Fig 1.
A sample rhythm pattern was presented at the beginning of each trial. Each rhythm pattern consisted of three repeats of a three-pure-tone sequence. The participants were required to memorize the rhythm pattern within 6 s (encoding phase), to maintain the rhythm information for 6–12 s (maintenance phase), and reproduce it by tapping with the right index finger, left index finger or right foot, or by articulation within 8 s (retrieval phase). We used 20 rhythmic patterns for each participant. The two out of three durations (SOA1, SOA2 and IUI) in each pattern were constantly same. The SOA and IUI were chosen from six possible durations (0.4, 0.5, 0.6, 0.8, 1.0 or 1.2 s) so that the duration of each rhythmic pattern ranged from 4 to 6 s. Abbreviations: SOA, stimulus onset asynchrony; IUI, inter-unit-interval.
Table 1.
Effector-independent and dependent regions revealed by conjunction analyses.
Fig 2.
Graph demonstrating the accuracy of the rhythm reproduction.
The graph shows the mean GIR index value in each effector condition of rhythm working memory tasks. A one-way ANOVA revealed that the accuracy of rhythm reproduction was consistent among these 4 effectors. Error bars indicate the standard error of the mean. See Methods for details of the GIR index. Abbreviations: Rt., right; Lt., left.
Fig 3.
Cortical brain regions exhibiting significant activations during the encoding (left panel) and the retrieval (right panel) phases of the rhythm working memory tasks compared to the number working memory tasks (threshold: P < 0.05; FWE-corrected of the clusters).
A) right finger, B) left finger, C) foot, and D) mouth conditions.
Fig 4.
A) Results of conjunction analysis. Cortical brain regions in common activated by all four effectors showed significant activations in the rhythm working memory tasks compared to the number working memory tasks (threshold: P < 0.05; FWE-corrected of the clusters). B) Response profile for each activation peak. Response profiles are shown for the right IPL and IFG in the encoding and retrieval phases. From these peaks, parameter estimates were extracted for all (rhythm—number) contrasts. Error bars indicate the standard error of the mean. Abbreviations: IFG, inferior frontal gyrus; IPL, inferior parietal lobule. REr represents Rhythm-Encoding-right finger condition. See Methods for details of the contrasts.
Fig 5.
A) Results of second conjunction analysis. Cortical brain regions demonstrating common brain activations by the right index finger, the left index finger and the right foot (except for the mouth), showed significant activations in the rhythm working memory tasks compared to the number working memory tasks (threshold: P < 0.05; FWE-corrected of the clusters). Red areas are consisted with the cortical regions, which were in common activated by all four effectors (Fig 4A). Yellow areas represent the additional regions in common activated by the three effectors, resulting from this second conjunction analysis. B) Response profiles are shown for the left IFG during rhythm encoding and the left IPL and the SMA during rhythm retrieval. The peak voxels of response profiles were obtained from the results of mean activations across all effectors using a contrast [(RXr—NXr) + (RXl—NXl) + (RXf—NXf) + (RXm—NXm)]. See Fig 4 legend for details of the response profile.
Fig 6.
Functional connectivity during rhythm encoding.
A) Three seed regions; right IPL (red), right IFG (blue), and left IFG (yellow) from left to right. B) Brain areas exhibiting significantly increased coupling with the seed regions in the presence of rhythm encoding. Colored areas show significant coupling with the seed region in the same color. Red, blue, and yellow regions show significant functional connectivity with the right IPL, right IFG, and let IFG, respectively. Threshold: P < 0.05; FWE-corrected of the clusters. Abbreviations: CB, cerebellum; IFG, inferior frontal gyrus; STG, superior temporal gyrus.
Table 2.
Functional connectivity analysis.
Fig 7.
Functional connectivity during rhythm retrieval.
A) Three seed regions; right IPL (red), left IPL (yellow), SMA (blue) from left to right. B) The brain areas exhibiting significantly increased coupling with the seed regions in the presence of rhythm retrieval. Red, yellow, and blue regions show significant functional connectivity with the right IPL, left IPL, and SMA, respectively. Right panel: left finger condition, left panel: foot condition. Threshold: P < 0.05; FWE-corrected of the clusters. Abbreviations: SMA, supplementary motor area; SMC, somatosensory/motor cortex.
Fig 8.
Hypothetical model of working memory of rhythm.
During rhythm encoding, the bilateral inferior frontal gyri (IFGs) and right inferior parietal lobule (IPL) receive auditory rhythm information from the superior temporal gyri (STGs), and encode the information as a temporal structured sequence. The temporal sequence of the rhythm is stored in the right fronto-parietal network. During rhythm retrieval, the IPLs and supplemental motor area (SMA) transform the temporal sequence into the motor sequence of the rhythm, and reproduce the rhythm through various effectors while monitoring the output with the feedback signals from the somatosensory/motor cortices. Red lines indicate functional connectivities during rhythm encoding, and the blue lines indicate those during rhythm retrieval. The filled rectangles indicate the estimated functional units of working memory of rhythm; encoding (red), retrieval (blue) and both (purple).