- Email: [email protected]
Neuropsychology: Time Out of Mind
Current Biology, Vol. 12, R9–R11, January 8, 2002, ©2002 Elsevier Science Ltd. All rights reserved.
Neuropsychology: Time Out of Mind
Penelope A. Lewis1 and Vincent Walsh2
We have always known that some form of clock is needed to measure time. It now seems that a variety of different neural clocks are involved in determining our temporal perceptions, some specialised for shorter and some for longer durations.
Immanuel Kant tried to explain the special status of time and space in perception by arguing that our understanding of the universe is limited by the way our brains process information. Specifically, he noted that we perceive all events as occurring in time and space, but it is not clear whether these dimensions exist in reality or are byproducts of our mental organisation [1]. For the neuroscientist, the question is slightly different: allowing that our perceptions are mental constructs and therefore often differ from, or ignore, physical reality (illusions for example), the question becomes how do brain structures and processes shape these perceptions. Within most sensory modalities there is a clear starting point because the dimensions being examined — size, colour, pitch, pressure and so on — can be measured using known receptor systems. For time, it is less clear how to approach the issue as we do not appear to have a set of peripheral time sensors or a primary time area. So how do we come to be aware of time, and what mechanisms do we use to measure it? Psychologists and physiologists have been investigating time measurement since the early 17th century. Approaches they have used fall into two main categories: examination of the psychophysical properties of temporal estimation data, and investigations aiming to isolate the necessary brain regions using focal lesions or, more recently, neuroimaging. An important fundamental concept which has emerged from this work is that of multiple neural clocks (Figure 1). Measurement of intervals with different durations, or for different behavioural purposes, appears to draw upon quite discrete mechanisms in many cases. A recent neuropsychological study [2] has provided further support for this notion that time perception involves multiple neural clocks. Take, for example, the circadian pacemaker, which measures 24 hour cycles. This clock has been shown to depend on the integrity of the suprachiasmatic nucleus (SCN) of the hypothalamus [3], and appears to 1School
of Psychology, University of Birmingham, Birmingham B15 2TT, UK. E-mail: [email protected] 2Dept of Experimental Psychology, University of Oxford, Oxford OX1, UK. E-mail: [email protected]
PII S0960-9822(01)00638-8
Dispatch
be distinct from the clock systems used to measure shorter intervals, such as ultradian activity cycles, which are commonly around 2 hours in the small rodents studied [4]. Lesions to the SCN fully abolish circadian rhythmicity, but do not perturb the ultradian pattern. Instead, ultradian rhythms are eliminated by lesions to other regions of the hypothalamus, the paraventricular and retrochiasmatic structures, suggesting that the clock mechanism may be housed in or at least connected via these areas. Similarly, the measurement of brief durations in the range of hundreds of milliseconds to multiple seconds appears to be completely independent of the hypothalamus, involving cerebral cortex as well as subcortical structures used in motor control [5]. Even within the range of hundreds of milliseconds to multiple seconds, it is likely that different timers are used for measurement of different durations [2,6–8]. Gibbon et al. [7], for example, suggest that the relative precision of temporal estimates varies in a predictable way, with subjects measuring time with a low relative standard deviation (coefficient of variation) in some duration ranges, and a much higher coefficient of variation in others. They propose that these varied levels of precision may be due to the use of different mechanisms. Using quite a different approach, Richard Ivry and colleagues [6] have shown that lesions to the cerebellum perturb the measurement of sub-second but not longer intervals, suggesting this structure’s involvement in timing at the short end of the scale alone. Thomas Rammsayer [8] has proposed a framework which encapsulates not only Gibbon’s and Ivry’s data, but also his own work showing that pharmacological manipulations can perturb the measurement of subsecond and longer intervals differentially. He suggests that briefer intervals are timed at a sub-cognitive level by parts of the motor system, while longer intervals are timed using cognitive processes such as working memory and attention [8]. Evidence supporting this view is rapidly emerging from new work using neuroimaging techniques, and suggests that the right hemispheric prefrontal and parietal areas are critical for measurement of supra-second intervals, while the motor cortex and sub-cortical components of the motor system, such as the basal ganglia and cerebellum, are involved in measuring briefer intervals. This proposed distinction between cognitively controlled and more automatic timing systems roughly parallels a similar distinction between ‘intentional’ and ‘automatic’ forms of memory [9]. If these frameworks are correct, both timing and memory may draw upon a common strategy in cognitive functions — that is, non-conscious automatic mechanisms are used to perform these tasks in more frequent or less demanding circumstances, but the flexible memory and attention modules of the prefrontal and parietal cortices
Dispatch R10
Multi-second intervals: Sub-second intervals:
Parietal cortex
Motor cortex
Frontal cortex
Cerebellum
Hypothalamus SCN: circadian timer (24 hours) PVN;RCN: ultradian timer Basal Ganglia: (may be involved in both multi and sub-second time measurements)
Figure 1. Different brain areas are believed to be involved in measurement of temporal intervals at different ranges. While the suprachiasmatic nucleus (SCN) of the hypothalamus is known to be involved in the measurement of 24 hour (circadian) cycles, the paraventricular (PVN) and retrochiasmatic (RCN) areas of the same structure are thought to be involved in measuring ultradian cycles of around 2 hours. For intervals of briefer duration: some evidence supports the involvement of right hemispheric prefrontal and parietal cortex in measurement of multisecond durations, while the cerebellum and motor cortex are often involved in measurement of sub-second intervals. The basal ganglia may be involved in both multi and sub-second systems.
Current Biology
can be recruited as alternative systems for tasks falling outside the scope of these more automatic algorithms. In a recent study supporting the suggestion of multiple clocks, Kargerer et al. [2] examined neuropsychological patients’ ability to reproduce intervals ranging from 1 to 5.5 seconds. Subjects experienced each interval, presented as either an auditory or visual stimulus, and then reproduced it by terminating a second presentation in the same modality at the correct time. The results show overproduction of intervals shorter than three seconds and underproduction Difference from accuracy (seconds)
A 0.20 0.8 1.0 1.2 0.00 0.5 0.7 —0.02 —0.04 —0.06 —0.08 Target time (seconds) —0.10
1.5
Difference from accuracy (seconds)
B 0.5 0.0 —0.5 —1.0 —1.5 —2.0 1
2
3
4
Actual values Right frontal
5
6 Control Current Biology
Figure 2. (A) A diagram illustrating the tendency to overestimate brief durations, underestimate long durations, and to estimate mean durations with comparative accuracy (the so-called ‘indifference point’). Adapted from Boring [11]. (B) A schematic representation of Kargerer et al.’s [2] data showing a similar pattern to that described in (A). Target intervals are shown on the X axis, error in estimates is shown on the Y axis. The excessive underestimation of short intervals by subjects with right hemispheric frontal lesions is also depicted.
of intervals longer than three seconds, leading to comparative accuracy at the ‘transition zone’ in between. This effect was observed in both patient and control groups. Patients with right hemispheric damage in the frontal cortex performed significantly worse than controls on the longer intervals, greatly underestimating these although their results appeared normal at the shorter range. This finding supports both the possibility that different mechanisms are used in measurement of intervals greater and less than about three seconds and the prior literature suggesting that cortical involvement in these tasks is largely right lateralised, and that the frontal cortex plays a critical role (see [4] for review). The most novel aspect of the finding is that this holds true for the longish intervals ranging from 3 to 5.5 second durations, which neither lesion studies nor the neuroimaging literature have examined extensively. Kargerer et al. [2] emphasise that the difference in slope observed for estimates of intervals shorter than 3 versus longer than 3.5 seconds (as shown in Figure 2B) may also suggest a duality of mechanisms. This interpretation should perhaps be treated with caution as perceptual judgements in many different domains, including time, size, weight, intensity and so forth, have been shown to gravitate toward a mean magnitude in the range examined [10]. Hence, stimuli below the mean tend to be overestimated and stimuli above it to be underestimated in a pattern very similar to that observed by Kargerer et al. [2]. Stimuli at the transition zone or ‘indifference point’, as it has been called in earlier work, show no constant error (Figure 2A). This effect was actually first observed for temporal estimates in 1864 [11] and the results were interpreted in a way very similar to those of Kargerer et al.’s [2], as suggesting the existence of different temporal mechanisms, or that a single temporal mechanism might have a preferred period. The flurry of ensuing experimentation, however, established that the indifference point varies in location from study to study, individual to individual and ‘attentive state’ to attentive state,
Current Biology R11
and depends upon the range of intervals examined (see [11] for a review). Because of these data, the old notion that the indifference point may provide information about the underlying mechanisms for time measurement was eventually set aside [10]. The emerging picture of various independent clocks raises several questions. Do these discrete timers interact? Is it important that they stay ‘in sync’ with each other, and if so, is there a primary timer by which they are all somehow set? The existing evidence suggests that, although they are independent in some ways, these mechanisms interact in others. For instance, it has been shown that the periods of circadian and ultradian pacemakers are not linked, as alterations in the circadian period do not lead to alterations in the ultradian period [3], but that the mechanisms do interact as the ultradian cycle appears to be reset by each circadian dawn [12]. At the level of shorter durations, the relationship is less clear: experimental work has shown that eliminating circadian rhythmicity via focal lesion to the SCN does not perturb the measurement of either a one second [13] or a ten second [14] interval, but the possibility that the clock mechanisms used for these judgements might be reset in some way by the circadian period has yet to be examined. Indeed, it is doubtful that the timers used to measure these intervals involve a pacemaker which could be reset in this way. Many challenges remain in this field. Foremost are questions about the precise locations, methods of function, and varieties of mechanisms available to measure time. These are closely followed by questions about how the different mechanisms influence one another. As for the philosophical question raised by Kant — only time, and continued investigation, will tell. References 1. Kant, I. (1788). Critique of Pure Reason. 2. Kargerer, F.A., Witmann, M., Szelag, E., and Steinbuchel, N.V. (2001). Cortical involvement in temporal reproduction: evidence for differential roles of the hemispheres. Neuropsychologia, 40, 357–366. 3. Hastings, M., and Maywood, E.S. (2000). Circadian clock in the mammalian brain. Bioessays 22, 23–31. 4. Gerkema, M.P., Groos, G.A., and Daan, S. (1990). Differential elimination of circadian and ultradian rhythmicity by hypothalamic lesions in the common vole, Microtus arvalis. J. Biol. Rhythms 5, 81–95. 5. Harrington, D.L., and Haaland, K.Y. (1999). Neural underpinnings of temporal processing. Rev. Neurosci. 10, 91–116. 6. Ivry, R.B. (1996). The representation of temporal information in perception and motor control. Curr. Opin. Neurobiol. 6, 851–857. 7. Gibbon, J., Malapani, C., Dale, C.L., and Gallistel, C.R. (1997). Towards a neurobiology of temporal cognition: advances and challenges. Curr. Opin. Neurobiol. 7, 170–184 8. Rammsayer, T.H. (1999). Neuropharmacological evidence for different timing mechanisms in humans. Q. J. Exp. Psychol. B 52, 273–286. 9. Petrides, M. (1994). In: Handbook of Neuropsychology. F. Boller and J. Grafman, eds. (Elsevier Science), pp. 59–82 10. Hollingworth, H.L. (1913). The central tendency of judgement. Arch Psychol (New York) 4, 44–52
11. Boring, E.G. (1942). In: Sensation and Perception in the History of Experimental Psychology. (New York, Appleton - Century - Crofts, INC.), pp. 574–607. 12. Gerkema, M.P., Daan, S., Wilbrink, M., Hop, M.W., and van der Leest, F. (1993). Phase control of ultradian feeding rhythms in the common vole (Microtus arvalis): The roles of light and the circadian system. J. Biol. Rhythms. 8, 151–171. 13. Innis, N.K., and Vanderwolf, C.H. (1981). Neural control of temporally organized behaviour in rats: the suprachiasmatic nucleus. Behav. Analysis Lett. 1, 53–62. 14. Lewis, P.A. (2000). In: Neural Mechanisms For Time Perception. (Oxford, Oxford University), 46–64.
Neuropsychology: Time Out of Mind
Penelope A. Lewis1 and Vincent Walsh2
We have always known that some form of clock is needed to measure time. It now seems that a variety of different neural clocks are involved in determining our temporal perceptions, some specialised for shorter and some for longer durations.
Immanuel Kant tried to explain the special status of time and space in perception by arguing that our understanding of the universe is limited by the way our brains process information. Specifically, he noted that we perceive all events as occurring in time and space, but it is not clear whether these dimensions exist in reality or are byproducts of our mental organisation [1]. For the neuroscientist, the question is slightly different: allowing that our perceptions are mental constructs and therefore often differ from, or ignore, physical reality (illusions for example), the question becomes how do brain structures and processes shape these perceptions. Within most sensory modalities there is a clear starting point because the dimensions being examined — size, colour, pitch, pressure and so on — can be measured using known receptor systems. For time, it is less clear how to approach the issue as we do not appear to have a set of peripheral time sensors or a primary time area. So how do we come to be aware of time, and what mechanisms do we use to measure it? Psychologists and physiologists have been investigating time measurement since the early 17th century. Approaches they have used fall into two main categories: examination of the psychophysical properties of temporal estimation data, and investigations aiming to isolate the necessary brain regions using focal lesions or, more recently, neuroimaging. An important fundamental concept which has emerged from this work is that of multiple neural clocks (Figure 1). Measurement of intervals with different durations, or for different behavioural purposes, appears to draw upon quite discrete mechanisms in many cases. A recent neuropsychological study [2] has provided further support for this notion that time perception involves multiple neural clocks. Take, for example, the circadian pacemaker, which measures 24 hour cycles. This clock has been shown to depend on the integrity of the suprachiasmatic nucleus (SCN) of the hypothalamus [3], and appears to 1School
of Psychology, University of Birmingham, Birmingham B15 2TT, UK. E-mail: [email protected] 2Dept of Experimental Psychology, University of Oxford, Oxford OX1, UK. E-mail: [email protected]
PII S0960-9822(01)00638-8
Dispatch
be distinct from the clock systems used to measure shorter intervals, such as ultradian activity cycles, which are commonly around 2 hours in the small rodents studied [4]. Lesions to the SCN fully abolish circadian rhythmicity, but do not perturb the ultradian pattern. Instead, ultradian rhythms are eliminated by lesions to other regions of the hypothalamus, the paraventricular and retrochiasmatic structures, suggesting that the clock mechanism may be housed in or at least connected via these areas. Similarly, the measurement of brief durations in the range of hundreds of milliseconds to multiple seconds appears to be completely independent of the hypothalamus, involving cerebral cortex as well as subcortical structures used in motor control [5]. Even within the range of hundreds of milliseconds to multiple seconds, it is likely that different timers are used for measurement of different durations [2,6–8]. Gibbon et al. [7], for example, suggest that the relative precision of temporal estimates varies in a predictable way, with subjects measuring time with a low relative standard deviation (coefficient of variation) in some duration ranges, and a much higher coefficient of variation in others. They propose that these varied levels of precision may be due to the use of different mechanisms. Using quite a different approach, Richard Ivry and colleagues [6] have shown that lesions to the cerebellum perturb the measurement of sub-second but not longer intervals, suggesting this structure’s involvement in timing at the short end of the scale alone. Thomas Rammsayer [8] has proposed a framework which encapsulates not only Gibbon’s and Ivry’s data, but also his own work showing that pharmacological manipulations can perturb the measurement of subsecond and longer intervals differentially. He suggests that briefer intervals are timed at a sub-cognitive level by parts of the motor system, while longer intervals are timed using cognitive processes such as working memory and attention [8]. Evidence supporting this view is rapidly emerging from new work using neuroimaging techniques, and suggests that the right hemispheric prefrontal and parietal areas are critical for measurement of supra-second intervals, while the motor cortex and sub-cortical components of the motor system, such as the basal ganglia and cerebellum, are involved in measuring briefer intervals. This proposed distinction between cognitively controlled and more automatic timing systems roughly parallels a similar distinction between ‘intentional’ and ‘automatic’ forms of memory [9]. If these frameworks are correct, both timing and memory may draw upon a common strategy in cognitive functions — that is, non-conscious automatic mechanisms are used to perform these tasks in more frequent or less demanding circumstances, but the flexible memory and attention modules of the prefrontal and parietal cortices
Dispatch R10
Multi-second intervals: Sub-second intervals:
Parietal cortex
Motor cortex
Frontal cortex
Cerebellum
Hypothalamus SCN: circadian timer (24 hours) PVN;RCN: ultradian timer Basal Ganglia: (may be involved in both multi and sub-second time measurements)
Figure 1. Different brain areas are believed to be involved in measurement of temporal intervals at different ranges. While the suprachiasmatic nucleus (SCN) of the hypothalamus is known to be involved in the measurement of 24 hour (circadian) cycles, the paraventricular (PVN) and retrochiasmatic (RCN) areas of the same structure are thought to be involved in measuring ultradian cycles of around 2 hours. For intervals of briefer duration: some evidence supports the involvement of right hemispheric prefrontal and parietal cortex in measurement of multisecond durations, while the cerebellum and motor cortex are often involved in measurement of sub-second intervals. The basal ganglia may be involved in both multi and sub-second systems.
Current Biology
can be recruited as alternative systems for tasks falling outside the scope of these more automatic algorithms. In a recent study supporting the suggestion of multiple clocks, Kargerer et al. [2] examined neuropsychological patients’ ability to reproduce intervals ranging from 1 to 5.5 seconds. Subjects experienced each interval, presented as either an auditory or visual stimulus, and then reproduced it by terminating a second presentation in the same modality at the correct time. The results show overproduction of intervals shorter than three seconds and underproduction Difference from accuracy (seconds)
A 0.20 0.8 1.0 1.2 0.00 0.5 0.7 —0.02 —0.04 —0.06 —0.08 Target time (seconds) —0.10
1.5
Difference from accuracy (seconds)
B 0.5 0.0 —0.5 —1.0 —1.5 —2.0 1
2
3
4
Actual values Right frontal
5
6 Control Current Biology
Figure 2. (A) A diagram illustrating the tendency to overestimate brief durations, underestimate long durations, and to estimate mean durations with comparative accuracy (the so-called ‘indifference point’). Adapted from Boring [11]. (B) A schematic representation of Kargerer et al.’s [2] data showing a similar pattern to that described in (A). Target intervals are shown on the X axis, error in estimates is shown on the Y axis. The excessive underestimation of short intervals by subjects with right hemispheric frontal lesions is also depicted.
of intervals longer than three seconds, leading to comparative accuracy at the ‘transition zone’ in between. This effect was observed in both patient and control groups. Patients with right hemispheric damage in the frontal cortex performed significantly worse than controls on the longer intervals, greatly underestimating these although their results appeared normal at the shorter range. This finding supports both the possibility that different mechanisms are used in measurement of intervals greater and less than about three seconds and the prior literature suggesting that cortical involvement in these tasks is largely right lateralised, and that the frontal cortex plays a critical role (see [4] for review). The most novel aspect of the finding is that this holds true for the longish intervals ranging from 3 to 5.5 second durations, which neither lesion studies nor the neuroimaging literature have examined extensively. Kargerer et al. [2] emphasise that the difference in slope observed for estimates of intervals shorter than 3 versus longer than 3.5 seconds (as shown in Figure 2B) may also suggest a duality of mechanisms. This interpretation should perhaps be treated with caution as perceptual judgements in many different domains, including time, size, weight, intensity and so forth, have been shown to gravitate toward a mean magnitude in the range examined [10]. Hence, stimuli below the mean tend to be overestimated and stimuli above it to be underestimated in a pattern very similar to that observed by Kargerer et al. [2]. Stimuli at the transition zone or ‘indifference point’, as it has been called in earlier work, show no constant error (Figure 2A). This effect was actually first observed for temporal estimates in 1864 [11] and the results were interpreted in a way very similar to those of Kargerer et al.’s [2], as suggesting the existence of different temporal mechanisms, or that a single temporal mechanism might have a preferred period. The flurry of ensuing experimentation, however, established that the indifference point varies in location from study to study, individual to individual and ‘attentive state’ to attentive state,
Current Biology R11
and depends upon the range of intervals examined (see [11] for a review). Because of these data, the old notion that the indifference point may provide information about the underlying mechanisms for time measurement was eventually set aside [10]. The emerging picture of various independent clocks raises several questions. Do these discrete timers interact? Is it important that they stay ‘in sync’ with each other, and if so, is there a primary timer by which they are all somehow set? The existing evidence suggests that, although they are independent in some ways, these mechanisms interact in others. For instance, it has been shown that the periods of circadian and ultradian pacemakers are not linked, as alterations in the circadian period do not lead to alterations in the ultradian period [3], but that the mechanisms do interact as the ultradian cycle appears to be reset by each circadian dawn [12]. At the level of shorter durations, the relationship is less clear: experimental work has shown that eliminating circadian rhythmicity via focal lesion to the SCN does not perturb the measurement of either a one second [13] or a ten second [14] interval, but the possibility that the clock mechanisms used for these judgements might be reset in some way by the circadian period has yet to be examined. Indeed, it is doubtful that the timers used to measure these intervals involve a pacemaker which could be reset in this way. Many challenges remain in this field. Foremost are questions about the precise locations, methods of function, and varieties of mechanisms available to measure time. These are closely followed by questions about how the different mechanisms influence one another. As for the philosophical question raised by Kant — only time, and continued investigation, will tell. References 1. Kant, I. (1788). Critique of Pure Reason. 2. Kargerer, F.A., Witmann, M., Szelag, E., and Steinbuchel, N.V. (2001). Cortical involvement in temporal reproduction: evidence for differential roles of the hemispheres. Neuropsychologia, 40, 357–366. 3. Hastings, M., and Maywood, E.S. (2000). Circadian clock in the mammalian brain. Bioessays 22, 23–31. 4. Gerkema, M.P., Groos, G.A., and Daan, S. (1990). Differential elimination of circadian and ultradian rhythmicity by hypothalamic lesions in the common vole, Microtus arvalis. J. Biol. Rhythms 5, 81–95. 5. Harrington, D.L., and Haaland, K.Y. (1999). Neural underpinnings of temporal processing. Rev. Neurosci. 10, 91–116. 6. Ivry, R.B. (1996). The representation of temporal information in perception and motor control. Curr. Opin. Neurobiol. 6, 851–857. 7. Gibbon, J., Malapani, C., Dale, C.L., and Gallistel, C.R. (1997). Towards a neurobiology of temporal cognition: advances and challenges. Curr. Opin. Neurobiol. 7, 170–184 8. Rammsayer, T.H. (1999). Neuropharmacological evidence for different timing mechanisms in humans. Q. J. Exp. Psychol. B 52, 273–286. 9. Petrides, M. (1994). In: Handbook of Neuropsychology. F. Boller and J. Grafman, eds. (Elsevier Science), pp. 59–82 10. Hollingworth, H.L. (1913). The central tendency of judgement. Arch Psychol (New York) 4, 44–52
11. Boring, E.G. (1942). In: Sensation and Perception in the History of Experimental Psychology. (New York, Appleton - Century - Crofts, INC.), pp. 574–607. 12. Gerkema, M.P., Daan, S., Wilbrink, M., Hop, M.W., and van der Leest, F. (1993). Phase control of ultradian feeding rhythms in the common vole (Microtus arvalis): The roles of light and the circadian system. J. Biol. Rhythms. 8, 151–171. 13. Innis, N.K., and Vanderwolf, C.H. (1981). Neural control of temporally organized behaviour in rats: the suprachiasmatic nucleus. Behav. Analysis Lett. 1, 53–62. 14. Lewis, P.A. (2000). In: Neural Mechanisms For Time Perception. (Oxford, Oxford University), 46–64.