Abstract

The inferior frontal gyrus (IFG) is active during both goal-directed action and while observing the same motor act, leading to the idea that also the meaning of a motor act (action understanding) is represented in this “mirror neuron system” (MNS). However, in the dual-loop model, based on dorsal and ventral visual streams, the MNS is thought to be a function of the dorsal steam, projecting to pars opercularis (BA44) of IFG, while recent studies suggest that conceptual meaning and semantic analysis are a function of ventral connections, projecting mainly to pars triangularis (BA45) of IFG. To resolve this discrepancy, we investigated action observation (AO) and imitation (IMI) using fMRI in a large group of subjects. A grasping task (GR) assessed the contribution from movement without AO. We analyzed connections of the MNS-related areas within IFG with postrolandic areas with the use of activation-based DTI. We found that action observation with imitation are mainly a function of the dorsal stream centered on dorsal part of BA44, but also involve BA45, which is dorsally and ventrally connected to the same postrolandic regions. The current finding suggests that BA45 is the crucial part where the MNS and the dual-loop system interact.

Introduction

The current concept of goal-directed movement is based on the integration of sensory information into a motor format. The related visual information, according to the model of Goodale and Milner, arises from the same early visual areas and takes 2 different pathways (Goodale and Milner 1992; Milner and Goodale 2008). The “ventral visual stream” (for “object semantics”) accesses the ventral temporal areas, and the “dorsal visual stream” (vision for action) reaches the parietal lobule (Milner and Goodale 2008). In recent years, this model was extended and more general functions of ventral and dorsal visual streams were described in a dual-loop model in different domains. This dual-loop model makes a distinction between a dorsal and a ventral stream (or pathway) with different properties, anatomically defined by their trajectories of the connecting long association tracts, which run either above or below the Sylvian fissure (Umarova et al. 2010, 2011; Rijntjes et al. 2012; Rijntjes and Weiller 2013). The dorsal stream is suggested to have the general capacity, independent from the domain, to analyze the sequence of segments, either in time or in space, as well as fast on-line integration between sensory event information and “internal models or emulators” (Rauschecker and Scott 2009). The ventral stream is thought to be responsible for the time-independent identification of an invariant set of categories that are related to semantic memory and meaning (Rauschecker and Scott 2009; Weiller et al. 2011; Rijntjes et al. 2012). It is of interest how the dual-loop model is related to the human “observation-execution matching system” or the human “mirror neuron system” (MNS) (Rizzolatti and Craighero 2004). The MNS is based on the “mirror” character of so-called mirror neurons within monkey's area F5 of the ventral premotor cortex (vPMC), which are active during both execution and observation of goal-directed motor acts (di Pellegrino et al. 1992; Gallese et al. 1996; Rizzolatti et al. 1996). Because of the multimodal character of mirror neurons, it is generally accepted that, in humans, they play a key role in imitation (Iacoboni et al. 1999), recognition of related context (Iacoboni et al. 2005), and attention toward the meaning more than toward the goal of an observed action (Hesse et al. 2009). Consequently, by observing an action, the MNS interprets the action itself as well as the underlying intention (e.g., (Fabbri-Destro et al. 2008)). However, current data in humans are inconclusive in reconciling these 2 models (MNS and dual-loop model). In the dual-loop model, the MNS is thought to be a function of the dorsal steam, projecting to pars opercularis (BA44) of the inferior frontal gyrus (IFG), while recent studies suggest that conceptual meaning and semantic analysis are a function of ventral connections, projecting mainly to pars triangularis (BA45) of IFG (Rauschecker and Scott 2009; Weiller et al. 2011; Rijntjes et al. 2012). However, 2 recent meta-analyses comparing action observation and imitation, one pooling data from a single laboratory (Molnar-Szakacs et al. 2005), another one a review of 139 studies in current literature (Caspers et al. 2010), both showed involvement of a dorsal part of BA44 (BA44d) of IFG, which is generally regarded as part of the dorsal system (Umarova et al. 2010, 2011; Rijntjes et al. 2012; Rijntjes and Weiller 2013; Vry et al. 2012; Saur et al. 2008; Yeterian et al. 2012), but not the involvement of BA45. Therefore, it seems unresolved where the recognition of the meaning of a movement takes place and whether, for this process, the ventral stream plays a role. Furthermore, the role of ventral BA44 in the MNS is unclear. Both recent meta-analyses found a functional parcellation of BA44, reflecting the parcellation of BA44 into ventral BA44 (BA44ventral, BA44v) and dorsal BA44 (BA44dorsal, BA44d) (Amunts et al. 2010). BA44d was involved in both action observation and imitation whereas BA44v was predominantly involved in imitation tasks (Caspers et al. 2010; Molnar-Szakacs et al. 2005).

To overcome shortcomings of meta-analysis, while very worthwhile in themselves, we investigated action observation and imitation in a large group of subjects to identify areas in the IFG involved in the MNS. Also, we included a grasping task without visual control to assess the contribution from movement without action observation to activation in the IFG. Thereafter, we analyzed the connection of the related seed regions of the MNS-related functional activated areas within postrolandic regions with those of the prefrontally activated areas. The combination of functional with connectional data might provide converging evidence for a more detailed understanding of the role of the frontal regions related to the MNS in the view of the dual-loop model.

Materials and Methods

Subjects

A total of 116 healthy volunteers (50 females; mean age: 25.7 years) were recruited for this study. All subjects provided written informed consent. The study was performed in accordance with the Declaration of Helsinki and was approved by the local Ethics Committee. All subjects were right-handed according to the Edinburgh handedness inventory (Oldfield 1971). Because of head movement, 14 volunteers were excluded and so 102 subjects were included into the final analysis.

MRI Acquisition

MRI was performed in a 3-Tesla whole-body MRI-System (TIM-TRIO, Siemens, Erlangen, Germany) equipped with a standard head-coil.

For fMRI, contiguous multislice echo-planar images (EPI; TE 60 ms) were obtained in axial orientation. Thirty slices (3 mm thickness) were acquired every 2.49 s (voxel size 3 × 3 × m3). For T1-weighted MRI, 160 sagittal slices (1 × 1 × m3 voxel size, TE 3.93 ms) were acquired.

For the probabilistic diffusion tensor imaging (probDTI) a diffusion-weighted Spin-Echo EPI sequence was used. The whole brain was covered with contiguous 2-mm slices with an in-plane resolution of 2 × 2 mm2. The diffusion encoding was performed in 61 different directions with an effective b-value of 1000 s/mm2.

The following fMRI conditions were applied in a block design in 1 session: during REST volunteers fixated a cross in the middle of the screen. Action observation (AO): Videos of a hand grasping objects were presented in 1 block with the right hand from the first person position. During active conditions objects were used (like a key) which were grasped with a pinch grip form. Imitation of object grasping (IMI): Volunteers had their right hand lying next to their body. They observed a video while a hand grasped an object and were requested to imitate the observed movement with their right hand according to the videos. Volunteers knew about the fMRI conditions and trained before starting scanning. With the start of the videos, volunteers started with their arm accordingly to the observed grasping movement of the videos (quasi online performance). Objects were fixed on velcro tape fastened around the hip before MRI acquisition and were not visible to the volunteers. To examine the contribution of movement execution during IMI, we also investigated grasping an object without visual control. Grasping objects (GR): The right hand of the volunteers was lying next to their body. Objects were fixed on velcro tape, so the way to grasp was very short. They grasped a single object with their right hand while fixating a cross in the middle of the screen. This signal for grasping during GR and IMI was delivered by an acoustic stimulus via headphones, controlled by a PC running “Presentation”-Software (Neurobehavioral-System) synchronized to the scanner. Each REST, GR, IMI, and AO block lasted 25 s. Each active block was repeated 4 times and alternated in a pseudo-randomized fashion with REST. Volunteers performed the grasping movement 5 times in each block of GR and IMI. Videos presented in IMI and AO were the same (each video lasted 5 s). In all conditions, the same objects were observed or grasped. Motor responses were controlled via inspection within the scanner and by bipolar surface electromyography (from both M. flexor digitorum superficialis and M. extensor digitorum communis). Visual stimuli were projected on a screen positioned in front of the bore of the scanner with a horizontal field of view of 19°. Volunteers viewed the visual stimuli through prism glasses.

Data Processing and Statistical Analysis

fMRI data processing was performed with SPM8 (Welcome Department of Cognitive Neurology, London, UK) running under Matlab 2007a (Mathworks, Sherborn, MA, USA). The first 4 images of each run were discarded to allow for equilibration of longitudinal magnetization. Calculation of on-line motion and distortion-corrected fMRI volumes was performed at the scanner (Zaitsev et al. 2006). Resulting volumes were spatially normalized to a symmetric template based on the Montreal Neurological Institute reference brain using the normalization parameters estimated during segmentation of the T1-anatomical scan. Normalized fMRI images were then smoothed with an isotropic 12-mm and 4-mm full-width half-maximum (FWHM) Gaussian kernel to allow valid statistical inference, according to the Gaussian random field theory (Friston et al. 2003). We used low smoothing filter of 4-mm FWHM to investigate whether maximum peaks of activated regions change their coordinates (see also (Gazzola and Keysers 2009)).

fMRI Statistical Analysis

Two complementary analysis methods were employed to fMRI data:

  1. A general linear model based on a model of the time course and the hemodynamic response function was employed. Stimuli onsets were convolved with a canonical hemodynamic response function, as implemented in SPM8. Low-frequency components of fMRI time series were removed by high-pass filtering. In the second-level analyses, volunteers were treated as a random effect, and the contrast images of the conditions of interest were entered into a two-sample t-test. For statistical analysis, we applied a threshold of P < 0.05 (corrected across the whole brain, familywise error, FWE).

  2. A “functional connectivity” (FC) analysis was used to examine the interaction between different region of interest (ROI) (Friston et al. 1997). We were interested in the difference of FC between 2 specific ROI from the fMRI group conjunction analysis of IMI and AO. Therefore, for each subject time series from ROIA and ROIB were extracted with a sphere of 5 mm. The time series from ROIA were included as regressor to subjects' fMRI images. From this analysis related to ROIA, one con-image for each subject was selected (this con-image includes FC between ROIA with the whole brain). This con-image was contrasted in a two-sample t-test to the con-image from ROIB. Consequently, FC of time series from ROIA which is prior to ROIB with the whole brain is analyzed (corrected P < 0.05, FWE). This analysis will elicit the difference between FC of ROIA in contrast to ROIB with whole brain. Anatomical description is based on the probabilistic cytoarchitectonic maps as implemented in the SPM anatomy toolbox (http://www.fz-juelich.de/inm/inm-1/DE/Forschung/_docs/SPMAnantomyToolbox/SPMAnantomyToolbox_node.html) (Eickhoff et al. 2005).

Probabilistic DTI-Based Fiber Tracking

To examine the anatomical contribution between 2 ROIs the probabilistic diffusion tensor imaging (DTI)-based fiber tracking was used. ROIs for the probabilistic fiber tracking were extracted from the fMRI group conjunction analysis of IMI and AO in a sphere of 5 mm. The tracking area was restricted to a white matter mask to avoid tracking across anatomical borders. To ensure contact between the cortical ROIs and the white matter, a rim of gray matter was included in the mask. The diffusion tensor was first computed (Basser et al. 1994) from the movement- and distortion-corrected diffusion-weighted imaging dataset. A Monte Carlo simulation of random walks similar to the probabilistic index of connectivity method (Parker et al. 2003) was used to calculate separate probabilistic maps for each ROI. Region-to-region anatomical connectivity between 2 ROIs was computed using a newly developed combination of probability maps without a priori knowledge about the presumed course (Kreher et al. 2008).

While probabilistic DTI detects anatomical connection between 2 selected seed regions, FC between ROIA and ROIB figure out their different FC with the whole brain that can be influenced by regions which are directly connected and/or areas which indirectly interact.

Results

fMRI Results

Conditions Versus Rest

Considering the aim of the study we only report activation within different regions of IFG.

During Imitation (IMI), 2 maximum peak voxels were found in BA44 (pars opercularis of IFG). Our 2 peak voxels corresponded to the recently described subdivision of BA44 (Amunts et al. 2010) with 1 peak voxel in BA44v (MNI coordinates x-, y-, and z-coordinates: −48, 6, 6; T = 20.06) and the other one in the dorsal part of BA44 (termed as BA44d; −54, 6, 30; T = 18.33) (Table 1). Another local maximum was found in BA45 (pars triangularis of IFG; MNI coordinates x-, y-, and z-coordinates: −45, 20, 24; T = 4.2). Since most previous imaging studies did not show activation of BA45 during IMI, we analyzed its activation in increasing subjects' number (group analysis of 10 subjects, 20 subjects, etc.). A group analysis of 40 subjects and more activated BA45 at the predefined P-value.

Table 1

Task-related fMRI activation (coordinates in MNI space)

Imitation (IMI) x y z 
BA44v −48 
BA44d −54 30 
BA45 −45 20 24 
BA6 −54 −6 39 
Action observation (AO) 
 BA45 −48 24 21 
 BA44d −45 33 
 PMC −45 −9 30 
Grasping (GR) 
 BA44v −48 
 BA6 −51 −3 36 
Conjunction IMI and AO 
 BA44d −45 33 
 BA45 −45 20 24 
 SLP (7A) −30 −57 57 
 V5 −42 −75 −3 
 hlp3 (AIP) −36 −45 51 
 Fusiform gyrus −42 −54 −15 
Imitation (IMI) x y z 
BA44v −48 
BA44d −54 30 
BA45 −45 20 24 
BA6 −54 −6 39 
Action observation (AO) 
 BA45 −48 24 21 
 BA44d −45 33 
 PMC −45 −9 30 
Grasping (GR) 
 BA44v −48 
 BA6 −51 −3 36 
Conjunction IMI and AO 
 BA44d −45 33 
 BA45 −45 20 24 
 SLP (7A) −30 −57 57 
 V5 −42 −75 −3 
 hlp3 (AIP) −36 −45 51 
 Fusiform gyrus −42 −54 −15 

During Action Observation (AO), the maximum peak voxels in IFG were in BA45 (−48, 24, 21; T = 7.05) and in BA44d (−45, 9, 33; T = 7.41). In BA44v no significant activation was found in the comparison AO versus Rest.

During Grasping (GR), the maximum peak voxel in IFG was found in BA44v (−48, 6, 6; T = 3.5), and there was no significant activation in BA44d and BA45.

Conditions Versus Each Other

For the comparisons between GR, IMI, and AO, we only focus on differences within IFG.

The comparison of IMI versus AO revealed significantly stronger activation for IMI within BA44v (−45, 3, 6; T = 17.64) and BA44d (−54, 6, 30; T = 9.66).

Comparing IMI versus GR, significantly stronger activation was found for IMI in all 3 parts of IFG (in BA44v, BA44d, and BA45). The reverse comparison of GR versus IMI did not show any significantly stronger activation in IFG. The comparison of AO and GR showed significantly stronger activation for AO in BA44d (−42, 9, 30; T = 9.26) and in BA45 (−48, 24, 21; T = 8.89). The reverse comparison showed stronger activation in BA44v (−48, 6, 6; T = 5.14).

Conjunction Analysis

The conjunction analysis of IMI and AO showed commonly activated areas of BA44d (−45, 9, 33; T = 7.41) and BA45 (−45, 20, 24; T = 5.72) in IFG and no activation of BA44v (see Fig. 1). Further areas were found within the posterior parietal cortex (PPC; according to anatomy toolbox SPL (superior parietal lobule), 7PC; −30, −57, 57), V5 (according to anatomy toolbox the left inferior occipital area; −42, −75, −3), and in the anterior intraparietal sulcus (aIPS; according to anatomy toolbox hlp3; −36, −45, 51). aIPS was suggested to be the homolog area of macaque monkeys AIP (Binkofski et al. 1998; Borra et al. 2008) (for the anterior intraparietal sulcus, we use the term AIP in the following text). Further activation was evident in the fusiform gyrus (−42, −54, –15) as described previously (Chao et al. 1999; Binkofski et al. 2004).

Figure 1.

Activation map during action observation (AO in green) and imitation (IMI in blue) showed among other regions activation of the ventral and dorsal part of the pars opercularis (BA44ventral and BA44dorsal) and the pars triangularis (BA45) of the inferior frontal gyrus (IFG). The maximum peak voxel of the conjunction analysis of IMI and AO is presented in yellow and includes BA44dorsal and BA45 of IFG as well as the posterior parietal cortex (PPC; according to anatomy toolbox (superior parietal lobule; SPL; 7PC), the anterior intraparietal sulcus (AIP; according to anatomy toolbox hlp3), the ventral temporal gyrus (VTG), and the visual cortex. Plot bars of activation from maximum activated peak voxel are demonstrated during grasping an object (GR), IMI, and AO.

Figure 1.

Activation map during action observation (AO in green) and imitation (IMI in blue) showed among other regions activation of the ventral and dorsal part of the pars opercularis (BA44ventral and BA44dorsal) and the pars triangularis (BA45) of the inferior frontal gyrus (IFG). The maximum peak voxel of the conjunction analysis of IMI and AO is presented in yellow and includes BA44dorsal and BA45 of IFG as well as the posterior parietal cortex (PPC; according to anatomy toolbox (superior parietal lobule; SPL; 7PC), the anterior intraparietal sulcus (AIP; according to anatomy toolbox hlp3), the ventral temporal gyrus (VTG), and the visual cortex. Plot bars of activation from maximum activated peak voxel are demonstrated during grasping an object (GR), IMI, and AO.

With the use of 4 mm smoothing, peaks in the conjunction analysis of IMI and AO within BA44 and BA45 remained stable and did not shift. Activation within the parietal lobe showed different maxima peak voxel: one in AIP (according to anatomy toolbox hlp3; −36, −45, 51), one in BA7a (according to anatomy toolbox SPL, 7PC; −30, −54, 57), and a further one in AIP (according to anatomy toolbox hlp3; −27, −60, 57).

FC of BA44d and BA45

The dorsal activation region of BA44 (BA44d) showed a superior FC with the primary sensorimotor cortex, the PPC, the superior, and inferior parietal lobule (SPL, IPL) as well as AIP and ventral temporal gyrus (VTG) in contrast to BA45. BA45 revealed superior FC with angular gyrus (−48, −66, 33) and several frontal areas (see Fig. 2).

Figure 2.

Functional connectivity of 2 regions of interest (ROIs) when contrasted against each other. In step 1, fMRI time series are extracted from ROIA (BA44dorsal) and ROIB (BA45) for each subject. In step 2, the extracted time series of each ROI are included as regressor to fMRI images for each subject. In step 3, the estimated model of step 2 is contrasted for each subject and each ROI. In the final analysis (step 4), a two-sample t-test was evaluated between the contrast of estimated model from ROIA (BA44dorsal) against ROIB (BA45) and vice versa. BA44dorsal shows more functional connectivity (in red) with the dorsal premotor cortex, the primary sensorimotor cortex, the posterior parietal cortex, the superior and inferior parietal lobule, as well as with the intraparietal sulcus and the ventral temporal gyrus in contrast to BA45. The BA45 revealed more functional connectivity with angular gyrus and more frontal areas (in green).

Figure 2.

Functional connectivity of 2 regions of interest (ROIs) when contrasted against each other. In step 1, fMRI time series are extracted from ROIA (BA44dorsal) and ROIB (BA45) for each subject. In step 2, the extracted time series of each ROI are included as regressor to fMRI images for each subject. In step 3, the estimated model of step 2 is contrasted for each subject and each ROI. In the final analysis (step 4), a two-sample t-test was evaluated between the contrast of estimated model from ROIA (BA44dorsal) against ROIB (BA45) and vice versa. BA44dorsal shows more functional connectivity (in red) with the dorsal premotor cortex, the primary sensorimotor cortex, the posterior parietal cortex, the superior and inferior parietal lobule, as well as with the intraparietal sulcus and the ventral temporal gyrus in contrast to BA45. The BA45 revealed more functional connectivity with angular gyrus and more frontal areas (in green).

Probabilistic DTI-based fiber tracking. ROIs were chosen from activated regions of the conjunction analysis of IMI and AO (BA44v, BA44d, BA45, PPC, AIP, and VTG). Anatomical contributions between all posterior regions (PPC, AIP, and VTG) with all frontal areas (BA44d, BA44v, and BA45) were calculated. The dorsal part of the BA44 (BA44d) was connected with AIP and PPC along the superior longitudinal fasciculus (SLF). The bundle of the arcuate fasciculus (AF) arising from the VTG blends with SLF and connects VTG with BA44d (the dorsal stream). In contrast, BA44v was connected with AIP and PPC via fibers passing through the extreme capsule (the fasciculus extreme capsule, EmC; the ventral stream) (see Fig. 3).

Figure 3.

Seed regions were extracted from the conjunction analysis of IMI and AO: PPC (according to anatomy toolbox (superior parietal lobule; SPL; 7PC), AIP (according to anatomy toolbox hlp3), VTG, BA44dorsal and BA45, as well as maximum peak voxel of IFG activated during IMI and GR (BA44ventral). Anatomical contributions of the same postrolandic ROIs (PPC, AIP, and VTG) were calculated with BA44ventral, BA44dorsal, and BA45. (a) BA44ventral was connected with AIP and PPC via fibers passing through the extreme capsule (the fasciculus extreme capsule, EmC; the ventral stream). (b) BA44dorsal was connected with AIP and PPC along the superior longitudinal fasciculus (SLF). The bundle of the arcuate fasciculus (AF) arising from the VTG blends with SLF and connects VTG with BA44dorsal (is not pictured). (c) BA45 was connected with AIP and PPC via both the SLF (the dorsal stream) and the EmC (the ventral stream).

Figure 3.

Seed regions were extracted from the conjunction analysis of IMI and AO: PPC (according to anatomy toolbox (superior parietal lobule; SPL; 7PC), AIP (according to anatomy toolbox hlp3), VTG, BA44dorsal and BA45, as well as maximum peak voxel of IFG activated during IMI and GR (BA44ventral). Anatomical contributions of the same postrolandic ROIs (PPC, AIP, and VTG) were calculated with BA44ventral, BA44dorsal, and BA45. (a) BA44ventral was connected with AIP and PPC via fibers passing through the extreme capsule (the fasciculus extreme capsule, EmC; the ventral stream). (b) BA44dorsal was connected with AIP and PPC along the superior longitudinal fasciculus (SLF). The bundle of the arcuate fasciculus (AF) arising from the VTG blends with SLF and connects VTG with BA44dorsal (is not pictured). (c) BA45 was connected with AIP and PPC via both the SLF (the dorsal stream) and the EmC (the ventral stream).

BA45 was connected with AIP and PPC via both the SLF (the dorsal stream) and the EmC (the ventral stream).

It must be considered that the probabilistic DTI does not have the resolution to discriminate axons and axonal terminations; therefore, we cannot rule out a polysynaptic pathway connecting postrolandic (PPC, AIP, and VTG) with prerolandic regions (BA44d, BA44v, and BA45) (see Fig. 4).

Figure 4.

A schema of different networks based on current functional–anatomical findings. The BA44dorsal and BA45 are active during conditions with additional task regarding action observation (IMI and AO). These 2 regions are connected to the VTG (the ventral temporal cortex) with the fasciculus arcuatus (AF) and the superior longitudinalis fasciculus (SLF; the dorsal stream). The BA44ventral is active during hand movement (IMI and GR). The extreme capsule (EmC) projects information from the anterior intraparietal sulcus (AIP) and the posterior parietal cortex (PPC) to BA44ventral (the ventral stream). We suppose that this network is responsible for action performance. Plus-sign in white circle represents strength of activation degree during the task. IMI, imitation; AO, action observation; GR, grasping.

Figure 4.

A schema of different networks based on current functional–anatomical findings. The BA44dorsal and BA45 are active during conditions with additional task regarding action observation (IMI and AO). These 2 regions are connected to the VTG (the ventral temporal cortex) with the fasciculus arcuatus (AF) and the superior longitudinalis fasciculus (SLF; the dorsal stream). The BA44ventral is active during hand movement (IMI and GR). The extreme capsule (EmC) projects information from the anterior intraparietal sulcus (AIP) and the posterior parietal cortex (PPC) to BA44ventral (the ventral stream). We suppose that this network is responsible for action performance. Plus-sign in white circle represents strength of activation degree during the task. IMI, imitation; AO, action observation; GR, grasping.

Discussion

Both action observation (AO) and imitation (IMI) involved BA44d, which is in line with 2 previous meta-analyses of Molnar-Szakacs and Caspers comparing AO and IMI (Molnar-Szakacs et al. 2005; Caspers et al. 2010). It must be considered that we did not control for eye movement, and our motor paradigm in GR and IMI irrespective of grasping movement differ with respect to their visual input. However, we did not find activation of the frontal eye fields in the individual comparisons, so that we assume that they might not have played a major role. However, both AO and IMI also involved BA45, both in the direct analysis of these conditions versus rest as well as in the conjunction analysis, which is a different finding from these 2 meta-analyses, where AO but not IMI was associated with activation in BA45. Analyzing IMI versus baseline in increasing number of subjects, we found that a group analysis of more than 40 subjects demonstrated activation of BA45 at predefined P-value (also in a low-smoothed single subject analysis (Gazzola and Keysers 2009)). First, this means that a large cohort of subjects can have an advantage over meta-analysis, where such a finding might be not detected if in each single study the threshold for activation is not reached. Second, the present finding suggests that arguments so far explaining the absence of activation of BA45 during imitation should be revised.

In recent years, a prominent role of BA45 in the time-independent analysis for semantic and conceptual analysis in different modalities was recognized, whereas BA44 as part of the dorsal system is thought to be involved in the time-dependent analysis of sensory percepts and internal models (Rauschecker and Scott 2009; Weiller et al. 2011; Rijntjes et al. 2012; Hickok 2009). Since the discovery of MNS, containing neurons that fire both during execution and observation of goal-directed motor acts, it is discussed to what extent the meaning of an observed action is also understood by the mirror neurons themselves, or whether hierarchically higher order areas area involved in this step (Buxbaum et al. 2005). In line with findings in other domains (Weiller et al. 2009, 2011; Rijntjes et al. 2012; Rijntjes and Weiller 2013), a recent view suggests that the “understanding” of action would mainly be a role of the ventral stream (Hickok and Hauser 2010). Therefore, an involvement of BA45 as found in the present study in both action observation and imitation would not be unexpected. However, since at least both dorsally (BA44d) and ventrally (BA45) connected areas are involved both in action observation and imitation, the present finding might lead to further studies investigating whether a functional subdivision between these areas with regard to “understanding” a movement exists.

BA45

BA45, as defined by the conjunction analysis of AO and IMI, was both dorsally and ventrally connected to parietal areas, in line with previous studies, with the dorsal connection corresponding to the SLF (Thiebaut de Schotten et al. 2011) (the dorsal stream) and the ventral connection to the EmC (Makris et al. 2009). This latter pathway, together with the cortical regions that it connects, corresponds to the ventral processing stream (Umarova et al. 2010, 2011; Rijntjes et al. 2012; Rijntjes and Weiller 2013; Vry et al. 2012; Saur et al. 2008; Weiller et al. 2009, 2011; Kravitz et al. 2011. Monkey's area 45 was subdivided into 45A and 45B with different connections and different functionality. 45A was suggested to be more involved in communication behavior and 45B more in action observation (Nelissen et al. 2011; Gerbella et al. 2010). With the current fMRI paradigm, we are not able to differentiate functional subdivision of BA45. However, the seed region we took for analysis of probable DTI connections was taken from the conjunction analysis of AO and IMI, so that at least for these conditions, both dorsal and ventral projections to the parietal lobe were found. Since we used an object during both IMI and AO, it cannot be excluded that activation within IFG is partially related to encoding of object semantics and shape (analog to canonical neurons in monkey), see also (Grafton et al. 1997; Chao and Martin 2000; Grezes et al. 2003; Weisberg et al. 2007), see also (Nelissen et al. 2005).

The Functional Connectivity of BA44d and BA45

The FC confirmed that BA44d is mainly associated with cortical areas in the dorsal stream, while BA45 showed a strong correlation with activation in the prefrontal lobe and in the angular gyrus (GA). The IPL in monkeys (PF, PFG, and PG, likely corresponding to the supramarginal gyrus and angular gyrus in humans, respectively) is connected to the ventro-lateral prefrontal and the superior and middle temporal cortex via the middle longitudinal fasciculus (MDLF (Seltzer and Pandya 1984)), while some parts of the IPL are connected via the EmC to the inferior frontal lobe (Caspers et al. 2011). A recent study also showed the differential connections between monkey area 45B and 44, where 45B receives mainly oculomotor input (Gerbella et al. 2010) and has connections stretching to area PG in the inferior parietal lobe and more extensive connections to the ventral pathway through the extreme capsule (Frey et al. 2014). As a multimodal association cortex directly connected to visual, somatosensory, and auditory association cortex (Geschwind 1965a, 1965b), GA serves as an interface between modalities and has been suggested to map perceptual input to distributed semantic knowledge (Binder et al. 2003). An alternative proposal is that GA is part of a fronto-parietal network engaged in top-down control of semantic processing in other regions (Corbett et al. 2009; Jefferies et al. 2006; Sharp et al. 2010). Thus, BA45 and GA may act in concordance as an interface between dorsal and ventral processing streams, enabling them to act in a highly coordinated way to ensure efficient interplay between cognitive control, perception and action (Vry et al. 2012; Rijntjes et al. 2012; Sweeney et al. 2007).

The Inferior Parietal Lobule

BA45 was both ventrally and dorsally connected with temporal and parietal areas, projecting to the same areas (PPC and AIP) in the parietal lobe and area VTG in the temporal lobe. Additionally, pathways from BA44d and BA44v were identified projecting to these areas in accordance with other studies, e.g., (Vry et al. 2012; Hoeren et al. 2013), although the same seed regions from the parietal and temporal areas were used. The dorsal stream of BA45 from IPL is suggested to serve as an “ideal interface, where feed-forward signals from motor preparatory networks in the inferior frontal and premotor cortex can be matched with feed-back signals from sensory areas” (Rauschecker and Scott 2009).

BA44d and BA44v

Combining DTI with the fMRI data showed a clear subdivision of BA44 in BA44v and BA44d, as proposed before in cytoarchitectonic (Amunts et al. 2010) and functional (Molnar-Szakacs et al. 2005; Caspers et al. 2010) studies. Inside the dorsal stream, a distinction is made between dorso-dorsal connection to the dorsal premotor cortex and ventro-dorsal connection to the ventral premotor cortex (Rizzolatti and Matelli 2003; Binkofski and Buxbaum 2013). We suggest that the current dorsal stream (connecting fibers of AIP and PPC to BA44d) might be the analog fibers of SLFIII (Vry et al. 2012) and is in line with Rizzolatti and Matteli's ventro-dorsal stream (Rizzolatti and Matelli 2003), which corresponds to the parieto-premotor pathway of Kravitz et al. (2011). This dorsal route may involve the strong connection between AIP and F5 (Luppino et al. 1999) and the connection between PFG of IPL with F5 (Nelissen et al. 2011), because all these interconnected regions (AIP, PFG, and F5) are involved in action observation (Rizzolatti et al. 1996; Rizzolatti and Matelli 2003; Fogassi et al. 2005; Nelissen et al. 2011; Murata et al. 2000).

BA44v as Part of the Ventral Stream

Only actual movements towards objects (IMI, GR) activated BA44v. It was proposed that this area is involved in forward modeling processes, as a source of an efferent copy during imitation (Molnar-Szakacs et al. 2005). This suggestion was supported by findings in other studies (Morin and Grezes 2008; Caspers et al. 2010; Vogt et al. 2007). These forward modeling processes would be therefore distinct from the internal models and sensorimotor integration processes associated with the dorsal pathway. The new information from the present study, however, is that BA44v, activated while making actual movements in the conditions GR and IMI, is connected with parietal areas via the ventral stream through the EmC. The general function of the ventral pathway, analogous to the ventral visual stream, is assumed to be the understanding of an action. However, when making an actual movement, we suggest that a matching with internal time-independent knowledge is required, additional to the spatial and sensorimotor inverse and feed-forward models of the dorsal stream. In this hypothesis, these calibrations are based on information about internal representations of the actual movement conferred by the ventral processing stream connecting parietal areas with BA44v. Interestingly, in another study contrasting the repetition of pseudo-words with listening to and understanding sentences, the repetition of pseudo-words, also an internally generated motor act without meaning, was associated with both a dorsal stream to BA44d and a ventral stream projecting to BA44v, but not with a connection to BA45, which was only activated and ventrally connected during the listening to and understanding sentences (Saur et al. 2008). However, the 2 explanations (forward modeling and matching meaning with actual movement) are not mutually exclusive and are possibly overlapping.

Conclusion

In sum, the present data show that action observation and imitation are a function mainly of the dorsal system centered on BA44d, but also involve BA45, which is dorsally and ventrally connected to postrolandic regions. Based on this knowledge, further studies should identify which parts of the brain network are crucial for the analysis and understanding of observed movements. Since both imitation and grasping also involve BA44v, studies investigating the network for imitation should control for actual movements.

Funding

F.H. is supported by the German Center for Sepsis Control & Care (CSCC), funded by the Ministry of Education and Research (BMBF), grant no. 01 E0 1002.

Notes

We are grateful to all individuals who participated in this study. We thank H. Mast, A. Vuck, J. Wanschura, M. Tepper, S. Hodrius, C. Schilling, and A. Schindler for their support during data acquisition. Conflict of Interest: None declared.

References

Amunts
K
,
Lenzen
M
,
Friederici
AD
,
Schleicher
A
,
Morosan
P
,
Palomero-Gallagher
N
,
Zilles
K
.
2010
.
Broca's region: novel organizational principles and multiple receptor mapping
.
PLoS Biol
 .
8
(9)
:
e1000489
.
Basser
PJ
,
Mattiello
J
,
LeBihan
D
.
1994
.
Estimation of the effective self-diffusion tensor from the NMR spin echo
.
J Magn Reson B
 .
103
(3)
:
247
254
.
Binder
JR
,
McKiernan
KA
,
Parsons
ME
,
Westbury
CF
,
Possing
ET
,
Kaufman
JN
,
Buchanan
L
.
2003
.
Neural correlates of lexical access during visual word recognition
.
J Cogn Neurosci
 .
15
(3)
:
372
393
.
Binkofski
F
,
Buccino
G
,
Zilles
K
,
Fink
GR
.
2004
.
Supramodal representation of objects and actions in the human inferior temporal and ventral premotor cortex
.
Cortex
 .
40
(1)
:
159
161
.
Binkofski
F
,
Buxbaum
LJ
.
2013
.
Two action systems in the human brain
.
Brain Lang
 .
127
(2)
:
222
229
.
Binkofski
F
,
Dohle
C
,
Posse
S
,
Stephan
KM
,
Hefter
H
,
Seitz
RJ
,
Freund
HJ
.
1998
.
Human anterior intraparietal area subserves prehension: a combined lesion and functional MRI activation study
.
Neurology
 .
50
(5)
:
1253
1259
.
Borra
E
,
Belmalih
A
,
Calzavara
R
,
Gerbella
M
,
Murata
A
,
Rozzi
S
,
Luppino
G
.
2008
.
Cortical connections of the macaque anterior intraparietal (AIP) area
.
Cereb Cortex
 .
18
(5)
:
1094
1111
.
Buxbaum
LJ
,
Johnson-Frey
SH
,
Bartlett-Williams
M
.
2005
.
Deficient internal models for planning hand-object interactions in apraxia
.
Neuropsychologia
 .
43
(6)
:
917
929
.
Caspers
S
,
Eickhoff
SB
,
Rick
T
,
von Kapri
A
,
Kuhlen
T
,
Huang
R
,
Shah
NJ
,
Zilles
K
.
2011
.
Probabilistic fibre tract analysis of cytoarchitectonically defined human inferior parietal lobule areas reveals similarities to macaques
.
Neuroimage
 .
58
(2)
:
362
380
.
Caspers
S
,
Zilles
K
,
Laird
AR
,
Eickhoff
SB
.
2010
.
ALE meta-analysis of action observation and imitation in the human brain
.
Neuroimage
 .
50
(3)
:
1148
1167
.
Chao
LL
,
Haxby
JV
,
Martin
A
.
1999
.
Attribute-based neural substrates in temporal cortex for perceiving and knowing about objects
.
Nat Neurosci
 .
2
(10)
:
913
919
.
Chao
LL
,
Martin
A
.
2000
.
Representation of manipulable man-made objects in the dorsal stream
.
Neuroimage
 .
12
(4)
:
478
484
.
Corbett
F
,
Jefferies
E
,
Ehsan
S
,
Lambon Ralph
MA
.
2009
.
Different impairments of semantic cognition in semantic dementia and semantic aphasia: evidence from the non-verbal domain
.
Brain
 .
132
(Pt 9)
:
2593
2608
.
di Pellegrino
G
,
Fadiga
L
,
Fogassi
L
,
Gallese
V
,
Rizzolatti
G
.
1992
.
Understanding motor events: a neurophysiological study
.
Exp Brain Res
 .
91
(1)
:
176
180
.
Eickhoff
SB
,
Stephan
KE
,
Mohlberg
H
,
Grefkes
C
,
Fink
GR
,
Amunts
K
,
Zilles
K
.
2005
.
A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data
.
Neuroimage
 .
25
(4)
:
1325
1335
.
Fabbri-Destro
M
,
Cattaneo
L
,
Boria
S
,
Rizzolatti
G
.
2008
.
Planning actions in autism
.
Exp Brain Res
 .
192
3
:
521
525
.
Fogassi
L
,
Ferrari
PF
,
Gesierich
B
,
Rozzi
S
,
Chersi
F
,
Rizzolatti
G
.
2005
.
Parietal lobe: from action organization to intention understanding
.
Science
 .
308
(5722)
:
662
667
.
Frey
S
,
Mackey
S
,
Petrides
M
.
2014
.
Cortico-cortical connections of areas 44 and 45B in the macaque monkey
.
Brain Lang
 .
131
:
36
55
.
Friston
KJ
,
Buechel
C
,
Fink
GR
,
Morris
J
,
Rolls
E
,
Dolan
RJ
.
1997
.
Psychophysiological and modulatory interactions in neuroimaging
.
Neuroimage
 .
6
(3)
:
218
229
.
Friston
KJ
,
Harrison
L
,
Penny
W
.
2003
.
Dynamic causal modelling
.
Neuroimage
 .
19
(4)
:
1273
1302
.
Gallese
V
,
Fadiga
L
,
Fogassi
L
,
Rizzolatti
G
.
1996
.
Action recognition in the premotor cortex
.
Brain
 .
119
(Pt 2)
:
593
609
.
Gazzola
V
,
Keysers
C
.
2009
.
The observation and execution of actions share motor and somatosensory voxels in all tested subjects: single-subject analyses of unsmoothed fMRI data
.
Cereb Cortex
 .
19
(6)
:
1239
1255
.
Gerbella
M
,
Belmalih
A
,
Borra
E
,
Rozzi
S
,
Luppino
G
.
2010
.
Cortical connections of the macaque caudal ventrolateral prefrontal areas 45A and 45B
.
Cereb Cortex
 .
20
(1)
:
141
168
.
Geschwind
N
.
1965a
.
Disconnexion syndromes in animals and man. I
.
Brain
 .
88
(2)
:
237
294
.
Geschwind
N
.
1965b
.
Disconnexion syndromes in animals and man. II
.
Brain
 .
88
(3)
:
585
644
.
Goodale
MA
,
Milner
AD
.
1992
.
Separate visual pathways for perception and action
.
Trends Neurosci
 .
15
(1)
:
20
25
.
Grafton
ST
,
Fadiga
L
,
Arbib
MA
,
Rizzolatti
G
.
1997
.
Premotor cortex activation during observation and naming of familiar tools
.
Neuroimage
 .
6
(4)
:
231
236
.
Grezes
J
,
Armony
JL
,
Rowe
J
,
Passingham
RE
.
2003
.
Activations related to “mirror” and “canonical” neurones in the human brain: an fMRI study
.
Neuroimage
 .
18
(4)
:
928
937
.
Hesse
MD
,
Sparing
R
,
Fink
GR
.
2009
.
End or means—the “what” and “how” of observed intentional actions
.
J Cogn Neurosci
 .
21
(4)
:
776
790
.
Hickok
G
.
2009
.
Eight problems for the mirror neuron theory of action understanding in monkeys and humans
.
J Cogn Neurosci
 .
21
(7)
:
1229
1243
.
Hickok
G
,
Hauser
M
.
2010
.
(Mis)understanding mirror neurons
.
Curr Biol
 .
20
(14)
:
R593
R594
.
Hoeren
M
,
Kaller
CP
,
Glauche
V
,
Vry
MS
,
Rijntjes
M
,
Hamzei
F
,
Weiller
C
.
2013
.
Action semantics and movement characteristics engage distinct processing streams during the observation of tool use
.
Exp Brain Res
 .
229
(2)
:
243
260
.
Iacoboni
M
,
Molnar-Szakacs
I
,
Gallese
V
,
Buccino
G
,
Mazziotta
JC
,
Rizzolatti
G
.
2005
.
Grasping the intentions of others with one's own mirror neuron system
.
PLoS Biol
 .
3
(3)
:
e79
.
Iacoboni
M
,
Woods
RP
,
Brass
M
,
Bekkering
H
,
Mazziotta
JC
,
Rizzolatti
G
.
1999
.
Cortical mechanisms of human imitation
.
Science
 .
286
(5449)
:
2526
2528
.
Jefferies
E
,
Frankish
C
,
Lambon Ralph
MA
.
2006
.
Lexical and semantic influences on item and order memory in immediate serial recognition: evidence from a novel task
.
Q J Exp Psychol (Hove)
 .
59
(5)
:
949
964
.
Kravitz
DJ
,
Saleem
KS
,
Baker
CI
,
Mishkin
M
.
2011
.
A new neural framework for visuospatial processing
.
Nat Rev Neurosci
 .
12
(4)
:
217
230
.
Kreher
BW
,
Schnell
S
,
Mader
I
,
Il'yasov
KA
,
Hennig
J
,
Kiselev
VG
,
Saur
D
.
2008
.
Connecting and merging fibres: pathway extraction by combining probability maps
.
Neuroimage
 .
43
(1)
:
81
89
.
Luppino
G
,
Murata
A
,
Govoni
P
,
Matelli
M
.
1999
.
Largely segregated parietofrontal connections linking rostral intraparietal cortex (areas AIP and VIP) and the ventral premotor cortex (areas F5 and F4)
.
Exp Brain Res
 .
128
(1–2)
:
181
187
.
Makris
N
,
Papadimitriou
GM
,
Kaiser
JR
,
Sorg
S
,
Kennedy
DN
,
Pandya
DN
.
2009
.
Delineation of the middle longitudinal fascicle in humans: a quantitative, in vivo, DT-MRI study
.
Cereb Cortex
 .
19
(4)
:
777
785
.
Milner
AD
,
Goodale
MA
.
2008
.
Two visual systems re-viewed
.
Neuropsychologia
 .
46
(3)
:
774
785
.
Molnar-Szakacs
I
,
Iacoboni
M
,
Koski
L
,
Mazziotta
JC
.
2005
.
Functional segregation within pars opercularis of the inferior frontal gyrus: evidence from fMRI studies of imitation and action observation
.
Cereb Cortex
 .
15
(7)
:
986
994
.
Morin
O
,
Grezes
J
.
2008
.
What is “mirror” in the premotor cortex? A review
.
Neurophysiol Clin
 .
38
(3)
:
189
195
.
Murata
A
,
Gallese
V
,
Luppino
G
,
Kaseda
M
,
Sakata
H
.
2000
.
Selectivity for the shape, size, orientation of objects for grasping in neurons of monkey parietal area AIP
.
J Neurophysiol
 .
83
(5)
:
2580
2601
.
Nelissen
K
,
Borra
E
,
Gerbella
M
,
Rozzi
S
,
Luppino
G
,
Vanduffel
W
,
Rizzolatti
G
,
Orban
GA
.
2011
.
Action observation circuits in the macaque monkey cortex
.
J Neurosci
 .
31
(10)
:
3743
3756
.
Nelissen
K
,
Luppino
G
,
Vanduffel
W
,
Rizzolatti
G
,
Orban
GA
.
2005
.
Observing others: multiple action representation in the frontal lobe
.
Science
 .
310
(5746)
:
332
336
.
Oldfield
RC
.
1971
.
The assessment and analysis of handedness: the Edinburgh inventory
.
Neuropsychologia
 .
9
(1)
:
97
113
.
Parker
GJ
,
Haroon
HA
,
Wheeler-Kingshott
CA
.
2003
.
A framework for a streamline-based probabilistic index of connectivity (PICo) using a structural interpretation of MRI diffusion measurements
.
J Magn Reson Imaging
 .
18
(2)
:
242
254
.
Rauschecker
JP
,
Scott
SK
.
2009
.
Maps and streams in the auditory cortex: nonhuman primates illuminate human speech processing
.
Nat Neurosci
 .
12
(6)
:
718
724
.
Rijntjes
M
,
Weiller
C
.
2013
.
Introduction. The dual loop model in language and other modalities: an interdisciplinary approach
.
Brain Lang
 .
127
(2)
:
177
180
.
Rijntjes
M
,
Weiller
C
,
Bormann
T
,
Musso
M
.
2012
.
The dual loop model: its relation to language and other modalities
.
Front Evol Neurosci
 .
4
:
9
.
Rizzolatti
G
,
Craighero
L
.
2004
.
The mirror-neuron system
.
Annu Rev Neurosci
 .
27
:
169
192
.
Rizzolatti
G
,
Fadiga
L
,
Gallese
V
,
Fogassi
L
.
1996
.
Premotor cortex and the recognition of motor actions
.
Brain Res Cogn Brain Res
 .
3
(2)
:
131
141
.
Rizzolatti
G
,
Matelli
M
.
2003
.
Two different streams form the dorsal visual system: anatomy and functions
.
Exp Brain Res
 .
153
(2)
:
146
157
.
Saur
D
,
Kreher
BW
,
Schnell
S
,
Kummerer
D
,
Kellmeyer
P
,
Vry
MS
,
Umarova
R
,
Musso
M
,
Glauche
V
,
Abel
S
et al
.
2008
.
Ventral and dorsal pathways for language
.
Proc Natl Acad Sci USA
 .
105
46
:
18035
18040
.
Seltzer
B
,
Pandya
DN
.
1984
.
Further observations on parieto-temporal connections in the rhesus monkey
.
Exp Brain Res
 .
55
(2)
:
301
312
.
Sharp
DJ
,
Awad
M
,
Warren
JE
,
Wise
RJ
,
Vigliocco
G
,
Scott
SK
.
2010
.
The neural response to changing semantic and perceptual complexity during language processing
.
Hum Brain Mapp
 .
31
(3)
:
365
377
.
Sweeney
JA
,
Luna
B
,
Keedy
SK
,
McDowell
JE
,
Clementz
BA
.
2007
.
fMRI studies of eye movement control: investigating the interaction of cognitive and sensorimotor brain systems
.
Neuroimage
 .
36
(Suppl 2)
:
T54
T60
.
Thiebaut de Schotten
M
,
Ffytche
DH
,
Bizzi
A
,
Dell'Acqua
F
,
Allin
M
,
Walshe
M
,
Murray
R
,
Williams
SC
,
Murphy
DG
,
Catani
M
.
2011
.
Atlasing location, asymmetry and inter-subject variability of white matter tracts in the human brain with MR diffusion tractography
.
Neuroimage
 .
54
(1)
:
49
59
.
Umarova
RM
,
Saur
D
,
Kaller
CP
,
Vry
MS
,
Glauche
V
,
Mader
I
,
Hennig
J
,
Weiller
C
.
2011
.
Acute visual neglect and extinction: distinct functional state of the visuospatial attention system
.
Brain
 .
134
(Pt 11)
:
3310
3325
.
Umarova
RM
,
Saur
D
,
Schnell
S
,
Kaller
CP
,
Vry
MS
,
Glauche
V
,
Rijntjes
M
,
Hennig
J
,
Kiselev
V
,
Weiller
C
.
2010
.
Structural connectivity for visuospatial attention: significance of ventral pathways
.
Cereb Cortex
 .
20
(1)
:
121
129
.
Vogt
S
,
Buccino
G
,
Wohlschlager
AM
,
Canessa
N
,
Shah
NJ
,
Zilles
K
,
Eickhoff
SB
,
Freund
HJ
,
Rizzolatti
G
,
Fink
GR
.
2007
.
Prefrontal involvement in imitation learning of hand actions: effects of practice and expertise
.
Neuroimage
 .
37
(4)
:
1371
1383
.
Vry
MS
,
Saur
D
,
Rijntjes
M
,
Umarova
R
,
Kellmeyer
P
,
Schnell
S
,
Glauche
V
,
Hamzei
F
,
Weiller
C
.
2012
.
Ventral and dorsal fiber systems for imagined and executed movement
.
Exp Brain Res
 .
219
(2)
:
203
216
.
Weiller
C
,
Bormann
T
,
Saur
D
,
Musso
M
,
Rijntjes
M
.
2011
.
How the ventral pathway got lost: and what its recovery might mean
.
Brain Lang
 .
118
(1–2)
:
29
39
.
Weiller
C
,
Musso
M
,
Rijntjes
M
,
Saur
D
.
2009
.
Please don't underestimate the ventral pathway in language
.
Trends Cogn Sci
 .
13
(9)
:
369
370
;
370
361
.
Weisberg
J
,
van Turennout
M
,
Martin
A
.
2007
.
A neural system for learning about object function
.
Cereb Cortex
 .
17
(3)
:
513
521
.
Yeterian
EH
,
Pandya
DN
,
Tomaiuolo
F
,
Petrides
M
.
2012
.
The cortical connectivity of the prefrontal cortex in the monkey brain
.
Cortex
 .
48
(1)
:
58
81
.
Zaitsev
M
,
Dold
C
,
Sakas
G
,
Hennig
J
,
Speck
O
.
2006
.
Magnetic resonance imaging of freely moving objects: prospective real-time motion correction using an external optical motion tracking system
.
Neuroimage
 .
31
(3)
:
1038
1050
.