G Model
ARTICLE IN PRESS
NSM 7208 1–9
Journal of Neuroscience Methods xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Journal of Neuroscience Methods journal homepage: www.elsevier.com/locate/jneumeth
Clinical Neuroscience
1
A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system
2
3
4
Q1
Andreas Bahmer a,∗ , Uwe Baumann b a
5
b
6
University of Würzburg, Clinic for Otolaryngology, Comprehensive Hearing Center, 97080 Würzburg, Germany University of Frankfurt Main, Clinic for Otolaryngology, Audiological Acoustics, 60590 Frankfurt, Germany
7
a r t i c l e
8 23
i n f o
a b s t r a c t
9
Article history: Received 15 September 2014 Received in revised form 2 March 2015 Accepted 2 April 2015 Available online xxx
10 11 12 13 14 15
22
Keywords: ASSR Jitter Psychophysic Electrophysiology Temporal coding SAM
24
1. Introduction
16 17 18 19 20 21
25Q4 26 27 28 29 30 31 32 33 34
Q2
Background: Simultaneous assessment of psychometric tasks and electrophysiological recordings is chal- Q3 lenging because each requires specific technical and physiological preconditions. Electrophysiological recordings require a comparatively long test time duration to gain sufficient signal-to-noise ratios, whereas test duration of psychometric measurements should be limited to prevent challenges to the attention of the subject. In order to investigate immediate correlation between both measurements a method is described, which combines electrophysiological and psychometrical measurements in a single test procedure. The test may be applied to subjects with deficits in temporal resolution (e.g. auditory neuropathy spectrum disorder, ANSD). New method: Auditory steady state responses (ASSR) and a pitch discrimination task were combined in a single procedure. The setup employed two short-time ASSR sub-stimuli with different fixed modulation frequencies but same carrier frequencies (signal 1 and 2). Simultaneously to the recording of ASSR, the test subject had to determine the signal interval which generated the perception of higher pitch. Results: The developed setup was successfully tested by means of an artificial EEG signal and in one human subject. ASSR signal as well as pitch discrimination performance. Comparison with existing methods: To our knowledge the presented method has not yet been described elsewhere. Conclusions: The feasibility of a setup to simultaneously perform a pitch discrimination task and electrophysiological measurements was demonstrated for the first time. The method provides the facility to apply sinusoidal amplitude modulated stimuli (SAM) with jittered modulation period lengths. © 2015 Published by Elsevier B.V.
The aim of simultaneous measurement of electrophysiological and a pitch discrimination task is to investigate the potential correlation between both entities. This aim may be achieved with sequential measurement as well, nevertheless a simultaneous measurement may contain a higher potential to reveal such a correlation. Provided there exists a high correlation between both, it is conceivable that complex and attention-dependent behavioral tasks might be substituted with objective electrophysiologic measurements (Fig. 1). The more peripheral the neural generators of the electrophysiological responses, the less neural processing takes
∗ Corresponding author. Tel.: +49 931 201 21779; fax: +49 3221 1348478. E-mail addresses: Bahmer [email protected] (A. Bahmer), [email protected] (U. Baumann).
place, and it is more likely that the electrophysiological measure reflects a basic step in the ascending auditory system for temporal processing and estimating pitch (Gockel et al., 2011; Krishnan and Plack, 2011; Greenberg et al., 1987). 1.1. Auditory steady state responses In auditory brainstem response (ABR) recordings transient acoustic stimuli are used which trigger neural responses in different parts of the auditory system (Jewett and Williston, 1971; Brown et al., 1994; Bahmer et al., 2008). The amplitude and latency of neural response peaks are evaluated for diagnostic purposes. The valuation of ASSRs is performed in the complex frequency domain in contrast to time domain based evaluation of ABRs (Lins and Picton, 1995). To generate ASSRs, continuous stimuli are presented, leading to summate periodic neuronal responses. Amplitude-modulated (AM) pure tones may be used as
http://dx.doi.org/10.1016/j.jneumeth.2015.04.003 0165-0270/© 2015 Published by Elsevier B.V.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
35 36 37 38
39
40 41 42 43 44 45 46 47 48 49
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
2
psychophysic
EEG ASSR signal strength
correct answers
decrasing stimulus regularity
task difficulty
Fig. 1. Both psychophysical performance and auditory steady state response (ASSR) signal strength decrease similarly with increasing task difficulty and corresponding decreasing stimulus regularity.
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
stimuli. Neural generators, related to the depth and frequency of the amplitude modulation, are triggered. A continuous pure tone with amplitude modulated envelope between 2 and 400 Hz evokes a steady-state response at the modulation frequency (Chambers et al., 1986; Kuwada et al., 1986; Rees et al., 1986; Lins et al., 1996; Picton et al., 1987). It is commonly reported that modulation rates between 75 and 110 Hz are best suited to assess hearing thresholds. At these modulation rates, the ASSR is not significantly influenced by sleep (Cohen et al., 1991; Lins et al., 1995; Mühler et al., 2014), and, in normal hearing adults, responses to a 1-kHz stimulus can be recorded down to a stimulus level of 26 dB SPL (Lins et al., 1996). Evoked responses derived with continuously presented AM stimuli are analyzed in terms of both spectral energy and phase at the modulation frequency and its integer multiples (Chambers et al., 1986; Kuwada et al., 1986; Picton et al., 1987; Bahmer and Baumann, 2010). At modulation rates between 75 and 110 Hz, there is little change in the amplitude of the responses when multiple stimuli are presented compared to single stimulus presentation and carrier frequencies differ by one octave (Lins and Picton, 1995). In order to save recoding time, an alternative to a single stimulus application is the simultaneous measurement of responses to several AM stimuli (named MASTER stimulus) with each stimulus with different modulation and carrier frequency (Regan and Cartwright, 1970; John and Picton, 2000; John et al., 1998; Lins and Picton, 1995; van Dun et al., 2008). Statistical procedures determine whether a response is significant by comparing the response level and/or the phase at the particular modulation frequency to the noise level and/or phase at adjacent frequencies. The reliability of repeated measurements can also be assessed by comparing the statistical analyses. In order to restrict AEP to a predefined tonotopic region, masking noise with a certain spectral gap (notched noise) has to be presented in addition to transient click stimuli. ASSR-stimuli do not need the application of masking since the spectral power of carrier and side bands is restricted to a narrow frequency region (Lins et al., 1996). ASSRs as well as tone pip evoked ABR allow the assessment of functional deficits in different frequency regions of the cochlea. By using frequency-specific stimuli, different frequency regions are activated and can be evaluated separately. In order to develop new AEP recording procedures or to test the impact of alternative stimuli, affordable research platforms have been introduced using LabView or C/C++ (John and Picton, 2000; van Dun et al., 2008). The method presented here is implemented in Matlab 7.6.0.324 R2008a (The MathWorks Inc., Natick, Massachusetts) interfaces with other hardware components. The software presented here (see Appendix for excerpts) differs from so far introduced other research platforms since ASSR recording and a pitch discrimination task are running in parallel. Commercially available platforms offer only stimuli restricted to clinical routine tasks. In addition, access to raw EEG or analysis data is not available which make research applications and evaluations difficult.
1.2. Pitch discrimination tests Frequently, pitch discrimination tests are designed as alternative forced choice experiments. Several signal alternatives are presented consecutively, and the listener has to detect the deviant interval, or the interval perceived higher or lower in pitch. In addition, adaptive procedures are often employed, which reduce or increase the frequency difference between standard and deviant interval depending on reliability of subjects’ decision (example application described in Bahmer and Baumann, 2013). At a certain frequency difference subjects will asymptotically reach a minimum frequency difference threshold which defines the “just noticeable difference” (JND). AM stimuli may be applied in pitch discrimination test (Lee, 1994). In addition, randomized alteration (jitter) of the frequency of the modulator can serve as a stimulus parameter that potentially influences both pitch discrimination performance as well as the signal strength of auditory steady state responses.
1.3. Tested hypotheses In the present report, it is tested whether the simultaneous measurement of electrophysiological responses in an ongoing psychophysical task is feasible. Specifically, it was tested whether: • signal presentation and simultaneous recording via external audio-interface (RME ADI-8 DS) is free of distortions, • the customized analysis procedures (Matlab routines) of the recorded signals operates successfully, • ASSR generated by stimuli consisting of sinusoidal amplitude modulated signals (SAMs) modulated with either 200 or 250 Hz are detectable in simultaneous a pitch discrimination experiment, • concatenated EEG recording windows as short as 2 s are sufficient to detect auditory steady state potentials, • continuous stimuli containing jittered modulation cycles can be generated and integrated in the setup.
2. Experiment 1: evaluation of simultaneous pitch discrimination and ASSR 2.1. Rationale The pitch discrimination experimental setup described here (see Section 2.2) applies a two alternative forced choice procedure with fixed stimuli (see Fig. 3). A fixed stimulus design was necessary because auditory steady state responses are recorded in parallel. A certain amount of recording time is necessary to detect the stimulus generated response signal in the EEG.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
101
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
117
118 119 120
121 122 123 124 125 126 127 128 129 130 131 132
133 134
135
136 137 138 139 140 141
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
PC
RME sound card input
output
touchscreen Which tone is higher?
biosignalamplifier
tone 1 headphones
tone 2
electrodes
Fig. 2. Hardware setup for ASSR recording and simultaneous assessment of pitch discrimination task. A standard PC with a Windows XP and Matlab installation is connected to an RME external sound card (ADI-8 DS). The external sound card receives input from a biosignal amplifier (one-channel self-made amplifier with analog medical grade isolation amplifier IA 297, Intronics, USA) and sends stimuli to the headphones (Sennheiser HDA 200). While recording ASSR, the subject performs a pitch discrimination task by pressing buttons on a touchscreen.
142
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
2.2. Method 2.2.1. Hard- and software The hardware setup consisted of a standard PC with a Windows XP and Matlab installation connected to an RME external sound card (ADI-8 DS). The external sound card received input from a single channel biosignal amplifier and sent stimuli to the headphones (HDA 200, Sennheiser, Wedemark, Germany) (Fig. 2). The self-made biosignal amplifier was provided by an analog medical grade isolation amplifier IA 297 (Intronics, USA) with one channel, and together with an analog display enabled to control a sufficient impedance range of surface EEG electrodes. The amplifier provided full patient protection from leakage currents and amplifier fault currents by optical isolation. Technical data are explained elsewhere (see (Bahmer et al., 2008)). The RME external sound card was chosen because it is inexpensive and provides a 24 bit dynamic range. The sound card is equipped with 8 input and 8 output channels that can be used simultaneously. EEG data was recorded on channel 1 of the sound card; the internally looped-back stimulus was recorded on channel 2 as a control. The latency between the beginning of the recording and the onset of the recorded loopedback stimulus signal was a constant 12.4 ms (548 points at 44.1 kHz sampling, measured with 10 repetitions). The EEG analysis using a sliding slit window was available via the ‘buffer’ function in Matlab (buffer length was set to the epoch length of 6 s consisting of tone 1, tone 2, and pause). The sliding was employed by a continuous shift of the EEG vector by one sampling point and subsequent feeding into the buffer. The sliding windows enable to find the exact time where the analysis windows perfectly match the time span of stimulus 1 and stimulus 2, respectively. The exact beginning of the EEG parts containing the neural response to the stimuli is unclear due to unknown inherent latencies. In experiment one, the functionality of the sliding window analysis was tested with stimulus 1 defined as an SAM signal with a carrier frequency of 2 kHz and a modulation frequency of 200 Hz; stimulus signal 2 was set to zero. The duration of each of stimulus 1, stimulus 2, and the pause was set to 2 s; altogether the epoch lasted 6 s. The amplitude of the sideband at 1800 Hz (carrier frequency minus modulation frequency) was plotted against the shift (increment 1000 sampling points, about 23 ms) of the sliding slit windows along the recorded input in Fig. 5. To enable temporal exact stimulus presentation and a synchronously EEG recording, the utilty ’playrec’ (Humphrey, 2008) was used (Matlab MEX file, Toolbox not necessary), which integrates ’PortAudio’, a free, open-source audio I/O library. ’PortAudio’ can access the AD/DA converters via different host application
3
programming interfaces (API) on different platforms (including ASIO, Audio Stream Input/Output protocol from Steinberg, Media Technologies, Hamburg; WMME, Windows Multimedia extension, and DirectSound under Windows, both Microsoft, Redmond). In the present study ASIO/Windows was used. Playrec provide control about the amount of skipped samples if, for example, the computer is slowed down by tasks running in parallel. The present recording system was developed because commercially available ASSR recording systems (e.g. Eclipse, Interacoustics, Assens, Denmark; Otometrics, ICS Chartr EP200, Taastrup, Denmark; Sentiero, Path Medical Solutions, Germering, Germany; Biologic Navigator Pro (MASTERII), Natus Medical Inc., Pleasanton, USA; GSI Audera, MN, USA) do not have the capability of presenting freely designed stimuli (e.g. with jitter) which is important for our research purposes. 2.2.2. Stimuli The stimuli consisted of sinusoidal amplitude modulated signals (SAM) with carrier fc , modulation frequency fm , and a subsequent pause. A SAM is described mathematically by SAM = sin(2fc t) · (1 + m · cos(2fm t))
(1)
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
201 202 203 204 205 206
1 SAM = sin(2fc t) + m · sin(2(fc + fm )t) 2 +
1 m · sin(2(fc − fm )t), 2
207
(2)
208 209
where fm and fc are the modulation and the carrier frequency, respectively. The frequency of the sidebands are denoted by fc ± fm , the modulation depth by m. The later described experiment presented two SAM tones with carrier frequency fc = 2 kHz and modulation frequencies fm1 = 200 Hz and fm2 = 250 Hz. The difference in terms of modulation frequency is perceived as a pitch difference. 2.2.3. Psychophysical test: pitch discrimination task The pitch discrimination task was a two alternative-forcedchoice experiment (Fig. 3). Two SAM stimuli with the same carrier frequency but different modulation frequencies were subsequently presented, which results in different pitches provided a sufficient frequency difference for the subject. Each SAM stimulus lasted 2 s, which was long enough to elicit an ASSR signal and a pitch percept, and not too long to exceed typical stimulus duration time in a pitch discrimination task. After the second stimulus was presented, a pause of 2 s followed. From the onset of the first stimulus until 80% of the block length (4.8 s; 6 s block with stimulus 1, stimulus 2, and pause) the subject could make a decision by pressing a button to indicate the interval which was perceived with higher pitch. After the pause, the next block of stimuli and pause followed independent of the subject’s decision. Stimuli were presented in a random order, which is considered in the analyses. The two SAM stimuli had a carrier frequency of 2 kHz, and modulation frequencies of 200 and 250 Hz, respectively. After finishing the test, the percentage of correct pitch difference identifications was calculated. The relatively high stimulus temporal pitch difference (fm1 − fm2 = 50 Hz) was chosen in order to test whether the psychoacoustic procedure was performable. In order to prove the described method, the sound pressure level of the stimulus was set to 84 dB SPL which generates a large spread of excitation on the basilar membrane. 2.3. Results 2.3.1. Phase preservation test of the setup using artificial signals The SAM stimulus was recorded to test whether the setup preserves the phase when recording and analysis is performed. The
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
210 211 212 213 214 215 216
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
241
242 243 244
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
4
Which tone is higher?
psychophysics
signal 1
signal 2
pause
signal 1
pause
signal 2
push button enabled
entry disabled, buttons released
EEG
1. mean 2.|FFT| @ fm1 or fm2
... ASSR signal strength shift of analysis windows along EEG
Fig. 3. Top: scheme of psychophysical experiment while an EEG is continuously recorded. Stimulus 1, stimulus 2, and the pause are continuously presented (duration 2 s each, stimulus sequence presented randomly). At 0.8 times of the period length (duration of two tones plus pause, about 6 s), the subject’s input is disabled and the decision is fixed. Afterward, both response buttons are released. Bottom: analysis of the recorded EEG. Concatenated sliding windows (only two/three windows are shown in one row, lengths corresponding to length of stimulus 1 or stimulus 2) are shifted along the recorded EEG (shift examples indicated by the three rows). The mean of the EEG data whose length correspond to the length of the sliding windows and subsequently the FFT are calculated. The FFT amplitude at frequency channel fm1 or fm2 of the mean (ASSR signal strength) is plotted against the shift of the sliding window. Periodic maxima are detected whenever the corresponding sections span the maximal neural response to stimulus 1 or stimulus 2. The ASSR signal strength can be calculated without knowledge of the absolute stimulus timing.
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278
SAM stimulus was generated by the software and sent to the AD/DA converter. The signal was looped-back to the input of the converter (direct hardware loop, feature of the RME sound card). The latency of the looped-back recorded signal was measured ten times in order to detect any jitter in the latency which may cause phase shifts between recorded signals. ASSR is sensitive to phase variations depending on analyses techniques (e.g. phase and averaged signal analyses, see (Cebulla et al., 2006)). Operating systems and software modules often cause uncontrolled delays. The looped-back recorded signal shows a constant latency of 548 sampling points at a sampling rate of 44.1 kHz from the beginning of the recording to the beginning of the recorded test signal. The calculated average of 70 recorded stimulus 1 signals (10 tests each containing 7 sliding windows, offset is 0, position of stimulus 1 directly after the pause in only 7 windows in all 10 tests) shows that the stimulus is exactly reproduced without any phase distortions (Fig. 4). Phase distortions would have resulted in destructive and constructive interference in an averaging process, and, consequently, amplitude fluctuations. In addition, the sliding slit window analysis was evaluated with test matrices (containing numbers zero, one, and two) and showed that the sorting of the recorded EEG snippets, which is mandatory due to the random order of stimulus presentation, works properly. In order to detect a (neural) signature as a response to the stimuli in the recorded data, an analysis of the recorded input by shifting sliding slit windows was performed (see Fig. 3, bottom). The software implementation was tested for this functionality by using the previously described looped-back stimulus recording (see Section 2.2). Fig. 5 shows, as expected, that the FFT amplitude of the sideband (fc − fm ) is maximal at shifts around 0, 6, and 12 s (plus latency shift of 548 sampling points due to looped-back signal, hidden by the figure scaling), which are multiples of the 6 s epoch length. The pattern repeats with a period of 6 s. The decrease from 0 to 2 s results from the fact that the buffer (i.e. the averaged windows) containing the full signal is incrementally substituted by
1
amplitude [a.u.]
245
0
−1
0
500
1000
1500
2000
time [sampling points] Fig. 4. Average of 70 recorded stimulus 1 signals (SAM, fc = 2000 Hz, fm = 200 Hz, 10 test runs each containing 7 sliding windows without shift or sliding, see Fig. 3; position of stimulus 1 directly after the pause in only 7 sliding windows in all tests). The test signal waveform is exactly reproduced which indicates perfect preservation. The latency of 548 sampling points (12 ms, sampling 44.1 kHz) is caused by loopback of the stimulus to the input of the sound card (direct internally hardware loop, feature of the RME sound card, see Section 2.2).
zeros (stimulus 2 is set to zero). At a shift of seconds the buffer is filled with zeros. As the analysis window set continues to slide along the recorded part containing the zero stimulus from the pause, the buffer remains at zero until the shift of 4 s. In the consecutive shift from 4 s until 6 s the buffer is gradually filled again with stimulus
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
279 280 281 282 283
G Model NSM 7208 1–9
ARTICLE IN PRESS A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
5
difference between the maximum and minimum FFT amplitude (correlated to stimulus on and stimulus off) indicates ASSR signal strength. For detecting obscure regularities in the EEG analysis an autocorrelation may be advantageous. 3. Experiment 2: Evaluation of jittered modulation cycle concatenation
284 285
286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323
1, resulting in an increase of the FFT amplitude from zero to one. A repetition of the described process occurs from 6 to 12 s and so on. 2.3.2. Timing of signal recording, stimulus playback, and touchscreen interface In 50 test sessions, signal recording, stimulus playback, and operation of the touchscreen interface were evaluated. The Matlab software package ’playrec’ (see Section 2.2 and Humphrey (2008)) enabled full duplex signal presentation and recording. In cases where the operation system scheduler prioritized parallel tasks that require a high CPU load, frames of samples were sometimes skipped. It was also observed that the movement of Matlab figures (i.e. the touchscreen interface) by the user frequently generated a loss of sample frames. During execution of the experiment, figures normally do not need to be moved, only interface buttons have to be pressed which does not result in a heavy processing load a subsequent sample skipping. In order to avoid that samples are skipped (e.g. with high CPU load), it is possible to enlarge the buffer length of recording and playback. The upper limit of acceptable software buffer enlargement which can be manipulated in the software is reached when the user interface shows a low sychronization or delay compared to the stimuli and recording. In our tests, the timing of button activation/inactivation was perfectly synchronized to the time at 0.8 of the epoch length (Fig. 3 top). 2.3.3. Test of the setup with human EEG recording After verification of the experimental setup (see Fig. 5), author A.B. served as a normal hearing subject (age 38). Stimuli were set as outlined in Section 2.2. The subject scored 100% percent correct in the pitch discrimination task. Fig. 6 shows the results of the subsequent EEG analysis. In both spectral analyses (stimulus 1 left, stimulus 2 right; FFT amplitude at 200 and 250 Hz, respectively) a distinct periodicity of about 6 s (stimulus repetition period) is detectable. For stimulus 1, maxima are at about 4 and 10 s (difference: 6 s), for stimulus 2 maxima are at about 1.5, 7.5, and 13.5 s (all differences: 6 s) after stimulus onset. The temporal regularities of 6 s found in the FFT analyses of the shifted EEG window for both SAM stimuli (Fig. 6 left and right) reflect the duration of stimulus 1, stimulus 2, and the pause (total 6 sec). The modulation frequency of the SAM stimuli only determines the FFT channel in which this regularity can be found by means of shifted time windows. The
327
330
3.2. Method
331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347
348
The modulation frequency may be jittered in order to disturb salient temporal information of pitch. The jitter is applied to the duration of an entire modulation cycle to ensure transmission of at least one cycle containing periodicity information (see Fig. 7 and Eqs. (3) and (4)). SAM = sin(2fc t) · (1 + m · cos(2fm (t)t)) fm (t) =
326
329
Temporal regularity, serving as pitch information, may be distorted by adding jitter. Consequently, the saliency of pitch is lowered and pitch discrimination ability decreases. On the other hand, AEP measures such as auditory steady state responses (ASSR) rely on the temporal regularity of an incoming signal. Pitch discrimination ability and ASSR responses are both negatively effected by the distortion generated by e.g. jitter of temporal regularity. If both entities are measured simultaneously, it is hypothesized that an increase of jitter will affect measurement results in both domains. For instance, in pitch discrimination tasks when two sinusoidal amplitude modulated signals have to be distinguished, jitter applied to the modulation frequency leads to a broadening of spectral energy at side bands and consecutive degradation of pitch salience until pitch discrimination completely vanishes (overlay of sidebands). An increase in modulation frequency jitter leads simultaneously to decreased spectral response amplitudes in ASSR modulation frequency.
325
328
3.1. Rationale
Fig. 5. FFT amplitude at sideband 1800 Hz of averaged recorded SAM signals (stimulus 1: SAM with fc = 2000 Hz and fm = 200 Hz at fixed first position, stimulus 2 set to zero, see Fig. 4) against the shift of the sliding windows along the recording (see Figs. 3 and 4). If the windows exactly span the EEG parts containing the SAM stimulus, the FFT amplitude of the sideband is maximal; if the windows are completely out of the range of the stimulus, the amplitude is zero. Only 8 sliding slit windows of 10 possible windows are averaged to avoid edge effects like buffer underrun when the window shift is larger than the corresponding EEG recording.
324
round|N(fm , )| : fm (t − 1) :
(1 + m · cos(2fm (t)t)) = 0
349 350 351 352 353
(3)
354
(4)
355
else
with fc : carrier frequency, fm (t): jittered modulation frequency, N: Gaussian distribution, fm : base modulation frequency, and : standard deviation (amount of jitter). Complete modulation cycles with jittered periods are concatenated (see Fig. 8). The challenge of concatenating continuously varying periods is that given a certain sampling frequency, only multiple integers of the sampling period can represent complete modulation cycles. If the length of the modulation cycles is not integer multiples of the sampling period, the overhang has to be transferred to the following modulation cycle. Otherwise, the signal is not uniformly continuous resulting in discontinuities. The modulation frequency differences are jittered within a certain probability distribution function (here Gaussian distribution). 3.3. Results It was tested whether the generated concatenated SAM is equal to a SAM generated without concatenation. For the SAM without concatenation a time vector (2 s) was generated and a concatenated SAM (2 s, fc 2 kHz, fm 200 Hz) according to Eq. (1) was calculated. There was no difference between the concatenated SAM and the non-concatenated SAM (both same fc and fm, no jitter). Additional tests with different fm and/or sampling rate gave no evidence for signal disruptions. In addition, a frequency analysis of the SAM with concatenated modulation cycles showed a peak only at the desired
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
356 357 358 359 360 361 362 363 364 365 366 367 368
369
370 371 372 373 374 375 376 377 378
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
6
Q5 Fig. 6. Analysis of test subject’s A.B. EEG (analysis procedure described in Fig. 5, 10 repetitions, stimulus 1: SAM with fm1 = 200 Hz, stimulus 2: SAM with fm2 = 250 Hz, both fc = 2000 Hz; duration stimulus 1, stimulus 2, pause: 2 s). The FFT maxima at the modulation frequencies are at multiples of the stimulus repetition period (6 s). Left: FFT of sliding analysis windows at fm1 = 200 Hz, right: FFT of sliding analysis windows at fm2 = 250 Hz. The increment of the shift is 1000 sampling points due to economy of time (green line: interpolation). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
SAM (fc, fm)
jittered SAM (fc, fm+ε)
|FFT|
|FFT|
fm fc-fm fc+fm
fc-(fm+ε) fc+(fm+ε)
fc
fc
fc
Fig. 7. SAM and jittered SAM with carrier frequency fc and modulation frequency fm in the temporal and frequency domain. The specific characteristic of the jittered SAM is containment of modulation cycles with different frequencies around fm (fm ±). These modulation cycles with different durations are concatenated to each other. This is reflected in the SAM frequency spectrum in a smearing of the sidebands. The jittered temporal pattern of the SAM defined by the jittered modulation (fm ±) is transferred to the auditory nerve at the tonotopic place defined by fc .
381
modulation frequency (Fig. 9 left). If the modulation periods were jittered, the corresponding frequency analysis of the modulation also showed a jitter (Fig. 9 right).
382
4. Discussion
379 380
383 384 385 386
The present report describes a setup which enables recording of ASSR while performing a two alternative forced choice pitch discrimination task. The setup may be augmented to a three or more alternative forced choice experiment. Testing with artificial signals
fm sampled first modulation cycle
showed that the hardware and software preserves the phase of the signal, and that the touchscreen timing is in synchrony with recording and stimulus playback. The developed analysis software is capable of detecting regularities in recorded signals and in a human EEG. In a standard setup, ASSR comprises the analysis of several minutes of EEG. This precondition cannot be used for psychophysical experiments as tones are mostly presented in the range of seconds or below. The challenge in a combination of both measures is to find a solution for the trade-off between preconditions for
fm+ε sampled next modulation cycles
concatenation of sampled modulation cycles
II.
I.
II.
I.
... Δt
Δt
... Δt
Fig. 8. Concatenation of modulation cycles with jittered period lengths (fm ±). If the length of a modulation cycle, e.g. cycle I., is not an integer multiple of the sampling period the overhang (t) has to be transferred to the next concatenated cycle (II.) and so forth. Only complete modulation cycles are concatenated.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
387 388 389 390 391 392 393 394 395 396
G Model NSM 7208 1–9
ARTICLE IN PRESS A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
7
Fig. 9. FFT amplitude of the concatenated SAM (sampling = 44.1 kHz, fc = 2 kHz, fm = 200 Hz, left: no jitter, =0, right: normally distributed jitter with = 10 · fm ( see Eqs. (3) and (4)), resolution of the frequency axis is 0.5 Hz). The concatenation algorithm yields exactly the desired modulation frequency and jittered modulation frequency.
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
both measurements. Electrophysiological recordings require a relatively long time to improve the signal-to-noise ratio, whereas in psychometric task measurements stimulus presentations are temporally limited to ensure continuous attention of the subject. Our setup is capable of combining both measurements. Testing showed that timing of recording, playback, and button control is highly precise (Figs. 4 and 5). A jitter may be applied to the modulation frequency of the SAM (Fig. 9) which potentially decreases temporal periodicity detected by pitch discrimination and/or ASSR. ASSR recording and analyses from averaged concatenated short intervals of EEG (here 2 s) opens a new field in the recording of ASSR. Further studies have to address, how the length of the stimuli, the repetition rate, and the length of silent intervals influence these responses. It is conceivable that mismatch negativity (Näätänen et al., 2007) may be used for the detection of pitch differences in electrophysiological responses. However, in contrast to ASSR, the generators of mismatch negativity are mostly related to higher cortical areas. As distortion of temporal patterns may be detected at a more peripheral site, the ASSR measurement is more directly linked to basic temporal properties. After recording and analysis of an artificial signal showed that the setup works without phase discontinuities, a human EEG was recorded while the subject performed a 2AIFC pitch discrimination test (Fig. 6). A potential application of the procedure is the assessment of pitch discrimination ability by decreasing the stimulus differences. In contrast to the results obtained with artificial EEG response signals (Fig. 5, in the real EEG data the amplitude maxima do not occur at 0, 6, and 12 s. A potential explanation might be, that the neural response is delayed by acoustic latency and neural network response properties. In addition, it can observed that the occurrence of the maxima seem to depend on modulation frequency. It has to be discussed whether a modulation specific responsiveness of the neural network generators is responsible for this effect – a question that has to be addressed in further studies more thoroughly. The maxima and minima in the shifted EEG response evoked by the stimulus differences and pauses (Fig. 6) were not disrupted by the motor system activation required to enter the decision on the touchscreen. Furthermore, despite the short duration of stimuli of 2 s and only 10 repetitions, a clear difference between stimuli is detected in the EEG. A drawback of the described method is the increased duration of the test due to the integration of the psychoacoustic test procedure. The prolongation of test duration is justified by the immediate relation between ASSR response pattern and psycho-acoustically derived measures, which shall be assessed by direct correlation (see Fig. 1). Further tests will
address the effect of stimulus duration, stimulus modulation frequency, stimulus carrier frequency, jitter, repetition rate/number, and sound pressure level on ASSR and psychophysical responses. In addition, statistical procedures may be implemented for detecting thresholds likewise statistical test procedures are used in standard ASSR (Chambers et al., 1986; Kuwada et al., 1986; Picton et al., 1987; Bahmer and Baumann, 2010). In conclusion, this study describes a feasibility test for a setup and procedure which is capable of simultaneously recording detectable ASSR while performing a pitch discrimination task.
5. Conclusions • The setup described is able to simultaneously record an EEG signal with high phase precision, continuously play SAM stimuli, and present, change, and read-out a touchscreen interface without phase discontinuity. • The described procedure reliably detects temporal periodicities in recorded data. • A preliminary experiment proved the detection of ASSR with unusual modulation frequencies (up to 250 Hz) used in a parallel pitch discrimination psychometric task. • Preliminary results demonstrate a concatenated recording time of 20 s (10 × 2 s stimulus interval)to be sufficient to discriminate the difference of temporal modulation induced by two different modulation frequencies in the ASSR. • The present procedure allows to calculate stimulus related ASSR signal strength without knowledge of the absolute stimulus timing, which is important for the analysis and integration of a psychometric test running in parallel. • The setup can be expanded to three or more alternative forced choice discrimination tasks. • The continuous stimulus containing jittered modulation cycles was developed and validated. The influence on both pitch discrimination tasks and ASSR measures will be subject to further studies.
Acknowledgements The work was supported by a grant of MED-EL, Innsbruck, Austria. We would like to acknowledge the contribution made by Rosalyn Tait whose comments helped to improve our English language usage. We would like to thank the two anonymous reviewers for the very helpful comments.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
444 445 446 447 448 449 450 451 452 453
454
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477
478
479 480 481 482 483
G Model NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
8 484
485 486 487
488 489 490
ARTICLE IN PRESS
Appendix. Matlab Code for concatenating and sampling complete modulation cycles with different periods (SAM with jittered modulation)
Matlab Code for sliding slit window analysis of the recorded auditory steady state responses including sorting of EEG snippets due to random stimulus representation
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
G Model NSM 7208 1–9
ARTICLE IN PRESS A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
491
492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
References Bahmer A, Baumann U. Recording and online analysis of auditory steady state responses (ASSR) in Matlab. J Neurosci Methods 2010];187:105–13. Bahmer A, Baumann U. New parallel stimulation strategies revisited: effect of synchronous multi electrode stimulation on rate discrimination in cochlear implant users. Cochlear Implants Int 2013];14:142–9. Bahmer A, Peter O, Baumann U. Recording of electrically evoked auditory brainstem responses (E-ABR) with an integrated stimulus generator in Matlab. J Neurosci Methods 2008];173:306–14. Brown CJ, Abbas PJ, Fryauf-Bertschy H, Kelsay D, Gantz J, Bruce J. Intraoperative and postoperative electrically evoked auditory brain stem responses in nucleus cochlear implant users: implications for the fitting process. Ear Hear 1994];15:168–76. Cebulla M, Stürzebecher E, Elberling C. Objective detection of auditory steady-state responses: comparison of one-sample and q-sample tests. J Am Acad Audiol 2006];17:93–103. Chambers RD, Feth LL, Burns EM. The relation between the human frequency following response and the low pitch of complex tones. J Acoust Soc Am 1986];80:1673–80. Cohen LT, Rickards FW, Clark GM. A comparison of steady-state evoked potentials to modulated tones in awake and sleeping humans. J Acoust Soc Am 1991];90:2467–79. Gockel HE, Carlyon RP, Mehta A, Plack CJ. The frequency following response (ffr) may reflect pitch-bearing information but is not a direct representation of pitch. JARO 2011];12:767–82. Greenberg S, Marsh JT, Brown WS, Smith JC. Neural temporal coding of low pitch. I. Human frequency-following responses to complex tones. Hear Res 1987];25:91–114. Humphrey R. ’playrec’. 2008. http://www.playrec.co.uk Jewett DL, Williston JS. Auditory-evoked far fields averaged from the scalp of humans. Brain 1971];94:681–96. John MS, Lins OG, Boucherand BL, Picton TW. Multiple auditory steadystate responses (MASTER): stimulus and recording parameters. Audiology 1998];37:59–82.
9
John MS, Picton TW. MASTER: a Windows program for recording multiple auditory steady-state responses. Comput Methods Progr Biomed 2000];61:125–50. Krishnan A, Plack CJ. Neural encoding in the human brainstem relevant to the pitch of complex tones. Hear Res 2011];275:110–9. Kuwada S, Batra R, Maher VL. Scalp potentials of normal and hearing-impaired subjects in response to sinusoidally amplitude-modulated tones. Hear Res 1986];21:179–92. Lee J. Amplitude modulation rate discrimination with sinusoidal carriers. J Acoust Soc Am 1994];96:2140–7. Lins OG, Picton PP, Picton TW, Champagne SC, Durieux-Smith A. Auditory steadystate responses to tones amplitude-modulated at 80–110 Hz. J Acoust Soc Am 1995];97:3051–63. Lins OG, Picton TW. Auditory steady-state responses to multiple simultaneous stimuli. Electroencephalogr Clin Neurophysiol 1995];96:420–32. Lins OG, Picton TW, Boucher BL, Durieux-Smith A, Champagne SC, Moran LM, et al. Frequency-specific audiometry using steady-state responses. Ear Hear 1996];17:81–96. Mühler R, Rahne T, Mentzel K, Verhey JL. 40-Hz multiple auditory steady-state responses to narrow-band chirps in sedated and anaesthetized infants. Int J Pediatr Otorhinolaryngol 2014];78:762–8. Näätänen R, Paavilainen P, Rinne T, Alho K. The mismatch negativity (MMN) in basic research of central auditory processing: a review. Clin Neurophysiol 2007];118:2544–90. Picton TW, Skinner CR, Champagne SC, Kellet AJ, Maiste AC. Potentials evoked by the sinusoidal modulation of the amplitude or frequency of a tone. J Acoust Soc Am 1987];82:165–78. Rees A, Green GGR, Kay RH. Steady-state evoked responses to sinusoidally amplitude-modulated sounds recorded in man. Hear Res 1986];23:123–33. Regan D, Cartwright RF. A method of measuring the potentials evoked by simultaneous stimulation of different retinal regions. Electroencephalogr Clin Neurophysiol 1970];28:314–9. van Dun B, Verstraeten S, Alaerts J, Luts H, Moonen M, Wouters J. A flexible research platform for multi-channel auditory steady-state response measurements. J Neurosci Methods 2008];169:239–48.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
ARTICLE IN PRESS
NSM 7208 1–9
Journal of Neuroscience Methods xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Journal of Neuroscience Methods journal homepage: www.elsevier.com/locate/jneumeth
Clinical Neuroscience
1
A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system
2
3
4
Q1
Andreas Bahmer a,∗ , Uwe Baumann b a
5
b
6
University of Würzburg, Clinic for Otolaryngology, Comprehensive Hearing Center, 97080 Würzburg, Germany University of Frankfurt Main, Clinic for Otolaryngology, Audiological Acoustics, 60590 Frankfurt, Germany
7
a r t i c l e
8 23
i n f o
a b s t r a c t
9
Article history: Received 15 September 2014 Received in revised form 2 March 2015 Accepted 2 April 2015 Available online xxx
10 11 12 13 14 15
22
Keywords: ASSR Jitter Psychophysic Electrophysiology Temporal coding SAM
24
1. Introduction
16 17 18 19 20 21
25Q4 26 27 28 29 30 31 32 33 34
Q2
Background: Simultaneous assessment of psychometric tasks and electrophysiological recordings is chal- Q3 lenging because each requires specific technical and physiological preconditions. Electrophysiological recordings require a comparatively long test time duration to gain sufficient signal-to-noise ratios, whereas test duration of psychometric measurements should be limited to prevent challenges to the attention of the subject. In order to investigate immediate correlation between both measurements a method is described, which combines electrophysiological and psychometrical measurements in a single test procedure. The test may be applied to subjects with deficits in temporal resolution (e.g. auditory neuropathy spectrum disorder, ANSD). New method: Auditory steady state responses (ASSR) and a pitch discrimination task were combined in a single procedure. The setup employed two short-time ASSR sub-stimuli with different fixed modulation frequencies but same carrier frequencies (signal 1 and 2). Simultaneously to the recording of ASSR, the test subject had to determine the signal interval which generated the perception of higher pitch. Results: The developed setup was successfully tested by means of an artificial EEG signal and in one human subject. ASSR signal as well as pitch discrimination performance. Comparison with existing methods: To our knowledge the presented method has not yet been described elsewhere. Conclusions: The feasibility of a setup to simultaneously perform a pitch discrimination task and electrophysiological measurements was demonstrated for the first time. The method provides the facility to apply sinusoidal amplitude modulated stimuli (SAM) with jittered modulation period lengths. © 2015 Published by Elsevier B.V.
The aim of simultaneous measurement of electrophysiological and a pitch discrimination task is to investigate the potential correlation between both entities. This aim may be achieved with sequential measurement as well, nevertheless a simultaneous measurement may contain a higher potential to reveal such a correlation. Provided there exists a high correlation between both, it is conceivable that complex and attention-dependent behavioral tasks might be substituted with objective electrophysiologic measurements (Fig. 1). The more peripheral the neural generators of the electrophysiological responses, the less neural processing takes
∗ Corresponding author. Tel.: +49 931 201 21779; fax: +49 3221 1348478. E-mail addresses: Bahmer [email protected] (A. Bahmer), [email protected] (U. Baumann).
place, and it is more likely that the electrophysiological measure reflects a basic step in the ascending auditory system for temporal processing and estimating pitch (Gockel et al., 2011; Krishnan and Plack, 2011; Greenberg et al., 1987). 1.1. Auditory steady state responses In auditory brainstem response (ABR) recordings transient acoustic stimuli are used which trigger neural responses in different parts of the auditory system (Jewett and Williston, 1971; Brown et al., 1994; Bahmer et al., 2008). The amplitude and latency of neural response peaks are evaluated for diagnostic purposes. The valuation of ASSRs is performed in the complex frequency domain in contrast to time domain based evaluation of ABRs (Lins and Picton, 1995). To generate ASSRs, continuous stimuli are presented, leading to summate periodic neuronal responses. Amplitude-modulated (AM) pure tones may be used as
http://dx.doi.org/10.1016/j.jneumeth.2015.04.003 0165-0270/© 2015 Published by Elsevier B.V.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
35 36 37 38
39
40 41 42 43 44 45 46 47 48 49
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
2
psychophysic
EEG ASSR signal strength
correct answers
decrasing stimulus regularity
task difficulty
Fig. 1. Both psychophysical performance and auditory steady state response (ASSR) signal strength decrease similarly with increasing task difficulty and corresponding decreasing stimulus regularity.
50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
stimuli. Neural generators, related to the depth and frequency of the amplitude modulation, are triggered. A continuous pure tone with amplitude modulated envelope between 2 and 400 Hz evokes a steady-state response at the modulation frequency (Chambers et al., 1986; Kuwada et al., 1986; Rees et al., 1986; Lins et al., 1996; Picton et al., 1987). It is commonly reported that modulation rates between 75 and 110 Hz are best suited to assess hearing thresholds. At these modulation rates, the ASSR is not significantly influenced by sleep (Cohen et al., 1991; Lins et al., 1995; Mühler et al., 2014), and, in normal hearing adults, responses to a 1-kHz stimulus can be recorded down to a stimulus level of 26 dB SPL (Lins et al., 1996). Evoked responses derived with continuously presented AM stimuli are analyzed in terms of both spectral energy and phase at the modulation frequency and its integer multiples (Chambers et al., 1986; Kuwada et al., 1986; Picton et al., 1987; Bahmer and Baumann, 2010). At modulation rates between 75 and 110 Hz, there is little change in the amplitude of the responses when multiple stimuli are presented compared to single stimulus presentation and carrier frequencies differ by one octave (Lins and Picton, 1995). In order to save recoding time, an alternative to a single stimulus application is the simultaneous measurement of responses to several AM stimuli (named MASTER stimulus) with each stimulus with different modulation and carrier frequency (Regan and Cartwright, 1970; John and Picton, 2000; John et al., 1998; Lins and Picton, 1995; van Dun et al., 2008). Statistical procedures determine whether a response is significant by comparing the response level and/or the phase at the particular modulation frequency to the noise level and/or phase at adjacent frequencies. The reliability of repeated measurements can also be assessed by comparing the statistical analyses. In order to restrict AEP to a predefined tonotopic region, masking noise with a certain spectral gap (notched noise) has to be presented in addition to transient click stimuli. ASSR-stimuli do not need the application of masking since the spectral power of carrier and side bands is restricted to a narrow frequency region (Lins et al., 1996). ASSRs as well as tone pip evoked ABR allow the assessment of functional deficits in different frequency regions of the cochlea. By using frequency-specific stimuli, different frequency regions are activated and can be evaluated separately. In order to develop new AEP recording procedures or to test the impact of alternative stimuli, affordable research platforms have been introduced using LabView or C/C++ (John and Picton, 2000; van Dun et al., 2008). The method presented here is implemented in Matlab 7.6.0.324 R2008a (The MathWorks Inc., Natick, Massachusetts) interfaces with other hardware components. The software presented here (see Appendix for excerpts) differs from so far introduced other research platforms since ASSR recording and a pitch discrimination task are running in parallel. Commercially available platforms offer only stimuli restricted to clinical routine tasks. In addition, access to raw EEG or analysis data is not available which make research applications and evaluations difficult.
1.2. Pitch discrimination tests Frequently, pitch discrimination tests are designed as alternative forced choice experiments. Several signal alternatives are presented consecutively, and the listener has to detect the deviant interval, or the interval perceived higher or lower in pitch. In addition, adaptive procedures are often employed, which reduce or increase the frequency difference between standard and deviant interval depending on reliability of subjects’ decision (example application described in Bahmer and Baumann, 2013). At a certain frequency difference subjects will asymptotically reach a minimum frequency difference threshold which defines the “just noticeable difference” (JND). AM stimuli may be applied in pitch discrimination test (Lee, 1994). In addition, randomized alteration (jitter) of the frequency of the modulator can serve as a stimulus parameter that potentially influences both pitch discrimination performance as well as the signal strength of auditory steady state responses.
1.3. Tested hypotheses In the present report, it is tested whether the simultaneous measurement of electrophysiological responses in an ongoing psychophysical task is feasible. Specifically, it was tested whether: • signal presentation and simultaneous recording via external audio-interface (RME ADI-8 DS) is free of distortions, • the customized analysis procedures (Matlab routines) of the recorded signals operates successfully, • ASSR generated by stimuli consisting of sinusoidal amplitude modulated signals (SAMs) modulated with either 200 or 250 Hz are detectable in simultaneous a pitch discrimination experiment, • concatenated EEG recording windows as short as 2 s are sufficient to detect auditory steady state potentials, • continuous stimuli containing jittered modulation cycles can be generated and integrated in the setup.
2. Experiment 1: evaluation of simultaneous pitch discrimination and ASSR 2.1. Rationale The pitch discrimination experimental setup described here (see Section 2.2) applies a two alternative forced choice procedure with fixed stimuli (see Fig. 3). A fixed stimulus design was necessary because auditory steady state responses are recorded in parallel. A certain amount of recording time is necessary to detect the stimulus generated response signal in the EEG.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
101
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116
117
118 119 120
121 122 123 124 125 126 127 128 129 130 131 132
133 134
135
136 137 138 139 140 141
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
PC
RME sound card input
output
touchscreen Which tone is higher?
biosignalamplifier
tone 1 headphones
tone 2
electrodes
Fig. 2. Hardware setup for ASSR recording and simultaneous assessment of pitch discrimination task. A standard PC with a Windows XP and Matlab installation is connected to an RME external sound card (ADI-8 DS). The external sound card receives input from a biosignal amplifier (one-channel self-made amplifier with analog medical grade isolation amplifier IA 297, Intronics, USA) and sends stimuli to the headphones (Sennheiser HDA 200). While recording ASSR, the subject performs a pitch discrimination task by pressing buttons on a touchscreen.
142
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
2.2. Method 2.2.1. Hard- and software The hardware setup consisted of a standard PC with a Windows XP and Matlab installation connected to an RME external sound card (ADI-8 DS). The external sound card received input from a single channel biosignal amplifier and sent stimuli to the headphones (HDA 200, Sennheiser, Wedemark, Germany) (Fig. 2). The self-made biosignal amplifier was provided by an analog medical grade isolation amplifier IA 297 (Intronics, USA) with one channel, and together with an analog display enabled to control a sufficient impedance range of surface EEG electrodes. The amplifier provided full patient protection from leakage currents and amplifier fault currents by optical isolation. Technical data are explained elsewhere (see (Bahmer et al., 2008)). The RME external sound card was chosen because it is inexpensive and provides a 24 bit dynamic range. The sound card is equipped with 8 input and 8 output channels that can be used simultaneously. EEG data was recorded on channel 1 of the sound card; the internally looped-back stimulus was recorded on channel 2 as a control. The latency between the beginning of the recording and the onset of the recorded loopedback stimulus signal was a constant 12.4 ms (548 points at 44.1 kHz sampling, measured with 10 repetitions). The EEG analysis using a sliding slit window was available via the ‘buffer’ function in Matlab (buffer length was set to the epoch length of 6 s consisting of tone 1, tone 2, and pause). The sliding was employed by a continuous shift of the EEG vector by one sampling point and subsequent feeding into the buffer. The sliding windows enable to find the exact time where the analysis windows perfectly match the time span of stimulus 1 and stimulus 2, respectively. The exact beginning of the EEG parts containing the neural response to the stimuli is unclear due to unknown inherent latencies. In experiment one, the functionality of the sliding window analysis was tested with stimulus 1 defined as an SAM signal with a carrier frequency of 2 kHz and a modulation frequency of 200 Hz; stimulus signal 2 was set to zero. The duration of each of stimulus 1, stimulus 2, and the pause was set to 2 s; altogether the epoch lasted 6 s. The amplitude of the sideband at 1800 Hz (carrier frequency minus modulation frequency) was plotted against the shift (increment 1000 sampling points, about 23 ms) of the sliding slit windows along the recorded input in Fig. 5. To enable temporal exact stimulus presentation and a synchronously EEG recording, the utilty ’playrec’ (Humphrey, 2008) was used (Matlab MEX file, Toolbox not necessary), which integrates ’PortAudio’, a free, open-source audio I/O library. ’PortAudio’ can access the AD/DA converters via different host application
3
programming interfaces (API) on different platforms (including ASIO, Audio Stream Input/Output protocol from Steinberg, Media Technologies, Hamburg; WMME, Windows Multimedia extension, and DirectSound under Windows, both Microsoft, Redmond). In the present study ASIO/Windows was used. Playrec provide control about the amount of skipped samples if, for example, the computer is slowed down by tasks running in parallel. The present recording system was developed because commercially available ASSR recording systems (e.g. Eclipse, Interacoustics, Assens, Denmark; Otometrics, ICS Chartr EP200, Taastrup, Denmark; Sentiero, Path Medical Solutions, Germering, Germany; Biologic Navigator Pro (MASTERII), Natus Medical Inc., Pleasanton, USA; GSI Audera, MN, USA) do not have the capability of presenting freely designed stimuli (e.g. with jitter) which is important for our research purposes. 2.2.2. Stimuli The stimuli consisted of sinusoidal amplitude modulated signals (SAM) with carrier fc , modulation frequency fm , and a subsequent pause. A SAM is described mathematically by SAM = sin(2fc t) · (1 + m · cos(2fm t))
(1)
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
201 202 203 204 205 206
1 SAM = sin(2fc t) + m · sin(2(fc + fm )t) 2 +
1 m · sin(2(fc − fm )t), 2
207
(2)
208 209
where fm and fc are the modulation and the carrier frequency, respectively. The frequency of the sidebands are denoted by fc ± fm , the modulation depth by m. The later described experiment presented two SAM tones with carrier frequency fc = 2 kHz and modulation frequencies fm1 = 200 Hz and fm2 = 250 Hz. The difference in terms of modulation frequency is perceived as a pitch difference. 2.2.3. Psychophysical test: pitch discrimination task The pitch discrimination task was a two alternative-forcedchoice experiment (Fig. 3). Two SAM stimuli with the same carrier frequency but different modulation frequencies were subsequently presented, which results in different pitches provided a sufficient frequency difference for the subject. Each SAM stimulus lasted 2 s, which was long enough to elicit an ASSR signal and a pitch percept, and not too long to exceed typical stimulus duration time in a pitch discrimination task. After the second stimulus was presented, a pause of 2 s followed. From the onset of the first stimulus until 80% of the block length (4.8 s; 6 s block with stimulus 1, stimulus 2, and pause) the subject could make a decision by pressing a button to indicate the interval which was perceived with higher pitch. After the pause, the next block of stimuli and pause followed independent of the subject’s decision. Stimuli were presented in a random order, which is considered in the analyses. The two SAM stimuli had a carrier frequency of 2 kHz, and modulation frequencies of 200 and 250 Hz, respectively. After finishing the test, the percentage of correct pitch difference identifications was calculated. The relatively high stimulus temporal pitch difference (fm1 − fm2 = 50 Hz) was chosen in order to test whether the psychoacoustic procedure was performable. In order to prove the described method, the sound pressure level of the stimulus was set to 84 dB SPL which generates a large spread of excitation on the basilar membrane. 2.3. Results 2.3.1. Phase preservation test of the setup using artificial signals The SAM stimulus was recorded to test whether the setup preserves the phase when recording and analysis is performed. The
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
210 211 212 213 214 215 216
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
241
242 243 244
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
4
Which tone is higher?
psychophysics
signal 1
signal 2
pause
signal 1
pause
signal 2
push button enabled
entry disabled, buttons released
EEG
1. mean 2.|FFT| @ fm1 or fm2
... ASSR signal strength shift of analysis windows along EEG
Fig. 3. Top: scheme of psychophysical experiment while an EEG is continuously recorded. Stimulus 1, stimulus 2, and the pause are continuously presented (duration 2 s each, stimulus sequence presented randomly). At 0.8 times of the period length (duration of two tones plus pause, about 6 s), the subject’s input is disabled and the decision is fixed. Afterward, both response buttons are released. Bottom: analysis of the recorded EEG. Concatenated sliding windows (only two/three windows are shown in one row, lengths corresponding to length of stimulus 1 or stimulus 2) are shifted along the recorded EEG (shift examples indicated by the three rows). The mean of the EEG data whose length correspond to the length of the sliding windows and subsequently the FFT are calculated. The FFT amplitude at frequency channel fm1 or fm2 of the mean (ASSR signal strength) is plotted against the shift of the sliding window. Periodic maxima are detected whenever the corresponding sections span the maximal neural response to stimulus 1 or stimulus 2. The ASSR signal strength can be calculated without knowledge of the absolute stimulus timing.
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278
SAM stimulus was generated by the software and sent to the AD/DA converter. The signal was looped-back to the input of the converter (direct hardware loop, feature of the RME sound card). The latency of the looped-back recorded signal was measured ten times in order to detect any jitter in the latency which may cause phase shifts between recorded signals. ASSR is sensitive to phase variations depending on analyses techniques (e.g. phase and averaged signal analyses, see (Cebulla et al., 2006)). Operating systems and software modules often cause uncontrolled delays. The looped-back recorded signal shows a constant latency of 548 sampling points at a sampling rate of 44.1 kHz from the beginning of the recording to the beginning of the recorded test signal. The calculated average of 70 recorded stimulus 1 signals (10 tests each containing 7 sliding windows, offset is 0, position of stimulus 1 directly after the pause in only 7 windows in all 10 tests) shows that the stimulus is exactly reproduced without any phase distortions (Fig. 4). Phase distortions would have resulted in destructive and constructive interference in an averaging process, and, consequently, amplitude fluctuations. In addition, the sliding slit window analysis was evaluated with test matrices (containing numbers zero, one, and two) and showed that the sorting of the recorded EEG snippets, which is mandatory due to the random order of stimulus presentation, works properly. In order to detect a (neural) signature as a response to the stimuli in the recorded data, an analysis of the recorded input by shifting sliding slit windows was performed (see Fig. 3, bottom). The software implementation was tested for this functionality by using the previously described looped-back stimulus recording (see Section 2.2). Fig. 5 shows, as expected, that the FFT amplitude of the sideband (fc − fm ) is maximal at shifts around 0, 6, and 12 s (plus latency shift of 548 sampling points due to looped-back signal, hidden by the figure scaling), which are multiples of the 6 s epoch length. The pattern repeats with a period of 6 s. The decrease from 0 to 2 s results from the fact that the buffer (i.e. the averaged windows) containing the full signal is incrementally substituted by
1
amplitude [a.u.]
245
0
−1
0
500
1000
1500
2000
time [sampling points] Fig. 4. Average of 70 recorded stimulus 1 signals (SAM, fc = 2000 Hz, fm = 200 Hz, 10 test runs each containing 7 sliding windows without shift or sliding, see Fig. 3; position of stimulus 1 directly after the pause in only 7 sliding windows in all tests). The test signal waveform is exactly reproduced which indicates perfect preservation. The latency of 548 sampling points (12 ms, sampling 44.1 kHz) is caused by loopback of the stimulus to the input of the sound card (direct internally hardware loop, feature of the RME sound card, see Section 2.2).
zeros (stimulus 2 is set to zero). At a shift of seconds the buffer is filled with zeros. As the analysis window set continues to slide along the recorded part containing the zero stimulus from the pause, the buffer remains at zero until the shift of 4 s. In the consecutive shift from 4 s until 6 s the buffer is gradually filled again with stimulus
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
279 280 281 282 283
G Model NSM 7208 1–9
ARTICLE IN PRESS A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
5
difference between the maximum and minimum FFT amplitude (correlated to stimulus on and stimulus off) indicates ASSR signal strength. For detecting obscure regularities in the EEG analysis an autocorrelation may be advantageous. 3. Experiment 2: Evaluation of jittered modulation cycle concatenation
284 285
286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323
1, resulting in an increase of the FFT amplitude from zero to one. A repetition of the described process occurs from 6 to 12 s and so on. 2.3.2. Timing of signal recording, stimulus playback, and touchscreen interface In 50 test sessions, signal recording, stimulus playback, and operation of the touchscreen interface were evaluated. The Matlab software package ’playrec’ (see Section 2.2 and Humphrey (2008)) enabled full duplex signal presentation and recording. In cases where the operation system scheduler prioritized parallel tasks that require a high CPU load, frames of samples were sometimes skipped. It was also observed that the movement of Matlab figures (i.e. the touchscreen interface) by the user frequently generated a loss of sample frames. During execution of the experiment, figures normally do not need to be moved, only interface buttons have to be pressed which does not result in a heavy processing load a subsequent sample skipping. In order to avoid that samples are skipped (e.g. with high CPU load), it is possible to enlarge the buffer length of recording and playback. The upper limit of acceptable software buffer enlargement which can be manipulated in the software is reached when the user interface shows a low sychronization or delay compared to the stimuli and recording. In our tests, the timing of button activation/inactivation was perfectly synchronized to the time at 0.8 of the epoch length (Fig. 3 top). 2.3.3. Test of the setup with human EEG recording After verification of the experimental setup (see Fig. 5), author A.B. served as a normal hearing subject (age 38). Stimuli were set as outlined in Section 2.2. The subject scored 100% percent correct in the pitch discrimination task. Fig. 6 shows the results of the subsequent EEG analysis. In both spectral analyses (stimulus 1 left, stimulus 2 right; FFT amplitude at 200 and 250 Hz, respectively) a distinct periodicity of about 6 s (stimulus repetition period) is detectable. For stimulus 1, maxima are at about 4 and 10 s (difference: 6 s), for stimulus 2 maxima are at about 1.5, 7.5, and 13.5 s (all differences: 6 s) after stimulus onset. The temporal regularities of 6 s found in the FFT analyses of the shifted EEG window for both SAM stimuli (Fig. 6 left and right) reflect the duration of stimulus 1, stimulus 2, and the pause (total 6 sec). The modulation frequency of the SAM stimuli only determines the FFT channel in which this regularity can be found by means of shifted time windows. The
327
330
3.2. Method
331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347
348
The modulation frequency may be jittered in order to disturb salient temporal information of pitch. The jitter is applied to the duration of an entire modulation cycle to ensure transmission of at least one cycle containing periodicity information (see Fig. 7 and Eqs. (3) and (4)). SAM = sin(2fc t) · (1 + m · cos(2fm (t)t)) fm (t) =
326
329
Temporal regularity, serving as pitch information, may be distorted by adding jitter. Consequently, the saliency of pitch is lowered and pitch discrimination ability decreases. On the other hand, AEP measures such as auditory steady state responses (ASSR) rely on the temporal regularity of an incoming signal. Pitch discrimination ability and ASSR responses are both negatively effected by the distortion generated by e.g. jitter of temporal regularity. If both entities are measured simultaneously, it is hypothesized that an increase of jitter will affect measurement results in both domains. For instance, in pitch discrimination tasks when two sinusoidal amplitude modulated signals have to be distinguished, jitter applied to the modulation frequency leads to a broadening of spectral energy at side bands and consecutive degradation of pitch salience until pitch discrimination completely vanishes (overlay of sidebands). An increase in modulation frequency jitter leads simultaneously to decreased spectral response amplitudes in ASSR modulation frequency.
325
328
3.1. Rationale
Fig. 5. FFT amplitude at sideband 1800 Hz of averaged recorded SAM signals (stimulus 1: SAM with fc = 2000 Hz and fm = 200 Hz at fixed first position, stimulus 2 set to zero, see Fig. 4) against the shift of the sliding windows along the recording (see Figs. 3 and 4). If the windows exactly span the EEG parts containing the SAM stimulus, the FFT amplitude of the sideband is maximal; if the windows are completely out of the range of the stimulus, the amplitude is zero. Only 8 sliding slit windows of 10 possible windows are averaged to avoid edge effects like buffer underrun when the window shift is larger than the corresponding EEG recording.
324
round|N(fm , )| : fm (t − 1) :
(1 + m · cos(2fm (t)t)) = 0
349 350 351 352 353
(3)
354
(4)
355
else
with fc : carrier frequency, fm (t): jittered modulation frequency, N: Gaussian distribution, fm : base modulation frequency, and : standard deviation (amount of jitter). Complete modulation cycles with jittered periods are concatenated (see Fig. 8). The challenge of concatenating continuously varying periods is that given a certain sampling frequency, only multiple integers of the sampling period can represent complete modulation cycles. If the length of the modulation cycles is not integer multiples of the sampling period, the overhang has to be transferred to the following modulation cycle. Otherwise, the signal is not uniformly continuous resulting in discontinuities. The modulation frequency differences are jittered within a certain probability distribution function (here Gaussian distribution). 3.3. Results It was tested whether the generated concatenated SAM is equal to a SAM generated without concatenation. For the SAM without concatenation a time vector (2 s) was generated and a concatenated SAM (2 s, fc 2 kHz, fm 200 Hz) according to Eq. (1) was calculated. There was no difference between the concatenated SAM and the non-concatenated SAM (both same fc and fm, no jitter). Additional tests with different fm and/or sampling rate gave no evidence for signal disruptions. In addition, a frequency analysis of the SAM with concatenated modulation cycles showed a peak only at the desired
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
356 357 358 359 360 361 362 363 364 365 366 367 368
369
370 371 372 373 374 375 376 377 378
G Model
ARTICLE IN PRESS
NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
6
Q5 Fig. 6. Analysis of test subject’s A.B. EEG (analysis procedure described in Fig. 5, 10 repetitions, stimulus 1: SAM with fm1 = 200 Hz, stimulus 2: SAM with fm2 = 250 Hz, both fc = 2000 Hz; duration stimulus 1, stimulus 2, pause: 2 s). The FFT maxima at the modulation frequencies are at multiples of the stimulus repetition period (6 s). Left: FFT of sliding analysis windows at fm1 = 200 Hz, right: FFT of sliding analysis windows at fm2 = 250 Hz. The increment of the shift is 1000 sampling points due to economy of time (green line: interpolation). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
SAM (fc, fm)
jittered SAM (fc, fm+ε)
|FFT|
|FFT|
fm fc-fm fc+fm
fc-(fm+ε) fc+(fm+ε)
fc
fc
fc
Fig. 7. SAM and jittered SAM with carrier frequency fc and modulation frequency fm in the temporal and frequency domain. The specific characteristic of the jittered SAM is containment of modulation cycles with different frequencies around fm (fm ±). These modulation cycles with different durations are concatenated to each other. This is reflected in the SAM frequency spectrum in a smearing of the sidebands. The jittered temporal pattern of the SAM defined by the jittered modulation (fm ±) is transferred to the auditory nerve at the tonotopic place defined by fc .
381
modulation frequency (Fig. 9 left). If the modulation periods were jittered, the corresponding frequency analysis of the modulation also showed a jitter (Fig. 9 right).
382
4. Discussion
379 380
383 384 385 386
The present report describes a setup which enables recording of ASSR while performing a two alternative forced choice pitch discrimination task. The setup may be augmented to a three or more alternative forced choice experiment. Testing with artificial signals
fm sampled first modulation cycle
showed that the hardware and software preserves the phase of the signal, and that the touchscreen timing is in synchrony with recording and stimulus playback. The developed analysis software is capable of detecting regularities in recorded signals and in a human EEG. In a standard setup, ASSR comprises the analysis of several minutes of EEG. This precondition cannot be used for psychophysical experiments as tones are mostly presented in the range of seconds or below. The challenge in a combination of both measures is to find a solution for the trade-off between preconditions for
fm+ε sampled next modulation cycles
concatenation of sampled modulation cycles
II.
I.
II.
I.
... Δt
Δt
... Δt
Fig. 8. Concatenation of modulation cycles with jittered period lengths (fm ±). If the length of a modulation cycle, e.g. cycle I., is not an integer multiple of the sampling period the overhang (t) has to be transferred to the next concatenated cycle (II.) and so forth. Only complete modulation cycles are concatenated.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
387 388 389 390 391 392 393 394 395 396
G Model NSM 7208 1–9
ARTICLE IN PRESS A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
7
Fig. 9. FFT amplitude of the concatenated SAM (sampling = 44.1 kHz, fc = 2 kHz, fm = 200 Hz, left: no jitter, =0, right: normally distributed jitter with = 10 · fm ( see Eqs. (3) and (4)), resolution of the frequency axis is 0.5 Hz). The concatenation algorithm yields exactly the desired modulation frequency and jittered modulation frequency.
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
both measurements. Electrophysiological recordings require a relatively long time to improve the signal-to-noise ratio, whereas in psychometric task measurements stimulus presentations are temporally limited to ensure continuous attention of the subject. Our setup is capable of combining both measurements. Testing showed that timing of recording, playback, and button control is highly precise (Figs. 4 and 5). A jitter may be applied to the modulation frequency of the SAM (Fig. 9) which potentially decreases temporal periodicity detected by pitch discrimination and/or ASSR. ASSR recording and analyses from averaged concatenated short intervals of EEG (here 2 s) opens a new field in the recording of ASSR. Further studies have to address, how the length of the stimuli, the repetition rate, and the length of silent intervals influence these responses. It is conceivable that mismatch negativity (Näätänen et al., 2007) may be used for the detection of pitch differences in electrophysiological responses. However, in contrast to ASSR, the generators of mismatch negativity are mostly related to higher cortical areas. As distortion of temporal patterns may be detected at a more peripheral site, the ASSR measurement is more directly linked to basic temporal properties. After recording and analysis of an artificial signal showed that the setup works without phase discontinuities, a human EEG was recorded while the subject performed a 2AIFC pitch discrimination test (Fig. 6). A potential application of the procedure is the assessment of pitch discrimination ability by decreasing the stimulus differences. In contrast to the results obtained with artificial EEG response signals (Fig. 5, in the real EEG data the amplitude maxima do not occur at 0, 6, and 12 s. A potential explanation might be, that the neural response is delayed by acoustic latency and neural network response properties. In addition, it can observed that the occurrence of the maxima seem to depend on modulation frequency. It has to be discussed whether a modulation specific responsiveness of the neural network generators is responsible for this effect – a question that has to be addressed in further studies more thoroughly. The maxima and minima in the shifted EEG response evoked by the stimulus differences and pauses (Fig. 6) were not disrupted by the motor system activation required to enter the decision on the touchscreen. Furthermore, despite the short duration of stimuli of 2 s and only 10 repetitions, a clear difference between stimuli is detected in the EEG. A drawback of the described method is the increased duration of the test due to the integration of the psychoacoustic test procedure. The prolongation of test duration is justified by the immediate relation between ASSR response pattern and psycho-acoustically derived measures, which shall be assessed by direct correlation (see Fig. 1). Further tests will
address the effect of stimulus duration, stimulus modulation frequency, stimulus carrier frequency, jitter, repetition rate/number, and sound pressure level on ASSR and psychophysical responses. In addition, statistical procedures may be implemented for detecting thresholds likewise statistical test procedures are used in standard ASSR (Chambers et al., 1986; Kuwada et al., 1986; Picton et al., 1987; Bahmer and Baumann, 2010). In conclusion, this study describes a feasibility test for a setup and procedure which is capable of simultaneously recording detectable ASSR while performing a pitch discrimination task.
5. Conclusions • The setup described is able to simultaneously record an EEG signal with high phase precision, continuously play SAM stimuli, and present, change, and read-out a touchscreen interface without phase discontinuity. • The described procedure reliably detects temporal periodicities in recorded data. • A preliminary experiment proved the detection of ASSR with unusual modulation frequencies (up to 250 Hz) used in a parallel pitch discrimination psychometric task. • Preliminary results demonstrate a concatenated recording time of 20 s (10 × 2 s stimulus interval)to be sufficient to discriminate the difference of temporal modulation induced by two different modulation frequencies in the ASSR. • The present procedure allows to calculate stimulus related ASSR signal strength without knowledge of the absolute stimulus timing, which is important for the analysis and integration of a psychometric test running in parallel. • The setup can be expanded to three or more alternative forced choice discrimination tasks. • The continuous stimulus containing jittered modulation cycles was developed and validated. The influence on both pitch discrimination tasks and ASSR measures will be subject to further studies.
Acknowledgements The work was supported by a grant of MED-EL, Innsbruck, Austria. We would like to acknowledge the contribution made by Rosalyn Tait whose comments helped to improve our English language usage. We would like to thank the two anonymous reviewers for the very helpful comments.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
444 445 446 447 448 449 450 451 452 453
454
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477
478
479 480 481 482 483
G Model NSM 7208 1–9
A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
8 484
485 486 487
488 489 490
ARTICLE IN PRESS
Appendix. Matlab Code for concatenating and sampling complete modulation cycles with different periods (SAM with jittered modulation)
Matlab Code for sliding slit window analysis of the recorded auditory steady state responses including sorting of EEG snippets due to random stimulus representation
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
G Model NSM 7208 1–9
ARTICLE IN PRESS A. Bahmer, U. Baumann / Journal of Neuroscience Methods xxx (2015) xxx–xxx
491
492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
References Bahmer A, Baumann U. Recording and online analysis of auditory steady state responses (ASSR) in Matlab. J Neurosci Methods 2010];187:105–13. Bahmer A, Baumann U. New parallel stimulation strategies revisited: effect of synchronous multi electrode stimulation on rate discrimination in cochlear implant users. Cochlear Implants Int 2013];14:142–9. Bahmer A, Peter O, Baumann U. Recording of electrically evoked auditory brainstem responses (E-ABR) with an integrated stimulus generator in Matlab. J Neurosci Methods 2008];173:306–14. Brown CJ, Abbas PJ, Fryauf-Bertschy H, Kelsay D, Gantz J, Bruce J. Intraoperative and postoperative electrically evoked auditory brain stem responses in nucleus cochlear implant users: implications for the fitting process. Ear Hear 1994];15:168–76. Cebulla M, Stürzebecher E, Elberling C. Objective detection of auditory steady-state responses: comparison of one-sample and q-sample tests. J Am Acad Audiol 2006];17:93–103. Chambers RD, Feth LL, Burns EM. The relation between the human frequency following response and the low pitch of complex tones. J Acoust Soc Am 1986];80:1673–80. Cohen LT, Rickards FW, Clark GM. A comparison of steady-state evoked potentials to modulated tones in awake and sleeping humans. J Acoust Soc Am 1991];90:2467–79. Gockel HE, Carlyon RP, Mehta A, Plack CJ. The frequency following response (ffr) may reflect pitch-bearing information but is not a direct representation of pitch. JARO 2011];12:767–82. Greenberg S, Marsh JT, Brown WS, Smith JC. Neural temporal coding of low pitch. I. Human frequency-following responses to complex tones. Hear Res 1987];25:91–114. Humphrey R. ’playrec’. 2008. http://www.playrec.co.uk Jewett DL, Williston JS. Auditory-evoked far fields averaged from the scalp of humans. Brain 1971];94:681–96. John MS, Lins OG, Boucherand BL, Picton TW. Multiple auditory steadystate responses (MASTER): stimulus and recording parameters. Audiology 1998];37:59–82.
9
John MS, Picton TW. MASTER: a Windows program for recording multiple auditory steady-state responses. Comput Methods Progr Biomed 2000];61:125–50. Krishnan A, Plack CJ. Neural encoding in the human brainstem relevant to the pitch of complex tones. Hear Res 2011];275:110–9. Kuwada S, Batra R, Maher VL. Scalp potentials of normal and hearing-impaired subjects in response to sinusoidally amplitude-modulated tones. Hear Res 1986];21:179–92. Lee J. Amplitude modulation rate discrimination with sinusoidal carriers. J Acoust Soc Am 1994];96:2140–7. Lins OG, Picton PP, Picton TW, Champagne SC, Durieux-Smith A. Auditory steadystate responses to tones amplitude-modulated at 80–110 Hz. J Acoust Soc Am 1995];97:3051–63. Lins OG, Picton TW. Auditory steady-state responses to multiple simultaneous stimuli. Electroencephalogr Clin Neurophysiol 1995];96:420–32. Lins OG, Picton TW, Boucher BL, Durieux-Smith A, Champagne SC, Moran LM, et al. Frequency-specific audiometry using steady-state responses. Ear Hear 1996];17:81–96. Mühler R, Rahne T, Mentzel K, Verhey JL. 40-Hz multiple auditory steady-state responses to narrow-band chirps in sedated and anaesthetized infants. Int J Pediatr Otorhinolaryngol 2014];78:762–8. Näätänen R, Paavilainen P, Rinne T, Alho K. The mismatch negativity (MMN) in basic research of central auditory processing: a review. Clin Neurophysiol 2007];118:2544–90. Picton TW, Skinner CR, Champagne SC, Kellet AJ, Maiste AC. Potentials evoked by the sinusoidal modulation of the amplitude or frequency of a tone. J Acoust Soc Am 1987];82:165–78. Rees A, Green GGR, Kay RH. Steady-state evoked responses to sinusoidally amplitude-modulated sounds recorded in man. Hear Res 1986];23:123–33. Regan D, Cartwright RF. A method of measuring the potentials evoked by simultaneous stimulation of different retinal regions. Electroencephalogr Clin Neurophysiol 1970];28:314–9. van Dun B, Verstraeten S, Alaerts J, Luts H, Moonen M, Wouters J. A flexible research platform for multi-channel auditory steady-state response measurements. J Neurosci Methods 2008];169:239–48.
Please cite this article in press as: Bahmer A, Baumann U. A setup for simultaneous measurement of electrophysiological and psychometric temporal encoding in the auditory system. J Neurosci Methods (2015), http://dx.doi.org/10.1016/j.jneumeth.2015.04.003
524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557
