Dichotic listening (DL) is a neuropsychological technique for the study of functional laterality. Based on behavioral patient studies, the “structural theory” states that lateralization of the auditory input during DL is allowed by an inhibition of the ipsilateral pathways. We aimed here at extending this theory to provide a neurophysiological basis of verbal DL. We investigated the magnetic responses of the primary auditory cortices elicited by dichotic consonant–vowel syllables. Dichotic stimuli consisted of 2 syllables pairs, a “competing” one composed by syllables with high spectral overlap (/da/ and /ba/) and a “noncompeting” pair (/da/ and /ka/). One of the syllables in each pair was delivered at 2 intensities, whereas the other did not change. A reduced increase of source intensity in response to dichotic pairs at the 2 levels was assumed to indicate pathway inhibition effects. We obtained that the left ipsilateral pathway (i.e., the left ipsilateral signal) was strongly inhibited by the right contralateral one. Conversely, the right ipsilateral pathway did not show an inhibition larger than the left contralateral one. These results extend the notion of auditory functional asymmetries by showing that beyond hemispheric functional specialization there is an asymmetry within the ascending auditory system, which is based on a competition mechanism. The larger the competition between the left and right ear stimuli, the larger are the inhibition effects, which determine the pathway asymmetry. These findings represent as well a neurophysiological basis for the “structural theory” explaining the right ear preference usually found in verbal DL tasks.
Dichotic listening (DL) is a neuropsychological technique for the study of functional laterality, which allows testing separately left and right auditory cortices (Bryden 1988). It consists in presenting to the subject 2 different simultaneous auditory stimuli to either ear. Being a completely noninvasive study method, DL has been broadly used in the investigation of hemispherical asymmetries. The main findings of DL studies are that, in general, subjects with left-hemispheric language lateralization are faster and more accurate in reporting verbal items presented at the right ear (Kimura 1961). Conversely, they exhibit a left ear advantage for tasks involving the recognition of musical or environmental sounds (Boucher and Bryden 1997; Brancucci and San Martini 1999, 2003).
Although earlier studies questioned the use of DL for the evaluation of cerebral laterality on single individuals (Teng 1981; Jancke et al. 1992), in the last years there was a growing interest on this topic. Functional neuroimaging studies of regional cerebral blood flow have elucidated spatial details of brain structures involved in DL, that is, bilateral primary auditory areas (Hugdahl et al. 1999, 2000; Lipschutz et al. 2002; Jäncke et al. 2003), orbitofrontal and hippocampal paralimbic belts (Pollmann et al. 2004), frontotemporal areas with particular importance of the connection between frontal cortex and planum temporale (Jancke et al. 2001; Jancke and Shah 2002), prefrontal cortex (Lipschutz et al. 2002; Thomsen et al. 2004), and splenium of the corpus callosum (Pollmann et al. 2002). These findings indicate that DL is a complex function that requires processing capacities involving a largely distributed neural network.
However, in spite of the amount of empirical and theoretical studies concerning DL topic, the neurophysiological bases of this technique remain somewhat unclear, in particular with respect to the mechanisms which allow the lateralization of verbal auditory inputs. It is known that the monaural input to each ear is represented in both cerebral hemispheres, with an advantage for contralateral over ipsilateral pathways in both latency and strength of the response (Hall and Goldstein 1968; Fujiki et al. 2002). Specifically, auditory evoked potentials or magnetic fields (AEFs) have shown that contralateral response to monaural stimulation starts earlier and is more sustained than the ipsilateral one (Reite et al. 1981; Romani et al. 1982; Papanicolaou et al. 1990). However, when different stimuli are presented dichotically, interactions between auditory pathways complicate the understanding of the neural response. According to the “structural theory” proposed by Kimura (1967), during DL the ipsilateral pathways are inhibited by the contralateral ones. This causes that the input from the right ear properly reaches the left but not the right auditory cortex. Vice versa, the input from the left ear reaches properly the right but not the left auditory cortex. This mechanism would be at the basis of the ear advantage effects usually observed in DL tasks. However, the input to one ear can reach, with some degree of depletion, the ipsilateral areas via the contralateral route passing first through the contralateral auditory cortex and then crossing the midline via corpus callosum (Pollmann et al. 2002). Of note, attentional mechanisms linked to priming effect of attending to a particular type of auditory stimulus play here also a role (Kinsbourne 1970; Hugdahl et al. 2000; Jäncke et al. 2003).
Evidence in favor of the “structural theory” came from studies testing split-brain patients with DL tasks (Milner et al. 1968; Sparks and Geschwind 1968; Springer and Gazzaniga 1975). These patients had no difficulty in reporting words or consonants–vowel (CV) syllables presented monaurally to either ear. But, when the same stimuli were presented dichotically, they failed to report items presented to the left ear. Differently from normal subjects, the lesion of the corpus callosum prevents the communication of the left ear with the left hemisphere via the indirect contralateral route passing through the right auditory cortex and crossing the callosal pathway. Thus, these patients show that when the sole route connecting left ear and left auditory cortex is the ipsilateral one, the input to the left ear can not be perceived if another stimulus is simultaneously (dichotically) presented to the right ear. This fact points to a suppression of the ipsilateral left pathway during DL. More recently, the “structural theory” has been supported by some evidence based on neuroimaging findings. A positron emission tomography study demonstrated that verbal dichotic stimuli induce stronger cortical responses in the left temporal lobe, whereas nonverbal dichotic stimuli induce a stronger activity in the right temporal lobe (Hugdahl et al. 1999). An event-related potentials study showed that the electrical activity in the supratemporal cortical plane begins earlier in the left compared with right hemisphere and its latency correlates with the ear advantage of the subjects (Eichele et al. 2005). Furthermore, a magnetoencephalography (MEG) study with complex tones demonstrated that responses to ipsilateral stimuli over the right auditory cortex are inhibited by the stimulus presented in the contralateral (left) ear (Brancucci et al. 2004).
At the present stage little is known about the interactions between ipsi- and contralateral auditory pathways during DL of speech sounds, which allows the lateralization of auditory input, as postulated by the “structural theory” of Kimura (1967). In this view, we used MEG to study signals arising from the auditory cortex (Hari 1990) in order to extend previous findings 1) to speech sounds, by using DL of CV-syllables, 2) to both right and left auditory cortical sources, and 3) to both ipsilateral and contralateral pathways. The working hypothesis is based on the fact that the amplitude of the cortical response is directly related to sound intensity (Hegerl et al. 1994), that is, sounds with higher intensities elicit stronger cortical responses. Indeed, the present experimental design is based on comparisons between cortical responses to repetitive dichotic stimulations in which one stimulus in the dichotic pair has constant intensity, whereas the other stimulus is delivered at 2 intensities. We assume that an inhibition of one auditory pathway is revealed by a reduced amplitude increment in cortical responses elicited by the pair containing the sound at higher intensity. A second assumption is that the interaction between ipsilateral and contralateral pathways (i.e., the ipsilateral and contralateral neural signals) is related to the spectral overlap of the stimuli constituting the dichotic pair (Springer et al. 1978; Sidtis 1981; Brancucci et al. 2004). We used CV-syllables having different spectral profile with the intent to reveal possible inhibition of the ipsilateral pathways and possible hemispheric asymmetries in the interactions between the auditory pathways.
Materials and Methods
Ten healthy subjects (7 females) aged between 20 and 31 years (mean age = 25) took part in the experiment. They were right handed (Edinburgh Inventory). None of them had auditory impairments as shown by auditory functional assessment (absolute hearing threshold < 20 dB). No differences (±5 dB) of hearing threshold were found between left and right ear in all subjects. All the experiments to be described in the following were undertaken with the understanding and written consent of each subject.
In a separate session, subjects underwent a behavioral verbal DL task. This task consisted of 60 items, which were continuously generated by a computer, with 3-s interitem interval. Subjects were provided with an earphone, a pencil, and a grid printed on a sheet of paper. He/she comfortably sat in front of the computer. After 30 items the positions of the earphones were reversed between the left and right ears, to avoid any bias due to the output channels. Each item consisted of a dichotic pair composed by 2 of the CV-syllables /ba/, /ka/, /da/, /ga/, /pa/, /ta/. The task of the subject was to indicate on the grid which CV-syllable he perceived at best, among the ones listed above. Subjects were asked to pay attention at both ears simultaneously, that is, without privileging one ear (Hugdahl et al. 2001). Data analysis was based on the number of correct reported syllables which were presented at the left versus right ear. That is, when the subject reported a syllable, which was actually present in the dichotic pair, a point was ascribed to the left (right) ear if the reported syllable was presented at the left (right) ear (Eichele et al. 2005). One-way analysis of variance (ANOVA) with ear of input (left, right) as factor was carried out on ear scores (dependent variable).
Stimulation for MEG Recordings
Dichotic stimuli consisted of 3 CV-syllables (/da/, /ba/, /ka/) recorded from a natural female voce. The intensity of the stimuli was adjusted at 2 levels, that is, 60 and 80 dBA, respectively. Stimulus recording and handling was performed by using the software “Wave 2.0” (Voyetra Turtle Beach Systems, Yonkers, NY) for Microsoft Windows on a PC Pentium III 550 MHz with audio card Sound Blaster AWE 32. The CV-syllables were recorded and handled with a sampling rate of 44 100 Hz and an amplitude resolution of 16 bit. Waveforms of the CV-syllables are plotted in Figure 1(A). Verification of the actual sound intensity was done by means of a Phonometer (Geass, Delta Ohm HD2110, Torino, Italy). The normalized power spectrum densities of the consonants forming the CV-syllables at 60 dBA are shown in Figure 1(B) (dashed black line). Only the time points during which the consonants were provided were used to compute the power spectra. These were based on a time window of 1024 points filtered by a Hamming window. A total of 5 CV-syllables were used to make up the stimuli (/da/ 60 dBA, /ba/ 60 dBA, /ka/ 60 dBA, /ba/ 80 dBA, /ka/ 80 dBA). These were arranged in 8 dichotic stimuli containing each the CV-syllable /da/ always at 60 dBA and the CV-syllable /ba/ or /ka/ at 60 or 80 dBA (see Fig. 1C). In this way, 2 competing CV-syllables (/da/ 60 dBA + /ba/ 60 dBA and /da/ 60 dBA + /ba/ 80 dBA) having high spectral overlap and 2 noncompeting CV-syllables (/da/ 60 dBA + /ka/ 60 dBA and /da/ 60 dBA + /ka/ 80 dBA) having low spectral overlap (control condition) were obtained. The amount of spectral overlap between the consonants of the CV-syllables forming the dichotic stimuli was estimated by computing the Euclidean distance between the consonant spectra. We obtained a spectral overlap of about 93% when comparing /d/ versus /b/ and about 23% for /d/ versus /k/.
Each of the 8 stimuli was presented 80 times for a total of 640 presentations. The stimuli were arranged in a randomized sequence, which was segmented in 2 blocks of 320 stimuli. The interstimulus interval varied randomly between 2500 and 3500 ms.
The AEFs were recorded by using the 165 channel MEG system (Della Penna et al. 2000) installed at the University of Chieti inside a high-quality magnetically shielded room. The system consists of 153 dc SQUID integrated magnetometers arranged on a helmet surface covering the whole head and 12 reference channels. The acoustic stimulation was provided by Sensorcom plastic ear tubes connected to a transducer which was placed inside a μ-metal box to avoid any artifact on the recordings. To get an estimate of how the transducer and the plastic ear tubes affected the acoustic stimuli we recorded the /ba/, /ka/ and /da/ syllables by means of a piezoelectric sensor connected to a PowerLab acquisition system (PowerLab 16/30, ADInstruments, Inc., Colorado Springs, CO). For each syllable, we then computed the Euclidean distance between the normalized power spectra of the consonants generated by the computer (before the transducer and plastic tubes) and the consonants recorded by the piezoelectric sensor. Both spectra are shown in Figure 1(B). The overlap between the generated and recorded consonants was 96% for /d/, 97% for /b/, and 96% for /k/. Of note, this estimate is affected by the measurement procedure (i.e., the coupling between the ear tube and piezoelectric sensor), thus we assume that the overlap is even larger than the above values. During the magnetic recordings, the subjects passively listened to the syllable sequences and did not perform any task. They were asked to pay attention at both ears simultaneously, that is, without privileging one ear. Two electrical channels acquired the EKG bioelectric signals simultaneously with magnetic recordings, to be used as a reference for the rejection of the heart artifact. The magnetic signals from the brain and EKG signals were band-pass filtered at 0.16–250 Hz and recorded at 1-kHz sampling rate.
The position of the subject's head with respect to the sensor was determined by a fit of the magnetic field generated by 4 coils placed on the scalp recorded before and after each sequence recording. The coil positions on the subject's scalp were digitized by means of a 3D digitizer (Polhemus, 3Space Fastrak), together with anatomical landmarks defining a coordinate system. A set of high resolution magnetic resonance images (MRIs) of subjects' heads were obtained by a Siemens Magnetom Vision 1.5 Tesla using an MPRAGE sequence (256 × 256, field of view 256, time repetition = 9.7 ms, time echo = 4 ms, flip angle 128, voxel size 1 mm3). To coregister MEG functional data with MRI anatomical images, spherical oil capsules were applied on the anatomical landmarks during the MPRAGE acquisition.
Dipole Source Analysis
The magnetic recordings were preprocessed to subtract the heart signal and to remove noisy trials. For each dichotic pair, about 70 artifact-free AEF trials were band-pass filtered (1–70 Hz) and averaged in a time period of 600 ms, including 50 ms prestimulus. For each channel, a baseline level was computed as the mean value of the magnetic field in the time interval [−10, +10] ms across the stimulus onset. Equivalent Current Dipoles (ECDs) representing the source activities were localized inside a homogeneous sphere modeling the head using multiple source analysis provided by the BESA software. The fitting interval was [70, 360] ms poststimulus. In this time interval, the position, orientation and amplitude of 2 free ECDs were fitted to model the sources in the primary auditory cortices using the dichotic pair comprising noncompeting syllables /da/ 60 dBA–/ka/ 80 dBA because the signal to noise ratio was higher in this condition. No regional constraints were applied to the fit of the 2 ECDs. For all the other dichotic conditions the locations of the auditory ECDs found in the competing condition /da/ 60 dBA–/ka/ 80 dBA were held fixed during the fit. For each subject, the ECD locations were checked on the MRI with the Brain-Voyager software. The choice to study the responses of the primary auditory cortex was based on the fact that this is the area in the cerebral cortex reached at first by auditory signals (Reite et al. 1981, 1988), it has been shown that it is involved in DL (Hugdahl et al. 1999, 2000; Lipschutz et al. 2002; Jäncke et al. 2003) and it is the auditory cortical area which has the lower degree of anatomical and functional asymmetries in the 2 hemispheres (Bogolepova and Belogrud' 2005; Dorsaint-Pierre et al. 2006; Firszt et al. 2006). A goodness of fit of 80% was the lower threshold to accept an ECD configuration. Differences between the ECD peak values measured at the corresponding dichotic pair of 80 and 60 dBA were computed to study the incremental response of the source strength to competing or noncompeting stimuli presented at increased intensities. As a matter of fact, it has been shown that the amplitude of the cortical response is directly related to sound intensity during monotic, dichotic and binaural verbal and nonverbal stimulations (Hegerl et al. 1994; Jancke et al. 1998; Brancucci et al. 2004).
For each source in the auditory cortices, we analyzed possible differences in the response growth (generated by CV-syllables of different intensities) in relation to the side of presentation of the increasing CV-syllable (ipsilateral or contralateral) and to the spectral content (competing or noncompeting). In the above comparisons, a reduced increase in the source strength was assumed to indicate a source inhibition. For each ECD in the left and right primary auditory cortices, 2-way repeated-measures ANOVA analysis was carried out to reveal possible different inhibition effects. Of note, the relatively small number of subjects (n = 10) prevented us to calculate a 3-way ANOVA including ECD in the left and right auditory cortex as a factor. For each statistical test, the dependent variable was the amplitude increase of the source, measured as the difference of the ECD peak amplitudes elicited by 2 dichotic pairs in which one CV-syllable was presented at 2 different intensity levels (60 vs. 80 dBA). We studied the amplitude increase instead of the amplitude to account for possible intersubject variability of the evoked response. One factor of the ANOVA analysis was the side of presentation of the increasing CV-syllable (ipsilateral or contralateral) and the other factor was the spectral content of the increasing CV-syllable (competing or noncompeting). Tukey's post hoc test was applied to the main effects. Mauchley's test evaluated the sphericity assumption. Correction of the degrees of freedom was made by the Greenhouse–Geisser procedure.
Individual laterality index (LI) scores are reported in Table 1. All subjects showed a right ear advantage in the verbal DL task. LI was computed as follows: LI = (R − L)/(R + L) × 100, where R is the number of correct reports of the right ear and L the number of correct reports of the left ear. Mean (±standard error) LI was 18.6 ± 2.9. LI ranged from 11.1 to 30.0. ANOVA indicated a statistically significant effect in favor of the right ear (F = 37.58, P < 0.001). This result indicates that the subjects perceived preferentially the syllable presented at the right ear, which, in DL conditions, sends its inputs mainly to the left hemisphere. However, it should be noted that the input to the left ear is sometimes correctly perceived by the subjects. This is due to the fact that the left ear input, whose (ipsilateral) route to the left hemisphere is inhibited during DL, can still reach, although depleted, left auditory speech areas via the contralateral route passing first through the right auditory cortex and then crossing the midline via corpus callosum (Pollmann et al. 2002).
Auditory Evoked Fields
For each stimulus condition, the average magnetic field showed a dipolar pattern over the bilateral auditory area at 2 peak latencies, the first being 130 ± 20 ms and the second 300 ± 50 ms after the stimulus onset. The field distribution and the isofield contour maps are shown in the upper part of Figure 2 for a representative subject. One ECD in each hemisphere was localized in the auditory cortex, as shown in the lower part of Figure 2 displaying the ECDs superimposed on the MRI of a representative subject's head. The average across all subjects of the ECD locations expressed in Talairach coordinates (Talairach and Tournoux 1988) resulted to be x = −52 ± 6 mm, y = −21 ± 7 mm, z = 7 ± 3 mm for the left auditory ECD, x = 48 ± 9 mm, y = −18 ± 7 mm, z = 7 ± 5 mm for the right auditory ECD. These locations correspond to the Heschl's gyrus (Brodmann area 41). In Figure 3, the average across subjects of the ECD waveforms are shown for the left (upper part of the figure) and the right (lower part) hemispheres. For each source and condition, 2 peaks can be detected in the waveform, the former at about 130 ms and the latter at about 300 ms from the stimulus onset. Comparing the control dichotic noncompeting stimuli (/da/ 60 dBA + /ka/ 60 dBA vs. /da/ 60 dBA + /ka/ 80 dBA) at the second peak latency, we observe that the peak intensities of both the right and left auditory sources increase, in spite of the presentation side of the increasing stimulus. Conversely, if we compare competing stimuli (/da/ 60 dBA + /ba/ 60 dBA vs. /da/ 60 dBA + /ba/ 80 dBA), a clearly larger inhibition effect (reduced increase) can be observed for the left auditory cortex when the increasing stimulus is ipsilateral.
For left and right ECDs in the primary auditory cortices at each peak latency (130 and 300 ms), the increase of source intensity following an increase of one of the syllables in the dichotic stimulus was separately studied by 2-ways ANOVA analysis. We aimed at evaluating the effect of the presentation side of the increasing CV-syllable and the effect of the stimulus spectral content (competing /da/ + /ba/ or noncompeting /da/ + /ka/) on possible ipsi/contra inhibitory pathway interactions. The averages across subjects of the strength increase for the right and left dipoles at both the first and the second peak latencies are summarized in Figure 4 for ipsilateral and contralateral presentation sides of the increasing CV-syllables and for competing or noncompeting stimuli.
No statistically significant effects, reflecting possible different inhibitions, were obtained at the first peak latency for both the left and right primary auditory cortices. Conversely, for the left source at the second peak latency a significant (F1,9 = 34.5; mean squared effect (Mse) = 674; P = 0.0002) main effect for the spectral content of the stimulus was found, indicating that dichotic competing stimuli (/da/ + /ba/) determined a stronger inhibition (only 3 nA m increase on average) than the noncompeting (/da/ + /ka/) stimuli (average increase 11 nA m) as measured by the reduced increase of the left source peak amplitude. We estimated a 100% statistical power associated with the significance of the spectral content, indicating that the sample size was adequate to disclose this effect. A significant interaction (F1,9 = 11.7; Mse = 95.5; P = 0.0077) was found between the 2 factors of the statistical analysis, namely side of presentation and spectral content, suggesting that the left auditory source was differently inhibited when the increasing competing or noncompeting CV-syllables were provided at the left (ipsilateral) or at the right (contralateral) ear. Specifically, Tukey post hoc analysis indicated that for competing stimuli the inhibition of the source increase in the left primary cortex was significantly stronger (P = 0.006) when the increasing CV-syllable was delivered at the ipsilateral ear (mean increase of the source amplitude 0.6 nA·m) than at the contralateral one (mean increase 5.2 nA·m). Conversely, no significantly different inhibitions were found when the increasing noncompeting CV-syllables were delivered at the left or right ear (ipsilateral mean increase 11.9 vs. 10.3 nA·m). Finally, the left auditory cortex was always significantly more inhibited (Pmax = 0.003) when the stimuli were competing than noncompeting, reflecting the ANOVA main effect. The statistical power associated with the significance of the interaction between the presentation side and the spectral content was estimated to be 89%, indicating that the sample size was adequate.
Always at the second peak latency, a significant effect associated with the spectral content of the stimulus (F1,9 = 29; Mse = 767; P = 0.0004, estimated statistical power 100%) was found also for the right primary auditory cortex, indicating that the noncompeting dichotic stimulus (/da/ + /ka/) determined a smaller inhibition (average amplitude increase 9.5 nA m) than the competing stimulus (/da/ + /ba/) (average amplitude increase 0.7 nA m). No other significant effects or interactions were found, indicating that the increasing CV-syllables did not produce different inhibitions regardless they were presented at the ipsi- or contralateral ear.
Pathway Inhibition Effects
The present MEG study investigated the responses of the primary auditory cortices elicited by dichotic CV-syllables in healthy subjects all showing right ear advantage in a verbal DL task. The main intent was to disclose interactions between ipsi- and contralateral auditory pathways during DL, which are at the basis of the lateralization of the auditory input as proposed by Kimura's (1967) “structural theory.” In the present research procedure, an inhibition of one auditory pathway is revealed by a reduction of source strength increase in response to dichotic CV-syllables presented at different intensities. More in detail, results were obtained by examining the responses of the primary auditory cortex to dichotic CV-syllables which were presented at increased intensities (60 and 80 dBA) at one ear, whereas at the other ear the stimulation intensity remained constant. With this stimulation pattern, if there were not interactions between the auditory pathways, one would expect that the source strength increased according to stimulation intensity (Hegerl et al. 1994; Jancke et al. 1998; Brancucci et al. 2004). If the increase of source strength is attenuated compared with a control condition (noncompeting dichotic pairs), then the pathway originating from the ear receiving the stimulation with increasing intensity could have been suppressed, yielding to a reduced amount of information transfer through that auditory pathway.
Dichotic Effects at Primary Auditory Cortex
The source of the auditory M100 component has been localized in the planum temporale of the Heschl's gyrus, containing the primary auditory cortex (Mäkelä et al. 1994; Reite et al. 1994; Huotilainen et al. 1998). Specifically, in the present study, due to the amplitude envelope of the dichotic stimuli (see Fig. 1) the real latency of the M100 component was about 130 ms poststimulus. The sources explaining the M100 component also accounted for a later peak at about 300 ms poststimulus. We assumed that this second peak represented the response to the vowel contained in the syllables making the dichotic stimuli, because this vowel occurred about 190 ms after the stimulus onset (Ostroff et al. 1998). The interaction between contralateral and ipsilateral sensory pathways would occur at the cortical response to the vowel (300 ms), after the consonant has been processed by the primary auditory cortex, and could not be detected at earlier latencies. Thus, the present results suggest that the inhibition mechanisms mentioned above should not occur at subcortical level. Indeed, a subcortical inhibition would have reduced the amplitude increment of all cortical AEF components, including both first and second AEF peaks.
As an alternative explanation, it could be argued that the first component of the cortical response, which is generated in the early stages of cortical auditory input processing, is more strongly influenced by sensory thalamocortical inputs than the second one (Reite et al. 1988, 1994). In contrast, the second component of the cortical response would be more affected by re-entrant cortical loops including inhibitory effects induced by DL of competing speech sounds. On the whole, it can be speculated that the dichotic inhibition is associated with information processing in the primary and secondary auditory cortices, which is mainly reflected by activity later than 100 ms (Pantev et al. 1990; Reite et al. 1994; Huotilainen et al. 1998).
Lateralization of Inhibition Effects
Activity of the primary auditory cortices is supposed to be strictly involved in DL and in the processing of acoustic material (Hari 1990; Hugdahl et al. 1999; Tervaniemi and Hugdahl 2003). Results showed an asymmetric interaction pattern between the 2 auditory pathways. Concerning the left auditory cortex, DL of competing compared with noncompeting CV-syllables caused an inhibition of both ipsilateral (i.e., stimulus presented at the left ear) and contralateral (i.e., stimulus presented at the right ear) pathways. Moreover, according to the statistical post hoc comparison of the (left) source increase elicited by CV-syllables of different intensities presented at the right compared with left ear, the inhibition of the ipsilateral pathway (i.e., the ipsilateral neural signal) was significantly stronger than the contralateral one. Conversely, concerning the right auditory cortex, ipsi- and contralateral pathways were inhibited to the same extent during DL of competing compared with noncompeting CV-syllables.
The results of the present study confirm and extend previous findings. Suppressive binaural interaction at the human auditory cortex has been observed in several MEG studies using nonverbal stimuli (Reite et al. 1981; Pantev et al. 1986; Tiihonen et al. 1989). In those recordings, binaural responses of the human auditory cortex were much smaller than the sum of responses to monaural left and right ear sounds and, at some time instants, even smaller than responses to contralateral sounds presented alone. Another study (Kaneko et al. 2003) found that during binaural stimulation, the ipsilateral input was suppressed significantly more than the contralateral one. The present results are in line with a MEG study aimed at selectively following the inputs from both ears up to the cortex during binaural hearing (Fujiki et al. 2002). In this report, neuromagnetic cortical responses to amplitude-modulated continuous tones were recorded with different modulation frequencies at the ears (“frequency tagging”). Results showed that during binaural hearing, the inputs to the left and right ear competed strongly in both auditory cortices. The responses of the right auditory cortex were symmetrically suppressed, compared with monaural stimulation, for sounds of both ears, whereas the responses of the left auditory cortex were suppressed significantly more for ipsilateral than contralateral sounds. If we consider binaural stimulation as a completely competing one (with a 100% spectral overlap), our results on speech competing DL complement previous findings on binaural stimuli. The results of the present MEG study extend the inhibition pattern of auditory pathways to DL and to speech sounds, and give a neurophysiological explanation to the right ear advantage usually found in DL tasks using verbal material. Moreover, they show that, beyond hemispheric functional specialization of auditory cortical areas (Tervaniemi and Hugdahl 2003) there is asymmetry as well within the ascending auditory system. On the basis of the present study on verbal DL and previous ones on complex tones (Fujiki et al. 2002), we suggest that the asymmetry of the auditory ascending pathways might be stimulus dependent. As mentioned in the Introduction, it should be not neglected that attention plays an important role in DL (Jancke et al. 2001). As an example it can not be excluded that some subjects habitually place their attentional focus on one of the two ears. To minimize possible attentional biases, the present experiment was designed to counterbalance possible attentional shifts to one of the ears. Indeed, whether a neural signal was named as contralateral or ipsilateral depended on which ear the stimulus at 2 intensities was delivered to. The test of both the left and right ear, implying that one hemisphere was contralateral and ipsilateral for the same number of stimuli, prevented the subjects from systematically favoring of one pathway. Moreover, we chose as a dependent variable the amplitude difference of 2 brain responses elicited by stimulations presented in a randomized order during the recordings, so as to avoid that a sequence of 2 same CV-syllables with increased intensities were contiguously presented. We thus assume it unlike that the present results are considerably biased by attentional factors.
Spectral Overlap of Syllables as a Basis of DL Effects
Regarding the role of the spectral overlap between the dichotic stimuli in the occlusion mechanism, the present study gives a physiological basis to previous reports using verbal (Springer et al. 1978) or tonal (Sidtis 1981) material. Springer et al. (1978) have shown that although report of CV-syllables presented to the left ear during dichotic testing was at chance, report of left ear digits under the same conditions was greater than 80% in 4 out of 5 split-brain patients. As the acoustic overlap between CV-syllables is greater than the overlap between digits, they concluded that the availability of information from the ipsilateral auditory pathway is a function of the spectral acoustic overlap between competing dichotic stimuli. Sidtis (1981) demonstrated that a nearly threefold difference in the magnitude of the laterality measure could be obtained by delivering dichotic tones with similar versus different fundamental frequencies. The musical fifth interval yielded minimal laterality effects, whereas intervals of a second, a minor third, or an octave as a special case, yielded maximal laterality effects. Similar results have been obtained with subjects who have undergone temporal lobectomy and hemispherectomy (Berlin et al. 1973). Thus, the present study and those previous reports indicate the existence of a strong relationship between the degree of spectral competition and the magnitude of the laterality effect. As the spectral overlap increases, stimulus competition increases and laterality effects are maximized by favoring the contralateral pathways. Another study (Rimol et al. 2006) showed that, in addition to the spectral overlap, a further parameter affecting the lateralization of dichotic stimuli is the voice onset time. A dichotic pair composed by a CV-syllable with short voice onset time in the left ear (such as /ba/) and by a CV-syllable with long voice onset time in the right ear (such as /ka/) produces a larger right ear advantage than the inverted dichotic pair. That is syllables with long voice onset time are “preferred” by the contralateral auditory pathway connecting right ear and left auditory cortex. These evidences suggest that the features of the stimulus and the effects of competition have to be adequately considered in the interpretation of laterality effects because not only different types of auditory materials (i.e., verbal vs. tonal) but also small changes in the degree of competition between ipsi- and contralateral information of the same type might significantly affect the magnitude of perceptual asymmetry. Future studies could generalize our results to other competing (i.e., /pa/-/ba/, /ta/-/da/) or noncompeting (i.e., /ga/-/ta/, /pa/-/ka/) dichotic CV-syllables pairs.
Extending Kimura's “Structural Theory”
The present findings validate and extend the structural theory proposed by Kimura (1967), which explained dichotic ear advantage by reference to the anatomy and physiology of the auditory system. This model emphasizes the notion that the contralateral auditory pathways are dominant, more numerous and more rapidly conducting than the ipsilateral ones. Kimura further suggested that such differences between contra- and ipsilateral pathways were exaggerated by an occlusion mechanism during dichotic stimulus application, whereby input from the contralateral ear would block the ipsilateral pathways and prevent information from reaching the auditory cortex via the direct ipsilateral route. The present results extend this notion to an asymmetry between right and left auditory cortices during DL of CV-syllables. Concerning the left hemisphere, dominant for speech, during DL of competing CV-stimuli the ipsilateral pathway is strongly inhibited, thus favoring the perception of the input to the right ear. Concerning the right hemisphere, both pathways are inhibited to the same extent. In this framework, the privileged ear is the right one, the input of which can reach the left hemisphere—dominant for language—via a preferential route that suppresses the ipsilateral left auditory pathway. In concomitance, the input to the right ear can reach the right auditory cortex without significant loss of information compared with the input of the left ear. The right auditory cortex plays not only a (minor) role in language but also allows the verbal input to join the dominant left auditory cortex via corpus callosum. As aforementioned, all the subjects of the present study showed right ear advantage. Future studies should investigate the neural interactions between left and right auditory pathways in subjects showing a reversed asymmetry outline (i.e., left ear advantage) for speech stimuli. It could be predicted that in these subjects the neural pathway inhibition pattern is to some extent mirrored being this feature one of the determinants of the behavioral asymmetry.
The present results can constitute a basis for the investigation of the neural interactions underlying auditory perception at cellular level. Relevant literature (reviewed in Bear et al. 1996) has shown that, beyond the main tonotopic organizational principle of the auditory cortex (Romani et al. 1982; Bilecen et al. 1998; Formisano et al. 2003), there is a second organizational principle based on ear preference. In cat primary auditory cortex, there are alternating patches of neurons, which respond preferentially to binaural stimulation (being inhibited by monaural stimulation) or monaural stimulation (being inhibited by binaural stimulation). These neuronal populations together with the well-know process of lateral inhibition may be a suitable low-level substrate for the DL effects reported in the present study.
Conflict of Interest: None declared.