Brain, Vol. 123, No. 7, 1391-1402,
July 2000
© 2000 Oxford University Press
Differential recruitment of the speech processing system in healthy subjects and rehabilitated cochlear implant patients
1 Laboratoire de Neurosciences et Systèmes Sensoriels, CNRS, Hôpital Edouard Herriot, 2 CERMEP, Lyon, France and 3 Wellcome Department of Cognitive Neurology, Institute of Neurology, London, UK
Correspondence to:
Dr Anne-Lise Giraud, Physiologisches Institut III, Universitätsklinikum, Theodor Stern Kai 7, 60590 Frankfurt/Main, Germany E-mail: Giraud{at}em.uni-frankfurt.de
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Differences in cerebral activation between control subjects and post-lingually deaf rehabilitated cochlear implant patients were identified with PET under various speech conditions of different linguistic complexity. Despite almost similar performance in patients and controls, different brain activation patterns were elicited. In patients, an attentional network including prefrontal and parietal modality-aspecific attentional regions and subcortical auditory regions was over-activated irrespective of the nature of the speech stimuli and during expectancy of speech stimuli. A left temporoparietal semantic region was responsive to meaningless stimuli (vowels). In response to meaningful stimuli (words, sentences, story), left middle and inferior temporal semantic regions and posterior superior temporal phonological regions were under-activated in patients, whereas anterior superior temporal phonological regions were over-activated. These differences in the recruitment of the speech comprehension system reflect the alternative neural strategies that permit speech comprehension after cochlear implantation.
cochlear implant; speech comprehension; brain imaging
BA = Brodmann area
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Cochlear implants allow restoration of hearing in cases of profound sensorineural deafness (Dijourno and Eyries, 1957
). This technique, based on the replacement of severed sensory cells by electrodes stimulating the auditory nerve directly, is very successful in patients with normal language acquisition prior to the onset of deafness and in deaf children if the implantation occurs around the age of 18 months, thus respecting the critical period for language development (House, 1995). Implants allow adults to recover speech comprehension and children to acquire language normally. In pre-lingually deaf adults, implants restore auditory alerting functions but only very seldom speech comprehension.
Previous functional neuroimaging studies have shown altered brain activation patterns in pre-lingually deaf patients compared with controls, whereas in post-lingually deaf patients, speech perception engages the same brain regions as in controls (Okazawa et al., 1996; Naito et al., 1997; Young et al., 1998). These differences between the two groups of patients illustrate the success of implantation and can mostly be accounted for by variations of performance level. However, as results in patients and controls were hitherto not directly compared, the congruence of brain activation patterns in controls and post-lingually deaf patients remains inconclusive.
In the present study we directly compared brain activation between healthy subjects and cochlear implant patients under experimental conditions where both groups of subjects understood correctly the speech stimuli. We used functional neuroimaging by PET to study the mechanisms underlying speech processing in patients who, with the aid of cochlear implants, had regained close to perfect speech comprehension in everyday life. This corresponds to the highest performing 5% of the overall adult cochlear implant patient population. With this patient selection we attempted to minimize the performance-related confound that obviously arises when studying incompletely rehabilitated patients. During scanning we achieved almost comparable performance in patients and normals and thus restricted the experimentally observed brain activation patterns to neural processes recruited for successful comprehension.
A different pattern in patients may, however, have various origins. It may result from neural reorganization; the cortical representation of speech sounds may have changed during deafness, as suggested by changes in the tonotopical organization of the auditory cortex (Ponton et al., 1993), and may not have returned to normal. Alternatively, it may arise from an increased effort needed for speech comprehension, conscious strategies developed during rehabilitation or implicit feedback modulation in sensory processing that compensates for degradation of input signals.
To maximize our sensitivity to detect differences in brain activation patterns between implant patients and normal subjects, we used various kinds of speech stimuli, namely vowels, words, sentences and a story, which require different speech processing and language comprehension mechanisms. Using these stimuli and a hierarchical analysis, we were able to distinguish differences between groups that depend on the nature of the stimulus from those that are common to all stimuli, such as differences related to non-specific attentional demands.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Subjects
Six cochlear implant patients and six normal subjects took part in the study which was approved by the local ethical committee (Hospices Civils de Lyon, France). Their written consent was obtained, in accordance with the declaration of Helsinki. All were right-handed. Each group comprised two females and four males, and the mean age was 37.5 ± 5 years in the patient group and 36.1 ± 6 years in the control group.
All patients presented excellent speech intelligibility (95–100% without lip-reading). They had similar histories in terms of duration of hearing loss and rehabilitation (mean duration of deafness prior to implantation = 2.33 years, 18–24 months of rehabilitation with speech therapy). Four patients were implanted on the right side, two on the left. Two had an MXM Digisonic implant, four had a Nucleus Spectra 22. All implants had between 15 and 22 functioning electrodes.
PET acquisition
Regional cerebral blood flow was assessed by recording cerebral radioactivity following the intravenous injection of H215O. Patients and controls received 14 injections of 9 mCi. The results described in this study are based on 10 injections per subject (see Experimental design). PET images were acquired parallel to the bicommissural (AC–PC) plane, using a Siemens CTI HR+ camera. The counts were integrated over a period of 60 s after the detected activity in the head had reached 400% of background noise. Stimuli started 10 s before the acquisition period and lasted 70 s.
Standardized procedures of realignment, normalization, smoothing (16 mm Gaussian filter) and statistical analysis were performed using SPM97 (Wellcome Department of Cognitive Neurology, London, UK) implemented in Matlab 4.0 software (Mathworks, Sherborne, Mass., USA).
Experimental design
We used five conditions: expecting vowels, listening to vowels, listening to words, listening to unrelated sentences, listening to a story. Each condition was repeated twice in random order with stimuli presented in a counterbalanced order. These conditions were chosen to test different speech processing levels and thereby maximize differential effects between groups. Across conditions there was an increasing complexity in the stimuli with an increasing demand to understand speech stimuli `on-line' to permit comprehension of a global meaning. We hypothesized that the brain activation patterns mainly follow hierarchical lines in controls, which is confirmed by the overlapping results when comparing each condition with vowels (see Table 2). The predominance of hierarchical patterns in controls is further underlined by the absence of a significant effect anywhere when modelling a parametric decrease across conditions. In patients, these hierarchical lines may be disrupted, e.g. understanding words in isolation requires more phonological processing to resolve phonetic ambiguities than understanding words in the context of a sentence. An analysis of differential activation between groups in a hierarchical way enables us to detect at which processing level disruptions in the hierarchy occurred in patients.
|
Stimuli
The first condition consisted of attentive rest. Subjects were instructed to listen to vowels presented at a random rate (ranging from 3/s to 1/min). The whole condition lasted 3 min, but data were collected during a 60 s period of silence inserted in the middle of the vowel series. In the next condition, subjects heard vowels delivered at a constant rate (2/s). Data were acquired during stimulus presentation. Vowels of the French language were used as stimuli (`par', `jeune', `fermé', `frais', `jeûne', `tapis', `fou', `mur'). We did not use syllables (like `ba', `da') because consonants are not clearly perceived by patients if presented outside the context of a word or a sentence.
The word condition involved unrelated bisyllabic words presented at 2 syllables/s. Simple sentences (subject, verb, object) were interrogative or affirmative. In the sentence and story conditions, normal prosody was respected (140 words/minute), but the mean rate of the whole sequence remained at 2 syllables/s (14 sentences, mean number of syllables/sentence = 10, one sentence every 5 s). A slow presentation rate was used to ensure comprehension in all patients. Stimuli were delivered in free field with loudspeakers located behind subjects in the PET camera. The sound level was set to be comfortable for each subject and to optimize speech comprehension. The stimuli were generated by a female speaker and recorded on a CD used in the experiment.
Data analysis
We used a fixed effects model with condition, subject and global effects, and appropriate contrasts to create statistical parametric maps of the t-statistics. The main contrasts were vowels against rest (Fig. 1), and speech conditions (words, sentences, story) against vowels. We also analysed the contrast between expecting (rest) and hearing vowels (Fig. 2). Results are presented in Tables 1 and 2 with interaction levels between groups.
|
|
|
A further analysis investigated interactions between groups, with a hierarchical approach to isolated condition-related differences between controls and patients (words–vowels, sentences–words, story–sentences in one group compared with the other).
Masks were applied with a threshold of P = 0.001 to restrict the interaction analysis to those voxels which showed a significant effect in one group, e.g. words > vowels in patients, and thereby to exclude voxels where the interaction resulted only from a relative deactivation in the other group, e.g. vowels > words in controls. Masking, however, allowed detection of crossover effects (increase in one group with decrease in the other group). Interactions were considered significant at P < 0.001. These results are presented in Table 3 and Figs 3–6.
|
|
|
|
|
To clarify the origin of an interaction (differential level of activity or crossover effect), blood flow variations relative to the mean of all conditions in both groups are displayed in the figures. To test disruptions in the hierarchy across conditions in a more specific way, we modelled a parametric increase of blood flow with conditions (vowels > words > sentences > story) in both groups and tested for an interaction between groups. This analysis was used to assign a statistical value to the crossover effects detected in interactions between groups by involving simple contrasts between conditions.
Using an additional analysis where we modelled each subject individually, we ensured that the interactions between groups reflected effects (patients > controls or controls > patients) that occurred in a systematic way in all subjects of the group of interest.
We additionally searched for interactions between groups that were common to all meaningful speech stimuli (words, sentences, story) against vowels. In this analysis, we masked with simple effects of words, sentences and story versus vowels (at P = 0.001).
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Behavioural data
Before data collection
Prior to the experiment we assessed speech intelligibility under scanning conditions (head in the PET camera) in both groups. Subjects were exposed to the stimuli (vowels, words and sentences presented in a different order from that during the experimental scans) and the percentage of correct phoneme recognition was calculated. Controls and patients showed 100% vowel recognition. Normal subjects achieved 100% phoneme recognition with words and sentences, patients 92.3% and 97% phoneme recognition, respectively. Four patients confounded several words of the word list and the sentences. The two others repeated perfectly. None of the patients reported not having understood a sentence. All subjects were pre-exposed once to the story. Patients and controls understood the general meaning of the story well.
During data collection
Just before data acquisition subjects were instructed to listen carefully to speech stimuli. No response was required during data collection but questions were asked after each scan.
The questions differed from one condition to the other and between the two replications of the same condition. The purpose of these questions was to assess comprehension and to ensure sustained attention during the whole scanning period. For all conditions there were two types of question (i) subjects were asked to recall some items, (ii) subjects were asked to indicate whether proposed items (one of three new, two of three used as stimuli) were present or not in the stimulus list.
After passive listening to vowels, recall of five of the presented vowels was tested and we asked whether five proposed vowels belonged to the stimulus list (some French vowels were not present in the stimulus list: `brin', `pont', `fort', `faon', `brun'). Vowels were identified without effort by the patients and the overall performance was comparable in controls and patients.
After the word conditions we asked the subjects to recall five words and proposed five words. After the sentence conditions we asked the subjects to recall three sentences. We read two incomplete sentences and asked for a missing word, and we proposed five sentences. In the story condition we asked for a summary of the story in five chronological steps and proposed five sentences with exact or inexact details.
We rated the accuracy of responses on a scale ranging from 1 to 10. Controls scored 9.8 (± 0.2) in the word condition and 10 in the sentence and story conditions. Patients achieved 9.3 (± 0.4) in the word condition, 9.6 (± 0.3) in the sentence condition and 9.5 (± 0.3) in the story condition. These observations indicate that patients and controls understood the stimuli almost comparably. This fulfilled our aim to match, as well as possible, patients and controls regarding the success of speech comprehension whatever the effort to achieve comprehension. Small (but insignificant) differences between groups indicate that the tasks were more easily achieved by controls than patients. This observation confirms that speech comprehension was reached with more effort in patients and that different neural strategies are probably used by the two groups. These observations are taken into account in the interpretation of our results.
Imaging data
Expecting and listening to vowels
In control subjects, hearing vowels activated a region located within the 50–75% contour of Penhune's probabilistic map of the primary auditory cortex (Penhune et al., 1996). In patients, a region of the primary auditory cortex (25–50%) responded to vowels, but several other regions also showed an increase in activity when they heard compared with when they expected vowels. The activation was located in the left posterior middle temporal gyrus (Brodmann area, BA 21), the left temporoparietal junction (BA 22/39), the right inferior parietal cortex (BA 40), the left superior frontal gyrus (BA 6) and the cerebellum (Table 1
). The left Heshl's gyrus, the left posterior middle temporal and temporoparietal regions, and the left cerebellum were significantly more activated in patients than in controls, as shown in Fig. 1. These regions were activated only when patients heard vowels, but not when controls heard or expected vowels or when patients expected vowels.
In control subjects, expecting vowels produced no greater activation than hearing vowels. In patients, this contrast was associated with activation of the left middle frontal (BA 6), the left occipital (BA 18/19) and the right superior parietal cortices (BA 7), and the lower brainstem, where interactions by group were significant. Figure 2 shows the brain regions that were more active in patients than in controls when they expected vowels. The left middle frontal and the right superior parietal regions show a crossover effect by group as they were also activated when controls heard vowels.
Listening to words
Compared with vowels, words produced activation in the right and left middle parts of the superior temporal gyri (BA 22) and in the anterior left middle temporal gyrus (BA 21). Patients activated the same regions but showed additional activation in the anterior right middle temporal gyrus (BA 21), the posterior right superior temporal gyrus (BA 42), the posterior left middle temporal gyrus (BA 21), the left dorsal occipital cortex, the thalamus, the lower brainstem and the cerebellum (see Fig. 3 and Table 2). Interactions by group were significant in the left superior temporal and dorsal occipital regions, the thalamus and the cerebellum. Figure 4 shows the condition-specific blood flow levels in right and left superior temporal regions.
Listening to unrelated sentences
Response to sentences relative to vowels.
In control subjects, sentences activated the same brain regions as words. Additional activations were observed in the posterior left superior temporal gyrus (BA 22), the anterior right superior temporal gyrus (BA 21) and the left inferior temporal gyrus (BA 21). These regions are co-localized with activation sources related to sentence congruity (Helenius et al., 1998) and sentence linguistic complexity (Carpenter et al., 1999). In patients, sentences produced activation of the right Heschl's gyrus, the middle and posterior parts of the left superior temporal gyrus (BA 22), the middle right superior temporal gyrus (BA 22), the anterior left middle temporal gyrus (BA 21), the right inferior parietal gyrus (BA 40), the right lateral premotor cortex (middle frontal, BA 6) and the left hippocampal gyrus (BA 36). Significant interactions by group were found in Heschl's gyrus, the posterior left superior temporal gyrus, and the right inferior parietal and premotor cortices.
We also observed activation in the posterior thalamus, the lower brainstem and the cerebellum (Table 2), where interactions were significant.
Response to sentences relative to words.
In control subjects, no brain region was significantly more active for sentences than for words. However, there was a subthreshold activation of the superior temporal regions (bilaterally) and the middle and inferior temporal region as shown in Fig. 3 (thresholded at P = 0.05). These regions were also recruited in patients, except the inferior temporal region which showed a strong interaction by group (Z = 4.26). In patients, sentences additionally produced a greater activation than words in the left superior frontal (BA 8) and the medial premotor cortex (middle frontal, BA 6) (Fig. 3). Activation in these regions did not significantly differ from the control group.
Understanding a coherent story
Response to a story relative to vowels.
In control subjects, listening to a story activated the middle part of the right and left superior temporal cortices (BA 22), the anterior and posterior left middle temporal gyrus (BA 21) and the left inferior temporal gyrus (BA 21/38). Interactions by group were significant in all these regions except the middle left superior temporal region, which was also recruited in patients. In patients, the same task activated the posterior right and left superior temporal gyri (BA 42) and the anterior left middle temporal gyri (BA 21). Activation was also observed in the right superior parietal (BA 7) and right superior and middle frontal (BA 6) cortices, the right and left hippocampus, the thalamus, the inferior colliculus, the lower brainstem and the cerebellum. In the group comparison, only the left hippocampus, the thalamus and the cerebellum showed a significant effect.
Response to a story relative to unrelated sentences.
In controls, but not in patients, listening to a story relative to listening to unrelated sentences activated the left middle (BA 21) and inferior temporal regions (BA 38 and BA 36) and the left cerebellum (Fig. 5). These regions showed a significant interaction by group. In the three temporal regions, the parametric effect across conditions revealed a significant interaction by group (middle Z = 3.23, anterior inferior Z = 3.15, posterior inferior Z = 3.04).
The same contrast (story relative to sentences) in patients revealed activations in the right medial premotor (BA 6), the right and left posterior hippocampus, the dorsal occipital cortex (BA 19) and the cerebellum. Interactions were significant in the right hippocampus and the left dorsal occipital cortex.
Non-specific over-activation in patients
We searched for interactions between groups irrespective of the nature of the speech stimuli presented (Fig. 3). Control subjects had greater activation in the right and left middle temporal gyri (BA 21) and in the left inferior temporal gyrus (BA 21/38/36). The right intraparietal sulcus (BA 40/7), the right premotor area (BA 6), the left hippocampus, the upper brainstem and the left cerebellum were significantly more activated in patients than in control subjects. Condition-related blood flow variation in the hippocampus is presented in Fig. 6.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
As previously reported by Okazawa and colleagues (Okazawa et al., 1996
), post-lingually deaf implanted patients presented activation of the classical language areas in response to verbal sounds. The detected regions were mainly co-localized with those observed in normal hearing subjects. However, we focused our analysis on interactions between groups that revealed many activation level differences located either in the language system or outside the language system and that depended, or not, on the nature of the speech stimuli used. A careful identification and functional classification of these differences suggest that successful speech comprehension resulted from different neural processes in the two groups. In patients, the contribution of more brain regions was required to process speech. Some of these over-activations may be specific, others less so, i.e. they may be related to speech processing or to non-specific attentional mechanisms. This in turn may reflect functional reorganization, instantaneous implicit compensation mechanisms or intentional listening strategies. We have endeavoured to interpret our results in light of these considerations.
Differential activation in the speech comprehension system
Across conditions, patients showed decreased activation in semantic regions and increased activation in phonological regions. Activations in language regions were larger in response to simple stimuli (words) in patients and in response to complex stimuli (story) in controls.
Increased low-level phonological processing
Patients had greater activations than controls in the posterior right superior temporal gyrus (BA 22/21), classically involved in phonological mechanisms (Petersen et al., 1989; Démonet et al., 1996; Fiez, 1997; Price et al., 1997), and in a right premotor region (BA 6) which is activated in syllable production (Price and Giraud, 1998). Over-activations in patients were associated with low activation levels in control subjects, reflecting the fact that word processing in normal-hearing subjects requires only minimal phonological processing. In patients, syllable analysis required more detailed phonological processing that probably increased interactions with regions supporting speech production and internal repetition of speech sounds.
Decreased semantic processing
Controls showed greater activations than patients in various regions of the inferior temporal cortex (basal temporal language areas), which are implicated in semantic knowledge (Alexander et al., 1989; Hodges and Patterson, 1997). These regions are activated when subjects perform semantic compared with phonological decisions (Vandenberghe et al., 1996; Mummery et al., 1998). Our observations suggest reduced semantic processing in patients that nonetheless remains sufficient to permit understanding of general meaning and most details. Reduced semantic activations while listening to complex speech stimuli might be related to the over-activation demonstrated in phonological regions for less complex stimuli. The difficulties of unmasking relevant speech signals from noise and compensating for missed transient speech events (such as consonants) might slow the early steps of speech processing and therefore prevent late processes from evolving as efficiently as in controls.
Compensation with memory
Memory-related activations in patients were increased, in particular, in the left posterior hippocampus, in association with decreased semantic processing. The left posterior hippocampus is involved in retrieving (LePage et al., 1998) in the context of semantic memory tasks (Maguire and Mummery, 1999). Understanding continuous speech requires on-line comprehension and recall; the meaning of perceived stimuli needs to be kept in mind while the next stimulus is being processed and concurrently recalled to generate a global meaning. This situation is particularly challenging for patients who experience perceptual difficulties. They may, therefore, try to retrieve the meaning of speech from memory, which was possible in our study since they had been familiarized with the stimuli. We suggest that patients compensate for impoverished speech signals by a strategy shift towards memory mechanisms (encoding and retrieval) to facilitate matching of phonological inputs to stored verbal representations.
Patients also showed greater activations than controls in a dorsal occipital region (precuneus, BA 19) that has been associated with episodic verbal retrieval (Grasby et al., 1994; Krause et al., 1998) and memory-related imagery (Fletcher et al., 1995). This region was activated during vowel expectancy, which corresponds to a brief period of total deafness for patients as the implant switches itself off below a certain signal to noise ratio. With the implant turned off, deaf patients are likely to produce abundant internal images. Activation was also observed in a more ventral region (BA 18). This activation may reflect a sensitivity gain in the visual modality expressed during deafness, as a result of cross-modal compensation (Rauschecker and Korte, 1993; Neville, 1995).
Activation of semantic areas for meaningless speech stimuli
In patients, hearing vowels induced greater activations (i) in the left superior temporal sulcus (BA 22/21) that responds to auditorily as well as visually presented words (Gorno-Tempini et al., 1998) and has been specifically implicated in word-form or word-meaning processing, and (ii) in the left temporoparietal junction (BA 39), a multimodal semantic region (Gorno-Tempini et al., 1998; Price, 1998). There are two possible interpretations for semantic areas responding to meaningless stimuli.
First, in patients, speech comprehension relies mainly on vowel detection, consonant detection being very poor. The perceptual processing of vowels may therefore be enhanced, accounting for over-activations in the phonological system. Moreover, the implicit screening of all possible words compatible with a perceived stimulus may be emphasized, accounting for activations in semantic and multimodal regions which reflect contextual (top-down) modulation that helps to reduce errors resulting from the paucity of perceptual (bottom-up) information.
The second interpretation implies a change in functional specialization. In patients, regions classically dedicated to semantic processing may revert to lower level processing, e.g. phonological analysis. This is unlikely in the temporoparietal semantic region as it is far removed from speech specific regions, but functional changes may occur in the superior temporal sulcus.
Attention and top-down modulation in speech comprehension with an implant
Activation of a cortical attentional network
Irrespective of the complexity of speech stimuli, and during expectancy of speech stimuli, patients activated a set of regions probably reflecting specific auditory attentional strategies (Benedict et al., 1998) to compensate for degraded speech signals and monaural stimulation (O'Leary et al., 1996). They are specific inasmuch as they involve attention to space (right intra-parietal sulcus) (Coull and Nobre, 1998), as patients pay particular attention to the implant side (from which stimuli are coming or expected) and attention to the temporal properties of auditory signals (left intra-parietal sulcus and left lateral premotor cortex) (Fiez et al., 1995; Platel et al., 1997; Tzourio et al., 1997; Coull and Nobre, 1998); these are cues to which patients are highly sensitive (Cazals et al., 1994; Lorenzi et al. 1997). Patients implanted on the left side showed a larger effect than right-side implanted patients in the left lateral premotor cortex and in the left intra-parietal region during vowel expectancy. This suggests that activations in these regions are indeed sensitive to the direction of attention.
During sentences, patients (but not control subjects) recruited Heschl's gyrus, although sentences were controlled for primary auditory processing by vowels presented at the same stimulation rate. This probably corresponds to top-down modulation of early auditory processing, which depends on task demand. Similar modulation of sensory input processing was observed during reading (Price et al., 1997) and while listening to sounds (Price and Giraud, 1998), when a response was required.
Subcortical implicit top-down modulation
In several of our conditions (including vowel expectancy), we observed in patients greater activation in the inferior colliculus and the lower brainstem. The latter activation focus is compatible with the location of the reticular formation or the superior olivary complex, which is a relay of the ascending and descending (top-down) auditory pathways. The olivocochlear neurons receive projections from the reticular formation (Thompson and Thompson, 1995) and the inferior colliculus (Huffman and Henson, 1990). They are involved in auditory arousal and attentional mechanisms (Giard et al., 1994) and more specifically in the extraction of speech signals from background noise (Giraud et al., 1997). Such subcortical activations were observed during vowel expectancy in the absence of any auditory stimulation, thus reflecting top-down attentional control.
Olivocochlear neurons are not functional in cochlear implant patients since their target (outer hair cells) is damaged during implantation. As cochlear function can no longer be controlled, descending modulation of the superior olivary complex becomes more important because it is the first stage at which the ascending auditory information can be modulated. These activations suggest that implicit modulation via feedback loops is involved after cochlear implantation to achieve speech comprehension.
We also found over-activation in the left lateral cerebellum of patients. The cerebellum is not only involved in some motor aspects of speech production but also in cognitive language tasks (Cabeza and Nyberg, 1997). Its lateroposterior part plays a specific role in spatial attention (Townsend et al., 1999). Through reciprocal connections with many cortical association areas, the cerebellum is likely to improve performance in a number of cognitive domains (Leiner et al., 1993). These activations in patients probably reflect the participation of the cerebellum in a general attentional effort to improve speech comprehension.
Adaptive changes or functional reorganization?
Our results suggest that speech comprehension can be subserved by variable cerebral mechanisms. Different activation patterns within the same neural network can lead to comparable performance. Short-term (instantaneous) adaptive strategies, that would also be used by normal subjects if submitted to degraded speech sounds or long-term reorganization, could explain differences in the relative levels of activation of the components of the speech comprehension system in controls and patients. The conceptual boundaries between short-term adaptation (instantaneous adaptation or tuning) and long-term plasticity, however, are not clear. A neural system that receives a pattern of new input can rapidly change its output due to changes in the effective connectivity between its components (Büchel et al., 1999). In patients, difficult hearing conditions could elicit a particular pattern of activation of the speech comprehension system as a result of the use of a strategy instantaneously recruited (Démonet et al., 1994). If abnormal input conditions continue, the system may then re-organize in terms of functional specialization. As cochlear implant patients receive degraded auditory inputs on a long-term basis, these inputs cannot really be considered new or abnormal. Our findings could therefore reflect true changes in functional specialization. The continuous need for emphasized processing dedicated to speech signal extraction from noise and syllable processing could redefine functional response properties across the areas forming the speech comprehension system, thus allocating more neuronal resources to acoustic and early phonological steps at the expense of late phonological and semantic processing. In any case, further studies are required to determine the origin of the differences we have established here.
Conclusion
We used a wide range of possible speech stimuli permitting different and natural speech processing mechanisms in controls and cochlear implant patients. This allowed us to show differential results across groups and conditions. Our results suggest three categories of effects in the cerebral activation patterns of patients: (i) invariant over-activation (common across conditions in the presence and absence of auditory stimuli) of attentional resources and subcortical top-down modulation loops; (ii) activation of semantic regions in the absence of meaning in the stimuli; and (iii) specific activation and deactivation within the speech comprehension system that are sensitive to stimulus complexity: over-activation of phonological regions and relative deactivation of semantic regions.
This study, by establishing differences in speech processing in implant patients with the most successful case of implantation and controls, is a prerequisite to an interpretation of studies involving heterogeneous groups of performers. The next step will address the longitudinal development of these patterns.
|
Acknowledgments |
|---|
We thank the patients who participated in this study and Andreas Kleinschmidt for help in data analysis and manuscript preparation. This work has been financed by The Wellcome Trust, Bonus Qualité Recherche 1997, Université Claude Bernard, Lyon I and Appel d'Offre Recherche Clinique 1996, Hospices Civils de Lyon, France. A.L.G. was initially supported by Fyssen Foundation and now by the European Commission, and R.S.J.F. by the Wellcome Trust
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Alexander MP, Hiltbrunner B, Fischer RS. Distributed anatomy of transcortical sensory aphasia. Arch Neurol 1989; 46: 885–92.
Benedict RH, Lockwood AH, Shucard JL, Shucard DW, Wack D, Murphy BW. Functional neuroimaging of attention in the auditory modality. Neuroreport 1998; 9: 121–6.[Web of Science][Medline]
Büchel C, Coull JT, Friston KJ. The predictive value of changes in effective connectivity for human learning. Science 1999; 283: 1538–41.
Cabeza R, Nyberg L. Imaging cognition: an empirical review of PET studies with normal subjects. J Cogn Neurosci 1997; 9: 1–26.[Web of Science]
Carpenter PA, Just MA, Keller TA, Eddy WF, Thulborn KR. Time course of fMRI-activation in language and spatial networks during sentence comprehension. Neuroimage 1999; 10: 216–24.[Web of Science][Medline]
Cazals Y, Pelizzone M, Saudan O, Boex C. Low-pass filtering in amplitude modulation detection associated with vowel and consonant identification in subjects with cochlear implants. J Acoust Soc Am 1994; 96: 2048–54.[Web of Science][Medline]
Coull JT, Nobre AC. Where and when to pay attention: the neural systems for directing attention to spatial locations and to time intervals as revealed by both PET and fMRI. J Neurosci 1998; 18: 7426–35.
Démonet J-F, Price C, Wise R, Frackowiak RS. A PET study of cognitive strategies in normal subjects during language tasks: influence of phonetic ambiguity and sequence processing on phoneme monitoring. Brain 1994; 117: 671–82.
Démonet JF, Fiez JA, Paulesu E, Petersen SE, Zatorre RJ. PET studies of phonological processing: a critical reply to Poeppel. Brain Lang 1996; 55: 352–79.[Web of Science][Medline]
Dijourno A, Eyries C. Prothèse auditive par excitation éléctrique du nerf sensoriel à l'aide d'un bobinage inclus à demeure. Presse Med 1957; 35: 14–7.
Fiez JA. Phonology, semantics, and the role of the left inferior prefrontal cortex [editorial]. [Review]. Hum Brain Mapp 1997; 5: 79–83.[Web of Science][Medline]
Fiez JA, Raichle ME, Miezin FM, Petersen SE, Tallal P, Katz WF. PET studies of auditory and phonological processing: effects of stimulus characteristics and task demands. J Cogn Neurosci 1995; 7: 357–75.
Fletcher PC, Frith CD, Baker SC, Shallice T, Frackowiak RS, Dolan RJ. The mind's eye – precuneus activation in memory-related imagery. Neuroimage 1995; 2: 195–200.[Web of Science][Medline]
Giard M-H, Collet L, Bouchet P, Pernier J. Auditory selective attention in the human cochlea. Brain Res 1994; 633: 353–6.[Web of Science][Medline]
Giraud A-L, Garnier S, Micheyl C, Lina G, Chays A, Chéry-Croze S. Auditory efferents involved in speech-in-noise intelligibility. Neuroreport 1997; 8: 1779–83.[Web of Science][Medline]
Gorno-Tempini ML, Price CJ, Josephs O, Vandenberghe R, Cappa SF, Kapur N, et al. The neural systems sustaining face and proper-name processing. Brain 1998; 121: 2103–18.
Grasby PM, Frith CD, Friston KJ, Simpson J, Fletcher PC, Frackowiak RS, et al. A graded task approach to the functional mapping of brain areas implicated in auditory-verbal memory. Brain 1994; 117: 1271–82.
Helenius P, Salmelin R, Service E, Connolly JF. Distinct time courses of word and context comprehension in the left temporal cortex. Brain 1998; 121: 1133–42.
Hodges JR, Patterson K. Semantic memory disorders. Trends Cogn Sci 1997; 1: 68–72.
House WF. Cochlear implants: my perspective; 1995. Available from: URL: http:/www.allhear.com.
Huffman RF, Henson OW Jr. The descending auditory pathway and acousticomotor systems: connections with the inferior colliculus. [Review]. Brain Res Brain Res Rev 1990; 15: 295–323.[Medline]
Krause BJ, Schmidt D, Mottaghy FM, Taylor J, Halsband U, Herzog H, et al. The precuneus is a major player in a network of distributed brain regions in episodic memory retrieval. Neuroimage 1998; 7 (4 Pt. 2): S828.
Leiner HC, Leiner AL, Dow RS. Cognitive and language functions of the human cerebellum. [Review]. Trends Neurosci 1993; 16: 444–7.[Web of Science][Medline]
Lepage M, Habib R, Tulving E. Hippocampal PET activations of memory encoding and retrieval: the HIPER model. Hippocampus 1998; 8: 313–22.[Web of Science][Medline]
Lorenzi C, Gallégo S, Patterson RD. Discrimination of temporal asymmetry in cochlear implantees. J Acoust Soc Am 1997; 102: 482–5.[Web of Science][Medline]
Maguire EA, Mummery CJ. Differential modulation of a common memory retrieval network revealed by positron emission tomography. Hippocampus 1999; 9: 54–61.[Web of Science][Medline]
Mummery CJ, Patterson K, Hodges JR, Price CJ. Functional neuroanatomy of the semantic system: divisible by what? J Cogn Neurosci 1998; 10: 766–77.[Web of Science][Medline]
Naito Y, Hirano S, Honjo I, Okazawa H, Ishizu K, Takahashi H, et al. Sound-induced activation of auditory cortices in cochlear implant users with post- and pre-lingual deafness demonstrated by positron emission tomography. Acta Otolaryngol (Stockh) 1997; 117: 490–6.[Medline]
Neville HJ. Developmental specificity in neurocognitive development in humans. In: Gazzaniga MS, editor. The cognitive neurosciences. Cambridge (MA): MIT press; 1995.
Okazawa H, Naito Y, Yonekura Y, Sadato N, Hirano S, Nishizawa S, et al. Cochlear implant efficiency in pre- and postlingually deaf subjects. A study with H215O and PET. Brain 1996; 119: 1297–306.
O'Leary DS, Andreason NC, Hurtig RR, Hichwa RD, Watkins GL, Ponto LL, et al. A positron emission tomography study of binaurally and dichotically presented stimuli: effects of level of language and directed attention. Brain Lang 1996; 53: 20–39.[Web of Science][Medline]
Penhune VB, Zatorre RJ, MacDonald JD, Evans AC. Interhemispheric anatomical differences in human primary auditory cortex: probabilistic mapping and volume measurement from magnetic resonance scans. Cereb Cortex 1996; 6: 661–72.
Petersen SE, Fox PT, Posner MI, Mintun M, Raichle ME. Positron emission tomographic studies of the processing of single words. J Cogn Neurosci 1989; 1: 153–70.
Platel H, Price C, Baron J-C, Wise R, Lambert J, Frackowiak RS. The structural components of music perception: a functional anatomical study. Brain 1997; 120: 229–43.
Ponton CW, Don M, Waring MD, Eggermont JJ, Masuda A. Spatio-temporal source modeling of evoked potentials to acoustic and cochlear implant stimulation. Electroencephalogr Clin Neurophysiol 1993; 88: 478–93.[Web of Science][Medline]
Price CJ. The functional anatomy of word comprehension and production. Trends Cogn Sci 1998; 2: 281–8.
Price CJ, Giraud A-L. The constraints functional anatomy places on cognitive models of auditory word processing, Neuroimage 1998; 7 (4 Pt. 2): S168.
Price CJ, Moore CJ, Humphreys GW, Wise RJS. Segregating semantic from phonological processes during reading. J Cogn Neurosci 1997; 9: 727–33.[Web of Science]
Rauschecker JP, Korte M. Auditory compensation for early blindness in cat cerebral cortex. J Neurosci 1993; 13: 4538–48.[Abstract]
Thompson AM, Thompson GC. Relationship of descending inferior colliculus projections to olivocochlear neurons. J Comp Neurol 1993; 335: 402–12.[Web of Science][Medline]
Townsend J, Courchesne E, Covington J, Westerfield M, Harris NS, Lyden P, et al. Spatial attention deficits in patients with acquired or developmental cerebellar abnormality. J Neurosci 1999; 19: 5632–43.
Tzourio N, El Massioui F, Crivello F, Joliot M, Renault B, Mazoyer B. Functional anatomy of human auditory attention studied with PET. Neuroimage 1997; 5: 63–77.[Web of Science][Medline]
Vandenberghe R, Price C, Wise R, Josephs O, Frackowiak RS. Functional anatomy of a common semantic system for words and pictures. Nature 1996; 383: 254–6.[Medline]
Young JP, O'Sullivan BT, Gibson WP, Mitchell T, Sanli H, Withall A, et al. Assessment of auditory cortical function in cochlear implant patients. Neuroimage 1998; 7 (4 Pt. 2) S384.
Received May 18, 1999. Revised August 18, 1999. Second revision on December 20, 1999. Accepted February 4, 2000.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. Sandmann, T. Eichele, M. Buechler, S. Debener, L. Jancke, N. Dillier, K. Hugdahl, and M. Meyer Evaluation of evoked potentials to dyadic tones after cochlear implantation Brain, July 1, 2009; 132(7): 1967 - 1979. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Coez, M. Zilbovicius, E. Ferrary, D. Bouccara, I. Mosnier, E. Ambert-Dahan, E. Bizaguet, A. Syrota, Y. Samson, and O. Sterkers Cochlear Implant Benefits in Deafness Rehabilitation: PET Study of Temporal Voice Activations J. Nucl. Med., January 1, 2008; 49(1): 60 - 67. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. M. Wilson, I. Molnar-Szakacs, and M. Iacoboni Beyond Superior Temporal Cortex: Intersubject Correlations in Narrative Speech Comprehension Cereb Cortex, January 1, 2008; 18(1): 230 - 242. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
U. Hasson, H. C. Nusbaum, and S. L. Small Brain Networks Subserving the Extraction of Sentence Information and Its Encoding to Memory Cereb Cortex, December 1, 2007; 17(12): 2899 - 2913. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Rouger, S. Lagleyre, B. Fraysse, S. Deneve, O. Deguine, and P. Barone From the Cover: Evidence that cochlear-implanted deaf patients are better multisensory integrators PNAS, April 24, 2007; 104(17): 7295 - 7300. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H.-J. Lee, A.-L. Giraud, E. Kang, S.-H. Oh, H. Kang, C.-S. Kim, and D. S. Lee Cortical Activity at Rest Predicts Cochlear Implantation Outcome Cereb Cortex, April 1, 2007; 17(4): 909 - 917. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-F. Demonet, G. Thierry, and D. Cardebat Renewal of the Neurophysiology of Language: Functional Neuroimaging Physiol Rev, January 1, 2005; 85(1): 49 - 95. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. L. Giraud, C. J. Price, J. M. Graham, and R. S. J. Frackowiak Functional plasticity of language-related brain areas after cochlear implantation Brain, July 1, 2001; 124(7): 1307 - 1316. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||