Acta Oto-Laryngologica
ISSN: 0001-6489 (Print) 1651-2251 (Online) Journal homepage: http://www.tandfonline.com/loi/ioto20
The Human Auditory Steady-State Evoked Potentials Gilles Plourde, David R. Stapells & Terence W. Picton To cite this article: Gilles Plourde, David R. Stapells & Terence W. Picton (1991) The Human Auditory Steady-State Evoked Potentials, Acta Oto-Laryngologica, 111:sup491, 153-160 To link to this article: http://dx.doi.org/10.3109/00016489109136793
Published online: 08 Jul 2009.
Submit your article to this journal
Article views: 86
View related articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ioto20 Download by: [New York University]
Date: 13 February 2016, At: 02:16
Acta Otolaryngol (Stockh) 1991; Suppl. 491: 153-160
The Human Auditory Steady-State Evoked Potentials GILLES PLOURDE, DAVID R. STAPELLS and TERENCE W. PICTON From the Human Neuroscience Research Unit. University of Ottawa, Ottawa, Canada KIH 8M5
Downloaded by [New York University] at 02:16 13 February 2016
Plourde G, Picton TW, Stapells D. The human auditory steady-state evoked potentials. Acta Otolaryngol (Stockh) 1991; Suppl 491: 153-160. When auditory stimuli are presented at rates near 40/s, they evoke a steady-state middle latency response. This results from the super-positon of the transient responses evoked by each of the rapidly presented stimuli. The steady-state evoked potentials are most appropriately analyzed using frequency-based techniques. The response is larger for stimuli of higher intensity and of lower tonal frequency. The amplitude of the response varies with the state of arousal of the subject. Sleep results in a decrease in the amplitude to between one third and one half of the amplitude during wakefulness. The response is even further attenuated by general anesthesia. This auditory steady-state evoked potential may therefore be helpful in monitoring the state of arousal of a patient undergoing anesthesia. Key words: analysis, audiometry, sleep, anesthesia.
INTRODUCTION The steady-state evoked potential is “a repetitive evoked potential whose constituent discrete frequency components remain constant in amplitude and phase over an infinitely long time period” (I). This evoked potential therefore differs from the “transient” evoked potential which can show marked variations from stimulus to stimulus. The steady-state evoked potential is usually recorded using stimulus-rates that are sufficiently fast for the potentials evoked by one stimulus to overlap with those evoked by the preceding stimuli (2). The response then becomes periodic at the rate of stimulation or some harmonic thereof (Fig. I ) . Many different auditory steady-state evoked potentials can be recorded (3). A cochlear microphonic is generated by the hair cells, and a later “frequency following potential” is generated in regions of the brainstem. The middle latency evoked potentials show steady-state waveforms at rates of stimulation near 40 Hz (4, 5). The slow evoked potentials also show some steady-state responses at rates of stimulation near 5 Hz (6). This paper will be concerned with some recent thoughts on the steady-state middle-latency response.
NATURE OF T H E 40 H Z RESPONSE The literature suggests that the 40 Hz potential may be generated in either the auditory cortices of the temporal lobe o r in some multi-modality brainstem area. Magnetic recordings (7,8) show patterns consistent with a source dipole on the supratemporal plane. The electrical fields (9) show polarity-inversion over the Sylvian fissure, a finding which again suggests a generator on the supratemporal plane. However, studies of patients with intracerebral lesions (10, 11) have shown that the 40 Hz response is affected by midline lesions but not by unilateral lesions of the temporal lobes. The relations between the transient and steady-state responses may provide some additional information concerning the generation of the steady-state response. If the generators contributing to the steady-state reponse are unaffected by the rate of stimulation, the steady-state response at a particular frequency can be completely predicted from the transient response. Physiological systems, however, may adapt at higher stimulus-rates or may resonate at particular rates of stimulation. Hari and her colleagues (8) have recently demonstrated that the magnetic 40 Hz response is reasonably well predicted from the super-position of repeating
154
G.Plourde et al.
Downloaded by [New York University] at 02:16 13 February 2016
Transient and Steady State Responses
,
1
.l.B yu
I 188 ms
,
1
8.2 (rU
I 288 Hz
Fig. 1. Transient and steady-state evoked potentials. On the upper left is shown the evoked potential to a 500 Hz 60 dB nHL tone with rise and fall times o f 4 ms and a total duration of 10 ms that was presented at a rate of IO/s. This is the grand mean waveform from 10 subjects. The response was recorded between vertex and mastoid with positivity at the vertex plotted upward. The lower tracing on the left represents the predicted steady-state response that would be obtained if stimuli were presented at a rate of 401s and if there were no alteration in the amplitudes or latencies of the evoked potentialt at these rates. On the right are shown the frequency spectra of these two waveforms. The frequency spectrum of the steady-state evoked potential shows activity only at the rate of stimulation and at harmonics thereof.
transient responses. We performed similar calculations with the electrical 40 Hz evoked potentials (12) and found that the steady-state response is quite well predicted from the superposition of the transient responses. Fig. 2 shows the results of calculating the predicted steadystate responses from data published several years ago ( I 3). The fact that the steady-state response can be predicted from the transient respone has some interesting implications. First, the generators of all of the components recorded in the transient response must be unaffected by rapid rates of stimulation. They may therefore represent a neuronal system that is important for the analysis of rapidly occurring stimulus changes. Second, the different waves of the middle-latency response that superimpose to give the steady-state response must be generated by different neuronal systems. For example, N b cannot represent the repetitive tiring of the same neurons that are contributing t o Na since in this case the same neurons would be doubly activated at the rapid rates of stimulation. Third, the electrical steady-state response must contain both brainstem and cortical contributions. Since the transient response contains components that are generated in both the brainstem (wave V) and the cortex (wave Pa) and since the steady-state response is predicted from the transient response, lesions to either the brainstem or the cortex should change the steady-state response. The fact that unilateral lesions of the temporal lobe d o not show clear effects on the electrical response can be explained by the opposite hemisphere still generating a response that is recordable at the vertex. ANALYSIS PROCEDURES
The steady-state responses are most appropriately analyzed using the techniques of frequency analysis. Probably the most effective way of enhancing the signal relative to the background
Steady-state responses
Downloaded by [New York University] at 02:16 13 February 2016
ASSR
-
Actual and Predicted
Fig, 2. The nature of the auditory steady-state response. The top tracing shows the transient evoked potential obtained when stimuli were presented at a rate of lO/s. The middle tracing represents the average evoked potential for 10 subjects for the stimuli when they were presented at a rate of 40/s. The lower waveform represents the predicted steady-state reponse that would be obtained if there were simple superposition of the transient responses at these rates of stimulation. The actual and predicted steady-state responses are very similar. Slight differences in the waveforms are probably as much related to the approximations necessary in transferring graphic waveforms to digital form as to any actual differences. (The previously recorded data were transferred from plotted waveforms by picking points on the waveforms and spline-fitting the intervening regions.)
noise is to record the steady-state response over a prolonged period of time (1). For a given rate of analog-digital (A-D) conversion, the frequency resolution varies with the duration of the recording. The frequency resolution equals l/(NAf) where N is the number of addresses in the recording and Af is the time per address (the reciprocal of the A-D conversion rate).
Rate of Stimulation Rnpl itude
8.4 pU
8.8 ~U Phase
368 deg
I--I 35.5
4 4 . 5 8Hz deg
Fig. 3. The apparent latency of a steady-state response. These data were obtained by calculating the predicted steady-state responses at different rates of stimulation from the transient response shown in Figs. I and 2. Between stimulus-rates of 35 and 45/s, the amplitude is reasonably constant (with a maximum near 40/s) and there is a linear change in phase with increasing stimulus-rate. The slope of this phase-rate relationship gives an apparent latency of 17 ms.
155
156
G.PIourde et al. TIME
TASK
FREQUENCY
EYES
count p r e s s c’osed
d
count closed
L I 4 L
p r e s s closed
Downloaded by [New York University] at 02:16 13 February 2016
count p r e s s open
read
open
, , , , 0
,
,
500 ms
0
,
,
,
1
0.375
YV
Fig. 4. Average responses from 10 normal subjects during different experimental conditions. In the first four conditions the subject counted and/or pressed a button in response to an occasional change in the tonal frequency of 40/s stimuli. In the last condition the subject ignored the stimuli. On the left are shown the average waveforms from the vertex with negativity at the vertex plotted upward. On the right are shown the amplitude spectra for the average waveforms.
64 Hz
There are several ways to measure the reliability of a steady-state evoked potential (14, 15). One method is to compare the amplitude of the response at the frequency of stimulation to the amplitudes at adjacent frequencies in the spectrum. If one makes repeated measurements of the response, one can use either the phase coherence or the Hotelling’s T 2test to determine whether the repeated measurements are significantly different from what might be expected by random variation. Recently, Victor & Mast (16) have reported an adaptation of the Hotelling’s T 2test that is particularly useful for small samples. Although the meaning of the amplitude of a steady-state response is quite obvious, the interpretation of phase is less clear. Regan has proposed a measure of the “apparent latency”
0.5
0
P
I
S
E
R
Fig. 5 . Auditory steady-state responses during anesthesia. This graph represents the amplitudes of the auditory steady-state response to a 500 Hz tone presented at a rate of 40/s and an intensity of 90 dB peak SPL before, during and after anesthesia in 7 subjects. The responses were measured as the root mean square amplitude. The recordings were obtained in the pre-anesthetic period ( P ) , during induction ( I ) , during surgical anesthesia (9,during emergence ( E ) and during recovery ( R ) . The anesthesia was with isoflurane and the induction was with thiopental. The auditory steady-state response is dramatically attenuated by anesthesia and returns toward the pre-anesthetic value during emergence and recovery.
Steady-state responses 157 of a steady-state response (1, 2 ) . When steady-state responses are recorded at different stimulus-rates, the apparent latency is determined by the slope of the phase versus stimulusrate. This is shown in Fig. 3. For this figure the data were not recorded but were constructed fromthe transient response of Figs. 1 and 2. The response is well defined and the phase changes linearly over this range of stimulus-rates. The slope of the phase versus stimulus-rate gives an apparent latency of 17 ms. Higher values were obtained if the range of stimulus-rates was increased, although the phase versus rate relation became less linear. The actual data recorded from the individual subjects at stimulus rates between 30 and 60/s gave apparent latencies of 34 ms (13). Unfortunately, as Hari and her colleagues (8) have pointed out, the apparent latency is difficult to relate either to the waveform of the transient response or to the physiologic processes that underly the generation of the transient and steady-state potentials.
Downloaded by [New York University] at 02:16 13 February 2016
STIMULUS RELATIONS The steady-state middle latency potential evoked by a repeating toneburst shows clear reLitions to the intensity of the stimulus and to the frequency of the tone (17). In general. the response decreases regularly with decreasing intensity and is larger for low-frequency tones than for higher frequency tones. With decreasing intensity the phase-delay of the response increases for low-frequency tones but remains fairly constant for higher frequency tones. These relations are probably related to the variable super-position of the middle latency components and the brainstem responses. In waking adult subjects, the steady-state response can provide an accurate assessment of threshold at the lower frequencies of stimulation (17). However, the response is less clearly related to auditory threshold during sleep ( 1 5) and is difficult to record in newborn infants (12).
Another important aspect of the steady-state response is that it can be evoked by modulations in the amplitude or frequency of a continuous tone ( 1 8). The response may therefore provide an objective measurement of suprathreshold discrimination. SUBJECT RELATIONS Unlike the earlier brainstem response, the middle latency response can vary with the level of arousal. These changes are particularly prominent in the steady-state response. There are minute-to-minute fluctuations in the response that may relate to cycles of alertness (19). When a subject falls asleep, the amplitude of the response decreases quite dramatically to between one third and one half the amplitude recorded during the waking state (20). Provided that the patient does not become drowsy, however, there is little effect of selective attention on the response (21). Fig. 4 shows the auditory steady-state evoked potential recorded from 10 subjects under several different conditions of attention. These recordings were obtained while the subjects listened t o a train of 40/s tones of 500 Hz which occasionally changed to 700 Hz for a brief period (75 ms). The study was designed t o determine the consistency of the response under different behavioral conditions in a waking subject. Thus, the subject either attended to the stimuli in order to count and/or press a button in response to the frequency-change or ignored the stimuli and read a book. Under some conditions the eyes were closed and in others they were open. The response was significantly smaller in the “read” condition than in the others, but this difference was very small (0.43 vs. 0.47 pV). It was difficult to determine whether this was related to attention or to some drowsiness (during a condition wherein no response was required). Because of the sensitivity t o the state of arousal and the relative insensitivity to attentional variables, we decided to investigate the effects of anesthesia on auditory steady-state evoked
158
G.Plourde et al.
potentials ( 2 2 ) .We found that the steady-state response was consistently and reliably recorded in patients just before an anesthetic. The response was eliminated by anesthesia with isoflurane (after thiopental induction) and then returned when the patient emerged from the anesthetic (Fig. 5). The response may therefore be useful in the monitoring of the state of anesthesia. Until now evoked potentials have been mainly used intra-operatively to monitor a sensory pathway in order to detect damage to this pathway. We suggest that the 40 Hz response might monitor the state of the brain in order to detect inappropriate lightening of the anesthetic during an operation.
Downloaded by [New York University] at 02:16 13 February 2016
CONCLUSION The auditory steady-state response evoked at stimulus rates near 40/s has intriguing relations to both the physical aspects of the stimulus and the arousal state of the subject. It should have important clinical uses in both the evaluation of hearing and the monitoring of consciousness.
REFERENCES I . Regan D. Human brain electrophysiology. Evoked potentials and evoked magnetic fields in science and medicine. New York: Elsevier, 1989. 2. Regan D. Comparison of transient and steady-state methods. Ann NY Acad Sci 1982; 388: 45-71. 3. Picton TW, Durieux-Smith A. Auditory evoked potentials in the assessment of hearing. Neurol Clin 1988; 6: 791-808. 4. Galambos R, Makeig S, Talmachoff PJ. A 40 Hz auditory potential recorded from the human scalp. Proc Natl Acad Sci USA 198 I ; 78: 2643-7. 5. Picton TW. Human auditory steady state responses. In: Barber C, Blum T, eds. Evoked potentials 111. New York: Butterworths, 1987: 119-24. 6. Maiste A, Picton T. Human auditory evoked potentials to frequency-modulated tones. Ear Hear 1989; 10: 153-60. 7. Makela JP, Hari R. Evidence for cortical origin of the 40 Hz auditory evoked response in man. Electroencephalogr Clin Neurophysiol 1987; 66: 539-46. 8. Hari R, Hamalainen M, Joutsiniemi SR. Neuromagnetic steady-state response to auditory stimuli. J Acoust SOCAm 1989; 86: 1033-9. 9. Johnson BW, Weinberg H, Ribary U, Cheyne DO. Topographic distribution of the 40 Hz auditory event-related potential in normal and ages subjects. Brain Topography 1988; 1: 117-21. 10. Spydell JD, Pattee G , Goldie WD. The 40 Hertz event-related potential: Normal values and effects of lesions. Electroencephalogr Clin Neurophysiol 1985; 62: 193-202. 1 1 . Rei G, Bao F. Diagnostic significance of the staggered spondaic word test and 40 Hz auditory eventrelated potentials. Audiology 1988; 27: 8-1 6. 12. Stapells DR, Galambos R, Costello A, Makeig S. Inconsistency of auditory middle latency and steadystate responses in infants. Electroencephalogr Clin Neurophysiol 1988; 71 : 289-95. 13. Stapells DR, Linden D, Suffield B, Hamel G, Picton TW. Human auditory steady state potentials. Ear Hear 1984; 5: 105-113. 14. Stapells DR, Makeig S, Galambos R. Auditory steady-state responses: threshold prediction using phase coherence. Electroencephalogr Clin Neurophysiol 1987; 67: 260-70. 15. Picton TW, Vajsar J, Rodriguez R, Campbell KB. Reliability estimates for steady-state evoked potentials. Electroencephalogr Clin Neurophysiol 1987; 68: I 19-3 I. 16. Victor JD, Mast J. A new statistic for steady-state evoked potentials. Electroencephalogr Clin Neurophysiol 1991; 78: 378-88. 17. Rodriguez R, Picton T, Linden D, Hamel G, Laframboise G. Human auditory steady-state responses: effects of intensity and frequency. Ear Hear 1986; 7: 300-13. 18. Picton TW, Skinner CR, Champagne SC, Kellett AJ, Maiste A. Potentials evoked by the sinusoidal modulation of the amplitude or frequency of a tone. J Acoust SOCAm 1987; 82: 165-78. 19. Galambos R, Makeig S. Dynamic changes in steady-state responses. In: Basar E, ed. Dynamics of sensory and cognitive processing by the brain. Berlin: Springer-Verlag, 1988: 103-22.
Steady-state responses 1 59 20. Linden RD, Campbell KB, Hamel G , Picton TW. Human auditory steady state evoked potentials during sleep. Ear Hear 1985; 6: 167-74. 21. Linden RD, Picton TW, Hamel G , Campbell KB. Human auditory steady state evoked potentials during selective attention. Electroencephalogr Clin Neurophysiol 1987; 66: 145-59. 22. Plourde G , Picton TW. Human auditory steady-state responses during general anesthesia. Anesth Analg 1990; 71: 460-8.
Downloaded by [New York University] at 02:16 13 February 2016
Address for correspondence: T. W. Picton, Division of Neurology, Ottawa General Hospital, 501 Smyth Road, Ottawa, Canada KI H 8L6
160 G.Plourde et al.
DISCUSSION
Downloaded by [New York University] at 02:16 13 February 2016
Participants: Bavar, Lutkenhoner, Picton, Scherg, Schreiner, Speckmann
Obtaining responses at 10 dB above threshold was typical of measurements at low frequencies (500 Hz) in subjects with normal hearing in a quiet background. In patients with hearing losses, 20 dB above threshold was more usual. Regan has described an implementation of his method that is very complicated; the recording has to be made on FM tape and then replayed. This could be done “on line” by taking all the epochs and doing a cross-correlation with a cosine of 40 Hz and of 40.1 Hz. Then one has only to take into account the phase shifts of subsequent epochs and d o the same high resolution analysis on sine waves at exactly 40 Hz. From this one should obtain a good criterion for threshold determination. The response was smaller during sleep but there were no significant differences in the responses between any of the stages of sleep. There may be small differences that could be detected using larger numbers of subjects. Bavar stated that the physiological implications of these results should be taken into account. There are evoked 40 Hz oscillations in several parts of the cortex. If the input stimulation is at a frequency of 40 Hz, the information contained in these 40 Hz oscillations is lost. Repetitive stimulation is not necessary, because every single evoked response contains 40 Hz oscillations. Scherg replied that experimentally, from both MEG and the stimulation described in this paper, there is no resonance enhancement. The term “oscillation” should be strictly defined. In his source analysis, Scherg finds a primary generator in the auditory cortex that generates N19 and P30,which would be two half sine waves. This could be called an oscillation but it is not a resonance phenomenon. Bavar said that in systems theory the time and frequency domains are equivalent. One can predict from the time domain the response in the frequency domain. To assess the physiological implications, one should use intracranial electrodes. Schreiner has studied the brainstem response t o clicks given in a pseudo-random maximum length sequence. The normal brainstem response could be extracted by deconvolution. This should be possible also for the middle latency responses. Then one would not be restricted t o looking at one cycle but would see the whole time course of the response. A disadvantage might be that the signal would be enhanced at particular frequencies.
ISSN: 0001-6489 (Print) 1651-2251 (Online) Journal homepage: http://www.tandfonline.com/loi/ioto20
The Human Auditory Steady-State Evoked Potentials Gilles Plourde, David R. Stapells & Terence W. Picton To cite this article: Gilles Plourde, David R. Stapells & Terence W. Picton (1991) The Human Auditory Steady-State Evoked Potentials, Acta Oto-Laryngologica, 111:sup491, 153-160 To link to this article: http://dx.doi.org/10.3109/00016489109136793
Published online: 08 Jul 2009.
Submit your article to this journal
Article views: 86
View related articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ioto20 Download by: [New York University]
Date: 13 February 2016, At: 02:16
Acta Otolaryngol (Stockh) 1991; Suppl. 491: 153-160
The Human Auditory Steady-State Evoked Potentials GILLES PLOURDE, DAVID R. STAPELLS and TERENCE W. PICTON From the Human Neuroscience Research Unit. University of Ottawa, Ottawa, Canada KIH 8M5
Downloaded by [New York University] at 02:16 13 February 2016
Plourde G, Picton TW, Stapells D. The human auditory steady-state evoked potentials. Acta Otolaryngol (Stockh) 1991; Suppl 491: 153-160. When auditory stimuli are presented at rates near 40/s, they evoke a steady-state middle latency response. This results from the super-positon of the transient responses evoked by each of the rapidly presented stimuli. The steady-state evoked potentials are most appropriately analyzed using frequency-based techniques. The response is larger for stimuli of higher intensity and of lower tonal frequency. The amplitude of the response varies with the state of arousal of the subject. Sleep results in a decrease in the amplitude to between one third and one half of the amplitude during wakefulness. The response is even further attenuated by general anesthesia. This auditory steady-state evoked potential may therefore be helpful in monitoring the state of arousal of a patient undergoing anesthesia. Key words: analysis, audiometry, sleep, anesthesia.
INTRODUCTION The steady-state evoked potential is “a repetitive evoked potential whose constituent discrete frequency components remain constant in amplitude and phase over an infinitely long time period” (I). This evoked potential therefore differs from the “transient” evoked potential which can show marked variations from stimulus to stimulus. The steady-state evoked potential is usually recorded using stimulus-rates that are sufficiently fast for the potentials evoked by one stimulus to overlap with those evoked by the preceding stimuli (2). The response then becomes periodic at the rate of stimulation or some harmonic thereof (Fig. I ) . Many different auditory steady-state evoked potentials can be recorded (3). A cochlear microphonic is generated by the hair cells, and a later “frequency following potential” is generated in regions of the brainstem. The middle latency evoked potentials show steady-state waveforms at rates of stimulation near 40 Hz (4, 5). The slow evoked potentials also show some steady-state responses at rates of stimulation near 5 Hz (6). This paper will be concerned with some recent thoughts on the steady-state middle-latency response.
NATURE OF T H E 40 H Z RESPONSE The literature suggests that the 40 Hz potential may be generated in either the auditory cortices of the temporal lobe o r in some multi-modality brainstem area. Magnetic recordings (7,8) show patterns consistent with a source dipole on the supratemporal plane. The electrical fields (9) show polarity-inversion over the Sylvian fissure, a finding which again suggests a generator on the supratemporal plane. However, studies of patients with intracerebral lesions (10, 11) have shown that the 40 Hz response is affected by midline lesions but not by unilateral lesions of the temporal lobes. The relations between the transient and steady-state responses may provide some additional information concerning the generation of the steady-state response. If the generators contributing to the steady-state reponse are unaffected by the rate of stimulation, the steady-state response at a particular frequency can be completely predicted from the transient response. Physiological systems, however, may adapt at higher stimulus-rates or may resonate at particular rates of stimulation. Hari and her colleagues (8) have recently demonstrated that the magnetic 40 Hz response is reasonably well predicted from the super-position of repeating
154
G.Plourde et al.
Downloaded by [New York University] at 02:16 13 February 2016
Transient and Steady State Responses
,
1
.l.B yu
I 188 ms
,
1
8.2 (rU
I 288 Hz
Fig. 1. Transient and steady-state evoked potentials. On the upper left is shown the evoked potential to a 500 Hz 60 dB nHL tone with rise and fall times o f 4 ms and a total duration of 10 ms that was presented at a rate of IO/s. This is the grand mean waveform from 10 subjects. The response was recorded between vertex and mastoid with positivity at the vertex plotted upward. The lower tracing on the left represents the predicted steady-state response that would be obtained if stimuli were presented at a rate of 401s and if there were no alteration in the amplitudes or latencies of the evoked potentialt at these rates. On the right are shown the frequency spectra of these two waveforms. The frequency spectrum of the steady-state evoked potential shows activity only at the rate of stimulation and at harmonics thereof.
transient responses. We performed similar calculations with the electrical 40 Hz evoked potentials (12) and found that the steady-state response is quite well predicted from the superposition of the transient responses. Fig. 2 shows the results of calculating the predicted steadystate responses from data published several years ago ( I 3). The fact that the steady-state response can be predicted from the transient respone has some interesting implications. First, the generators of all of the components recorded in the transient response must be unaffected by rapid rates of stimulation. They may therefore represent a neuronal system that is important for the analysis of rapidly occurring stimulus changes. Second, the different waves of the middle-latency response that superimpose to give the steady-state response must be generated by different neuronal systems. For example, N b cannot represent the repetitive tiring of the same neurons that are contributing t o Na since in this case the same neurons would be doubly activated at the rapid rates of stimulation. Third, the electrical steady-state response must contain both brainstem and cortical contributions. Since the transient response contains components that are generated in both the brainstem (wave V) and the cortex (wave Pa) and since the steady-state response is predicted from the transient response, lesions to either the brainstem or the cortex should change the steady-state response. The fact that unilateral lesions of the temporal lobe d o not show clear effects on the electrical response can be explained by the opposite hemisphere still generating a response that is recordable at the vertex. ANALYSIS PROCEDURES
The steady-state responses are most appropriately analyzed using the techniques of frequency analysis. Probably the most effective way of enhancing the signal relative to the background
Steady-state responses
Downloaded by [New York University] at 02:16 13 February 2016
ASSR
-
Actual and Predicted
Fig, 2. The nature of the auditory steady-state response. The top tracing shows the transient evoked potential obtained when stimuli were presented at a rate of lO/s. The middle tracing represents the average evoked potential for 10 subjects for the stimuli when they were presented at a rate of 40/s. The lower waveform represents the predicted steady-state reponse that would be obtained if there were simple superposition of the transient responses at these rates of stimulation. The actual and predicted steady-state responses are very similar. Slight differences in the waveforms are probably as much related to the approximations necessary in transferring graphic waveforms to digital form as to any actual differences. (The previously recorded data were transferred from plotted waveforms by picking points on the waveforms and spline-fitting the intervening regions.)
noise is to record the steady-state response over a prolonged period of time (1). For a given rate of analog-digital (A-D) conversion, the frequency resolution varies with the duration of the recording. The frequency resolution equals l/(NAf) where N is the number of addresses in the recording and Af is the time per address (the reciprocal of the A-D conversion rate).
Rate of Stimulation Rnpl itude
8.4 pU
8.8 ~U Phase
368 deg
I--I 35.5
4 4 . 5 8Hz deg
Fig. 3. The apparent latency of a steady-state response. These data were obtained by calculating the predicted steady-state responses at different rates of stimulation from the transient response shown in Figs. I and 2. Between stimulus-rates of 35 and 45/s, the amplitude is reasonably constant (with a maximum near 40/s) and there is a linear change in phase with increasing stimulus-rate. The slope of this phase-rate relationship gives an apparent latency of 17 ms.
155
156
G.PIourde et al. TIME
TASK
FREQUENCY
EYES
count p r e s s c’osed
d
count closed
L I 4 L
p r e s s closed
Downloaded by [New York University] at 02:16 13 February 2016
count p r e s s open
read
open
, , , , 0
,
,
500 ms
0
,
,
,
1
0.375
YV
Fig. 4. Average responses from 10 normal subjects during different experimental conditions. In the first four conditions the subject counted and/or pressed a button in response to an occasional change in the tonal frequency of 40/s stimuli. In the last condition the subject ignored the stimuli. On the left are shown the average waveforms from the vertex with negativity at the vertex plotted upward. On the right are shown the amplitude spectra for the average waveforms.
64 Hz
There are several ways to measure the reliability of a steady-state evoked potential (14, 15). One method is to compare the amplitude of the response at the frequency of stimulation to the amplitudes at adjacent frequencies in the spectrum. If one makes repeated measurements of the response, one can use either the phase coherence or the Hotelling’s T 2test to determine whether the repeated measurements are significantly different from what might be expected by random variation. Recently, Victor & Mast (16) have reported an adaptation of the Hotelling’s T 2test that is particularly useful for small samples. Although the meaning of the amplitude of a steady-state response is quite obvious, the interpretation of phase is less clear. Regan has proposed a measure of the “apparent latency”
0.5
0
P
I
S
E
R
Fig. 5 . Auditory steady-state responses during anesthesia. This graph represents the amplitudes of the auditory steady-state response to a 500 Hz tone presented at a rate of 40/s and an intensity of 90 dB peak SPL before, during and after anesthesia in 7 subjects. The responses were measured as the root mean square amplitude. The recordings were obtained in the pre-anesthetic period ( P ) , during induction ( I ) , during surgical anesthesia (9,during emergence ( E ) and during recovery ( R ) . The anesthesia was with isoflurane and the induction was with thiopental. The auditory steady-state response is dramatically attenuated by anesthesia and returns toward the pre-anesthetic value during emergence and recovery.
Steady-state responses 157 of a steady-state response (1, 2 ) . When steady-state responses are recorded at different stimulus-rates, the apparent latency is determined by the slope of the phase versus stimulusrate. This is shown in Fig. 3. For this figure the data were not recorded but were constructed fromthe transient response of Figs. 1 and 2. The response is well defined and the phase changes linearly over this range of stimulus-rates. The slope of the phase versus stimulus-rate gives an apparent latency of 17 ms. Higher values were obtained if the range of stimulus-rates was increased, although the phase versus rate relation became less linear. The actual data recorded from the individual subjects at stimulus rates between 30 and 60/s gave apparent latencies of 34 ms (13). Unfortunately, as Hari and her colleagues (8) have pointed out, the apparent latency is difficult to relate either to the waveform of the transient response or to the physiologic processes that underly the generation of the transient and steady-state potentials.
Downloaded by [New York University] at 02:16 13 February 2016
STIMULUS RELATIONS The steady-state middle latency potential evoked by a repeating toneburst shows clear reLitions to the intensity of the stimulus and to the frequency of the tone (17). In general. the response decreases regularly with decreasing intensity and is larger for low-frequency tones than for higher frequency tones. With decreasing intensity the phase-delay of the response increases for low-frequency tones but remains fairly constant for higher frequency tones. These relations are probably related to the variable super-position of the middle latency components and the brainstem responses. In waking adult subjects, the steady-state response can provide an accurate assessment of threshold at the lower frequencies of stimulation (17). However, the response is less clearly related to auditory threshold during sleep ( 1 5) and is difficult to record in newborn infants (12).
Another important aspect of the steady-state response is that it can be evoked by modulations in the amplitude or frequency of a continuous tone ( 1 8). The response may therefore provide an objective measurement of suprathreshold discrimination. SUBJECT RELATIONS Unlike the earlier brainstem response, the middle latency response can vary with the level of arousal. These changes are particularly prominent in the steady-state response. There are minute-to-minute fluctuations in the response that may relate to cycles of alertness (19). When a subject falls asleep, the amplitude of the response decreases quite dramatically to between one third and one half the amplitude recorded during the waking state (20). Provided that the patient does not become drowsy, however, there is little effect of selective attention on the response (21). Fig. 4 shows the auditory steady-state evoked potential recorded from 10 subjects under several different conditions of attention. These recordings were obtained while the subjects listened t o a train of 40/s tones of 500 Hz which occasionally changed to 700 Hz for a brief period (75 ms). The study was designed t o determine the consistency of the response under different behavioral conditions in a waking subject. Thus, the subject either attended to the stimuli in order to count and/or press a button in response to the frequency-change or ignored the stimuli and read a book. Under some conditions the eyes were closed and in others they were open. The response was significantly smaller in the “read” condition than in the others, but this difference was very small (0.43 vs. 0.47 pV). It was difficult to determine whether this was related to attention or to some drowsiness (during a condition wherein no response was required). Because of the sensitivity t o the state of arousal and the relative insensitivity to attentional variables, we decided to investigate the effects of anesthesia on auditory steady-state evoked
158
G.Plourde et al.
potentials ( 2 2 ) .We found that the steady-state response was consistently and reliably recorded in patients just before an anesthetic. The response was eliminated by anesthesia with isoflurane (after thiopental induction) and then returned when the patient emerged from the anesthetic (Fig. 5). The response may therefore be useful in the monitoring of the state of anesthesia. Until now evoked potentials have been mainly used intra-operatively to monitor a sensory pathway in order to detect damage to this pathway. We suggest that the 40 Hz response might monitor the state of the brain in order to detect inappropriate lightening of the anesthetic during an operation.
Downloaded by [New York University] at 02:16 13 February 2016
CONCLUSION The auditory steady-state response evoked at stimulus rates near 40/s has intriguing relations to both the physical aspects of the stimulus and the arousal state of the subject. It should have important clinical uses in both the evaluation of hearing and the monitoring of consciousness.
REFERENCES I . Regan D. Human brain electrophysiology. Evoked potentials and evoked magnetic fields in science and medicine. New York: Elsevier, 1989. 2. Regan D. Comparison of transient and steady-state methods. Ann NY Acad Sci 1982; 388: 45-71. 3. Picton TW, Durieux-Smith A. Auditory evoked potentials in the assessment of hearing. Neurol Clin 1988; 6: 791-808. 4. Galambos R, Makeig S, Talmachoff PJ. A 40 Hz auditory potential recorded from the human scalp. Proc Natl Acad Sci USA 198 I ; 78: 2643-7. 5. Picton TW. Human auditory steady state responses. In: Barber C, Blum T, eds. Evoked potentials 111. New York: Butterworths, 1987: 119-24. 6. Maiste A, Picton T. Human auditory evoked potentials to frequency-modulated tones. Ear Hear 1989; 10: 153-60. 7. Makela JP, Hari R. Evidence for cortical origin of the 40 Hz auditory evoked response in man. Electroencephalogr Clin Neurophysiol 1987; 66: 539-46. 8. Hari R, Hamalainen M, Joutsiniemi SR. Neuromagnetic steady-state response to auditory stimuli. J Acoust SOCAm 1989; 86: 1033-9. 9. Johnson BW, Weinberg H, Ribary U, Cheyne DO. Topographic distribution of the 40 Hz auditory event-related potential in normal and ages subjects. Brain Topography 1988; 1: 117-21. 10. Spydell JD, Pattee G , Goldie WD. The 40 Hertz event-related potential: Normal values and effects of lesions. Electroencephalogr Clin Neurophysiol 1985; 62: 193-202. 1 1 . Rei G, Bao F. Diagnostic significance of the staggered spondaic word test and 40 Hz auditory eventrelated potentials. Audiology 1988; 27: 8-1 6. 12. Stapells DR, Galambos R, Costello A, Makeig S. Inconsistency of auditory middle latency and steadystate responses in infants. Electroencephalogr Clin Neurophysiol 1988; 71 : 289-95. 13. Stapells DR, Linden D, Suffield B, Hamel G, Picton TW. Human auditory steady state potentials. Ear Hear 1984; 5: 105-113. 14. Stapells DR, Makeig S, Galambos R. Auditory steady-state responses: threshold prediction using phase coherence. Electroencephalogr Clin Neurophysiol 1987; 67: 260-70. 15. Picton TW, Vajsar J, Rodriguez R, Campbell KB. Reliability estimates for steady-state evoked potentials. Electroencephalogr Clin Neurophysiol 1987; 68: I 19-3 I. 16. Victor JD, Mast J. A new statistic for steady-state evoked potentials. Electroencephalogr Clin Neurophysiol 1991; 78: 378-88. 17. Rodriguez R, Picton T, Linden D, Hamel G, Laframboise G. Human auditory steady-state responses: effects of intensity and frequency. Ear Hear 1986; 7: 300-13. 18. Picton TW, Skinner CR, Champagne SC, Kellett AJ, Maiste A. Potentials evoked by the sinusoidal modulation of the amplitude or frequency of a tone. J Acoust SOCAm 1987; 82: 165-78. 19. Galambos R, Makeig S. Dynamic changes in steady-state responses. In: Basar E, ed. Dynamics of sensory and cognitive processing by the brain. Berlin: Springer-Verlag, 1988: 103-22.
Steady-state responses 1 59 20. Linden RD, Campbell KB, Hamel G , Picton TW. Human auditory steady state evoked potentials during sleep. Ear Hear 1985; 6: 167-74. 21. Linden RD, Picton TW, Hamel G , Campbell KB. Human auditory steady state evoked potentials during selective attention. Electroencephalogr Clin Neurophysiol 1987; 66: 145-59. 22. Plourde G , Picton TW. Human auditory steady-state responses during general anesthesia. Anesth Analg 1990; 71: 460-8.
Downloaded by [New York University] at 02:16 13 February 2016
Address for correspondence: T. W. Picton, Division of Neurology, Ottawa General Hospital, 501 Smyth Road, Ottawa, Canada KI H 8L6
160 G.Plourde et al.
DISCUSSION
Downloaded by [New York University] at 02:16 13 February 2016
Participants: Bavar, Lutkenhoner, Picton, Scherg, Schreiner, Speckmann
Obtaining responses at 10 dB above threshold was typical of measurements at low frequencies (500 Hz) in subjects with normal hearing in a quiet background. In patients with hearing losses, 20 dB above threshold was more usual. Regan has described an implementation of his method that is very complicated; the recording has to be made on FM tape and then replayed. This could be done “on line” by taking all the epochs and doing a cross-correlation with a cosine of 40 Hz and of 40.1 Hz. Then one has only to take into account the phase shifts of subsequent epochs and d o the same high resolution analysis on sine waves at exactly 40 Hz. From this one should obtain a good criterion for threshold determination. The response was smaller during sleep but there were no significant differences in the responses between any of the stages of sleep. There may be small differences that could be detected using larger numbers of subjects. Bavar stated that the physiological implications of these results should be taken into account. There are evoked 40 Hz oscillations in several parts of the cortex. If the input stimulation is at a frequency of 40 Hz, the information contained in these 40 Hz oscillations is lost. Repetitive stimulation is not necessary, because every single evoked response contains 40 Hz oscillations. Scherg replied that experimentally, from both MEG and the stimulation described in this paper, there is no resonance enhancement. The term “oscillation” should be strictly defined. In his source analysis, Scherg finds a primary generator in the auditory cortex that generates N19 and P30,which would be two half sine waves. This could be called an oscillation but it is not a resonance phenomenon. Bavar said that in systems theory the time and frequency domains are equivalent. One can predict from the time domain the response in the frequency domain. To assess the physiological implications, one should use intracranial electrodes. Schreiner has studied the brainstem response t o clicks given in a pseudo-random maximum length sequence. The normal brainstem response could be extracted by deconvolution. This should be possible also for the middle latency responses. Then one would not be restricted t o looking at one cycle but would see the whole time course of the response. A disadvantage might be that the signal would be enhanced at particular frequencies.
