- Email: [email protected]
A method for the generation of complex vibrotactile stimuli
Journal of Neuroscience Methods, 38 ( 1991) 47 - 5 0
47
© 1991 Elsevier Science Publishers B.V. 0165-0270/91/$03.50
NSM 01234
A method for the generation of complex vibrotactile stimuli J.W. M o r l e y , E.N. C r a w f o r d , D . G . F e r r i n g t o n , J.S. A r c h e r , A.B. T u r m a n a n d M.J. R o w e School of Physiology and Pharmacology, University of New South Wales, Sydney, NSW 2033 (Australia) (Received 3 January 1991) (Revised version received 19 February 1991) (Accepted 20 February 1991)
Key words: Complex waveform; Vibrotactile stimulus; Beat A method is described utilizing computer-generated sine wave data and purpose-built hardware to generate a complex vibrotactile stimulus. Two sine waves of different frequency were s u m m e d to produce a complex waveform with two temporal components, a high frequency component and a low frequency beat component. T h e computer-generated data points for each of the two c o m p o n e n t sine waves were downloaded to two banks of static memory in a dual synchronous arbitrary function generator. The data points in memory were fed to two 12-bit digital-to-analogue converters which sent the two analogue sine wave signals to a s u m m i n g amplifier where the two sine waves were added. This method provides a complex waveform that can be gated on and off, has a fixed frequency ratio of the c o m p o n e n t sine waves and no phase drift between the c o m p o n e n t waves. Addition of the separate sine waves in a s u m m i n g amplifier allows for easy alteration of the amplitude ratio of the sine waves. The output of the s u m m i n g amplifier is sent to a feedback controlled mechanical stimulator, thereby allowing the stimulus to be presented to the skin of h u m a n subjects and experimental animals.
Introduction
Most studies on tactile responsiveness of cutaneous sensory fibers in both human subjects and experimental animals have been confined to the use of step stimuli or simple sinusoidal vibration of the skin (Talbot et al., 1968; J~inig et al., 1968; Mountcastle et al., 1972; Ferrington and Rowe, 1980; Johansson et al., 1982; Ferrington et al., 1984). We have recently carried out a series of experiments using a vibratory stimulus that is more complex than a simple sinusoidal stimulus and mimics to some extent the types of vibratory disturbances that may occur in the skin in the
Correspondence: Dr. J.W. Morley, School of Physiology and Pharmacology, University of New South Wales, P.O. Box 1, Kensington, NSW 2033, Australia. Phone: (02) 697 2549; Fax: (02) 313 6043.
tactile exploration of simple textures (Morley and Goodwin, 1987). The complex vibrotactile stimulus consists of the sum of two sine waves of different frequency, resulting in a vibratory stimulus with two temporal components. The stimulus resembles the type of waveform employed in studies of the neural coding of combination tones in the auditory system (Brugge et al., 1969; Rose, 1980). The waveform in the auditory studies was generated by combining the output of a sine wave oscillator and a function generator. For our experiments, generation of the complex waveform needed to satisfy three criteria. First, the waveform needed to be gated on and off; second, the frequency ratio of each of the component sine waves needed to be accurately fixed; and third, phase drift between the component sine waves had to be avoided. In this paper we present an account of the
4~
method we have developed to generate a complex vibrotactile stimulus, satisfying each of the above criteria. We have employed the complex stimulus
to investigate: (1) the ability of' human subjects to detect and discriminate the separate temporal components of the stimulus, and (2) the neural
8 J
/
Data In END-COUNT REGISTER
Variable Frequency
Out Count Enable
z
COMMAND
IN
8
/
o
[ !
Latch Select Address
I
CLOCK
IN
from Function
I 14
Write
E-, Z O t)
CK
Generator
In 1
l-~
ADDRESS
1o
COMPARATOR
!
I
Count Control Losic
ln~
O
_
GATE IN
8
L__q~
Data l~
CK .__.~MEMORY ADDRESS
I
Trl-State I BUFFER
COUNTER I,
/--
out
I/4
/~ 4
14 /
Tri-State BUFFER
¢4 ...J O DATA
IN
~ (
Tri-State , BUFFER I '
i~ " \]
ew
1
Analogue OUT 2
Tri-State BUFFER Analogue
o
e~ e~ BUFFER
"
OUT 1
k
"
\ Fig. 1. Simplified block diagram of the dual synchronous arbitrary function generator. The data points for each component sine wave, generated on an Apple lie computer, were passed on separate channels to two banks of static memory. The memory address counter cycled through the data points, which were then sent to two 12 bit DACs. W h e n the last data point was reached the memory address counter was reset to zero and the data cycle repeated. The analogue signals from the two D A C s were passed to a s u m m i n g amplifier (not shown) where the two signals were added.
4 t)
representation of each temporal component of the stimulus in the activity of single mechanoreceptive fibers and central neurons in the anesthetised cat (Ferrington et al., 1987, 1988; Morley et a1.,1989).
Complex waveform The complex waveform was generated by electronically summing two sine waves of different frequency. The waveform has a frequency that is the average of the two component frequencies and an amplitude that varies with time (Halliday and Resnick, 1981). In the case of an equivalent sound stimulus this amplitude variation gives rise to an auditory sensation termed a beat. It was essential that the two waveforms were accurately synchronized and able to be gated on and off. To satisfy these criteria we used a combination of computer generation of the component sine wave data and purpose-built hardware, which we have termed a dual synchronous arbitrary function generator. To generate acceptable sine waves on the computer we used between 20 and 50 data points per sine wave. Each component wave was downloaded on separate output channels to the hardware unit that contained memory, two digital-to-analogue converters (DACs) and logic. In our experiments we used vibration frequencies up to a maximum of about 2000 Hz and therefore required an upper sampling limit of 100 kHz per channel. The hardware logic allowed data sampling rates of greater than 1 MHz. Fig. 1 is a simplified block diagram of the dual synchronous arbitrary function generator. The high data sampling rates of the dual synchronous arbitrary function generator allowed us to utilize an inexpensive computer (Apple lie) to generate the sine wave data. Software to generate the sine waves was written in Apple II Pascal and machine language. The software was menu-driven and the sine wave p a r a m e t e r s could be saved on disk, thereby allowing rapid retrieval and downloading of data when the frequency p a r a m e t e r s of the component sine waves were changed during experiments. Full details of both hardware and software are available from the authors.
The two frequencies of the component sine waves had a ratio N 1 : N 2, where N l and N, are integers. It follows that N 1 cycles of one wave and N~ cycles of the second wave take the same time (T). The time T is the period of the complex wave (beat cycle) and is the period for which the data are generated. If both waves begin at zero (time = 0), data are generated from time = 0 to the last sample before time = T, when the wholc cycle (and data) is repeated. Data points for this period for each wave were fed to two banks of memory (one for each wave, each 16 K × 12 bits). A memory address counter cycles through the memory points and feeds the successive points to the two 12 bit DACs. A separate address limiting register is loaded with the address of the final data point; at this value the memory, address is reset to zero, and the cycle repeats. The data rate for feeding the DACs is controlled by a variable frequency source (standard function generator). The software calculates the required data transfer rate and indicates the setting to be used. The waveform is gated on and off by providing a trigger signal to the computer on a binary input. When the system is gated ON the above cycle commences. When it is gated OFF, it continues running until the end of the current run through the data and back to address zero (i.e., the first point where both waves are zero). This ensures that both waves always start at precisely the same phase. The two sine waves are brought out separately from each D A C and are added in a summing amplifier. The amplifier has separate potentiometers to allow the amplitude ratio of each wave to be varied. The summed complex waveform is sent to a feedback controlled mechanical stimulator, allowing the stimulus to be presented to the skin of human subjects in psychophysical experiments and the anesthetised cat in electrophysiological experiments.
Discussion The combination of computer-generated waveform data and a purpose-built hardware unit allowed us to produce a complex vibrotactile stimu-
50
lus which satisfied the three requirements specified in the Introduction. A fixed ratio of the sine wave frequencies was essential to avoid irregular wave patterns; for example, if a one second stimulus at a frequency of 1 kHz has an error in frequency of only 0.1%, this will result in a 360 ° phase shift (one complete cycle) during the one second stimulus period. There are a number of ways in which a similar waveform to the one we employed could be generated. One such way is to have two digitally controlled function generators gated together and sum the outputs. This method is not suitable however, as even very small errors in the frequency accuracy (see above) result in irregular waveforms. Alternatively, two frequencies could be accurately locked by combining frequency multipliers a n d / o r dividers and phase-locked loops. However, such systems are difficult to gate and have a tendency to take some time to settle down after being gated on. In this method the oscillator could be run continuously and only the output gated on and off, however the phase of the two waves at gate O N would be random. The computer synthesis technique is the most effective means of satisfying all our requirements. We chose to download the two separate sine waves to separate D A C channels and add them in a summing amplifier. Alternatively, the two waves could be summed in the computer and the complex waveform fed out through a single DAC. This would be a simpler method if the frequencies, and the ratio of the frequencies, of the component sine waves were varied. However, if this method were to be used, the waveform would need to be recalculated each time a change in the amplitude ratio was required, and so the method would be time consuming and less convenient.
Acknowledgements This work was supported by the National Health and Medical Research Council of Australia and the Australian Research Council.
References Brugge, J.F., Anderson, D.J., Hind, J.E. and Rose, J.E. (1969) Time structure of discharges in single auditory nerve fibers of the squirrel monkey in response to complex periodic sounds, J. Neurophysiol., 32: 386-401. Ferrington, D.G. and Rowe, M.J. (1980) Functional capacities of tactile afferent fibres in neonatal kittens. J. Physiol. (Lond.), 307: 335-353. Ferrington, D.G., Hora, M.O.H. and Rowe, M~J. (1984) Development of coding capacities in tactile afferent fibers of the kitten, J. Neurophysiol., 52: 74-85. Ferrington, D.G., Archer, J.S. and Rowe. M.J. (1987) Impulse patterning in neural responses to complex vibrotactile stimuli, Proc. Aust. Physiol. Pharmacol. Soc., 19: 104P. Ferrington, D.G., Archer, J.S. and Rowe. M.J. (1988) Coding of complex vibrotactile stimuli in the impulse patterns of afferent fibres, Proc. Aust. Physiol. Pharmacol. Soc. 19: 196P, Halliday, D. and Resnick, R. (1981) Fundamentals of Physics, 2nd edn., Wiley and Sons, New York. Johansson, R.S., Landstr6m, V. and Lundstr6m, R. (1982) Responses of mechanoreceptive afferent units in the glabrous skin of the human hand to sinusoidal skin displacements, Brain Res., 244: 17-25. J~inig, W., Schmidt, R.F. and Zimmermann, M. (1968) Single unit responses and the total afferent outflow from the cat's foot pad upon mechanical stimulation, Exp. Brain Res., 6: 100-115. Morley, J.W. and Goodwin, A.W. (1987) Sinusoidal movement of a grating across the monkey's fingerpad: temporal pattern of afferent fiber responses, J. Neurosci., 7: 2181-2191. Morley, J.W., Ferrington, D.G., Rowe, M.J. and Turman, A.B. (1990) Representation of complex vibrotactile stimuli in the responses of neurones in the cuneate nueleus of the cat, Neurosci. Lett., 34: S125. Mountcastle, V.B., LaMotte, R.H. and Carli, G. (1972) Detection thresholds for stimuli in humans and monkeys: comparison with threshold events in mechanoreceptive afferent nerve fibres innervating the monkey hand, J. Neurophysiol., 35: 122-136. Rose, J.E. (1980) Neural correlates of some psychoacoustic experiences, In D. McFadden (Ed.), Neural Mechanisms in Behavior, Springer-Verlag, New York. Talbot, W.H., Darian-Smith, I., Kornhuber, H.H. and Mountcastle, V.B. (1968) The sense of flutter-vibration~ comparison of the human capacity with response patterns of mechanoreceptive afferents from the monkey hand, J. Neurophysiol., 31:301-334.
47
© 1991 Elsevier Science Publishers B.V. 0165-0270/91/$03.50
NSM 01234
A method for the generation of complex vibrotactile stimuli J.W. M o r l e y , E.N. C r a w f o r d , D . G . F e r r i n g t o n , J.S. A r c h e r , A.B. T u r m a n a n d M.J. R o w e School of Physiology and Pharmacology, University of New South Wales, Sydney, NSW 2033 (Australia) (Received 3 January 1991) (Revised version received 19 February 1991) (Accepted 20 February 1991)
Key words: Complex waveform; Vibrotactile stimulus; Beat A method is described utilizing computer-generated sine wave data and purpose-built hardware to generate a complex vibrotactile stimulus. Two sine waves of different frequency were s u m m e d to produce a complex waveform with two temporal components, a high frequency component and a low frequency beat component. T h e computer-generated data points for each of the two c o m p o n e n t sine waves were downloaded to two banks of static memory in a dual synchronous arbitrary function generator. The data points in memory were fed to two 12-bit digital-to-analogue converters which sent the two analogue sine wave signals to a s u m m i n g amplifier where the two sine waves were added. This method provides a complex waveform that can be gated on and off, has a fixed frequency ratio of the c o m p o n e n t sine waves and no phase drift between the c o m p o n e n t waves. Addition of the separate sine waves in a s u m m i n g amplifier allows for easy alteration of the amplitude ratio of the sine waves. The output of the s u m m i n g amplifier is sent to a feedback controlled mechanical stimulator, thereby allowing the stimulus to be presented to the skin of h u m a n subjects and experimental animals.
Introduction
Most studies on tactile responsiveness of cutaneous sensory fibers in both human subjects and experimental animals have been confined to the use of step stimuli or simple sinusoidal vibration of the skin (Talbot et al., 1968; J~inig et al., 1968; Mountcastle et al., 1972; Ferrington and Rowe, 1980; Johansson et al., 1982; Ferrington et al., 1984). We have recently carried out a series of experiments using a vibratory stimulus that is more complex than a simple sinusoidal stimulus and mimics to some extent the types of vibratory disturbances that may occur in the skin in the
Correspondence: Dr. J.W. Morley, School of Physiology and Pharmacology, University of New South Wales, P.O. Box 1, Kensington, NSW 2033, Australia. Phone: (02) 697 2549; Fax: (02) 313 6043.
tactile exploration of simple textures (Morley and Goodwin, 1987). The complex vibrotactile stimulus consists of the sum of two sine waves of different frequency, resulting in a vibratory stimulus with two temporal components. The stimulus resembles the type of waveform employed in studies of the neural coding of combination tones in the auditory system (Brugge et al., 1969; Rose, 1980). The waveform in the auditory studies was generated by combining the output of a sine wave oscillator and a function generator. For our experiments, generation of the complex waveform needed to satisfy three criteria. First, the waveform needed to be gated on and off; second, the frequency ratio of each of the component sine waves needed to be accurately fixed; and third, phase drift between the component sine waves had to be avoided. In this paper we present an account of the
4~
method we have developed to generate a complex vibrotactile stimulus, satisfying each of the above criteria. We have employed the complex stimulus
to investigate: (1) the ability of' human subjects to detect and discriminate the separate temporal components of the stimulus, and (2) the neural
8 J
/
Data In END-COUNT REGISTER
Variable Frequency
Out Count Enable
z
COMMAND
IN
8
/
o
[ !
Latch Select Address
I
CLOCK
IN
from Function
I 14
Write
E-, Z O t)
CK
Generator
In 1
l-~
ADDRESS
1o
COMPARATOR
!
I
Count Control Losic
ln~
O
_
GATE IN
8
L__q~
Data l~
CK .__.~MEMORY ADDRESS
I
Trl-State I BUFFER
COUNTER I,
/--
out
I/4
/~ 4
14 /
Tri-State BUFFER
¢4 ...J O DATA
IN
~ (
Tri-State , BUFFER I '
i~ " \]
ew
1
Analogue OUT 2
Tri-State BUFFER Analogue
o
e~ e~ BUFFER
"
OUT 1
k
"
\ Fig. 1. Simplified block diagram of the dual synchronous arbitrary function generator. The data points for each component sine wave, generated on an Apple lie computer, were passed on separate channels to two banks of static memory. The memory address counter cycled through the data points, which were then sent to two 12 bit DACs. W h e n the last data point was reached the memory address counter was reset to zero and the data cycle repeated. The analogue signals from the two D A C s were passed to a s u m m i n g amplifier (not shown) where the two signals were added.
4 t)
representation of each temporal component of the stimulus in the activity of single mechanoreceptive fibers and central neurons in the anesthetised cat (Ferrington et al., 1987, 1988; Morley et a1.,1989).
Complex waveform The complex waveform was generated by electronically summing two sine waves of different frequency. The waveform has a frequency that is the average of the two component frequencies and an amplitude that varies with time (Halliday and Resnick, 1981). In the case of an equivalent sound stimulus this amplitude variation gives rise to an auditory sensation termed a beat. It was essential that the two waveforms were accurately synchronized and able to be gated on and off. To satisfy these criteria we used a combination of computer generation of the component sine wave data and purpose-built hardware, which we have termed a dual synchronous arbitrary function generator. To generate acceptable sine waves on the computer we used between 20 and 50 data points per sine wave. Each component wave was downloaded on separate output channels to the hardware unit that contained memory, two digital-to-analogue converters (DACs) and logic. In our experiments we used vibration frequencies up to a maximum of about 2000 Hz and therefore required an upper sampling limit of 100 kHz per channel. The hardware logic allowed data sampling rates of greater than 1 MHz. Fig. 1 is a simplified block diagram of the dual synchronous arbitrary function generator. The high data sampling rates of the dual synchronous arbitrary function generator allowed us to utilize an inexpensive computer (Apple lie) to generate the sine wave data. Software to generate the sine waves was written in Apple II Pascal and machine language. The software was menu-driven and the sine wave p a r a m e t e r s could be saved on disk, thereby allowing rapid retrieval and downloading of data when the frequency p a r a m e t e r s of the component sine waves were changed during experiments. Full details of both hardware and software are available from the authors.
The two frequencies of the component sine waves had a ratio N 1 : N 2, where N l and N, are integers. It follows that N 1 cycles of one wave and N~ cycles of the second wave take the same time (T). The time T is the period of the complex wave (beat cycle) and is the period for which the data are generated. If both waves begin at zero (time = 0), data are generated from time = 0 to the last sample before time = T, when the wholc cycle (and data) is repeated. Data points for this period for each wave were fed to two banks of memory (one for each wave, each 16 K × 12 bits). A memory address counter cycles through the memory points and feeds the successive points to the two 12 bit DACs. A separate address limiting register is loaded with the address of the final data point; at this value the memory, address is reset to zero, and the cycle repeats. The data rate for feeding the DACs is controlled by a variable frequency source (standard function generator). The software calculates the required data transfer rate and indicates the setting to be used. The waveform is gated on and off by providing a trigger signal to the computer on a binary input. When the system is gated ON the above cycle commences. When it is gated OFF, it continues running until the end of the current run through the data and back to address zero (i.e., the first point where both waves are zero). This ensures that both waves always start at precisely the same phase. The two sine waves are brought out separately from each D A C and are added in a summing amplifier. The amplifier has separate potentiometers to allow the amplitude ratio of each wave to be varied. The summed complex waveform is sent to a feedback controlled mechanical stimulator, allowing the stimulus to be presented to the skin of human subjects in psychophysical experiments and the anesthetised cat in electrophysiological experiments.
Discussion The combination of computer-generated waveform data and a purpose-built hardware unit allowed us to produce a complex vibrotactile stimu-
50
lus which satisfied the three requirements specified in the Introduction. A fixed ratio of the sine wave frequencies was essential to avoid irregular wave patterns; for example, if a one second stimulus at a frequency of 1 kHz has an error in frequency of only 0.1%, this will result in a 360 ° phase shift (one complete cycle) during the one second stimulus period. There are a number of ways in which a similar waveform to the one we employed could be generated. One such way is to have two digitally controlled function generators gated together and sum the outputs. This method is not suitable however, as even very small errors in the frequency accuracy (see above) result in irregular waveforms. Alternatively, two frequencies could be accurately locked by combining frequency multipliers a n d / o r dividers and phase-locked loops. However, such systems are difficult to gate and have a tendency to take some time to settle down after being gated on. In this method the oscillator could be run continuously and only the output gated on and off, however the phase of the two waves at gate O N would be random. The computer synthesis technique is the most effective means of satisfying all our requirements. We chose to download the two separate sine waves to separate D A C channels and add them in a summing amplifier. Alternatively, the two waves could be summed in the computer and the complex waveform fed out through a single DAC. This would be a simpler method if the frequencies, and the ratio of the frequencies, of the component sine waves were varied. However, if this method were to be used, the waveform would need to be recalculated each time a change in the amplitude ratio was required, and so the method would be time consuming and less convenient.
Acknowledgements This work was supported by the National Health and Medical Research Council of Australia and the Australian Research Council.
References Brugge, J.F., Anderson, D.J., Hind, J.E. and Rose, J.E. (1969) Time structure of discharges in single auditory nerve fibers of the squirrel monkey in response to complex periodic sounds, J. Neurophysiol., 32: 386-401. Ferrington, D.G. and Rowe, M.J. (1980) Functional capacities of tactile afferent fibres in neonatal kittens. J. Physiol. (Lond.), 307: 335-353. Ferrington, D.G., Hora, M.O.H. and Rowe, M~J. (1984) Development of coding capacities in tactile afferent fibers of the kitten, J. Neurophysiol., 52: 74-85. Ferrington, D.G., Archer, J.S. and Rowe. M.J. (1987) Impulse patterning in neural responses to complex vibrotactile stimuli, Proc. Aust. Physiol. Pharmacol. Soc., 19: 104P. Ferrington, D.G., Archer, J.S. and Rowe. M.J. (1988) Coding of complex vibrotactile stimuli in the impulse patterns of afferent fibres, Proc. Aust. Physiol. Pharmacol. Soc. 19: 196P, Halliday, D. and Resnick, R. (1981) Fundamentals of Physics, 2nd edn., Wiley and Sons, New York. Johansson, R.S., Landstr6m, V. and Lundstr6m, R. (1982) Responses of mechanoreceptive afferent units in the glabrous skin of the human hand to sinusoidal skin displacements, Brain Res., 244: 17-25. J~inig, W., Schmidt, R.F. and Zimmermann, M. (1968) Single unit responses and the total afferent outflow from the cat's foot pad upon mechanical stimulation, Exp. Brain Res., 6: 100-115. Morley, J.W. and Goodwin, A.W. (1987) Sinusoidal movement of a grating across the monkey's fingerpad: temporal pattern of afferent fiber responses, J. Neurosci., 7: 2181-2191. Morley, J.W., Ferrington, D.G., Rowe, M.J. and Turman, A.B. (1990) Representation of complex vibrotactile stimuli in the responses of neurones in the cuneate nueleus of the cat, Neurosci. Lett., 34: S125. Mountcastle, V.B., LaMotte, R.H. and Carli, G. (1972) Detection thresholds for stimuli in humans and monkeys: comparison with threshold events in mechanoreceptive afferent nerve fibres innervating the monkey hand, J. Neurophysiol., 35: 122-136. Rose, J.E. (1980) Neural correlates of some psychoacoustic experiences, In D. McFadden (Ed.), Neural Mechanisms in Behavior, Springer-Verlag, New York. Talbot, W.H., Darian-Smith, I., Kornhuber, H.H. and Mountcastle, V.B. (1968) The sense of flutter-vibration~ comparison of the human capacity with response patterns of mechanoreceptive afferents from the monkey hand, J. Neurophysiol., 31:301-334.