Edited by: Gina Kuperberg, Tufts University, USA
Reviewed by: David Corina, UC Davis, USA; Anna Simmonds, Imperial College London, UK; Roel M. Willems, Donders Institute for Brain, Cognition and Behaviour, Netherlands; Jonathan Henry Venezia, University of California, Irvine, USA
*Correspondence: Stefan Heim, Department of Psychiatry, Psychotherapy and Psychosomatics, Medical School, RWTH Aachen University, Pauwelsstraße 30, 52074 Aachen, Germany. e-mail:
This article was submitted to Frontiers in Language Sciences, a specialty of Frontiers in Psychology.
This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third-party graphics etc.
Research on the evolutionary basis of the human language faculty has proposed the mirror neuron system as a link between motor processing and speech development. Consequently, most work has focused on the left inferior frontal cortex, in particular Broca’s region, and the left inferior parietal cortex. However, the direct link between planning of hand motor and speech actions has yet to be elucidated. Thus, the present study investigated whether motor sequencing of hand vs. speech actions has a common neural denominator. For the hand motor task, 25 subjects performed single, repeated, or sequenced button presses with either the left or right hand. The speech task was in analogy; the same subjects produced the syllable “po” once or repeatedly, or a sequence of different syllables (“po-pi-po”). Speech motor vs. hand motor effectors resulted in increased perisylvian activation including Broca’s region (left area 44 and areas medially adjacent to left area 45). In contrast, common activation for sequenced vs. repeated production of button presses and syllables revealed the effector-independent involvement of left area 7A in the superior parietal lobule (SPL) in sequencing. These data demonstrate that sequencing of vocal gestures, an important precondition for ordered utterances and ultimately human speech, shares area 7A, rather than inferior parietal regions, as a common cortical module with hand motor sequencing. Interestingly, area 7A has previously also been shown to be involved in the observation of hand and non-hand actions. In combination with the literature, the present data thus suggest a distinction between area 44, which is specifically recruited for (cognitive aspects of) speech, and SPL area 7A for general aspects of motor sequencing. In sum, the study demonstrates a previously underspecified role of the SPL in the origins of speech, and may be discussed in the light of embodiment of speech and language in the motor system.
“Every journey begins with a first step” – This Chinese proverb poetically characterizes a process most relevant in many domains of human life: sequencing. Sequencing means that complex actions such as speaking or grasping are broken down into a series of individual steps or sub-processes, which are taken in a particular order as time unfolds. As we grasp a key to unlock a door, we perform a series of arm, hand, and finger movements targeted toward the key, to move it toward the door lock, and to turn it. As we speak, we create a syntactic sentence structure in our mind, fill its slots with words, serially retrieve the words’ phonemes and realize them sequentially with our articulators as series of vocal gestures.
The outstanding question is therefore whether “sequencing” is one common process which supports the domains of motor actions and speech production alike, or whether each domain has its own sequencing affordances and modules. About 40 years ago, Doreen Kimura addressed this question in a number of well-designed and, at that time, technically well advanced studies (e.g., Kimura,
A more recent analysis of cognitive and neural components commonly involved in speech and praxis indeed suggests that both domains rely on shared resources (Roby-Brami et al.,
The concept of “sequencing,” however, may actually comprise several aspects, in particular one “cognitive,” i.e., the mental planning of a sequence, and one “motor” focusing on the actual serial execution of units or chunks of actions. Following the two-stage model by Klapp (
Several groups have approached the issue of shared sequencing resources in motor praxis and speech with functional neuroimaging studies on healthy volunteers. Reviewing the available evidence, Fiebach and Schubotz (
However, there are a couple of potential arguments against the idea of a shared neural sequencing module in motor actions and speech that should be considered, one neurofunctional and one neuroanatomical. First, neurofunctionally, there is growing consensus that higher cognitive functions such as sequencing are supported by distributed networks, rather than isolated functionally specified regions (for the phonology-to-articulation loop cf. Eickhoff et al.,
To summarize, the current view on “sequencing” regions in the motor and speech domains appears as follows. (1) At a cognitive level, sequencing may be a component that is shared, as module, in different variants of motor and speech processing. Given the heterogeneous findings listed above, a distinction between “cognitive sequencing” (involving a planning stage) and “motor sequencing” (involving the ordered execution), and consequent focus on the one or other, might further clarify matters. (2) At a neurofunctional level, there are hints at potentially shared neural systems in frontal and parietal regions. (3) At a neuroanatomical level, it is far from clear whether such systems recruit the same, or rather neighboring, cytoarchitectonic areas.
The present study was therefore designed to investigate motor sequencing in finger movements and in speech production in healthy controls using functional magnetic resonance imaging (fMRI) in combination with the Düsseldorf-Jülich cytoarchitectonic brain atlas.
In particular, we addressed the following question: are there brain regions involved in sequencing that are shared in finger movements and speech production?
Twenty-five healthy volunteers (average age: 35 years; age range 30–61 years; 11 women) with no reported history of neurological or psychiatric disorders participated in the study. All were right-handed according to the Edinburgh Handedness Inventory (Oldfield,
The functional scan consisted of six blocks, two for the right hand (R), two for the left hand (L), and two for overt speech (S). In each block, the participants completed trials for three different modes which were presented in a pseudo-randomized order: trials with a single response (ONE), trials in which the same response was repeated three times (REP), and trials in which a sequence of different responses was performed (SEQ). Combining three response modes (right hand, left hand, speech) with three gave a total of nine experimental conditions (R-ONE, R-REP, R-SEQ, L-ONE, L-REP, L-SEQ, S-ONE, S-REP, S-SEQ). A visual cue at the beginning of each trial indicated which response was required. In R-ONE trials, the participants pressed a response button once with the index finger of the right hand. For R-REP, they pressed the same button three times repeatedly. Finally, for R-SEQ, they pressed an alternating sequence with their right index, middle, and index finger. The same was done with the index finger (or index and middle fingers) of left hand for L-ONE, L-REP, and L-SEQ. Finally, for spoken responses, S-ONE required saying the syllable “po,” S-REP saying the same three times (“po-po-po”), and S-SEQ producing a diadochokinesis-like sequence “po-pi-po.” In each of the six blocks, there were 30 trials for each response mode (ONE, REP, SEQ) plus 10 null events, amounting to a total of 540 task events completed by each participant.
Each trial started with an empty screen for 1900 ms, followed by the presentation of a fixation cross for 1000 ms. Then, the visual task cue appeared for 1000 ms, and the participants gave their (hand motor or spoken) response while the task cue remained on the screen for the rest of the trial. In the null events, a blank was shown instead of the cue. Responses were restricted to a time window of 1000 ms (see below). Practice trials before scanning ensured that the subjects were familiar with all conditions and responded correctly and with constant speed to the cue.
Scanning was performed on a Trio 3T scanner (Siemens, Erlangen, Germany) located at the Research Centre in Jülich. A time series of 616 T2*-weighted EPI images (flip angle FA: 90°; echo time TE: 30 ms; field of view FOV: 200 mm; matrix: 64 × 64) were acquired from 49 sagittal slices with a thickness of 3 mm. The sagittal orientation of the slices was chosen in order to better correct for nodding head movements in the
After the functional scan, a T1-weighted MPRAGE sequence (FOV 256 mm; TR 2250 ms; TE 3,03 ms; FA: 9°) was run in order to obtain structural images of each participant’s brain at an isotropic resolution of 1 mm × 1 mm × 1 mm.
Analysis of the functional data was performed with SPM5 (Wellcome Trust Centre for Neuroimaging, UK) running on Matlab 7.0 (The MathWorks, Inc., USA). Images were first corrected for head movement by affine registration using a two-pass procedure, by which images were initially realigned to the first image and subsequently to the mean of the realigned images. After realignment, the mean EPI image for each subject was spatially normalized using the “unified segmentation” approach (Ashburner and Friston,
The first level event-related statistical analysis of each individual data set consisted of convolving the onset vectors for each of the nine conditions with the canonical hemodynamic response function. Estimates of the regressors for each condition were obtained by contrasting each condition against the implicit resting baseline (contrasts of the type “1 0 0 0 …”). These contrast images were then entered into a random-effects group analysis at the second level with “subject” as repetition factor and “condition” as experimental factor. In this analysis, we used the following contrasts in order to identify, and juxtapose, effector-independent networks for motor sequencing vs. repetition:
Main effect left hand sequencing (L-SEQ > L-REP)
Main effect right hand sequencing (R-SEQ > R-REP)
Main effect speech sequencing (S-SEQ > S-REP)
Conjunction analysis (SPM option “conjunction null”) of the three main effects for SEQUENCING, revealing regions commonly involved in sequencing in all three effectors, i.e., independently of effector which are assigned the least of the
All individual contrasts were calculated at
For the anatomical localization of the activations we used cytoarchitectonic probability maps, which are based on an observer-independent analysis of the cytoarchitecture in a sample of 10 post-mortem brains (Zilles et al.,
The results for sequencing effects (SEQ > REP) are displayed in Figure
Broca’s region was not involved in the sequencing in the repetition contrast (a finding that is being discussed below). Therefore, two additional analyses were computed. First, in order to investigate whether the Fiebach and Schubotz (
With respect to the first analysis, the upper part of Figure
With respect to Speech > Hand Motor processing, the lower part of Figure
The present study investigated whether the process of motor sequencing is a shared resource in the speech domain and in the hand motor domain, and whether there is a common neurofunctional substrate which is distinct from brain regions distinct from brain regions merely supporting repeated movements. Indeed, by using comparisons between sequences, repetitions, or single speech or motor acts, we could identify one such region. This region was not located in the left frontal or premotor cortex, but rather in cytoarchitectonic area 7A in the SPL. This effect for sequencing was clearly distinct from that for repetitive actions, which commonly involved bilateral SMA (area 6) and bilateral IPL (areas PFcm), as well as the cerebellum.
The one region that was significantly stronger involved in sequencing rather than repetition of right hand movements, left hand movements, and speech was area 7A in the left SPL. This shared involvement is a novel finding which, however, resonates well with previous reports on the neural correlates of sequential hand/finger movements. Van Oostende et al. (
The current observation that the left SPL is involved in sequencing across modalities across modalities is thus well in line with its previously presumed (more specific) role in sequencing hand motor actions. Its left-lateralization shows that it is not related to secondary sensory-motor processing, which would depend on the use of left vs. right hand and thus recruit right and left regions, respectively. The stronger involvement of the left SPL (as compared to right SPL) within an otherwise bilateral motor network has recently been demonstrated by Otten et al. (
At this point, it should be noted that, in general, the role of SPL area 7A is distinct from that of its neighboring area 5. Whereas area 7A has been shown to be involved in sequencing in the hand motor domain, and in the present study also in the speech domain, the function of area 5 is best defined in the context of reaching gestures in the hand motor domain. For instance, McGuire and Sabes (
As discussed above, the finding that area 7A the left SPL plays a crucial role for sequencing also in speech nicely mirrors findings from the hand motor domain. Still, as outlined in the introduction, this result is somewhat in contrast to the prevalent hypothesis that it should be frontal, rather than parietal, areas that support sequencing. One reason can be found in the (again) rather coarse anatomical concept of a “frontal” cortex which, as discussed above, consists of several cytoarchitectonically (and presumably also functionally) distinct areas. Fiebach and Schubotz (
A number of studies on aspects of cognitive processing observed the involvement of Broca’s region in the processing of actions (e.g., Binkofski et al.,
In line with this argument that area 44 is not directly involved in motor sequencing, but may rather be involved in cognitive sequencing for all required responses (ONE; REP; SEQ), is the fact that the conjunction of all sequencing conditions (when these were contrasted only against the implicit resting baseline) again yielded a cluster overlapping with left area 44. In this contrast, cognitive sequencing, i.e., planning, does not cancel out because it is present in the task but not during rest. This may be the reason why area 44 shows up in the baseline contrast but not in the contrast explicitly tapping into motor sequencing.
Having argued for a shared sequencing module in speech and hand motor processing, the question arises how generalizable this finding is – i.e., does this result extend to other fields of sequencing? As outlined above, the speech production process involves sequencing at a number of different stages, from the coordination of the articulators over the sequential retrieval and realization of phonemes and syllables to ordered words and sentences. Thus, in its most extreme form, one question would be whether complex hand and finger movements and hierarchical phrase-structure grammars (e.g., Friederici et al.,
A link that can be conceived of more easily is that to writing. Writing is an ordered finger-hand action requiring sequencing. This sequencing is based on the phoneme-to-grapheme conversion by which letters are sequentially assigned to the speech sounds of a word, and then realized as sequential hand-finger movements. Indeed, a recent study by Segal and Petrides (
Another link worth following is that to sign language and gesturing. Indeed, sign language, like spoken language, requires substantial amounts of cognitive sequencing. Interestingly, and in contrast to spoken language, motor sequencing does not involve speech articulators but rather the hands. Thus, sign language processing represents a model in which cognitive-linguistic sequencing is linked to hand motor sequencing. In line with the present findings of the role of left SPL area 7A, a neuroimaging study by Emmorey et al. (
One limitation of the study is that only motor sequencing for hand and speech action was investigated, leaving the comparison to cognitive sequencing for future research. Another potential limitation could be seen in the choice of the stimuli, which consisted of sequences of one single unit (po) or two different units (po/pi), resulting in just two sequences (popopo vs. popipo). One might raise the objection that producing either popopo or popipo can be seen as producing one pre-learned item out of a set of two alternatives, with no requirements for sequencing whatsoever. Whereas it is true that a bigger set of units to be potentially combined would, from this perspective, be favorable, it might, on the other hand, have led to increased confusion of the subjects who would have had to learn more cue-response combinations in order to perform the correct response – a setting with higher cognitive demands we did not wish to include. Whether or not a larger number of stimuli would be useful, in particular to investigate cognitive sequencing, the fact remains that, at the motor level, sequencing was required for both popipo and popopo, with higher demands (due to the alternation of the units po and pi) for what we conceptualized as the “sequencing” condition. Moreover, if both popopo and popipo merely represented two templates (as instances of item learning), one would expect both conditions to elicit comparable brain activation – which was clearly not the case, in particular not in SPL area 7A. Rather, evidence from a study by Bohland and Guenther (
To conclude, the present study provided evidence that motor sequencing in the hand and in the speech domain shares a neural component in area 7A in the left SPL. This is a novel finding, extending data from earlier studies related only to the hand motor domain. The nature of this link needs to be investigated further with respect to hierarchies in speech sounds (e.g., complex phoneme clusters), syntax, finger movements, and complex higher-order actions not only from the motor, but also from a cognitive perspective. Moreover, the present study dissociated this sequencing finding from areas involved in repetition both in speech and finger actions, showing that sequencing of syllable production and alternating finger tapping require more than repeated access to, or execution of, movement plans. In sum, the findings are encouraging to further pursue investigations of the language-motor interface, with findings that may have implications for interventions for patients with deficits affecting one but sparing the other domain.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The results for repetition effects (REP > ONE) are displayed in Figure
Moreover, the inferior parietal lobule also showed shared effects of repetition in both hemispheres. One cluster was located in the right supramarginal gyrus with its local maximum (
While the present study primarily focused on areas supporting sequencing in the speech and the hand motor domain, it was also designed to identify brain regions commonly involved in repetition of actions (as compared to single actions). Such distinction between sequencing and repetition further underscores the uniqueness of the findings for sequencing. At the same time, it enhances the understanding of the link between the speech and the hand motor system in the brain at a more general level.
In the repetition contrast, a bilateral network was observed which included areas PFcm in the supramarginal gyri and areas 6 in the left and right SMA. These findings again extend knowledge from the hand motor literature which reported the involvement of the SMA (e.g., Sadato et al.,