Time in the Mind

Research on temporal perception goes back to the nineteenth century. In 1860, Karl Ernst von Baer introduced the notion of perceptual moment suggesting that different durations of the moments result in a different flow of subjective time. In 1865, Ernst Mach looked for Weber's law in temporal perception, and he observed that 30 ms is apparently the lowest limit for subjective durations. Then, in 1868 Frans Cornelis Donders presented the reaction time paradigm, which remains the basis for chronometric analyses of mental processes. In the same year, Karl von Vierordt investigated temporal integration using the paradigm of stimulus reproduction. These ideas embedded in the conceptual framework of PSYCHOPHYSICS as promoted by Gustav Theodor Fechner, set the stage for many decades, but then interest in temporal mechanisms of perception and cognition in general declined. Only recently has temporal perception become a central issue again, because cognitive processes cannot be understood without their temporal dynamics.

At least two independent temporal processing systems have been described that are basic for perceptual and cognitive processes and that are characterized by discrete time sampling. Both these systems are fundamental for the instantiation of perceptual acts, of cognitive processing, or of volitional MOTOR CONTROL. First, some observations on a high-frequency processing system generating discrete time quanta of 30 ms duration are mentioned. Then, a low-frequency processing system setting up functional states of approximately 3 seconds, which is believed to be the operative basis for what we refer to as "subjective present," is addressed. In between those two temporal processing levels additional timing mechanisms come into play that are fundamental for motor control, especially for the repeated initiation of movements.

Evidence for a high-frequency processing system derives from studies on temporal order threshold. If the temporal sequence of two stimuli has to be indicated, independent of sensory modality a threshold of approximately 30 ms is observed. Sensory data picked up within 30 ms are treated as cotemporal, that is, a relationship of separate stimuli with respect to the before-after dimension cannot be established (Pöppel 1997). Temporal order threshold being identical in different sensory modalities (Hirsh and Sherrick 1961), thus, also indicates a lower limit for event identification. Temporal order analysis of the speech signal seems to be the basis for phoneme identification, a disturbance of temporal acuity being associated with language disorders such as APHASIA and LANGUAGE IMPAIRMENT (Tallal, Miller, and Fitch 1993; see also DYSLEXIA).

Support for distinct processing stages comes from a variety of studies using qualitatively different paradigms. Under stationary experimental conditions response distributions of reaction times (Harter and White 1968) and of pursuit (Pöppel and Logothetis 1986) or saccadic eye movements (see EYE MOVEMENTS AND VISUAL ATTENTION and OCULOMOTOR CONTROL) show multimodal characteristics with a 30 ms separation of distinct response modes. These multimodalities can be explained on the basis of neuronal oscillations. After the transduction of a stimulus a relaxation oscillation with a period of 30 ms is initiated, which is triggered instantaneously by the stimulus. Such an oscillatory mechanism being under environmental stimulus control allows integration of information from different sensory modalities, that is, data from various sources can be collected within one period, which defines a basic system state (Pöppel 1997). Possibly, the separate response modes represent similar successive and discrete decision-making stages as are assumed in high-speed scanning of short-term memory.

Some neurophysiological observations support the notion of discrete temporal processing on the basis of system states implemented by oscillations. The auditory evoked potential (see ELECTROPHYSIOLOGY, ELECTRIC AND MAGNETIC EVOKED FIELDS) in the midlatency region shows an oscillatory component with a period of 30 ms. This component is a sensitive marker for the anesthetic state because it selectively disappears during general anesthesia. Oscillations with a period of 30 ms represent functional system states that are prerequisites for the establishment of events. If temporal coherence within a neuronal network as expressed by oscillations is removed as with general anesthetics conscious representation is interrupted (Schwender et al. 1994). Events cannot be implemented as functional "building blocks" of conscious activity.

A low-frequency mechanism, independent of a high- frequency mechanism implementing system states for event identification, binds successive events up to approximately 3 seconds into perceptual and action units (Fraisse 1984; Pöppel 1997). Support for such a binding operation comes from studies on spontaneous alteration rates of ambiguous figures (see ILLUSIONS). If stimuli can be perceived with two perspectives (like the Necker cube), there is an automatic shift of perceptual content after 3 seconds (Schleidt and Kien 1997; Gomez et al. 1995). Such a perceptual shift is true also for ambiguous auditory material like the phoneme sequence CU-BA-CU . . . where one hears either CUBA or BACU. The spontaneous alteration rate in the two modalities suggests that after an exhaust period of 3 seconds, attentional mechanisms (see ATTENTION and ATTENTION AND THE HUMAN BRAIN) are elicited that open sensory channels for new information; if the physical stimulus remains the same, the alternative interpretation of the stimulus will gain control. Metaphorically, the brain asks every 3 seconds "what is new?" and with unusual stimuli like the ambiguous material the temporal eigen-operations of the brain are unmasked.

Temporal integration up to 3 seconds is also observed in sensorimotor behavior. If a subject is requested to synchronize a regular sequence of auditory stimuli with finger taps, stimuli are anticipated by some tens of milliseconds. Stimulus anticipation with high temporal precision is, however, possible only up to interstimulus intervals of 3 seconds. If the next stimulus lies too far in the future, that is, more than 3 seconds, it is not possible to program an anticipatory movement that is precisely related to the stimulus (Mates et al. 1994).

Because the experiments referred to (and others) employ qualitatively different paradigms covering perceptual processes, cognitive evaluations, or movement control, it is proposed that temporal integration up to 3 seconds is a general principle of the neurocognitive machinery. This integration is automatic and presemantic, that is, the temporal limit is not determined by what is processed, but by intrinsic time constants. Because of the omnipresence of temporal integration, it can be used for a pragmatic definition of the subjective present, which is characterized phenomenally by a feeling of nowness, or one can relate temporal integration to singular single states of being conscious (see CONSCIOUSNESS and CONSCIOUSNESS, NEUROBIOLOGY OF).

Additionally, on a different timing level control of motor performance can be registered. Two categorically distinct speed modes with frequencies of 2 Hz and 5 Hz in the sequential initiation of motor behavior are most prominent and can be assessed in simple finger tapping tasks. Nevertheless, they represent basic temporal movement characteristics. Fast automatic movements in the maximum speed in finger tapping can be performed with interresponse intervals of 150 to 200 ms, representing a frequency of approximately 5 Hz. The speed in a personally chosen finger- tapping task is performed with interresponse intervals around 500 ms, representing a frequency of approximately 2 Hz (Fraisse 1982). These two different frequency modes are also seen in other movement tasks and are associated with distinct sensorimotor control processes, the 2-Hz movement being under voluntary control and allowing the collection of somatosensory information, the maximum speed 5-Hz performance requiring only coarse preattentive control (Kunesch et al. 1989). In sensorimotor synchronization where the frequency of a pacer signal has to be reproduced accurately by finger taps, the notion of the categorical difference of the two frequency modes is complemented. The subjective representation of every single finger tap is possible only when a subject is tapping to interstimulus intervals of above 300 ms (Peters 1989). The single taps cannot be temporally resolved in somatosensory perception with interstimulus intervals below 300 ms. This threshold of approximately 300 ms marks the categorical change in motor performance, dividing the aforementioned two motor control processes into automatized movement and voluntarily controlled behavior.

See also

-- Ernst Pöppel and Marc Wittmann


Fraisse, P. (1982). Rhythm and tempo. In D. Deutsch, Ed., The Psychology of Music. New York: Academic Press, pp. 149-180.

Fraisse, P. (1984). Perception and estimation of time. Annual Review of Psychology 35:1-36

Gomez, C., E. D. Argandona, R. G. Solier, J. C. Angulo, and M. Vazquez. (1995). Timing and competition in networks representing ambiguous figures. Brain and Cognition 29:103-114.

Harter, M. R., and C. T. White. (1968). Periodicity within reaction time distributions and electromyograms. Quarterly Journal of Experimental Psychology 20:157-166.

Hirsh, I. J., and C. E. Sherrick. (1961). Perceived order in different sense modalities. Journal of Experimental Psychology 62:423-432.

Kunesch, E., F. Binkofski, and H.-J. Freund. (1989). Invariant temporal characteristics of manipulative hand movements. Experimental Brain Research 78:539-546.

Mates, J., U. Müller, T. Radil, and E. Pöppel. (1994). Temporal integration in sensorimotor synchronization. J. Cogn. Neuroscience 6:332-340.

Peters, M. (1989). The relationship between variability of intertap intervals. Psychological Research 51:38-42.

Pöppel, E. (1997). A hierarchical model of temporal perception. Trends in Cognitive Sciences 1:56-61.

Pöppel, E., and N. Logothetis. (1986). Neural oscillations in the brain. Discontinuous initiations of pursuit eye movements indicate a 30-Hz temporal framework for visual information processing. Naturwissenschaften 73:267-268.

Schleidt, M., and J. Kien. (1997). Segmentation in behavior and what it can tell us about brain function. Human Nature 8:77-111.

Schwender, D., C. Madler, S. Klasing, K. Peter, and E. Pöppel. (1994). Anaesthetic control of 40-Hz brain activity and implicit memory. Consciousness and Cognition 3:129-147

Tallal, P., S. Miller, and R. Fitch. (1993). Neurobiological basis of speech: A case for the preeminence of temporal processing. In P. Tallal, A. Galaburda, R. Llinas, and C. von Euler, Eds., Temporal Information Processing in the Nervous System. Special Reference to Dyslexia and Dysphasia. New York: The New York Academy of Sciences, pp. 27-47.

Further Readings

Atmanspacher, H., and E. Ruhnau, Eds. (1997). Time, Temporality, Now. Berlin: Springer.

Block, R., Ed. (1990). Cognitive Models of Psychological Time. Hillsdale, NJ: Erlbaum.

Hazeltine, E., L. Helmuth, and R. Ivry. (1997). Neural mechanisms of timing. Trends in Cognitive Sciences 1:163-169.

Nichelli, P. (1993). The neuropsychology of human temporal information processing. In F. Boiler and J. Grafman, Eds., Handbook of Neuropsychology 8. Amsterdam: Elsevier Science, pp. 337-369.

Pastor, M. A., and J. Artieda, Eds. (1996). Time, Internal Clocks, and Movement. Amsterdam: Elsevier Science.