Task- and performance-related modulation of domain-specific auditory short-term memory representations in the gamma-band

The short-term retention of information has been related to oscillatory activity in the gamma-band. In recent auditory spatial short-term memory studies we have found stimulus-specific components of parieto-occipital gamma-band activity (GBA) which might reflect the activation of local networks tuned to task-relevant stimulus features. The present magnetoencephalography study (N = 22) tested this interpretation by assessing whether the topography of stimulus-specific GBA depends on task demands. Sample sounds were characterized by both a variable interaural time delay and a variable central frequency. In separate task blocks, either the lateralization or the frequency of the same stimuli had to be maintained. Statistical probability mapping of differences in oscillatory responses to the retention of sample sounds replicated the contralateral posterior topography for GBA components distinguishing between medial and lateral sounds in the spatial memory task. In contrast, lower- and higher-frequency stimuli were accompanied by frontal GBA components in the frequency task. Memory for lateralization versus frequency selectively enhanced oscillatory activity for these posterior versus frontal components, directly demonstrating their modulation by task demands. Incorrect “non-match” responses were negatively correlated with delay-phase GBA to the relevant feature, whereas incorrect “match” responses correlated positively with GBA to the irrelevant feature. In summary, the topography of stimulus-specific GBA to identical stimuli reflected the selective representation of task-relevant features. Task performance was predicted by both enhanced stimulus-specific GBA for the task-relevant stimulus attribute and reduced gamma-band representations of the task-irrelevant stimulus feature. Stimulus-specific GBA may reflect the memory representation that is used in subsequent recognition.     

Dynamics of gamma-band activity in human magnetoencephalogram during auditory pattern working memory

Both electrophysiological research in animals and human brain imaging studies have suggested that, similar to the visual system, separate cortical ventral "what" and dorsal "where" processing streams may also exist in the auditory domain. Recently we have shown enhanced gamma-band activity (GBA) over posterior parietal cortex belonging to the putative auditory dorsal pathway during a sound location working memory task. Using a similar methodological approach, the present study assessed whether GBA would be increased over auditory ventral stream areas during an auditory pattern memory task. Whole-head magnetoencephalogram was recorded from N = 12 subjects while they performed a working memory task requiring same-different judgments about pairs of syllables S1 and S2 presented with 0.8-s delays. S1 and S2 could differ either in voice onset time or in formant structure. This was compared with a control task involving the detection of possible spatial displacements in the background sound presented instead of S2. Under the memory condition, induced GBA was enhanced over left inferior frontal/anterior temporal regions during the delay phase and in response to S2 and over prefrontal cortex at the end of the delay period. gamma-Band coherence between left frontotemporal and prefrontal sensors was increased throughout the delay period of the memory task. In summary, the memorization of syllables was associated with synchronously oscillating networks both in frontotemporal cortex, supporting a role of these areas as parts of the putative auditory ventral stream, and in prefrontal, possible executive regions. Moreover, corticocortical connectivity was increased between these structures.     

The involvement of ipsilateral temporoparietal cortex in tactile pattern working memory as reflected in beta event-related desynchronization

Cortical oscillatory activity in various frequency bands has been shown to reflect working memory processes operating on visual and auditory stimulus information. Here we use magnetoencephalography to investigate cortical oscillatory activity related to working memory for tactile patterns. Right-handed subjects made same-different judgements on two dot patterns sequentially applied with a 3-s delay to the right middle fingertip. Spectral analysis revealed beta desynchronization (17+/-2.5 Hz) at contralateral postcentral and ipsilateral temporoparietal regions preceding and during the presentation of both tactile stimuli as well as during the early and late delay periods. Whereas contralateral beta desynchronization preceding tactile stimulation may reflect anticipation of incoming stimuli, ipsilateral beta desynchronization may underlie working memory maintenance of tactile patterns. The later hypothesis is supported by a significant positive correlation between subjects' performance and the amplitude of ipsilateral beta desynchronization 800 ms to 500 ms before the onset of the second pattern stimulus. Thus, our results suggest that ipsilateral temporoparietal cortex contributes to the maintenance of tactile pattern information in working memory.     

Gamma-band activity dissociates between matching and nonmatching stimulus pairs in an auditory delayed matching-to-sample task

Electro- and magnetoencephalography studies have suggested that increased gamma-band activity (GBA) is a correlate of activated neural stimulus representations. In this study, a delayed matching-to-sample paradigm for auditory spatial information was employed to investigate the role of magnetoencephalographic gamma-band activity in the differentiation between matching and nonmatching stimulus pairs. Twelve subjects made same-different judgments about the lateralization angle of pairs of filtered noise stimuli (S1 and S2) presented with 0.8-s delays. One half of the subjects had to respond to matching stimulus pairs, the other half to nonmatching stimulus pairs. Cortical oscillatory activity in the memory task was compared to a control task requiring the detection of background noise intensity changes. Memory-related GBA increases were revealed over midline parietal areas in the middle of the delay phase and during the presentation of S2 and over frontocentral areas at the end of the delay phase. This replicated previous findings. In addition, nonmatching trials were associated with increased GBA over right parietal areas in response to S2. The midline parietal GBA increase during S2 in the memory condition may have reflected the representation of S1 needed for a comparison between S1 and S2. When S1 and S2 were identical, no further representation was required. In contrast, for nonmatching pairs, a second representation was activated over right parietal areas.     

Delayed match to object or place: an event-related fMRI study of short-term stimulus maintenance and the role of stimulus pre-exposure

Recent delayed matching studies have demonstrated that maintaining trial-unique stimuli in working memory modulates activity in temporal lobe structures. In contrast, most previous studies that focused on the role of the prefrontal cortex (PFC) used familiar stimuli. We combined fMRI with a delayed-match-to-sample (DMS) task in humans that allowed us to manipulate stimulus pre-exposure (trial-unique vs. familiar objects) and stimulus domain (object vs. location). A visually guided saccade task was used to localize the frontal eye fields (FEF). We addressed two questions: First, we examined whether delay-period activity within PFC regions was more strongly engaged when stimuli were familiar (pre-exposed) than when they were not seen previously (trial-unique). Second, we examined the role of regions within the PFC in object vs. location working memory. Subjects were instructed to remember one stimulus domain while ignoring the other over an 8-s delay period. Object-specific delay-period activity was greatest in the posterior orbitofrontal cortex (OFC) bilaterally, and was stronger for familiar than trial-unique objects. In addition, consistent with previous findings, right posterior superior frontal sulcus, and the FEF were specifically active during the delay period of the location DMS task. These activations outside FEF were not related to saccadic eye movements. In contrast to previous reports, object-specific delay activity was more prominent in the posterior OFC than in the ventrolateral PFC, and was found to be greater for familiar than for trial-unique objects. These results suggest a critical role for the orbitofrontal cortex for maintaining object information in working memory.     

Gamma-band activity over early sensory areas predicts detection of changes in audiovisual speech stimuli

Oscillatory activity in the gamma-band range in human magneto- and electroencephalogram is thought to reflect the oscillatory synchronization of cortical networks. Findings of enhanced gamma-band activity (GBA) during cognitive processes like gestalt perception, attention and memory have led to the notion that GBA may reflect the activation of internal object representations. However, there is little direct evidence suggesting that GBA is related to subjective perceptual experience. In the present study, magnetoencephalogram was recorded during an audiovisual oddball paradigm with infrequent visual (auditory /ta/ + visual /pa/) or acoustic deviants (auditory /pa/ + visual /ta/) interspersed in a sequence of frequent audiovisual standard stimuli (auditory /ta/ + visual /ta/). Sixteen human subjects had to respond to perceived acoustic changes which could be produced either by real acoustic or illusory (visual) deviants. Statistical probability mapping served to identify correlations between oscillatory activity in response to visual and acoustic deviants, respectively, and the detection rates for either type of deviant. The perception of illusory acoustic changes induced by visual deviants was closely associated with gamma-band amplitude at approximately 80 Hz between 250 and 350 ms over midline occipital cortex. In contrast, the detection of real acoustic deviants correlated positively with induced GBA at approximately 42 Hz between 200 and 300 ms over left superior temporal cortex and negatively with evoked gamma responses at approximately 41 Hz between 220 and 240 ms over occipital areas. These findings support the relevance of high-frequency oscillatory activity over early sensory areas for perceptual experience.     

Functional topography of working memory for face or voice identity

We used functional magnetic resonance imaging (fMRI) to investigate whether the neural systems for nonspatial visual and auditory working memory exhibits a functional dissociation. The subjects performed a delayed recognition task for previously unfamiliar faces and voices and an audiovisual sensorimotor control task. During the initial sample and subsequent test stimulus presentations, activation was greater for the face than for the voice identity task bilaterally in the occipitotemporal cortex and, conversely, greater for voices than for faces bilaterally in the superior temporal sulcus/gyrus (STS/STG). Ventral prefrontal regions were activated by both memory delays in comparison with the control delays, and there was no significant difference in direct voxelwise comparisons between the tasks. However, further analyses showed that there was a subtle difference in the functional topography for two delay types within the ventral prefrontal cortex. Face delays preferentially activate the dorsal part of the ventral prefrontal cortex (BA 44/45) while voice delays preferentially activate the inferior part (BA 45/47), indicating a ventral/dorsal auditory/visual topography within the ventral prefrontal cortex. The results confirm that there is a modality-specific attentional modulation of activity in visual and auditory sensory areas during stimulus presentation. Moreover, within the nonspatial information-type domain, there is a subtle across-modality dissociation within the ventral prefrontal cortex during working memory maintenance of faces and voices.     

Reactivation of visual cortex during memory retrieval: content specificity and emotional modulation

Studies on memory retrieval suggest a reactivation of cortical regions engaged during encoding, such that visual or auditory areas reactivate for visual or auditory memories. The content specificity and any emotion dependency of such reactivations are still unclear. Because distinct visual areas are specialized in processing distinct stimulus categories, we tested for face and word specific reactivations during a memory task using functional magnetic resonance imaging (fMRI). Furthermore, because visual processing and memory are both modulated by emotion, we compared reactivation for stimuli encoded in a neutral or emotionally significant context. In the learning phase, participants studied pairs of stimuli that consisted of either a scene and a face, or a scene and a word. Scenes were either neutral or negative, but did not contain faces or words. In the test phase scenes were presented alone (one in turn), and participants indicated whether it was previously paired with a face, a word, or was new. Results from the test phase showed activation in a functionally defined face-responsive region in the right fusiform gyrus, as well as in a word-responsive region in the left inferior temporal gyrus, for scenes previously paired with faces and words, respectively. Reactivation tended to be larger in both the face- and word-responsive regions when the associated scene was negative as compared to neutral. However, relative to neutral context, the recall of faces and words paired with a negative context produced smaller activations in brain regions associated with social and semantic processing, respectively, as well as poorer memory performance overall. Taken together, these results support the idea of cortical memory reactivations, even at a content-specific level, and further suggest that emotional context may produce opposite effects on reactivations in early sensory areas and more elaborate processing in higher-level cortical areas.     

Evidence for a frontal cortex role in both auditory and somatosensory habituation: a MEG study

Auditory and somatosensory responses to paired stimuli were investigated for commonality of frontal activation that may be associated with gating using magnetoencephalography (MEG). A paired stimulus paradigm for each sensory evoked study tested right and left hemispheres independently in ten normal controls. MR-FOCUSS, a current density technique, imaged simultaneously active cortical sources. Each subject showed source localization, in the primary auditory or somatosensory cortex, for the respective stimuli following both the first (S1) and second (S2) impulses. Gating ratios for the auditory M50 response, equivalent to the P50 in EEG, were 0.54+/-0.24 and 0.63+/-0.52 for the right and left hemispheres. Somatosensory gating ratios were evaluated for early and late latencies as the pulse duration elicits extended response. Early gating ratios for right and left hemispheres were 0.69+/-0.21 and 0.69+/-0.41 while late ratios were 0.81+/-0.41 and 0.80+/-0.48. Regions of activation in the frontal cortex, beyond the primary auditory or somatosensory cortex, were mapped within 25 ms of peak S1 latencies in 9/10 subjects during auditory stimulus and in 10/10 subjects for somatosensory stimulus. Similar frontal activations were mapped within 25 ms of peak S2 latencies for 75% of auditory responses and for 100% of somatosensory responses. Comparison between modalities showed similar frontal region activations for 17/20 S1 responses and for 13/20 S2 responses. MEG offers a technique for evaluating cross modality gating. The results suggest similar frontal sources are simultaneously active during auditory and somatosensory habituation.     

Magnetoencephalographic evidence of the interhemispheric asymmetry in echoic memory lifetime and its dependence on handedness and gender

The echoic memory trace (EMT) refers to neuronal activity associated with the short-term retention of stimulus-related information, especially within the primary and association auditory cortex. Using magnetoencephalography it is possible to determine quantitatively the lifetime of the EMT. Previous studies assumed that each new stimulus drives the EMT to its full strength, which then passively decays. In this study we show the limitations of this assumption using trains of auditory stimuli designed specifically for computing the EMT lifetime and its contextual sensitivity. We estimated a time-dependent EMT using a data-driven approach, which allows contributions from a relatively wide area around the auditory cortex in our quantitative measures. We identified: (1) internally generated cortical activations during the silent period between stimuli well separated in time from each other, which had influence on the morphology of the neuromagnetic response to the next external stimulus; and (2) EMTs with different lifetimes that modulate the amplitude of the evoked responses at different latencies, suggesting the existence of multiple neural delay lines. Long EMT lifetimes were observed on the descending part of the M100 complex, which showed handedness and gender-dependent interhemispheric asymmetry. Specifically, all subjects showed longer EMT lifetimes on the left hemisphere, except left-handed males. Distributed source analysis of the data for one left- and one right-handed male subject identified a secondary generator in the right-handed subject, which was located posterior to the early primary generator and dominated the auditory response at late latencies, where EMT lifetime asymmetry was high. The identified multiple neural delay lines and their laterality may provide a link between macroneuronal activity and left hemisphere specialization for processing linguistic material.     

Evidence for Dissociation of Spatial and Nonspatial Auditory Information Processing

Several lines of evidence suggest that visual information processing is segregated into the ventral "what" and dorsal "where" pathways. But the question whether information processing in the auditory system is also parceled to spatial and nonspatial domains remains open. In the present study, we performed simultaneous EEG and MEG recordings during auditory location and pitch delayed matching-to-sample tasks to find out whether working memory processing of the auditory stimulus attribute affects the transient components of the evoked potentials. In both tasks, identical blocks of tone stimuli of one of two frequencies were presented in one of two locations; the only difference between the tasks was the instruction to attend either to the frequency or to the location. In the match condition, the N1 latency was shorter and the N1m amplitude larger in the location task compared to the pitch task. Furthermore, the right-hemisphere generator of N1m elicited in the match condition of the location task was situated significantly medially to the N1m generator in the match condition of the pitch task. Latency and amplitude task-related differences in the N1/N1m components as well as the source location differences indicate at least partial segregation of neuronal mechanisms involved in working memory processing of spatial and nonspatial auditory information.     

Parietal activation during retrieval of abstract and concrete auditory information

Successful memory retrieval has been associated with a neural circuit that involves prefrontal, precuneus, and posterior parietal regions. Specifically, these regions are active during recognition memory tests when items correctly identified as "old" are compared with items correctly identified as "new." Yet, as nearly all previous fMRI studies have used visual stimuli, it is unclear whether activations in posterior regions are specifically associated with memory retrieval or if they reflect visuospatial processing. We focus on the status of parietal activations during recognition performance by testing memory for abstract and concrete nouns presented in the auditory modality with eyes closed. Successful retrieval of both concrete and abstract words was associated with increased activation in left inferior parietal regions (BA 40), similar to those observed with visual stimuli. These results demonstrate that activations in the posterior parietal cortex during retrieval cannot be attributed to bottom-up visuospatial processes but instead have a more direct relationship to memory retrieval processes.     

The representation of shape in the context of visual object categorization tasks

To investigate the role of human fusiform gyrus in shape processing, we determined the effect of shape degradation on BOLD contrast in this region with fMRI during three tasks requiring subjects to determine either whether two successively presented nonsense shapes had the same global orientation (OR task); whether two successively presented meaningful objects belonged to the same basic level category (CAT task); or whether two successively presented objects represented the same exemplar of a category (EX task). On the behavioral level, shape degradation by locally shifting the pixels constituting the lines of stimuli had no effect on performance in the OR task, while it was detrimental to performance in the CAT and EX tasks. In comparison to the OR task, both the CAT and EX tasks were associated with activations in the occipitotemporal and parietal cortex. When shape degradation was applied, activation in the middle fusiform gyrus was reduced in all tasks. The occurrence of this effect in the OR task indicates that it is independent of memory representations. The persistence of the effect in both tasks that showed a behavioral effect of degradation suggests that it does not reflect the amount of shape processing performed on the stimuli, but rather the specificity of the final perceptual representation that can be built from the shape information that is available. Other studies have shown effects of stimulus familiarity and task requirements in the fusiform gyrus, suggesting that there is no need to assume different modules for perceptual representation and representation in memory.     

Exploring the temporal dynamics of the spatial working memory n-back task using steady state visual evoked potentials (SSVEP)

The neural networks associated with spatial working memory (SWM) are well established. However, the temporal dynamics of SWM-related brain activity are less clear. This study examined changes in temporal neurophysiology during the spatial n-back task using steady state probe topography (SSPT) to record cortical steady state visual evoked potentials (SSVEPs) at 64 scalp locations. Twenty healthy male volunteers participated in the study. The findings identified three different time periods of significance during the spatial n-back task--an early perceptual/encoding period (approximately 0-500 ms), an early delay period just following the stimulus disappearing from view (approximately 850-1400 ms), and a late period lasting the final second of the delay and anticipation of the new stimulus (approximately 2500-3500 ms). The delay period was associated with increases in frontal and occipital region amplitude, consistent with previous findings in more basic working memory tasks. The two different SSVEP components during the delay appear reflective of the additional "executive" demands associated with the n-back and may suggest variable roles for the PFC during different stages of the delay. All three n-back levels demonstrated a relative consistent electrophysiological profile, indicating that this pattern is specific to the spatial n-back task. Nevertheless, these findings supported the hypothesis that memory load modulates activity within the networks identified, consistent with previous neuroimaging studies. The current findings may offer a framework in which to further investigate the temporal aspects of SWM.     

Give it time: Neural evidence for distorted time perception and enhanced memory encoding in emotional situations

Time perception is compromised in emotional situations, yet our ability to remember these events is enhanced. Here we suggest how the two phenomena might be functionally linked and describe the neural networks that underlie this association. We found that participants perceived an emotionally aversive stimulus longer than it was, compared to an immediately following neutral stimulus. These time estimation errors were in the same trials associated with better recognition memory for the emotionally aversive stimuli and poorer memory for the neutral stimuli. Functional imaging revealed that the superior frontal gyrus was activated during time perception with aversive stimuli, and the amygdala, putamen and insula showed activations that are specific to time estimation errors in this aversive context. We further found that activity in the insula and putamen was correlated with memory performance but only during over-estimation of time with the aversive stimuli. We suggest that processing is accelerated during the experience of emotionally aversive events, presumably in the service of memory-related operations, resulting in better encoding but at the expense of time perception accuracy.     

Functional anatomy of pitch memory--an fMRI study with sparse temporal sampling

Auditory functional magnetic resonance imaging tasks are challenging since the MR scanner noise can interfere with the auditory stimulation. To avoid this interference a sparse temporal sampling method with a long repetition time (TR = 17 s) was used to explore the functional anatomy of pitch memory. Eighteen right-handed subjects listened to a sequence of sine-wave tones (4.6 s total duration) and were asked to make a decision (depending on a visual prompt) whether the last or second to last tone was the same or different as the first tone. An alternating button press condition served as a control. Sets of 24 axial slices were acquired with a variable delay time (between 0 and 6 s) between the end of the auditory stimulation and the MR acquisition. Individual imaging time points were combined into three clusters (0-2, 3-4, and 5-6 s after the end of the auditory stimulation) for the analysis. The analysis showed a dynamic activation pattern over time which involved the superior temporal gyrus, supramarginal gyrus, posterior dorsolateral frontal regions, superior parietal regions, and dorsolateral cerebellar regions bilaterally as well as the left inferior frontal gyrus. By regressing the performance score in the pitch memory task with task-related MR signal changes, the supramarginal gyrus (left>right) and the dorsolateral cerebellum (lobules V and VI, left>right) were significantly correlated with good task performance. The SMG and the dorsolateral cerebellum may play a critical role in short-term storage of pitch information and the continuous pitch discrimination necessary for performing this pitch memory task.     

The brain network associated with acquiring semantic knowledge

There is ongoing debate about how semantic information is acquired, whether this occurs independently of episodic memory, and what role, if any, brain areas such as hippocampus are required to play. We used auditory stimuli and functional MRI (fMRI) to assess brain activations associated with the incidental acquisition of new and true facts about the world of the sort we are exposed to day to day. A control task was included where subjects heard sentences that described novel scenarios involving unfamiliar people, but these did not convey general knowledge. The incidental encoding task was identical for two stimulus types; both shared the same episodic experience (lying in the brain scanner) and conveyed complex information. Despite this, and considering only those stimuli successfully encoded, compared to a baseline task, a more extensive network of brain regions was found to be associated with exposure to new facts including the hippocampus. Direct comparison between the two stimulus types revealed greater activity in dorsal, ventrolateral and dorsomedial prefrontal cortex, medial dorsal nucleus of the thalamus, and temporal cortex for fact stimuli. The findings suggest that successful encoding is not invariably associated with activation of one particular brain network. Rather, activation patterns may depend on the type of materials being acquired, and the different processes they engender when subjects encode. Qualitatively, from postscan debriefing sessions, it emerged that the factual information was found to be potentially more useful. We suggest that current or prospective utility of incoming information may be one factor that influences the processes engaged during encoding and the concomitant neuronal responses.     

Sensory gating of auditory evoked and induced gamma band activity in intracranial recordings

Oscillatory activity in the gamma band range (30-50 Hz) and its functional relation to auditory evoked potentials (AEPs) is yet poorly understood. In the current study, we capitalized on the advantage of intracranial recordings and studied gamma band activity (GBA) in an auditory sensory gating experiment. Recordings were obtained from the lateral surface of the temporal lobe in 34 epileptic patients undergoing presurgical evaluation. Two kinds of activity were differentiated: evoked (phase locked) and induced (not phase locked) GBA. In 18 patients, an intracranial P50 was observed. At electrodes with maximal P50, evoked GBA occurred with a similar peak latency as the P50. However, the intensities of P50 and evoked GBA were only modestly correlated, suggesting that the intracranial P50 does not represent a subset of evoked GBA. The peak frequency of the intracranial evoked GBA was on average relatively low (approximately 25 Hz) and is, therefore, probably not equivalent to extracranially recorded GBA which has normally a peak frequency of approximately 40 Hz. Induced GBA was detected in 10 subjects, nearly exclusively in the region of the superior temporal lobe. The induced GBA was increased after stimulation for several hundred milliseconds and encompassed frequencies up to 200 Hz. Single-trial analysis revealed that induced GBA occurred in relatively short bursts (mostly <100 ms), indicating that the duration of the induced GBA in the averages originates from summation effects. Both types of gamma band activity showed a clear attenuation with stimulus repetition.     

Enhancing cognitive control through neurofeedback: A role of gamma-band activity in managing episodic retrieval

Neural synchronization has been proposed to be the underlying mechanism for exchanging and integrating anatomically distributed information and has been associated with a myriad of cognitive domains, including visual feature binding, top–down control, and long-term memory. Moreover, it seems that separate frequency bands have different functions in these cognitive processes. Here we studied whether neurofeedback training designed either to increase local gamma band activity (GBA+; 36–44 Hz), or local beta band activity (BBA+; 12–20 Hz), would have an impact on performance of behavioral tasks measuring short-term and long-term episodic binding. Our results show that GBA-enhancing neurofeedback training increased occipital GBA within sessions, and occipital and frontal GBA across sessions. Both groups showed an increase of GBA coherence between frontal and occipital areas, but the BBA+ group increased BBA coherence between these areas as well. Neurofeedback training had profound effects on behavior. First, we replicated earlier findings that enhancing GBA led to greater flexibility in handling (selectively retrieving) episodic bindings, which points to a role of GBA in top–down control of memory retrieval. Moreover, the long-term memory task revealed a double dissociation: GBA-targeted training improved recollection, whereas BBA-targeted training improved familiarity memory. We conclude that GBA is important for controlling and organizing memory traces of relational information in both short-term binding and long-term memory, while frontal–occipital coherence in the beta band may facilitate familiarity processes.     

The dorsal auditory pathway is involved in performance of both visual and auditory rhythms

We used functional magnetic resonance imaging to investigate the effect of two factors on the neural control of temporal sequence performance: the modality in which the rhythms had been trained, and the modality of the pacing stimuli preceding performance. The rhythms were trained 1-2 days before scanning. Each participant learned two rhythms: one was presented visually, the other auditorily. During fMRI, the rhythms were performed in blocks. In each block, four beats of a visual or auditory pacing metronome were followed by repetitive self-paced rhythm performance from memory. Data from the self-paced performance phase was analysed in a 2x2 factorial design, with the two factors Training Modality (auditory or visual) and Metronome Modality (auditory or visual), as well as with a conjunction analysis across all active conditions, to identify activations that were independent of both Training Modality and Metronome Modality. We found a significant main effect only for visual versus auditory Metronome Modality, in the left angular gyrus, due to a deactivation of this region after auditory pacing. The conjunction analysis revealed a set of brain areas that included dorsal auditory pathway areas (left temporo-parietal junction area and ventral premotor cortex), dorsal premotor cortex, the supplementary and presupplementary premotor areas, the cerebellum and the basal ganglia. We conclude that these regions are involved in controlling performance of well-learned rhythms, regardless of the modality in which the rhythms are trained and paced. This suggests that after extensive short-term training, all rhythms, even those that were both trained and paced in visual modality, had been transformed into auditory-motor representations. The deactivation of the angular cortex following auditory pacing may represent cross-modal auditory-visual inhibition.