Mismatch negativity and psychophysical detection of

0 downloads 0 Views 2MB Size Report
Feb 5, 2018 - rising/falling sounds with gradual or stepwise change of intensity. The MMN ... Changes in sound level, along with spectral content, can produce changes in ... sound level produced by approaching sources in natural environments has been ..... were placed on ear lobes and at the tip of the nose. Horizontal ...
Biological Psychology 133 (2018) 99–111

Contents lists available at ScienceDirect

Biological Psychology journal homepage: www.elsevier.com/locate/biopsycho

Mismatch negativity and psychophysical detection of rising and falling intensity sounds

T



Lidia B. Shestopalova , Ekaterina A. Petropavlovskaia, Varvara V. Semenova, Nikolai I. Nikitin I.P. Pavlov Institute of Physiology, Russian Academy of Sciences, Saint-Petersburg, Russia

A R T I C L E I N F O

A B S T R A C T

Keywords: Mismatch negativity Behavioral performance Sound intensity Increment Decrement Approaching (looming) sound

Human subjects demonstrate a perceptual priority for rising level sounds compared with falling level sounds. The aim of the present study was to investigate whether or not the perceptual preference for rising intensity can be found in the preattentive processing indexed by mismatch negativity (MMN). Reversed oddball stimulation was used to produce MMNs and to test the behavioral discrimination of rising, falling and constant level sounds. Three types of stimuli served as standards or deviants in different blocks: constant level sounds and two kinds of rising/falling sounds with gradual or stepwise change of intensity. The MMN amplitudes were calculated by subtracting ERPs to identical stimuli presented as standard in one block and deviant in another block. Both rising and falling level deviants elicited MMNs which peaked after 250 ms and did not overlap with N1 waves. MMN was elicited by level changes even when the deviants were not discriminated behaviorally. Most importantly, we found dissociation between earlier and later stages of auditory processing: the MMN responses to the level changes were mostly affected by the direction of deviance (increment or decrement) in the sequence, whereas behavioral performance depended on the direction of the level change within the stimuli (rising or falling).

1. Introduction Changes in sound level, along with spectral content, can produce changes in the apparent distance to a source (Coleman, 1963, 1968; Strybel & Perrot, 1984). Signals with rising or falling sound level are generally perceived as approaching or receding sound sources (Bronkhorst & Houtgast, 1999; Zahorik, 2005). The salience of rising sound level produced by approaching sources in natural environments has been supported by a number of behavioral studies which suggested high-priority processing for approaching sounds in human and nonhuman primates. The subjects overestimated increasing compared to decreasing sound levels (Ghazanfar, Neuhoff, & Logothetis, 2002; Neuhoff, 1998; Stecker & Hafter, 2000) and underestimated the time to contact approaching sound sources (Rosenblum, Carello, & Pastore, 1987; Schiff & Oldak, 1990). Similar asymmetry in the discrimination of intensity increments and decrements was found in an oddball experiment (Rinne, Särkkä, Degerman, Schröger, & Alho, 2006). The perceptual bias for approaching sound objects seems to be reflected in the pattern of neural activity (Bach et al., 2008; Hall & Moore, 2003; Lu, Liang, & Wang, 2001). Generally, converging experimental evidence suggests that brain activations due to spatial auditory processing are centered in the posterior superior temporal gyrus and planum temporale (for a review, see Alho, Rinne, Herron, & Woods,



2014). Furthermore, only the right-hemispheric auditory cortex has shown significant differences in the loci for spatial processing in passive and active listening conditions: the median locus of spatial attentionrelated modulations have been found in the superior temporal sulcus, significantly inferior to the median locus for passive spatial processing. The fMRI study of Seifritz et al. (2002) demonstrated that rising and falling sound levels activated the right temporal plane more than constant level sounds. Rising compared to falling levels activated a widely distributed network of activity subserving auditory spatial perception and attention. A more recent fMRI study reported activity of the right amygdala and left temporal areas in response to rising compared to falling sound level (Bach et al., 2008). Previous electrophysiological studies of auditory processing of dynamic acoustical information accumulated an ample body of data concerning cortical evoked responses elicited by changes of sound level. Two components of event-related potentials (ERPs) are thought to reflect the first stage of passive (preattentive) auditory processing: 1) the N1 wave elicited by an onset and simple change detection process (for a review, see Näätänen & Picton, 1987) 2) the mismatch negativity (MMN) elicited by a sensory memory-based deviance detection process (for a review, see Näätänen, Paavilainen, Rinne, & Alho, 2007). The N1 component reflects differential activation of neural elements sensitive to various stimulus features, and the MMN indexes the process of

Corresponding author at: I.P. Pavlov Institute of Physiology, Russian Academy of Sciences, 199034 Makarova Emb., 6, Saint-Petersburg, Russia. E-mail address: [email protected] (L.B. Shestopalova).

https://doi.org/10.1016/j.biopsycho.2018.01.018 Received 1 June 2017; Received in revised form 26 January 2018; Accepted 31 January 2018 Available online 05 February 2018 0301-0511/ © 2018 Elsevier B.V. All rights reserved.

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

predicts that behavioral performance would be better 1) for rising deviants among constant standards as compared to falling deviants among constant standards 2) for constant deviants among rising standards as compared to constant deviants among falling standards. If the same bias can be found in MMN responses, the sequences containing rising stimuli would produce higher MMNs (compared to the sequences with falling ones), no matter what the functional role of a rising level stimulus will be, standard or deviant. Then the MMN amplitude elicited by rising deviants among constant standards would be in similar proportions to the MMNs elicited in the two opposite configurations as described above for the behavioral performance. On the contrary, if the MMNgenerating mechanisms are related to more general sequence properties aside from rising or falling stimulus level, the proportions between MMN responses would not parallel the psychophysical data. The question addressed by the present study is whether or not the perceptual preference for rising intensity can be found in the preattentive processing indexed by MMN. The data accumulated by now suggest that behavioral deviance detection may be at least partially governed by the processes underlying MMN generation, and the MMN mechanism may serve as an “alarm signal” which can initiate an attentional switch to a deviant event and exert influences observable at the behavioral level (Paavilainen, 2013; Winkler, 2007). According to another assumption, rising sound level may serve as an intrinsic, unconditioned warning cue which may enhance activation of early preattentive processes related to stimulus detection (Bach et al., 2008). We expected therefore to find a priority for the processing of rising intensity in the behavioral and electrophysiological measures. In the current experiment we have used the reversed oddball stimulation to produce the MMNs and to test the behavioral discrimination of rising, falling and constant level sounds. During psychophysical measurements, our subjects were required to detect the deviant sounds in the oddball sequences similar to those used to elicit the MMNs. We expected that rise and fall of signal intensity should exert significant effect on the behavioral responses. The speed of the level change was varied using two temporal patterns of rising and falling intensity (gradual and stepwise change). In regard to our earlier MMN study which employed gradual and stepwise change within the stimuli (Shestopalova et al., 2015), we anticipated that the stepwise change of intensity would elicit higher MMNs relative to gradual one, and that both increment and decrement intensity MMNs would be easily separable from the N1 wave. Thus, we contrasted the effects of the level change direction within the stimuli and of the deviance direction in the sequence (i.e., of the configuration reversal) on the MMN amplitude and latency.

comparison of incoming stimuli to representations generated on the basis of temporal regularities extracted from the auditory input (e.g., Horváth, Winkler, & Bendixen, 2008). According to a more general interpretation of MMN response, the ultimate function of the MMNgenerating process is to adjust neuronal models underlying detection of auditory objects by initiating changes in those particular models whose predictions were mismatched by the acoustic input (Winkler, 2007). An important dissimilarity between the N1 and MMN components is that N1 amplitude diminishes when the stimulus intensity is decreased (Beagley & Knight, 1967; Picton, Goodman, & Bryce, 1970; Rapin, Schimmel, Tourk, Krasnegor, & Pollack, 1966), whereas MMN is elicited by both sound level increases and decreases, its magnitude following the magnitude of level changes irrespective of the direction of change (Näätänen, 1992). An MMN can be elicited by either an increment or decrement in certain acoustical dimension (e.g., intensity or frequency). If the deviant stimulus has higher intensity or frequency than the standard, the MMN generated is called an increment MMN. In the opposite direction of deviance (if the deviant has lower intensity or frequency than the standard), the MMN generated is called a decrement MMN. The effect of the direction of deviance was described for duration MMN (Colin et al., 2009; Okazaki, Kanoh, Takaura, Tsukada, & Oka, 2006; Peter, McArthur, & Thompson, 2010; Takegata, Tervaniemi, Alku, Ylinen, & Näätänen, 2008), for frequency MMN (Jacobsen & Schröger, 2001; Karanasiou et al., 2011; Peter et al., 2010) and for sound velocity MMN (Shestopalova, Petropavlovskaia, Vaitulevich, & Nikitin, 2015). These authors shared the view that increment MMNs were larger in magnitude than decrement MMNs. However, a few studies comparing the MMNs produced by increments and decrements of intensity still have not come to a definite conclusion. Some studies have reported no differences between increment and decrement intensity MMNs (Altmann et al., 2013; Näätänen, 1992), while Rinne et al. (2006) have found that decrement MMN was higher and peaked later. A comprehensive study of Jacobsen, Horenkamp, and Schröger (2003) designed in order to separate memory-comparison-related effects of intensity from refractoriness-related ones has revealed that sound level increments elicited MMNs which were higher than decrement MMNs but could hardly be separable from the N1 wave, whereas decrement MMNs were free from this contamination due to their longer latency. This discrepancy between the properties of increment and decrement MMNs is not typical for MMNs elicited by cues apart from intensity (e.g., by frequency or duration). It should be also noted that most of the above mentioned MMN studies employed constant level stimuli as standards and deviants. The only MMN experiment which used smooth intensity changes has not found any differences in MMN amplitudes and latencies for rising and falling intensity (Altmann et al., 2013). The effect of deviance direction can be easily modeled by the reversal of the functional roles of the stimuli within the oddball sequence. The influence of standard-deviant reversals on preattentive processing of various sound features was explored in a number of studies using a reversed oddball paradigm (e.g., Jacobsen & Schröger, 2003; Peter et al., 2010). When the roles of standards and deviants are reversed, an increment sequence turns into a decrement one, and vice versa. Hence, the change of the context within which a rising/falling level sound appears (i.e., the change of the sequence structure) makes it possible to obtain the same direction of deviance (for instance, increment) using opposite directions of sound level change: the sequences containing rising intensity deviants in the context of constant standards or constant deviants in the context of falling level standards both represent the increment configurations. So, the reversed oddball stimulation can be used to separate the effect of the deviance direction from the effect of the direction of sound level change. Our working hypothesis about the differences between direction of level change (rising or falling) and direction of deviance (increment or decrement sequence) can be exemplified by an increment sequence containing rising deviants among constant standards and two possible opposite configurations. The perceptual priority for rising intensity

2. Methods 2.1. Participants Nine paid volunteers (1 male, 8 females, aged 27.3 ± 5.8 years, mean ± SD, all right-handed) participated in the experiments. All subjects had normal hearing (self-reported) and no history of neurological or otological disease. Research protocols were approved by the Ethical Committee (IRB) of St.-Petersburg State University (N02-79). Written informed consent from the subjects was obtained prior to the study. 2.2. Apparatus and stimuli The subjects were seated in a sound-attenuated and electrically shielded chamber and were diotically presented with blocks of auditory stimuli. Each subject participated in a complete experimental cycle consisting of electrophysiological and psychophysical parts. At the beginning of each experimental session the hearing thresholds of the listener’s left and right ear were measured by a simplified staircase procedure, using noise bursts of the same bandwidth as in the main 100

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Fig. 1. Schematic patterns of the intensity changes in the sound signals. Top: constant level signals and signals with rising level. Bottom: constant and falling level signals. Dashed line indicates the intensity level of the constant stimuli.

In each block, one of the three stimuli (“const”, “grad” or “step”) served as the standard while the other two were used as deviants. In other blocks, the standards and deviants were reversed. The deviants are referred to by capital letters (CONST, GRAD, STEP) and the standards are referred to by small letters (const, grad, step) throughout the manuscript. Each block contained 420 standards and 80 deviants presented in a pseudorandom order with 1 s onset-to-onset interval. Therefore, a single block was about 8 min long. In the const-standard blocks, 40 GRAD-deviants and 40 STEP-deviants had the same direction of deviance (that is, rising or falling in intensity). The grad-standard blocks contained 40 STEP-deviants which had the same direction of level change (either rising or falling) as the standards, and also 40 CONST-deviants. The step-standard blocks contained 40 GRAD-deviants which had the same direction of level change (either rising or falling) as the standards, and also 40 CONST-deviants. This resulted in a total of 6 block types (3 types of standards*2 directions of level change) presented in a pseudorandom order. The stimulus probabilities were 0.84, 0.08 and 0.08 for the standard and for two types of deviants, respectively. Each block was presented 6–7 times on different days to each subject, so that minimum 240 deviants were presented for each experimental condition. Each experimental session consisted of 7–8 blocks presented with short breaks and lasted about 1.5 h. Each subject completed all stimulus blocks over 5–6 experimental sessions carried out on different days.

experiment. During the EEG recording, the subjects read books of their choice. During the psychophysical tests, they performed tasks on sound discrimination, where they were asked to press buttons on a special keyboard. Before the psychophysical experiments, all subjects were given preliminary training until their performance was stabilized. Three types of sound signals were employed in the experiment: constant level sounds and two kinds of rising/falling level sounds with gradual or stepwise change of intensity (Fig. 1). All stimuli were binaurally presented low-frequency noise bursts (bandwidth 100–1300 Hz). The signals of 96 kHz sampling rate were delivered through the 24-digit sound board GINA 24 (96 kHz, Echo Audio, USA) and were presented directly into the subject's auditory meatus via insert earphones ER-2 (Etymotic Research Inc., USA). The earphones provided at least 30 dB external noise attenuation; their frequency responses, measured with Zwislocki coupler, were flat in the range of 0.2–10 kHz. The duration of all stimuli was 200 ms (apart from 10 ms fade-in/fadeout fronts smoothed by the cosine function), which corresponded to the loudness integration time (Uppenkamp & Röhl, 2014). All stimuli were created on the basis of the same initial signal by means of custom-made scripts. The stimuli of constant level (referred to as “const”) were diotically presented initial signals, and were perceived by all listeners as constant fused sounds. The intensity of the initial sound was adjusted to the level of 45 dB above the hearing threshold of each ear. The two kinds of rising/falling noise bursts differed in their temporal intensity patterns so as to simulate either gradual increase/ decrease of loudness, or the abrupt change of loudness within the same range (Fig. 1). Their intensity level was changed by 5 dB either linearly (from stimulus onset to offset) or instantly (at 100 ms post-onset, that is, at the middle of the signal). These stimuli are referred to as “grad” or “step”, respectively. The sound onset-offset difference of 5 dB is sufficient to elicit a MMN (Näätänen et al., 2007; Rinne et al., 2006) and to create the perceptual asymmetry for rising and falling intensity (Seifritz et al., 2002). Both the original and modified signals were then bandpass filtered between 100 and 1300 Hz and presented diotically.

2.3.2. Psychophysical experiment The behavioral measurements were interleaved with MMN recordings. The stimuli were presented in the oddball blocks similar to those used in the MMN experiment. The only difference was that each block contained 210 standards and 40 deviants of one kind, so that the stimulus probabilities were 0.84 for the standard and 0.16 for the deviant. In the grad-standard and step-standard blocks, the STEP- or GRAD-deviants had the same direction of level change (rising or falling) as the standards. A single block was about 4 min long. In order to minimize the putative influence of our interleaved design on the MMN responses, the order of blocks was pseudorandomized in a way that the behavioral and EEG stimulus blocks containing similar configurations could not be presented one after another. Subjects performed a “yes-no” task. They were instructed to press a button when they detected a deviant sound in the standard sequence. Hit rates (HR), False alarms (FA) and mean

2.3. Experimental procedures 2.3.1. MMN experiment The MMN experiment included six types of oddball blocks: three types of blocks with rising sounds and three types with falling sounds. 101

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

using one-tailed paired t-tests. To test for significant effects, mean MMN amplitudes at Fz site were calculated for each subject in the 50-ms wide windows centered on the MMN peak latency, defined as the time of the grand mean difference waveform maximum for each stimulus configuration. The mean MMN amplitudes, peak MMN latencies and the behavioral measures (HRs and RTs) were subjected to 3-way repeated measures ANOVAs, with factors: Direction of level change (Dir: Rise, Fall), Configuration (Config: Direct, Reversed), Modulation pattern in the deviant-standard pair (Mod: Step, Gradual, Step&Gradual). The configurations STEP/const, GRAD/const and STEP/grad were referred to as direct, and CONST/step, CONST/grad and GRAD/step as reversed. The modulation pattern was referred to as Step for the STEP/const and CONST/step pairs, as Gradual for the GRAD/const and CONST/grad pairs, and as Step&Gradual for the STEP/grad and GRAD/step pairs. To test the interhemispheric asymmetry of the MMN responses, the individual ERPs to each stimulus were averaged over two clusters: Fp1, AF3, F3, F7, FC1, FC5, C3, C5 in the left hemisphere and Fp2, AF4, F4, F8, FC2, FC6, C4, C6 in the right hemisphere. The individual difference waves were computed for each electrode cluster according to the samestimulus method described above. The mean MMN amplitudes in the 50-ms windows were entered into 4-way repeated measures ANOVAs, with factors: Hemisphere (Left, Right), Direction of level change (Dir: Rise, Fall), Configuration (Config: Direct, Reversed), Modulation pattern in the deviant-standard pair (Mod: Step, Gradual, Step&Gradual). The Bonferroni adjustment for the follow-up contrasts (p < .05) was applied in all ANOVAs to correct for multiple comparisons. Correlations between Fz MMN amplitudes and behavioral data were analyzed by Spearman correlation test (p < .05).

reaction times (RTs) were calculated separately for each stimulus configuration. Responses occurring between 100 ms and 1100 ms after stimulus onset were accepted in the analysis. True Hit rates (HRs) were calculated using Blackwell correction: HRs = (HR − FA)/(1 − FA) (Blackwell, 1953). 2.4. Electrophysiological recordings Continuous EEG acquisition was performed using the Active 2 system (Biosemi, The Netherlands) equipped with 32 sintered Ag/AgCl electrodes. Electrode placement was based on the modified 10–20 system with C5/C6 mounted instead of PO3/PO4. Additional electrodes were placed on ear lobes and at the tip of the nose. Horizontal and vertical eye movements were recorded using electrodes placed below the left and above the right eye. EEG data were sampled at 2048 Hz, lowpass-filtered (102 Hz cutoff), and decimated to 512 Hz sample rate for computational efficiency. 2.5. Data analysis Continuous EEG recordings were segmented offline into 1000 ms epochs including a 300 ms pre-stimulus interval set as the baseline. These epochs were then referenced to the ear lobes. The trials affected by eye movements or blink artifacts of more than 150 μV amplitude were rejected (< 25% of trials in total). The epochs were averaged separately for each stimulus type and bandpass-filtered (2–30 Hz bandwidth). Recordings from the Fz and Cz sites were used to calculate ERPs and difference waveforms, as the recording from these locations typically yielded the largest-amplitude MMN responses (see Näätänen et al., 2007 for review). For each individual standard ERP waveform, the N1 and P2 components were characterized by their mean amplitudes measured in the 50-ms wide windows set at ± 25 ms relative to the peak latency, defined as the time of the corresponding grand-average ERP maximum or minimum for each stimulus configuration. The mean voltage of the later negative deflections, peaking at about 250 ms, was calculated within the 50-ms wide windows set around the corresponding peaks of grandaverage ERPs. The three components were then analyzed statistically using separate 2-way repeated measures ANOVAs with factor Standard (const, grad, step) and Direction of level change (rise, fall). When the constant stimulus served as standard in a sequence with rising GRADand STEP-deviants, it was assigned “rise” direction. The constant standards used in a sequence with falling level deviants were assigned “fall” direction. The Greenhouse–Geisser correction for degrees of freedom was applied when the sphericity assumption was not met. In the follow-up contrasts, the Bonferroni adjustment for multiple comparisons was applied (p < .05). MMN was calculated by subtracting the ERP for the same stimulus when it was presented as standard in one block from when it was presented as deviant in another block. The MMN notations used in the analysis reflect the standard-deviant configurations in which the ERPs were recorded. The ERP to the const-standard was subtracted from the ERP to the CONST-deviant obtained in the grad-standard (CONST/ grad) or the step-standard blocks (CONST/step). The ERP to the gradstandard was subtracted from the ERP to the GRAD-deviant obtained in the const-standard (GRAD/const) or the step-standard blocks (GRAD/ step). The ERP to the step-standard was subtracted from the ERP to the STEP-deviant obtained in the grad-standard (STEP/grad) or the conststandard blocks (STEP/const). Using this method, six individual same-stimulus difference waveforms were obtained for each direction of the level change and then averaged across all subjects to produce grand-mean MMNs. To determine the time intervals where MMNs were significantly different from 0, a moving window analysis was performed on the amplitudes of the individual difference waveforms averaged over 50-ms intervals in 2ms steps. Statistical analysis on the difference scores was carried out by

3. Results 3.1. Behavioral task HRs and RTs were affected mainly by the direction of intensity change: the discrimination performance was better for the rising as compared to the falling intensity sounds (Fig. 7, lower panels). Results of 3-way repeated measures ANOVAs with the Direction of level change (Dir: Rise, Fall), Configuration (Config: Direct, Reversed) and Modulation pattern in the deviant-standard pair (Mod: Step, Gradual, Step& Gradual) as factors are given in Table 1. Apart from the main effects, the ANOVAs indicated significant 2-way interactions for HRs and RTs, but no further interaction. The reversals in the stimulus pairs (STEP/ const – CONST/step, or GRAD/const – CONST/grad, or STEP/grad – GRAD/step) did not change the HRs (interactions Dir*Config: F(1, 8) = 3.38, p > .05 and Config*Mod: F(2, 16) = 0.77, p > .05). According to the post-hocs, HRs were significantly higher for the rising level sounds in all configurations (p < .01). The significant interaction Dir*Mod (F(2, 16) = 6.43, p < .05) was due to the fact that the modulation pattern had no effect on the HRs in the case of rising level sounds (p > .05), whereas in the case of falling sounds the highest HRs were obtained for Step and the lowest for Step&Gradual modulation (p < .05 for both). The effects obtained for the RTs generally mirrored those described for the HRs. The significant interaction Dir*Mod (F(2, 16) = 3.66, p < .05) was due to the fact that the modulation pattern had no effect on the RTs in the case of falling level sounds (p > .05), whereas in the case of rising sounds the highest RTs were obtained for Step&Gradual modulation (p < .01). The ANOVAs revealed also the significant interactions of Dir*Config (F(1, 8) = 10.98, p < .05) and Config*Mod (F (2, 16) = 4.07, p > .05). According to the post-hocs, the RTs were significantly lower for the rising level stimuli in STEP/const and GRAD/ const configurations (p < .001) but not in the reversed ones.

102

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Table 1 Results (F, p and epsilon values) of 3-ways ANOVAs for the mean Fz MMN amplitudes, peak MMN latencies, Hit rates (Hits) and Reaction times (RT). The ANOVA factors were Direction of level change (Dir: Rise, Fall), Configuration (Config: Direct, Reversed), Modulation pattern in the deviant-standard pair (Mod: Step, Gradual, Step&Gradual). Degrees of freedom (df) are given in the row titles. Greenhouse-Geisser epsilons (when applied) are given next to the p values. ANOVA factors (df)

Dir (1, 8)

Config (1, 8)

Mod (2, 16)

Dir*Config (1, 8)

Dir*Mod (2, 16)

Config*Mod (2, 16)

Dir*Config*Mod (2, 16)

MMN amp

5.91 < .05 67.06 < .001 47.28 < .001 13.37 < .01

2.83 .13 22.13 < .01 8.23 < .05 0.3 .6

9.59 < .01 ε = 0.80 12.67 < .01 ε = 0.74 59.81 < .001 ε = 0.58 6.77 < .05 ε = 0.57

22.51 < .001 45.76 < .001 3.38 .1 10.98 < .05

5.24 < .05 ε = 0.80 64.54 < .001 ε = 0.92 6.43 < .05 ε = 0.55 3.66 < .05 ε = 0.77

8.24 < .05 ε = 0.58 20.16 < .001 ε = 0.63 0.28 .68 ε = 0.70 4.07 < .05 ε = 0.82

26.78 < .001 ε = 0.97 9.13 < .01 ε = 0.96 0.77 .47 ε = 0.88 0.62 .46 ε = 0.54

MMN lat Hits RT

F p F p F p F p

deviants at each site were averaged in running 50 ms windows and submitted to the one-tailed paired-sample t-tests. The MMN amplitudes plotted across time are given in Figs. 4 and 5. Statistically significant MMNs (p < .05) for increment and decrement configurations are shown by red and blue coloured lines and areas, which corresponds to the black-and white scales in the printed version of the article. The effects of rising/falling intensity, of deviant-standard reversals and of modulation pattern were explored by 3-ways repeated measures ANOVA performed on the mean MMN amplitudes and peak MMN latencies, with factors Direction of level change (Dir: Rise, Fall), Configuration (Config: Direct, Reversed) and Modulation pattern in the deviant-standard pair (Mod: Step, Gradual, Step&Gradual) (see Table 1). Apart from the main effects, the analysis revealed strong 3ways interactions Dir*Config*Mod (F(2, 16) = 26.78, p < .001 for mean MMN amplitudes and F(1.91, 15.32) = 9.13, p < .01 for peak MMN latencies). The post-hoc tests indicated that only STEP/const and CONST/step combinations of rising level stimuli had a significant effect on the MMN magnitude. This was due to a higher MMN elicited by rising STEP deviants in the context of constant standards as compared to other combinations of stimuli (p < .01 in the post-hocs). In particular, the direction of intensity change exerted no effect on MMN magnitude in any of other configurations (p > .05) (Fig. 7, upper panels). The pairwise comparisons of the MMN latencies confirmed that earlier MMNs were elicited by the increment deviants represented by rising level sounds in the direct configurations (p < .001) and by falling level sounds in the reversed configurations (p < .01). The MMN topographies (Fig. 6) were dominated by frontocentral negativity, with higher activation over the right frontal areas. To test for the putative right-hemispheric dominance, the mean MMN amplitudes averaged in clusters were analyzed by 4-way repeated measures ANOVAs, with factors: Hemisphere (Left, Right), Direction of level change (Dir: Rise, Fall), Configuration (Config: Direct, Reversed), Modulation pattern in the deviant-standard pair (Mod: Step, Gradual, Step&Gradual). The effect of Hemisphere was revealed only in the significant 3-way interaction Hemisphere*Direction*Configuration (F (1, 8) = 5.78, p < .05). According to the post-hocs, the right-hemispheric MMN responses were higher than their left-side counterparts in most combinations of stimuli. This difference was stronger in the increment configurations and reached maximal significance for the rising GRAD/const MMN (p < .06) and for the falling CONST/step MMN (p < .05). To illustrate the interhemispheric asymmetry, the mean MMN amplitudes in both electrode clusters are shown in the bar plots placed next to each panel of Figs. 4 and 5. Spearman correlation test between Fz MMN amplitudes and behavioral measures revealed only one significant result: MMNs elicited by rising GRAD/const combination were positively correlated with HRs (ρ = 0.683, p < .05). All other sequences produced non-correlated MMN and behavioral data.

3.2. The EEG results 3.2.1. The ERPs to standards and deviants Left panels of Figs. 2 and 3 demonstrate the grand-average ERPs elicited by const, grad and step standards and by the CONST, GRAD and STEP deviants. In responses to standards (Figs. 2 and 3, left panels, thin lines), the late negative deflections peaking at around 250 ms elicited by rising level stimuli were higher than those elicited by constant stimuli. The amplitudes of N1, P2 and late negativity elicited by the standards were entered into separate 2-way repeated measures ANOVAs with factor Standard (const, grad, step) and Direction of level change (rise, fall). The analysis yielded significant main effects of Direction for both negative deflections (F(1, 8) = 5.75, p < .05 for N1 and F(1, 8) = 5.63, p < .05 for the late negativity) in the absence of interactions. This was due to lower N1 and higher late negativity elicited by rising compared to falling standard stimuli (note the upward negativity displayed on Figs. 2 and 3). However, both effects did not exceed 0.2 μV on the average and were not revealed in the post-hoc comparisons (p > .05). The P2 amplitudes did not show any significant effects. Responses to the deviants demonstrated enhanced negativities peaking between 200 and 300 ms (Figs. 2 and 3, left panels, thick lines) which were analyzed after the calculation of the MMN components (see below).

3.2.2. The MMNs The difference waveforms obtained at the Fz recording site and the intervals of significance for MMN components are shown in right panels of Figs. 2 and 3. The schemes below the curves represent the configurations of the stimulus sequences. A configuration can be labeled as increment when the deviant stimulus has a higher level than the standard and as decrement when the deviant has a lower level than the standard. The combinations of gradual and step stimuli can be labeled increment when the speed of intensity increase in the deviant is higher than in the standard (STEP/grad, rising level) or when the speed of intensity decrease in the deviant is lower than in the standard (GRAD/ step, falling level). Note that the direction of level change (rise or fall) does not always coincide with the direction of deviance (increment, decrement). The most pronounced properties of the MMNs were, firstly, that the increment and decrement MMNs differed in latency, both in direct (Fig. 2) and reversed (Fig. 3) configurations: the increment MMNs peaked much earlier than the decrement ones so that their intervals of significance (p < .05 in the t-tests) did not overlap. The decrement MMNs’ onsets, determined as the start points of their intervals of significance, were delayed by 70–90 ms relative to onsets of the increment MMNs. Secondly, the highest MMN peak was elicited by the rising intensity (increment) deviant in the STEP/const configuration, all other MMNs being of similarly low magnitude. To explore the intervals of significance of the MMNs recorded at different sites, the individual amplitudes of ERPs to standards and 103

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Fig. 2. Grand-average ERPs to standards and deviants and difference waveforms recorded at the Fz site in the direct configurations. Thin and thick lines on the left panels show ERPs to standards and deviants, respectively. Note that the ERPs given on the left panels represent the responses used to calculate the “same-stimulus” MMNs shown on the right panels. The increment and decrement oddball configurations are displayed below each row. Combinations are labeled as increment when the deviant gets a higher intensity than the standard, or when intensity increases faster in the deviant than in the standard, or when intensity decreases slower in the deviant than in the standard. Combinations are labeled as decrement when the deviant gets a lower intensity than the standard, or when intensity decreases faster in the deviant than in the standard, or when intensity increases slower in the deviant than in the standard. Red lines indicate the rise of intensity; blue lines indicate the fall of intensity. Colored areas show the time intervals within which the MMN components were statistically significant (p < .05).

104

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Fig. 3. Grand-average ERPs to standards and deviants and difference waveforms recorded at the Fz site in the reversed configurations. The indications are the same as in Fig. 2.

4. Discussion

amplitude appears to be governed by the difference magnitude in terms of perceptual features as opposed to acoustic parameters (Tiitinen, May, Reinikainen, & Näätänen, 1994; Winkler, 2007; Winkler et al., 1997). The MMNs elicited in the role-reversal paradigm where the magnitude of acoustic deviance is kept the same, but the roles of standards and deviants change, indicate the perceptual effect of stimulus contexts on preattentive auditory processing. We investigated whether the response bias for rising intensity can be found in earlier processing indexed by

A vast body of evidence supports the notion that MMN amplitude represents standard-deviant acoustical differences (Näätänen et al., 2007). Minimal MMNs can be elicited by the acoustical differences approximately equal to the just noticeable difference of the given parameter (Amendo & Escera, 2000; Kraus, Koch, McGee, Nicol, & Cunningham, 1999; for a review, see Näätänen & Alho, 1997), and its 105

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Fig. 4. Mean MMN amplitudes recorded at different electrodes in the direct configurations and plotted as a function of time (X-axes, ms). Gradient scales indicate the statistically significant MMN amplitudes (t-tests, p < .05). Left column: increment sequences; right column: decrement sequences. Red and blue colors indicate the rise and fall of intensity. The bar plots next to each panel show the MMN amplitudes averaged across sites Fp1, AF3, F3, F7, FC1, FC5, C3, C5 in the left hemisphere (LH) and Fp2, AF4, F4, F8, FC2, FC6, C4, C6 in the right hemisphere (RH). Vertical lines indicate the standard errors. Significant interhemispheric difference is shown by the asterisk (p < .05).

The main finding for the behavioral measures was that the HRs and RTs were strongly affected by the direction of the level change within the stimuli, contrary to what we have found for the MMNs (see below). In the sequences with rising intensity, the deviants were discriminated more effectively than in the sequences with the falling intensity, regardless of the standard-deviant context. This is in line with the earlier data suggesting better discrimination for the rising intensity produced by looming sound sources (e.g., Bach et al., 2008; Hall & Moore, 2003;

MMN. We expected that 1) the behavioral responses (HRs and RTs) would be affected by direction of intensity change, 2) both increment and decrement MMNs would be easily separable from the N1 wave, and 3) the MMN responses would be affected by direction of deviance and/ or by level change. We found an interesting dissociation between earlier and later stages of processing: the MMN responses were mostly affected by direction of deviance, whereas behavioral performance depended on direction of the intensity change. 106

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Fig. 5. Mean MMN amplitudes recorded at different electrodes in the reversed configurations and plotted as a function of time (X-axes, ms). The indications are the same as in Fig. 4.

the abrupt change of intensity in the STEP signal entrained the activity of a wide range of neural populations. This finding propones the question of a possible shift in the brain location of different cortical neurons specialized for certain ranges of sound intensity. Available experimental evidence suggests that an amplitopic arrangement in human auditory cortex is apparently rather weak, if at all existing (see for review Uppenkamp & Röhl, 2014). It is more likely that the enhanced negativity in the ERP to the rising step standard (and also in the

Tajadura-Jiménez, Väljamäe, & Vastfjall, 2008). It is also worth noting that both behavioral indices showed similar properties in the combinations of step and gradual stimuli with constant ones, that is, the step and gradual stimuli seemed to belong to the same perceptual category. The ERPs elicited by all types of standards exhibited the N1 wave peaking at around 100 ms and the late negative deflections peaking at around 250 ms. Late negativities were higher in responses to rising as compared to constant and falling level standards. It seems plausible that 107

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Fig. 6. Scalp distributions of the MMNs taken at the time points indicated above the topograms. The direction of deviance is given in the column titles; the direction of the level change is indicated below each topogram. Note the difference in scaling. Negativity is depicted by blue and positivity by red color.

increments could be due to more complex time patterns of level change used in our study. The dynamic change of sound intensity provided a delay of deviation relative to the stimulus onset, similar to the delay of spatial change used in our earlier studies (Shestopalova et al., 2012), which resulted in a delay of the MMN response. Interestingly, the MMN latencies were strongly affected by the direction of deviance: the increment stimulus sequences elicited significantly earlier MMNs as compared to the decrement ones. The time delay of the decrement MMNs relative to increment ones was so large that their intervals of statistical significance did not overlap. It should be kept in mind that the increment configurations were formed by both rising and falling level stimuli, as is also the case with decrement ones. Thus, the onset of MMN most likely reflected the direction of deviance in the sequence but not the direction of level change within the stimuli. These findings show no agreement with those of Altmann et al. (2013) who reported no difference between MMN responses to approaching and receding sounds. It seems probable that the roving oddball paradigm used in Altmann’s experiment was entirely appropriate for detecting the deviations but was not adequate for extracting more complex information necessary to estimate the direction of deviance. The effect of the deviance direction on MMN latencies was similar at different recording sites (Figs. 4 and 5). The scalp distributions of the

STEP/const MMN) may be explained through activation of a larger cortical volume in the case of abrupt level increase. Since the acoustical features of the rising level stimuli had a significant effect on the ERPs, it could have confounded the difference waves, traditionally calculated by subtracting ERPs of standards and deviants presented within the same block. To avoid misestimating the MMN features, we used an alternative, “same-stimulus” method of calculation (e.g., Schröger, 2001, 2003;). MMN amplitudes were calculated by subtracting ERPs to identical stimuli presented as standard in one block and deviant in another block. The MMN components emerged substantially later than N1 deflections and peaked after 250 ms. Though it cannot be ruled out that the neural processes underlying the N1 component and MMN persist in the overlapping time intervals, the MMNs obtained in the current study could still be easily separated from the N1 wave. This is in contrast to the MMN study of Rinne et al. (2006) which employed stimuli of constant level differing by −3, −6, −9, +3, +6, or +9 dB, and reported overlapping N1 and MMN components elicited by intensity increments. The authors argue that the change-related response to infrequent intensity increments consists of enhanced N1 and MMN while the response to intensity decrements is dominated by the MMN response. Distinct time windows we found for the N1 and MMN to level 108

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Fig. 7. Mean MMN amplitudes and behavioral measures. A: Fz and Cz MMN amplitudes for rising and falling level stimuli. B: mean hit rates (HRs) and reaction times (RTs).

initiation of an involuntary attention switch to the potentially important changes in the auditory environment (Paavilainen, 2013, a review). These brief attention switches may be reflected in the positive P3a component following MMN. This is exactly what we have found with the abrupt increase of intensity (in the present work) or with the abrupt increase of angular velocity (Shestopalova et al., 2015) in the deviant sound. When the deviant parameters changed gradually within the same range, the MMNs were lower in amplitude and were not followed by prominent P3a deflections, indicating smaller deviance magnitude (Figs. 2 and 3). However, the strong difference in the MMNs to abrupt and gradual change of intensity was not paralleled by the behavioral indices. Though the performance rate was higher for the deviants with stepwise than with gradual level change, the difference in HRs was not as striking as in MMN amplitudes. Further, the direct preattentive discrimination of rising intensity patterns has shown that STEP/grad increment MMN was significant (though of a low magnitude), the behavioral performance being slightly above threshold (HR = 0.57). Significant MMN was produced by STEP/grad decrement combination as well, though the corresponding performance rate was very low (HR = 0.09). Thus, the sensitivity of the MMN-generating system to the dynamic changes within sound stimuli with the same onset/offset intensities could be higher than that of conscious discrimination. Similar properties of early cortical auditory processing were found earlier for moving sounds (Shestopalova et al., 2012). Interpreting the results of parallel EEG and psychophysical study needs to address the issue of interrelation between sound intensity and its perceptual measures. The perceptual correlate of intensity is loudness (Florentine, 2011). The transformation of sound intensity into perceived loudness is not straightforward, since loudness can be determined not only by intensity but also by the bandwidth of an acoustic signal, by its duration and modulations, etc. A number of fMRI studies indicate that the relationship between sound level and BOLD signal intensity is nearly linear, while the activated cortical volume grows in a nonlinear fashion with sound intensity (see for review Uppenkamp &

MMN responses demonstrated a tendency to the right-hemispheric dominance in the increment configurations. However, the MMN asymmetry was rather weak and did not correlate with the modulation type. Our finding of high HRs and short RTs for rising level stimuli (Fig. 7, lower panels) adds further evidence to the assumption that rising sound intensity may serve as an intrinsic, unconditioned warning cue (Bach et al., 2008), which might be revealed at preattentive level. However, the short latency of increment MMNs most likely reflected the temporal advantage in the processing of increasing deviance (but not of rising intensity) at earlier stages of auditory analysis. Further research is needed to clarify possible interrelation of the hypothesized warning signal with preattentive processes related to deviance detection. The significant effect of context reversals was revealed with one configuration only: the increment STEP/const MMN (rising STEP deviants) was much higher than the increment CONST/step MMN (falling step standards). Though physical deviant-standard differences were the same, and the direction of deviance was increment in both direct and reversed sequence, the large magnitude of the STEP/const MMN suggests that the rising STEP stimulus was the most salient deviant when presented in the context of the constant standards. For other configurations, the magnitude of deviance was likely to be too low for the effect of the role reversal to become apparent. The pattern of the deviant level change affected the amplitudes of the increment MMNs: STEP/const MMN was higher than GRAD/const MMN. The unique feature of the stepwise change of sound level is that it represents an infinitely large acoustic change per time. Its auditory processing may therefore be compared with processing of abrupt shift in sound location (Shestopalova et al., 2015). The proportion between STEP/const MMN and GRAD/const MMNs elicited by moving sounds was similar to that obtained in the current study, suggesting that the effect of the time pattern of acoustic change on MMN amplitude reflects a more general property of preattentive auditory processing, rather than a specific feature of a certain stimulus parameter. MMN generation in the traditional passive oddball conditions supposedly reflects the 109

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

apparently reflects high-level perceptual processes with longer timescales. This could be the reason why behavioral measures estimate the onset/offset values of stimulus parameters in a categorical manner, whereas the preattentive indices are more sensitive to dynamic properties of the stimuli, aside from onset/offset information.

Röhl, 2014). The neural activity in auditory cortex appears to be a direct linear reflection of subjective loudness sensation, rather than a display of physical sound pressure level. For the intensity range of 80 ± 20 dB, a high correlation was found between the sound level dependent change of the extent of fMRI activation measured as number of activated voxels (but not of the activation magnitude) and the corresponding changes of the mean current source density within the same region of the primary auditory cortex (Mulert et al., 2005). However, the experiments which investigated the transformation of the sensory coding of acoustic stimuli in the auditory system into the actual perception of auditory events involved constant level stimuli in a wide range of intensities. Conversely, our study was focused at the discrimination of sounds with dynamic level changes with the overall level difference of ± 5 dB used in all sequences. The behavioral responses to the rising and falling sounds are known to show an asymmetry related to their different biological salience. It cannot be ruled out that this asymmetry may interfere to some extent with non-linearity of brain representation of the perceived loudness, but it is most unlikely that the latter effect were substantial in the narrow intensity range used in the current experiment. Taken together, our data imply that the preattentive discrimination mostly depended on the direction of deviance in the sequence, whereas behavioral performance depended on the direction of the intensity change within the stimuli. The MMN bias for increment combinations could be considered as a phenomenon driven by sequence structure as opposed to the effect of stimulus parameters exhibited by behavioral measures. Furthermore, the present findings together with our earlier data on moving sounds (Shestopalova et al., 2015) show that early cortical mechanisms could detect the standard-deviant difference even when it was not possible for the later conscious processes. Another example of standard-deviant configurations that are difficult to consciously discriminate can be implemented in complex paradigms based on high-order invariances. MMN elicited by deviant stimuli violating an abstract rule is referred to as “higher-order” or “abstract-feature” MMN (in contrast to “first-order” or “simple” MMN produced by traditional oddball sequences). Elicitation of higher-order MMN which reflects automatic encoding abstract attributes corresponding to sequence structure favors the functioning of preattentive perceptual–cognitive processing termed “automatic auditory intelligence”. Supposedly, its biological significance consists in causing involuntary attention switch to potentially important events. This attention-switching mechanism may facilitate rapid conscious evaluation of significance of the environmental change and a prompt behavioral response to it (reviewed in Näätänen, Astikainen, Ruusivirta, & Huotilainen, 2010; Näätänen, Kujala, & Winkler, 2011; Paavilainen, 2013). The qualitative discrepancy which we have found between the properties of MMN response and conscious deviance detection may serve an indirect illustration of dissociation between functioning of the “brain” and “mind”. Basically, the preconscious neural representations of sound changes are not necessarily read at higher levels of sound processing. The behavioral indices may be susceptible to a variety of factors and may not always be sensitive to effects at earlier stages of processing that precede attentional switch, motor preparation, and execution (Hopfinger & Parks, 2012; Paavilainen, 2013). The present study has shown high sensitivity of MMN-generating process to the sequence structure and to the velocity of level change within the stimulus as well. On the contrary, the mechanism of conscious discrimination seemed to notably discard this information and to subsume the incoming stimuli under larger categories of “rising” and “falling” sounds. The categorical perception likely enables the auditory processing system to allocate the available resources towards potentially hazardous approaching objects. It is also likely that the early and later discrimination mechanisms operate on different timescales. Preattentive processing relies on integration of sensory information within a relatively narrow time window corresponding to faster timescale. Active discrimination

5. Conclusion The MMN-generating system demonstrated its high sensitivity to the dynamic changes within rising and falling level sounds with the same onset/offset intensities. We found that the MMN could be elicited by intensity changes even when the deviants were not discriminated behaviorally. This finding provides further support to the view that preattentive deviance discrimination can be more effective than conscious detection. More importantly, our data indicate that the preattentive neural mechanisms were mostly driven by the direction of deviance (i.e. by the sequence structure), whereas behavioral performance depended on the direction of the intensity change within the stimuli. This dissociation between earlier and later stages of processing may stem from possible timescale differences and also from considerable biological salience of rising sound intensity. The information about the sequence structure seems to be discarded by mechanisms of conscious discrimination which is intended to allocate the attentional resources towards potentially hazardous events; therefore the perceptual priority which we have found for rising intensity was not reflected in the MMN data. Acknowledgements We wish to express our gratitude to the anonymous reviewers for their valuable comments and constructive critique. This study was supported by Russian Academy of Sciences (Program I.26П for the fundamental research) and by Saint-Petersburg Committee on Science and Higher Education (personal grant allocated to Lidia B. Shestopalova). The authors declare no conflict of interest. References Alho, K., Rinne, T., Herron, T. J., & Woods, D. L. (2014). Stimulus-dependent activations and attention-related modulations in the auditory cortex: A meta-analysis of fMRI studies. Hearing Research, 307, 29–41. http://dx.doi.org/10.1016/j.heares.2013.08. 001. Altmann, C. F., Hiraumi, H., Terada, S., Adachi, T., Votinov, M., Ono, K., et al. (2013). Preattentive processing of horizontal motion, radial motion, and intensity changes of sounds. Neuroreport, 24, 861–865. http://dx.doi.org/10.1097/WNR. 0000000000000006. Bach, D. R., Schächinger, H., Neuhoff, J. G., Esposito, F., Di Salle, F., Lehmann, C., et al. (2008). Rising sound intensity: An intrinsic warning cue activating the amygdala. Cerebral Cortex, 18, 145–150. http://dx.doi.org/10.1093/cercor/bhm040. Beagley, H. A., & Knight, J. J. (1967). Changes in auditory evoked response with intensity. Journal of Laryngology and Otology, 81, 861–873. Blackwell, H. R. (1953). Psychophysical thresholds: Experimental studies of method of measurement. University of Michigan: Engineering Research Institute Bulletin №36. Bronkhorst, A. W., & Houtgast, T. (1999). Auditory distance perception in rooms. Nature, 397, 517–520. Coleman, P. D. (1963). An analysis of cues to auditory depth perception in free space. Psychological Bulletin, 60, 302–315. Coleman, P. D. (1968). Dual role of frequency spectrum in determination of auditory distance. The Journal of the Acoustical Society of America, 44, 631–632. Colin, C., Hoonhorst, I., Markessis, E., Radeau, M., de Tourtchaninoff, M., Foucher, A., et al. (2009). Mismatch Negativity (MMN) evoked by sound duration contrasts: An unexpected major effect of deviance direction on amplitudes. Clinical Neurophysiology, 120, 51–59. http://dx.doi.org/10.1016/j.clinph.2008.10.002. Florentine, M. (2011). Loudness. In M. Florentine, A. N. Popper, & R. R. Fay (Vol. Eds.), Springer handb. auditory res: Vol. 37. New York: Springer. Ghazanfar, A. A., Neuhoff, J. G., & Logothetis, N. K. (2002). Auditory looming perception in rhesus monkeys. Proceedings of the National Academy of Sciences of the United States of America, 99, 15755–15757. Hall, D. A., & Moore, D. R. (2003). Auditory Neuroscience: The Salience of Looming Sounds. Current Biology, 13, R91–R93. Hopfinger, J. B., & Parks, E. L. (2012). Involuntary attention. In G. R. Mangun (Ed.). The neuroscience of attention: Attentional control and selection (pp. 30–53). New York: Oxford University Press. Horváth, J., Winkler, I., & Bendixen, A. (2008). Do N1/MMN, P3a, and RON form a

110

Biological Psychology 133 (2018) 99–111

L.B. Shestopalova et al.

Neuroscience Letters, 482, 71–75. http://dx.doi.org/10.1016/j.neulet.2010.07.010. Picton, T. W., Goodman, W. S., & Bryce, D. P. (1970). Amplitude of evoked responses to tones of high intensity. Acta Otolaryngologica, 79, 77–82. Rapin, I., Schimmel, H., Tourk, L. M., Krasnegor, L. M., & Pollack, C. (1966). Evoked potentials to clicks and tones of varying intensity of waking adults. Electroencephalography and Clinical Neurophysiology, 21, 335–344. Rinne, T., Särkkä, A., Degerman, A., Schröger, E., & Alho, K. (2006). Two separate mechanisms underlie auditory change detection and involuntary control of attention. Brain Research, 1077, 135–143. http://dx.doi.org/10.1016/j.brainres.2006.01.043. Rosenblum, L. D., Carello, C., & Pastore, R. E. (1987). Relative effectiveness of three stimulus variables for locating a moving sound source. Perception, 16, 175–186. Schiff, W., & Oldak, R. (1990). Accuracy of judging time to arrival: Effects of modality, trajectory and gender. The Journal of Experimental Psychology: Human Perception and Performance, 16, 303–316. Seifritz, E., Neuhoff, J. G., Bilecen, D., Scheffler, K., Mustovic, H., Schächinger, H., et al. (2002). Neural processing of auditory looming in the human brain. Current Biology, 12, 2147–2151. Shestopalova, L. B., Petropavlovskaia, E. A., Vaitulevich, S. P., Vasilenko, Y. A., Nikitin, N. I., & Altman, J. A. (2012). Discrimination of auditory motion patterns: Mismatch negativity study. Neuropsychologia, 50, 2720–2729. Shestopalova, L. B., Petropavlovskaia, E. A., Vaitulevich, S. P., & Nikitin, N. I. (2015). Contextual effects on preattentive processing of sound motion as revealed by spatial MMN. International Journal of Psychophysiology, 96, 49–56. http://dx.doi.org/10. 1016/j.ijpsycho.2015.02.021. Stecker, G. C., & Hafter, E. R. (2000). An effect of temporal asymmetry on loudness. The Journal of the Acoustical Society of America, 107, 3358–3368. Strybel, T. Z., & Perrot, D. R. (1984). Discrimination of relative distance in the auditory modality. The Journal of the Acoustical Society of America, 76, 316–320. Tajadura-Jiménez, A., Väljamäe, A., & Vastfjall, D. (2008). Emotional bias for the perception of rising tones. The Journal of the Acoustical Society of America, 123(5), 1597–1602. http://dx.doi.org/10.1121/1.2933507. Takegata, R., Tervaniemi, M., Alku, P., Ylinen, S., & Näätänen, R. (2008). Parameterspecific modulation of the mismatch negativity to duration decrement and increment: Evidence for asymmetric processes. Clinical Neurophysiology, 119, 1515–1523. http:// dx.doi.org/10.1016/j.clinph.2008.03.025. Uppenkamp, S., & Röhl, M. (2014). Human auditory neuroimaging of intensity and loudness. Hearing Research, 307, 65–73. Winkler, I. (2007). Interpreting the mismatch negativity. Journal of Psychophysiology, 21(3–4), 147–163. Zahorik, P. (2005). Auditory distance perception in humans: A summary of past and present research. Acta Acustica, 91, 409–420.

strongly coupled chain reflecting the three stages of auditory distraction? Biological Psychology, 79, 139–147. Jacobsen, T., & Schröger, E. (2001). Is there pre-attentive memory-based comparison of pitch? Psychophysiology, 38, 723–727. Jacobsen, T., & Schröger, E. (2003). Measuring duration mismatch negativity. Clinical Neurophysiology, 114, 1133–1143. Jacobsen, T., Horenkamp, T., & Schröger, E. (2003). Preattentive memory-based comparison of sound intensity. Audiology and Neurootology, 8, 338–346. Karanasiou, I. S., Papageorgiou, C., Kyprianou, M., Tsianaka, E. I., Matsopoulos, G. K., Ventouras, E. M., et al. (2011). Effect of frequency deviance direction on performance and mismatch negativity. Journal of Integrative Neuroscience, 10(4), 525–536. Lu, T., Liang, L., & Wang, X. (2001). Neural representations of temporally asymmetric stimuli in the auditory cortex of awake primates. Journal of Neurophysiology, 85, 2364–2380. Mulert, C., Jäger, L., Propp, S., Karch, S., Störmann, S., Pogarell, O., et al. (2005). Sound level dependence of the primary auditory cortex: Simultaneous measurement with 61-channel EEG and fMRI. NeuroImage, 28, 49–58. Näätänen, R. (1992). Attention and Brain Function. Hillsdale, NJ: Lawrence Erlbaum Associates. Näätänen, R., & Picton, T. W. (1987). The N1 wave of the human electric and magnetic response to sound: A review and an analysis of the component structure. Psychophysiology, 24, 375–425. Näätänen, R., Paavilainen, P., Rinne, T., & Alho, K. (2007). The mismatch negativity (MMN) in basic research of central auditory processing: A review. Clinical Neurophysiology, 118, 2544–2590. Näätänen, R., Astikainen, P., Ruusivirta, T., & Huotilainen, M. (2010). Automatic auditory intelligence: An expression of the sensory–cognitive core of cognitive processes. Brain Research Reviews, 64, 123–136. http://dx.doi.org/10.1016/j.brainresrev.2010.03. 001. Näätänen, R., Kujala, T., & Winkler, I. (2011). Auditory processing that leads to conscious perception: A unique window to central auditory processing opened by the mismatch negativity and related responses. Psychophysiology, 48, 4–22. http://dx.doi.org/10. 1111/j.1469-8986.2010.01114.x. Neuhoff, J. G. (1998). Perceptual bias for rising tones. Nature, 395, 123–124. Okazaki, S., Kanoh, S. I., Takaura, K., Tsukada, M., & Oka, K. (2006). Change detection and difference detection of tone duration discrimination. Neuroreport, 17, 395–399. Paavilainen, P. (2013). The mismatch-negativity (MMN) component of the auditory event-related potential to violations of abstract regularities: A review. International Journal of Psychophysiology, 88, 109–123. http://dx.doi.org/10.1016/j.ijpsycho. 2013.03.015. Peter, V., McArthur, G., & Thompson, W. F. (2010). Effect of deviance direction and calculation method on duration and frequency mismatch negativity (MMN).

111