five foreperiods were repeated eight times with equal probability in a randomized order, giving a sequence of 40 values. Five duplications of the sequence generated 200 foreperiods which were stored in the computer and used in the same order for all subjects in both experiments. Measures.-In Exp. 1, the measurement of reaction time (RT) started from the onset of the stimulus and terminated after a 0.9 mm downward movement of the response button (denoted RT,) simultaneously, as the stimulus disappeared, the measurement of the key-press time (KT,) started. KT, continued during the last 0.7 mm phase of the downward key travel, during the down position and the first 0.8 mrn upwards movement, whereupon KT, ceased and the next foreperiod began. The 0.1-mm difference in distance before release and depression in the upward and downward key travel was made possible by the spring suspension in the buttons. I n Exp. 2, the subjects held the 'ready' button down before presentation of the stimulus, otherwise the computer went into a resting state. RT was measured as the time from onset of the stimulus until 0 . 8 mm release of the ready button, denoted RT,. Simultaneously, movement time from Button 1 to Button 2 started (MT, -,). MT,_, ceased after a 0.9-mm depression of the target button, at the same time as measurement of the key-press time of Button 2 started (KT,). The last phase of the response from Button 2 to Button 1 (MT,_,) began when target button travelled upwards and terminated when the ready button moved down, concurrently with the start of the next foreperiod. Values from all measurements were stored in the computer for later statistical analysis by the SPSS/PC + 3.0 package (Norusis, 1988) in a design with repeated measures, with the means of each subject at the five foreperiods as cells. Procedure Each subject received extensive training in both conditions. They start-
667
RT AND MT AT KEY PRESS AND KEY RELEASE
ed with Exp. 1 and continued on Exp. 2 after 15 min. rest. It was assumed that the design counteracted negative within-subjects transfer of response movements between the two experiments. The subject was seated with the forearms placed on a table in a shielded room (3.1 x 3.2 x 3.0 m). The response unit was located upon the table. In Exp. 1, the index finger of the writing hand rested just upon the response key before onset of each stimulus. In Exp. 2, the same key (now acting as a ready button) was held depressed between trials. Before both experiments the subject was instructed to react as fast as possible immediately upon observing the stimulus. RESULTS Table 1 summarizes the main findings of Exp. 1 and Exp. 2. Mean RT, in the key-press condition (Exp. 1) was 244 msec. and mean RT, was 267 msec. in the release condition (Exp. 2). The Pearson correlation between the two means was significant (Table 2). TABLE 1 MEANS,MEDIANS, STANDARD ERRORS OF MEANS,STANDARD DEVIATIONS, AHD RANGES MEASUREMENTS (MSEC.) OBTAINED FORMEASURES UNDERTWOCONDITIONS Measures (key press) RT, (key release) (key press) KT, MT,. , (key release) (key release) KT, MT,. , (key release) RT,
M
Mdn
SEM
SD
244 267 110 126 73 101
237 24 1 113 115 64 98
8.5 10.1 6.3 6.9 4.6 8.6
37.8 45.1 28.3 31.1 20.6 35.6
OF
Range 198-327 221-367 60-162 64-182 46-123 48-182
Fig. 2 depicts the correlations of RT, and RT, for each of the 20 subjects. Linear regression analysis yielded a significant positive slope between the two measures (RT, = RT, x .7132 + 53.8, r = .851, p < .001). TABLE 2 PEARSON CORRELATION COEFFICIENTSAMONGM ~ S U R E(SEE S TULE 1) Measures
RT,
RT2
.85lt ,268 .031 ,222 -.277
KT, KT, MT, - 1
RT2
KT,
KT,
MT,.,
.lo3 -.042 ,267 -.361
,491 ,394 .43 1
,338 .398
.63St
MT,., Note.-One-tailed significance: ' p < .01, t p < ,001.
The effect of foreperiod on RT, and RT, is shown in Fig. 3. In both experiments, RT increased at the long foreperiods. The variance analysis with repeated measures showed significant within-subjects difference of 23
R. A. BJBRKLUND
FIG.2. Mean RT, against mean RT, for 20 subjects
rnsec. on RT between the two conditions (Table 3). The variance analysis also gave a significant effect of foreperiod in both experiments. From Fig. 3 it can be seen that the two lines representing RT, and RT, parallel each other as a function of foreperiod.
220 400
1000 FOREPERIOD (ms)
4000
FIG.3. Mean RT, (key press) and mean RT, (key release) at five foreperiods
RT AND MT AT KEY PRESS AND KEY RELEASE
669
The values in Table 1 show a considerable range in the measures obtained for the four motor variables, and Table 2 are the correlation matrices for the corresponding mean values. No significant correlations appeared between the perceptual and motor phase in the two experimental conditions. Table 2 shows a positive significant correlation between the two movement components (MT,_, and MT,_ ,) in Exp. 2 ( r = .64, p < .01) and between the sum of the three motor phases in the key-release condition and KT, in key-press condition (r = .53, p < .01), indicating that measure of the simple key press in Exp. 1 to some extent predicts the response pattern in Exp. 2 . Fig. 4 depicts means from the four motor measures related to the length of foreperiod.
400
1000 FOREPERIOD (ms)
4000
FIG. 4. The motor components at five foreperiods (KT, =key press time in Exp. 1, MT,., =movement time from ready button to tar et button in Exp. 2, KT, =key press rime of target button in Exp. 2, MT,. , =movement time &m target button to ready button in Exp. 2)
The KT, results in Fig. 4 were not affected by foreperiod, although the other three motor components all increased with the longer foreperiods. Table 3 summarizes a within-subject analysis based upon the four motor measures. KT, held constant across the five foreperiods, but KT, increased at the longer foreperiods. Movement time from the ready button to the target button, MT,.,, was longer than the reverse travel M T 2 _ ,(p
