Physiology&Behavior,Vol. 52, pp. 839-841, 1992
0031-9384/92 $5.00 + .00 Copyright© 1992PergamonPressLtd.
Printed in the USA.
Time of Day Effects on a Human Force Discrimination Task L. S. MILLER, *1 T. W. L O M B A R D O *
A N D S. C. F O W L E R * t
Departments of*Psychology and 7"Pharmacology, University of Mississippi, University, M S 38677 R e c e i v e d 24 D e c e m b e r 1991 MILLER, L. S., T. W. LOMBARDO AND S. C. FOWLER. Time of day effects on a humanforce discrimination task. PHYSIOL BEHAV 52(5) 839-841, 1992.--Although numerous studies have demonstrated reliable relationships between various human performance measures and time of day, disagreement exists concerning the shape of these relationships and their dependence on task variables. Most perceptual-motor tasks emphasize responsiveness to exteroceptive stimuli. We used a multiple force-band discrimination task that requires responsiveness to both exteroceptive and proprioceptive information. Results for a response duration measure showed a quadratic time of day trend similar to previously reported performance tasks. Response latency to the force emission cue and number of correct inband force emissions showed cubic time of day trends not typically reported in the time of day and performance literature. These results have implications for time-of-day effects on real world perceptual performance. Force discrimination Proprioception
Time of day
Circadian
Human
A multitude of methods has been used to study the relationship between human performance and time of day. Tasks have ranged from such simple, highly repetitive tasks as reaction time, card dealing and sorting, time estimation, and hand steadiness (2,13) to more cognitively complex tasks with strong memory requirements such as vigilance tasks, lengthy calculation tasks, and logical reasoning tasks (3,6,7,10, l 1). These experiments have produced a variety of patterns of performance effects, whose shapes typically depend on the complexity of the task. Early studies (l 3) found that performance on simple repetitive tasks tended to improve throughout the morning, followed by a performance decline during the afternoon. This curve seemed to parallel the diurnal body temperature curve, but peaked somewhat earlier. Further studies indicated even greater correspondence between temperature and reaction time, with performance peaking later in the day and closely following the temperature curve. Since then, body temperature has been used as a marker of a general physiological circadian rhythm because of its consistent 24-h period and ease of measurement. Typically, it rises in the early morning, peaks around 2000-2100 h, and then begins to drop, creating an inverted U-shaped curve for the diurnal phase. Body temperature has been used to help interpret circadian variations in performance and to make general inferences about differences in the phase and amplitude of performance rhythms. Later studies (2,4) found that performance in most simple repetitive tasks closely paralleled changes in body temperature,
Response duration
Response latency
peaking at around 2100 h. The relationship between performance and body temperature appeared to persist even when subjects were adjusting to different sleep/wake schedules. However, experiments using memory tasks instead of speed tasks have produced somewhat different results. Immediate memory task performance tends to be negatively correlated with time of day (1,2). Other studies have indicated that short-term memory tasks, such as digit span, show a slight rise in performance in the morning followed by a general decrease over the rest of the day (6,8,1 l). Also, performance likely varies with the relative working load of memory (9). In a visual search paradigm, working load was systematically varied by altering the number of embedded letters that needed to be identified in a task. Low working memory load performance was positively correlated with temperature, medium load performance was uncorrelated, and high load performance was negatively correlated with circadian changes in temperature. This suggested that the way information is processed may be a major factor in predicting time-of-day effects on performance. While performance appears to be affected by time of day, the parallelism between temperature and performance seems to hold for only a small range of tasks. The variability in results from different tasks suggests that there may be more task dimensions that influence trends in performance across the day. Increased energies need to be directed toward identifying the salient properties of different tasks to enable us to understand why performance varies over the day.
A preliminary version of this paper was presented at the Annual Convention of The American Psychological Association, Division 1, Atlanta, GA, August 12-16, 1988. t Reprint requests should be addressed to L. Stephen Miller, Ph.D., at his present address: Department of Psychology, The University of Georgia, Athens, GA 30602. 839
840
MILLER, L O M B A R D O A N D F O W L E R 34
We studied time-of-day effects on a multiple force discrimination task (MFDT) (5). This task appears to have many advantages in the study of time of day and performance. The M F D T is relatively simple to present and can be easily recorded. It is highly sensitive to small differences in response characteristics, and performance can be broken down into a variety of individual properties for detailed analysis. Finally, the task requires the subject to respond to both exteroceptive and proprioceptive information, the latter being a response dimension largely neglected in the time of day and performance literature.
28
METHOD
20
The subjects were 90 male Psychology students, tested in 45min sessions, at one of six randomly assigned times of the day: 0800, 1100, 1400, 1700, 2000, or 2300 h. Force was measured with a Grass Instruments F T - l 0 force-displacement transducer and a Model 7D polygraph with a 7Pl DC preamplifier. Data was recorded by an Apple IIe computer using an analog to digital converter. The measuring procedure was modeled after previous work in this laboratory (14). Variables measured were total number of correct responses, mean peak force, mean response latency, and mean response duration. Oral temperature was measured with a digital electronic thermometer.
18
Procedure Subjects received two types of practice before the experimental task. First, they received continuous, immediate digital visual force feedback while pressing the transducer using the distal phalanx of the forefinger. Practice continued for approximately 10 min at three specified force ranges. They then practiced for 6 rain receiving feedback only for peak force while attempting to emit peak forces within the three target ranges. The range for each trial was indicated by a computer-controlled light emitting diode (LED) signal. The three ranges were 1) 192260 g; 2) 392-468 g; and 3) 620-796 g. Ranges were chosen for equal discriminability based on data gathered in a separate experiment. In the experimental task which followed, feedback was reduced to a postresponse LED signal for responses that fell within the specified target range. Subjects received 30 trials within each of the three bands for a total of 90 trials. All subjects received the same randomly generated sequence of force band presentations. At the end of the session, oral temperature was measured.
99.0
0.50 rgl
~r 0.45
~ 0.35
- 98.5 ~ Z m - 98.0
m ~ 0.30 0
- 97.5
Z 0
0.40
m
~ 0.25 T 97.0
0.20 800
1 1 0 0 1400
1700
2000
2300
TIME OF DAY FIG. 1. Mean number of correct responses across the three force bands and body temperature over six time-of-day measurements. Correct responding consisted of accurately pressing the force transducer within the signaled targeted band. Performance on this measure required that subjects process proprioceptive stimuli.
F 99.0
32 983 m
30 i
I
z
i
26
98.0
24 22
c~
I
- 97.5 ~ b,
i
97.0 800
1 1 0 0 1400 1700
2000
2300
TIME OF DAY FIG. 2. Mean response duration across the three force bands and body temperature over six time-of-day measurements. Response duration was the time between onset and offset of force on the transducer. There were no time limitations on the subjects. RESULTS Individual responses of all subjects were separated into groups by force band and by time of day. Visual inspection of the data indicated high variability in the distribution ofseores. To control for this variability, which could obscure the subtle effects of time of day on our force discrimination task, outliers were removed prior to analysis to increase the precision of central tendency estimation (12). This was accomplished by approximating the interquartile range and removing the three most extreme values at each end of the distribution (15). This was done in a consistent format across all variables. This resulted in six groups, nine subjects per group, for each dependent variable. Response distributions were unimodal and not skewed, indicating that the mean was a reasonable statistic to represent the distributions. One-way between-groups analyses of variance were performed on all dependent variables, with time of day as the independent variable. Between-groups sums of squares were partitioned into linear, quadratic, and cubic trends. Results showed significant main effects for the following variables: oral temperature, F(5, 48) = 33.9, p < 0.0001; correct responding, F(5, 48) = 3.2, p < 0.02; response duration, F(5, 48) = 8.4, p < 0.0001; and response latency, F(5, 48) = 7.2, p < 0.0001. Peak force approached significance, F(5, 48) = 2.1, p < 0.08. There were several significant time-of-day trends: oral temperature [linear, F(1, 4) = 93.6, p < 0.0001; quadratic, F(1, 3) = 59.9, p < 0.0001]; response duration [linear, F(I, 4) = 10.9, p < 0.002; quadratic, F(I, 3) = 5.9, p < 0.02]; correct responses [cubic, F(1, 2) = 13.0, p < 0.001]; and response latency [cubic, F(1, 2) = 9.1, p < 0.005]. Oral temperature (Fig. 1) showed an upward sloping inverted U-shaped curve with a steady rise from 0800, peaking at 2000 h, and beginning to drop by 2300 h. Response duration (Fig. 1) peaked at 1700 h, then quickly dropped to its lowest point between 2000 and 2300 h during the oral temperature peak. Correct responding (Fig. 2) peaked at 1100 h when oral temperature was lowest and again at 2300 h when oral temperature began to drop. Correct responding was at its lowest during the oral temperature peak. Response latency (Fig. 3) showed a minor peak at 1100 h and a major peak at 2300 h. This latter peak coincided with the oral temperature peak. Response latency was lowest at 1400 h. Generally, these functions were maintained across the individual band widths. DISCUSSION The results indicated that performance on the M F D T was affected by time of day. Oral temperature showed a typical qua-
TIME OF DAY AND FORCE
841
0.8
Z
l
0.7
99.0
- 98.5
2; 0.6
r~ Z
98.0 m 0.5 a~ 97.5 ~
0.4 [-
0.3
i
i
i
i
1
800
1100
1400
1700
2000
TIME OF
97.0 2300
DAY
FIG. 3. Mean response latency across the three force bands and body temperature over six time-of-day measurements. Response latency was the time from the initial force band cuing signal to the onset of the subject's response. There were no response-time limitations on the subjects.
dratic function in agreement with most other studies. Thus, as a group, our subjects showed the expected daytime circadian temperature rhythm of most persons living on a standard sleep/ wake cycle. Response duration produced a curve similar to that exhibited by speeded or simple repetitive tasks found in other studies (2,4), and peaking in advance of oral temperature by about 3 h. However, in this case, a peak suggested a longer time to complete a response. Response duration was the time between onset of contact with the transducer until its offset, presumably the end of the decision-making process. This can be thought of as a measure of speed of decision making, rather than speed of responding. Thus, results indicate that the relative speed of the decision-making process to respond with a particular force is at its longest in the late afternoon and at its shortest in the late evening. Response latency fit a more complex, cubic time of day function which has not been typically reported using other performance tasks. This resulted in both morning and evening peaks for this measure. Response latency showed a small morning peak
unrelated to temperature change and a large evening peak which coincided with the rise in body temperature. Again, a peak in this instance suggested a longer period of time to initiate a response. Thus, it can be viewed as a measure of reaction time to the cuing stimulus. Results indicated that the relative speed of responding was at its slowest in the evening and possibly in the late morning, while at its fastest in the early afternoon and perhaps again in the late evening. Correct responding was the primary performance measure of the experimental task and was the result of the entire decisionmaking process for responding to the stimulus cue. It was also the variable most dependent on proprioceptive stimuli. As with response latency, a complex, cubic time of day pattern was found. Correct responding showed a high morning peak appearing unrelated to temperature and a high evening peak which lagged the oral temperature rhythm by 3 h. Thus, subjects were most accurate at the experimental task in the late morning and again in the late evening. Taken together, these varied time-of-day patterns suggest that our subjects were not only most accurate, but fastest, in terms of their decision-making abilities, in the late evening while their reaction time to the stimulus cue was at its slowest. This latter finding of slower response speed coupled with increased accuracy was also seen in the morning response latency and morning correct responding scores. These results emphasize the relatively limited range of tasks that appear to be related only to simple changes in body temperature and, thus, to arousal. The different peak times of these measures, as compared to the oral temperature peak, suggest that there may be differential time-of-day influences, depending on the task. We suggest that the varied time-of-day trends found in our study may be due, at least in part, to the unique proprioceptive requirements of the MFDT and its relation to direct response-produced sensory phenomena. While previous tasks may be tapping time-of-day influences on cognitive information processing functions, the MFDT may tap proprioceptive processing as well. This may be influenced by time of day differently than exteroceptive information. More detailed analyses of individual task components may yield a better understanding of the differences in processing proprioceptive and exteroceptive tasks and may elucidate how time of day interacts with the way information is processed to influence task performance.
REFERENCES 1. Baddeley,A. D.; Hatter, J.; Scott, D.; Snashall,A. Memory and time of day. Br. J. Psychol. 22:605-609; 1970. 2. Blake, M. J. F. Time of day effects on performance in a range of tasks. Psychon. Sci. 9:349-350; 1967. 3. Colquhoun, W. P. Circadian variations in mental efficiency. In: Colquhoun, W. B., ed. Biologicalrhythms and human performance. New York: Academic Press; 1971:39-107. 4. Colquhoun, W. P.; Blake, M. J. F.; Edwards, R. S. Experimental studies of shift work III: Stabilized 12-hourshift systems. Ergonomics 12:856-882; 1969. 5. F'dion,R. L.; Fowler,S. C.; Notterman,J. M. Psychophysicalevaluation of feedback phenomenaas related to precisionof force emission:Some methodologicalcon~derations.Am. J. Psychol. 82:266-271; 1969. 6. Folkard, S. Diurnal variation in logical reasoning. Br. J. Psychol. 66:1-8; 1975. 7. Folkard, S. Time of day and level of processing. Mem. Cognit. 7: 247-252; 1979. 8. Folkard, S. Circadian rhythms and human memory. In: Brown, F. M.; Graeber, R. C., eds. Rhythmic aspects of behavior. Hillsclale, NJ: Erlbaum; 1982:241-272.
9. Folkard, S.; Knauth, P.; Monk, T. H.; Rutenfranz, J. The effect of memory load on the circadian variation in performance efficiency under a rapidly rotating shift system. Ergonomics 19:479-488; 1976. 10. Folkard, S.; Monk, T. H. Time of day and processing strategy in free recall. Q. J. Exp. Psychol. 31:461-475; 1979. 11. Folkard, S.; Monk, T. H. Circadian rhythms in human memory. Br. J. Psychol. 71:295-307; 1980. 12. Hampel, F. H.; Ronchetti, E. M.; Rousseauw, P. J.; Stahel, W. A. Robust statistics: The approach based on influence functions. New York: Wiley and Sons; 1985:56-70. 13. Kleitman, N. Sleep and wakefulness. Chicago: Chicago University Press; 1963. 14. Klitzke, M. J.; Lombardo, T. W.; Fowler, S. C. The effects of smoking on discriminative force emission in humans. Paper presented at The 96th Annual Convention of The American Psychological Association, Division 28, Psychopharmacology, August, 1987. 15. Rohatgi, V. K. Statistical inference. New York: Wiley and Sons; 1984:164.
0031-9384/92 $5.00 + .00 Copyright© 1992PergamonPressLtd.
Printed in the USA.
Time of Day Effects on a Human Force Discrimination Task L. S. MILLER, *1 T. W. L O M B A R D O *
A N D S. C. F O W L E R * t
Departments of*Psychology and 7"Pharmacology, University of Mississippi, University, M S 38677 R e c e i v e d 24 D e c e m b e r 1991 MILLER, L. S., T. W. LOMBARDO AND S. C. FOWLER. Time of day effects on a humanforce discrimination task. PHYSIOL BEHAV 52(5) 839-841, 1992.--Although numerous studies have demonstrated reliable relationships between various human performance measures and time of day, disagreement exists concerning the shape of these relationships and their dependence on task variables. Most perceptual-motor tasks emphasize responsiveness to exteroceptive stimuli. We used a multiple force-band discrimination task that requires responsiveness to both exteroceptive and proprioceptive information. Results for a response duration measure showed a quadratic time of day trend similar to previously reported performance tasks. Response latency to the force emission cue and number of correct inband force emissions showed cubic time of day trends not typically reported in the time of day and performance literature. These results have implications for time-of-day effects on real world perceptual performance. Force discrimination Proprioception
Time of day
Circadian
Human
A multitude of methods has been used to study the relationship between human performance and time of day. Tasks have ranged from such simple, highly repetitive tasks as reaction time, card dealing and sorting, time estimation, and hand steadiness (2,13) to more cognitively complex tasks with strong memory requirements such as vigilance tasks, lengthy calculation tasks, and logical reasoning tasks (3,6,7,10, l 1). These experiments have produced a variety of patterns of performance effects, whose shapes typically depend on the complexity of the task. Early studies (l 3) found that performance on simple repetitive tasks tended to improve throughout the morning, followed by a performance decline during the afternoon. This curve seemed to parallel the diurnal body temperature curve, but peaked somewhat earlier. Further studies indicated even greater correspondence between temperature and reaction time, with performance peaking later in the day and closely following the temperature curve. Since then, body temperature has been used as a marker of a general physiological circadian rhythm because of its consistent 24-h period and ease of measurement. Typically, it rises in the early morning, peaks around 2000-2100 h, and then begins to drop, creating an inverted U-shaped curve for the diurnal phase. Body temperature has been used to help interpret circadian variations in performance and to make general inferences about differences in the phase and amplitude of performance rhythms. Later studies (2,4) found that performance in most simple repetitive tasks closely paralleled changes in body temperature,
Response duration
Response latency
peaking at around 2100 h. The relationship between performance and body temperature appeared to persist even when subjects were adjusting to different sleep/wake schedules. However, experiments using memory tasks instead of speed tasks have produced somewhat different results. Immediate memory task performance tends to be negatively correlated with time of day (1,2). Other studies have indicated that short-term memory tasks, such as digit span, show a slight rise in performance in the morning followed by a general decrease over the rest of the day (6,8,1 l). Also, performance likely varies with the relative working load of memory (9). In a visual search paradigm, working load was systematically varied by altering the number of embedded letters that needed to be identified in a task. Low working memory load performance was positively correlated with temperature, medium load performance was uncorrelated, and high load performance was negatively correlated with circadian changes in temperature. This suggested that the way information is processed may be a major factor in predicting time-of-day effects on performance. While performance appears to be affected by time of day, the parallelism between temperature and performance seems to hold for only a small range of tasks. The variability in results from different tasks suggests that there may be more task dimensions that influence trends in performance across the day. Increased energies need to be directed toward identifying the salient properties of different tasks to enable us to understand why performance varies over the day.
A preliminary version of this paper was presented at the Annual Convention of The American Psychological Association, Division 1, Atlanta, GA, August 12-16, 1988. t Reprint requests should be addressed to L. Stephen Miller, Ph.D., at his present address: Department of Psychology, The University of Georgia, Athens, GA 30602. 839
840
MILLER, L O M B A R D O A N D F O W L E R 34
We studied time-of-day effects on a multiple force discrimination task (MFDT) (5). This task appears to have many advantages in the study of time of day and performance. The M F D T is relatively simple to present and can be easily recorded. It is highly sensitive to small differences in response characteristics, and performance can be broken down into a variety of individual properties for detailed analysis. Finally, the task requires the subject to respond to both exteroceptive and proprioceptive information, the latter being a response dimension largely neglected in the time of day and performance literature.
28
METHOD
20
The subjects were 90 male Psychology students, tested in 45min sessions, at one of six randomly assigned times of the day: 0800, 1100, 1400, 1700, 2000, or 2300 h. Force was measured with a Grass Instruments F T - l 0 force-displacement transducer and a Model 7D polygraph with a 7Pl DC preamplifier. Data was recorded by an Apple IIe computer using an analog to digital converter. The measuring procedure was modeled after previous work in this laboratory (14). Variables measured were total number of correct responses, mean peak force, mean response latency, and mean response duration. Oral temperature was measured with a digital electronic thermometer.
18
Procedure Subjects received two types of practice before the experimental task. First, they received continuous, immediate digital visual force feedback while pressing the transducer using the distal phalanx of the forefinger. Practice continued for approximately 10 min at three specified force ranges. They then practiced for 6 rain receiving feedback only for peak force while attempting to emit peak forces within the three target ranges. The range for each trial was indicated by a computer-controlled light emitting diode (LED) signal. The three ranges were 1) 192260 g; 2) 392-468 g; and 3) 620-796 g. Ranges were chosen for equal discriminability based on data gathered in a separate experiment. In the experimental task which followed, feedback was reduced to a postresponse LED signal for responses that fell within the specified target range. Subjects received 30 trials within each of the three bands for a total of 90 trials. All subjects received the same randomly generated sequence of force band presentations. At the end of the session, oral temperature was measured.
99.0
0.50 rgl
~r 0.45
~ 0.35
- 98.5 ~ Z m - 98.0
m ~ 0.30 0
- 97.5
Z 0
0.40
m
~ 0.25 T 97.0
0.20 800
1 1 0 0 1400
1700
2000
2300
TIME OF DAY FIG. 1. Mean number of correct responses across the three force bands and body temperature over six time-of-day measurements. Correct responding consisted of accurately pressing the force transducer within the signaled targeted band. Performance on this measure required that subjects process proprioceptive stimuli.
F 99.0
32 983 m
30 i
I
z
i
26
98.0
24 22
c~
I
- 97.5 ~ b,
i
97.0 800
1 1 0 0 1400 1700
2000
2300
TIME OF DAY FIG. 2. Mean response duration across the three force bands and body temperature over six time-of-day measurements. Response duration was the time between onset and offset of force on the transducer. There were no time limitations on the subjects. RESULTS Individual responses of all subjects were separated into groups by force band and by time of day. Visual inspection of the data indicated high variability in the distribution ofseores. To control for this variability, which could obscure the subtle effects of time of day on our force discrimination task, outliers were removed prior to analysis to increase the precision of central tendency estimation (12). This was accomplished by approximating the interquartile range and removing the three most extreme values at each end of the distribution (15). This was done in a consistent format across all variables. This resulted in six groups, nine subjects per group, for each dependent variable. Response distributions were unimodal and not skewed, indicating that the mean was a reasonable statistic to represent the distributions. One-way between-groups analyses of variance were performed on all dependent variables, with time of day as the independent variable. Between-groups sums of squares were partitioned into linear, quadratic, and cubic trends. Results showed significant main effects for the following variables: oral temperature, F(5, 48) = 33.9, p < 0.0001; correct responding, F(5, 48) = 3.2, p < 0.02; response duration, F(5, 48) = 8.4, p < 0.0001; and response latency, F(5, 48) = 7.2, p < 0.0001. Peak force approached significance, F(5, 48) = 2.1, p < 0.08. There were several significant time-of-day trends: oral temperature [linear, F(1, 4) = 93.6, p < 0.0001; quadratic, F(1, 3) = 59.9, p < 0.0001]; response duration [linear, F(I, 4) = 10.9, p < 0.002; quadratic, F(I, 3) = 5.9, p < 0.02]; correct responses [cubic, F(1, 2) = 13.0, p < 0.001]; and response latency [cubic, F(1, 2) = 9.1, p < 0.005]. Oral temperature (Fig. 1) showed an upward sloping inverted U-shaped curve with a steady rise from 0800, peaking at 2000 h, and beginning to drop by 2300 h. Response duration (Fig. 1) peaked at 1700 h, then quickly dropped to its lowest point between 2000 and 2300 h during the oral temperature peak. Correct responding (Fig. 2) peaked at 1100 h when oral temperature was lowest and again at 2300 h when oral temperature began to drop. Correct responding was at its lowest during the oral temperature peak. Response latency (Fig. 3) showed a minor peak at 1100 h and a major peak at 2300 h. This latter peak coincided with the oral temperature peak. Response latency was lowest at 1400 h. Generally, these functions were maintained across the individual band widths. DISCUSSION The results indicated that performance on the M F D T was affected by time of day. Oral temperature showed a typical qua-
TIME OF DAY AND FORCE
841
0.8
Z
l
0.7
99.0
- 98.5
2; 0.6
r~ Z
98.0 m 0.5 a~ 97.5 ~
0.4 [-
0.3
i
i
i
i
1
800
1100
1400
1700
2000
TIME OF
97.0 2300
DAY
FIG. 3. Mean response latency across the three force bands and body temperature over six time-of-day measurements. Response latency was the time from the initial force band cuing signal to the onset of the subject's response. There were no response-time limitations on the subjects.
dratic function in agreement with most other studies. Thus, as a group, our subjects showed the expected daytime circadian temperature rhythm of most persons living on a standard sleep/ wake cycle. Response duration produced a curve similar to that exhibited by speeded or simple repetitive tasks found in other studies (2,4), and peaking in advance of oral temperature by about 3 h. However, in this case, a peak suggested a longer time to complete a response. Response duration was the time between onset of contact with the transducer until its offset, presumably the end of the decision-making process. This can be thought of as a measure of speed of decision making, rather than speed of responding. Thus, results indicate that the relative speed of the decision-making process to respond with a particular force is at its longest in the late afternoon and at its shortest in the late evening. Response latency fit a more complex, cubic time of day function which has not been typically reported using other performance tasks. This resulted in both morning and evening peaks for this measure. Response latency showed a small morning peak
unrelated to temperature change and a large evening peak which coincided with the rise in body temperature. Again, a peak in this instance suggested a longer period of time to initiate a response. Thus, it can be viewed as a measure of reaction time to the cuing stimulus. Results indicated that the relative speed of responding was at its slowest in the evening and possibly in the late morning, while at its fastest in the early afternoon and perhaps again in the late evening. Correct responding was the primary performance measure of the experimental task and was the result of the entire decisionmaking process for responding to the stimulus cue. It was also the variable most dependent on proprioceptive stimuli. As with response latency, a complex, cubic time of day pattern was found. Correct responding showed a high morning peak appearing unrelated to temperature and a high evening peak which lagged the oral temperature rhythm by 3 h. Thus, subjects were most accurate at the experimental task in the late morning and again in the late evening. Taken together, these varied time-of-day patterns suggest that our subjects were not only most accurate, but fastest, in terms of their decision-making abilities, in the late evening while their reaction time to the stimulus cue was at its slowest. This latter finding of slower response speed coupled with increased accuracy was also seen in the morning response latency and morning correct responding scores. These results emphasize the relatively limited range of tasks that appear to be related only to simple changes in body temperature and, thus, to arousal. The different peak times of these measures, as compared to the oral temperature peak, suggest that there may be differential time-of-day influences, depending on the task. We suggest that the varied time-of-day trends found in our study may be due, at least in part, to the unique proprioceptive requirements of the MFDT and its relation to direct response-produced sensory phenomena. While previous tasks may be tapping time-of-day influences on cognitive information processing functions, the MFDT may tap proprioceptive processing as well. This may be influenced by time of day differently than exteroceptive information. More detailed analyses of individual task components may yield a better understanding of the differences in processing proprioceptive and exteroceptive tasks and may elucidate how time of day interacts with the way information is processed to influence task performance.
REFERENCES 1. Baddeley,A. D.; Hatter, J.; Scott, D.; Snashall,A. Memory and time of day. Br. J. Psychol. 22:605-609; 1970. 2. Blake, M. J. F. Time of day effects on performance in a range of tasks. Psychon. Sci. 9:349-350; 1967. 3. Colquhoun, W. P. Circadian variations in mental efficiency. In: Colquhoun, W. B., ed. Biologicalrhythms and human performance. New York: Academic Press; 1971:39-107. 4. Colquhoun, W. P.; Blake, M. J. F.; Edwards, R. S. Experimental studies of shift work III: Stabilized 12-hourshift systems. Ergonomics 12:856-882; 1969. 5. F'dion,R. L.; Fowler,S. C.; Notterman,J. M. Psychophysicalevaluation of feedback phenomenaas related to precisionof force emission:Some methodologicalcon~derations.Am. J. Psychol. 82:266-271; 1969. 6. Folkard, S. Diurnal variation in logical reasoning. Br. J. Psychol. 66:1-8; 1975. 7. Folkard, S. Time of day and level of processing. Mem. Cognit. 7: 247-252; 1979. 8. Folkard, S. Circadian rhythms and human memory. In: Brown, F. M.; Graeber, R. C., eds. Rhythmic aspects of behavior. Hillsclale, NJ: Erlbaum; 1982:241-272.
9. Folkard, S.; Knauth, P.; Monk, T. H.; Rutenfranz, J. The effect of memory load on the circadian variation in performance efficiency under a rapidly rotating shift system. Ergonomics 19:479-488; 1976. 10. Folkard, S.; Monk, T. H. Time of day and processing strategy in free recall. Q. J. Exp. Psychol. 31:461-475; 1979. 11. Folkard, S.; Monk, T. H. Circadian rhythms in human memory. Br. J. Psychol. 71:295-307; 1980. 12. Hampel, F. H.; Ronchetti, E. M.; Rousseauw, P. J.; Stahel, W. A. Robust statistics: The approach based on influence functions. New York: Wiley and Sons; 1985:56-70. 13. Kleitman, N. Sleep and wakefulness. Chicago: Chicago University Press; 1963. 14. Klitzke, M. J.; Lombardo, T. W.; Fowler, S. C. The effects of smoking on discriminative force emission in humans. Paper presented at The 96th Annual Convention of The American Psychological Association, Division 28, Psychopharmacology, August, 1987. 15. Rohatgi, V. K. Statistical inference. New York: Wiley and Sons; 1984:164.
