Time-Lagged Multidimensional Pattern Connectivity (TL-MDPC): An EEG/MEG pattern transformation based functional connectivity metric.

Functional and effective connectivity methods are essential to study the complex information flow in brain networks underlying human cognition. Only recently have connectivity methods begun to emerge that make use of the full multidimensional information contained in patterns of brain activation, rather than unidimensional summary measures of these patterns. To date, these methods have mostly been applied to fMRI data, and no method allows vertex-to-vertex transformations with the temporal specificity of EEG/MEG data. Here, we introduce time-lagged multidimensional pattern connectivity (TL-MDPC) as a novel bivariate functional connectivity metric for EEG/MEG research. TL-MDPC estimates the vertex-to-vertex transformations among multiple brain regions and across different latency ranges. It determines how well patterns in ROI X at time point tx can linearly predict patterns of ROI Y at time point ty. In the present study, we use simulations to demonstrate TL-MDPC's increased sensitivity to multidimensional effects compared to a unidimensional approach across realistic choices of number of trials and signal-to-noise ratios. We applied TL-MDPC, as well as its unidimensional counterpart, to an existing dataset varying the depth of semantic processing of visually presented words by contrasting a semantic decision and a lexical decision task. TL-MDPC detected significant effects beginning very early on, and showed stronger task modulations than the unidimensional approach, suggesting that it is capable of capturing more information. With TL-MDPC only, we observed rich connectivity between core semantic representation (left and right anterior temporal lobes) and semantic control (inferior frontal gyrus and posterior temporal cortex) areas with greater semantic demands. TL-MDPC is a promising approach to identify multidimensional connectivity patterns, typically missed by unidimensional approaches.

Distinct roles for the anterior temporal lobe and angular gyrus in the spatiotemporal cortical semantic network.

Semantic knowledge is supported by numerous brain regions, but the spatiotemporal configuration of the network that links these areas remains an open question. The hub-and-spokes model posits that a central semantic hub coordinates this network. In this study, we explored distinct aspects that define a semantic hub, as reflected in the spatiotemporal modulation of neural activity and connectivity by semantic variables, from the earliest stages of semantic processing. We used source-reconstructed electro/magnetoencephalography, and investigated the concreteness contrast across three tasks. In a whole-cortex analysis, the left anterior temporal lobe (ATL) was the only area that showed modulation of evoked brain activity from 100 ms post-stimulus. Furthermore, using Dynamic Causal Modeling of the evoked responses, we investigated effective connectivity amongst the candidate semantic hub regions, that is, left ATL, supramarginal/angular gyrus (SMG/AG), middle temporal gyrus, and inferior frontal gyrus. We found that models with a single semantic hub showed the highest Bayesian evidence, and the hub region was found to change from ATL (within 250 ms) to SMG/AG (within 450 ms) over time. Our results support a single semantic hub view, with ATL showing sustained modulation of neural activity by semantics, and both ATL and AG underlying connectivity depending on the stage of semantic processing.

Towards an objective evaluation of EEG/MEG source estimation methods - The linear approach.

The spatial resolution of EEG/MEG source estimates, often described in terms of source leakage in the context of the inverse problem, poses constraints on the inferences that can be drawn from EEG/MEG source estimation results. Software packages for EEG/MEG data analysis offer a large choice of source estimation methods but few tools to experimental researchers for methods evaluation and comparison. Here, we describe a framework and tools for objective and intuitive resolution analysis of EEG/MEG source estimation based on linear systems analysis, and apply those to the most widely used distributed source estimation methods such as L2-minimum-norm estimation (L2-MNE) and linearly constrained minimum variance (LCMV) beamformers. Within this framework it is possible to define resolution metrics that define meaningful aspects of source estimation results (such as localization accuracy in terms of peak localization error, PLE, and spatial extent in terms of spatial deviation, SD) that are relevant to the task at hand and can easily be visualized. At the core of this framework is the resolution matrix, which describes the potential leakage from and into point sources (point-spread and cross-talk functions, or PSFs and CTFs, respectively). Importantly, for linear methods these functions allow generalizations to multiple sources or complex source distributions. This paper provides a tutorial-style introduction into linear EEG/MEG source estimation and resolution analysis aimed at experimental (rather than methods-oriented) researchers. We used this framework to demonstrate how L2-MNE-type as well as LCMV beamforming methods can be evaluated in practice using software tools that have only recently become available for routine use. Our novel methods comparison includes PLE and SD for a larger number of methods than in similar previous studies, such as unweighted, depth-weighted and normalized L2-MNE methods (including dSPM, sLORETA, eLORETA) and two LCMV beamformers. The results demonstrate that some methods can achieve low and even zero PLE for PSFs. However, their SD as well as both PLE and SD for CTFs are far less optimal for all methods, in particular for deep cortical areas. We hope that our paper will encourage EEG/MEG researchers to apply this approach to their own tasks at hand.

Task modulation of spatiotemporal dynamics in semantic brain networks: An EEG/MEG study.

How does brain activity in distributed semantic brain networks evolve over time, and how do these regions interact to retrieve the meaning of words? We compared spatiotemporal brain dynamics between visual lexical and semantic decision tasks (LD and SD), analysing whole-cortex evoked responses and spectral functional connectivity (coherence) in source-estimated electroencephalography and magnetoencephalography (EEG and MEG) recordings. Our evoked analysis revealed generally larger activation for SD compared to LD, starting in primary visual area (PVA) and angular gyrus (AG), followed by left posterior temporal cortex (PTC) and left anterior temporal lobe (ATL). The earliest activation effects in ATL were significantly left-lateralised. Our functional connectivity results showed significant connectivity between left and right ATL, PTC and right ATL in an early time window, as well as between left ATL and IFG in a later time window. The connectivity of AG was comparatively sparse. We quantified the limited spatial resolution of our source estimates via a leakage index for careful interpretation of our results. Our findings suggest that the different demands on semantic information retrieval in lexical and semantic decision tasks first modulate visual and attentional processes, then multimodal semantic information retrieval in the ATLs and finally control regions (PTC and IFG) in order to extract task-relevant semantic features for response selection. Whilst our evoked analysis suggests a dominance of left ATL for semantic processing, our functional connectivity analysis also revealed significant involvement of right ATL in the more demanding semantic task. Our findings demonstrate the complementarity of evoked and functional connectivity analysis, as well as the importance of dynamic information for both types of analyses.

Recurrence is required to capture the representational dynamics of the human visual system.

The human visual system is an intricate network of brain regions that enables us to recognize the world around us. Despite its abundant lateral and feedback connections, object processing is commonly viewed and studied as a feedforward process. Here, we measure and model the rapid representational dynamics across multiple stages of the human ventral stream using time-resolved brain imaging and deep learning. We observe substantial representational transformations during the first 300 ms of processing within and across ventral-stream regions. Categorical divisions emerge in sequence, cascading forward and in reverse across regions, and Granger causality analysis suggests bidirectional information flow between regions. Finally, recurrent deep neural network models clearly outperform parameter-matched feedforward models in terms of their ability to capture the multiregion cortical dynamics. Targeted virtual cooling experiments on the recurrent deep network models further substantiate the importance of their lateral and top-down connections. These results establish that recurrent models are required to understand information processing in the human ventral stream.

The Neural Time Course of Semantic Ambiguity Resolution in Speech Comprehension.

Semantically ambiguous words challenge speech comprehension, particularly when listeners must select a less frequent (subordinate) meaning at disambiguation. Using combined magnetoencephalography (MEG) and EEG, we measured neural responses associated with distinct cognitive operations during semantic ambiguity resolution in spoken sentences: (i) initial activation and selection of meanings in response to an ambiguous word and (ii) sentence reinterpretation in response to subsequent disambiguation to a subordinate meaning. Ambiguous words elicited an increased neural response approximately 400-800 msec after their acoustic offset compared with unambiguous control words in left frontotemporal MEG sensors, corresponding to sources in bilateral frontotemporal brain regions. This response may reflect increased demands on processes by which multiple alternative meanings are activated and maintained until later selection. Disambiguating words heard after an ambiguous word were associated with marginally increased neural activity over bilateral temporal MEG sensors and a central cluster of EEG electrodes, which localized to similar bilateral frontal and left temporal regions. This later neural response may reflect effortful semantic integration or elicitation of prediction errors that guide reinterpretation of previously selected word meanings. Across participants, the amplitude of the ambiguity response showed a marginal positive correlation with comprehension scores, suggesting that sentence comprehension benefits from additional processing around the time of an ambiguous word. Better comprehenders may have increased availability of subordinate meanings, perhaps due to higher quality lexical representations and reflected in a positive correlation between vocabulary size and comprehension success.

Multi-dimensional connectivity: a conceptual and mathematical review.

The estimation of functional connectivity between regions of the brain, for example based on statistical dependencies between the time series of activity in each region, has become increasingly important in neuroimaging. Typically, multiple time series (e.g. from each voxel in fMRI data) are first reduced to a single time series that summarises the activity in a region of interest, e.g. by averaging across voxels or by taking the first principal component; an approach we call one-dimensional connectivity. However, this summary approach ignores potential multi-dimensional connectivity between two regions, and a number of recent methods have been proposed to capture such complex dependencies. Here we review the most common multi-dimensional connectivity methods, from an intuitive perspective, from a formal (mathematical) point of view, and through a number of simulated and real (fMRI and MEG) data examples that illustrate the strengths and weaknesses of each method. The paper is accompanied with both functions and scripts, which implement each method and reproduce all the examples.

Adaptive cortical parcellations for source reconstructed EEG/MEG connectomes.

There is growing interest in the rich temporal and spectral properties of the functional connectome of the brain that are provided by Electro- and Magnetoencephalography (EEG/MEG). However, the problem of leakage between brain sources that arises when reconstructing brain activity from EEG/MEG recordings outside the head makes it difficult to distinguish true connections from spurious connections, even when connections are based on measures that ignore zero-lag dependencies. In particular, standard anatomical parcellations for potential cortical sources tend to over- or under-sample the real spatial resolution of EEG/MEG. By using information from cross-talk functions (CTFs) that objectively describe leakage for a given sensor configuration and distributed source reconstruction method, we introduce methods for optimising the number of parcels while simultaneously minimising the leakage between them. More specifically, we compare two image segmentation algorithms: 1) a split-and-merge (SaM) algorithm based on standard anatomical parcellations and 2) a region growing (RG) algorithm based on all the brain vertices with no prior parcellation. Interestingly, when applied to minimum-norm reconstructions for EEG/MEG configurations from real data, both algorithms yielded approximately 70 parcels despite their different starting points, suggesting that this reflects the resolution limit of this particular sensor configuration and reconstruction method. Importantly, when compared against standard anatomical parcellations, resolution matrices of adaptive parcellations showed notably higher sensitivity and distinguishability of parcels. Furthermore, extensive simulations of realistic networks revealed significant improvements in network reconstruction accuracies, particularly in reducing false leakage-induced connections. Adaptive parcellations therefore allow a more accurate reconstruction of functional EEG/MEG connectomes.

Frontal Cortex Supports the Early Structuring of Multiple Solution Steps in Symbolic Problem-solving.

problem-solving relies on a sequence of cognitive steps involving phases of task encoding, the structuring of solution steps, and their execution. On the neural level, metabolic neuroimaging studies have associated a frontal-parietal network with various aspects of executive control during numerical and nonnumerical problem-solving. We used EEG-MEG to assess whether frontal cortex contributes specifically to the early structuring of multiple solution steps. Basic multiplication ("3 × 4" vs. "3 × 24") was compared with an arithmetic sequence rule ("first add the two digits, then multiply the sum with the smaller digit") on two complexity levels. This allowed dissociating demands of early solution step structuring from early task encoding demands. Structuring demands were high for conditions that required multiple steps, that is, complex multiplication and the two arithmetic sequence conditions, but low for easy multiplication that mostly relied on direct memory retrieval. Increased right frontal activation in time windows between 300 and 450 msec was observed only for conditions that required multiple solution steps. General task encoding demands, operationalized by problem size (one-digit vs. two-digit numbers), did not predict these early frontal effects. In contrast, parietal effects occurred as a function of problem size irrespectively of structuring demands in early phases of task encoding between 100 and 300 msec. We here propose that frontal cortex subserves domain-general processes of problem-solving, such as the structuring of multiple solution steps, whereas parietal cortex supports number-specific early encoding processes that vary as a function of problem size.

Frontal and Parietal Cortices Show Different Spatiotemporal Dynamics across Problem-solving Stages.

Arithmetic problem-solving can be conceptualized as a multistage process ranging from task encoding over rule and strategy selection to step-wise task execution. Previous fMRI research suggested a frontal-parietal network involved in the execution of complex numerical and nonnumerical tasks, but evidence is lacking on the particular contributions of frontal and parietal cortices across time. In an arithmetic task paradigm, we evaluated individual participants' "retrieval" and "multistep procedural" strategies on a trial-by-trial basis and contrasted those in time-resolved analyses using combined EEG and MEG. Retrieval strategies relied on direct retrieval of arithmetic facts (e.g., 2 + 3 = 5). Procedural strategies required multiple solution steps (e.g., 12 + 23 = 12 + 20 + 3 or 23 + 10 + 2). Evoked source analyses revealed independent activation dynamics within the first second of problem-solving in brain areas previously described as one network, such as the frontal-parietal cognitive control network: The right frontal cortex showed earliest effects of strategy selection for multistep procedural strategies around 300 msec, before parietal cortex activated around 700 msec. In time-frequency source power analyses, memory retrieval and multistep procedural strategies were differentially reflected in theta, alpha, and beta frequencies: Stronger beta and alpha desynchronizations emerged for procedural strategies in right frontal, parietal, and temporal regions as function of executive demands. Arithmetic fact retrieval was reflected in right prefrontal increases in theta power. Our results demonstrate differential brain dynamics within frontal-parietal networks across the time course of a problem-solving process, and analyses of different frequency bands allowed us to disentangle cortical regions supporting the underlying memory and executive functions.

Early Parallel Activation of Semantics and Phonology in Picture Naming: Evidence from a Multiple Linear Regression MEG Study.

The time course of brain activation during word production has become an area of increasingly intense investigation in cognitive neuroscience. The predominant view has been that semantic and phonological processes are activated sequentially, at about 150 and 200-400 ms after picture onset. Although evidence from prior studies has been interpreted as supporting this view, these studies were arguably not ideally suited to detect early brain activation of semantic and phonological processes. We here used a multiple linear regression approach to magnetoencephalography (MEG) analysis of picture naming in order to investigate early effects of variables specifically related to visual, semantic, and phonological processing. This was combined with distributed minimum-norm source estimation and region-of-interest analysis. Brain activation associated with visual image complexity appeared in occipital cortex at about 100 ms after picture presentation onset. At about 150 ms, semantic variables became physiologically manifest in left frontotemporal regions. In the same latency range, we found an effect of phonological variables in the left middle temporal gyrus. Our results demonstrate that multiple linear regression analysis is sensitive to early effects of multiple psycholinguistic variables in picture naming. Crucially, our results suggest that access to phonological information might begin in parallel with semantic processing around 150 ms after picture onset.

Early Visual Word Processing Is Flexible: Evidence from Spatiotemporal Brain Dynamics.

Visual word recognition is often described as automatic, but the functional locus of top-down effects is still a matter of debate. Do task demands modulate how information is retrieved, or only how it is used? We used EEG/MEG recordings to assess whether, when, and how task contexts modify early retrieval of specific psycholinguistic information in occipitotemporal cortex, an area likely to contribute to early stages of visual word processing. Using a parametric approach, we analyzed the spatiotemporal response patterns of occipitotemporal cortex for orthographic, lexical, and semantic variables in three psycholinguistic tasks: silent reading, lexical decision, and semantic decision. Task modulation of word frequency and imageability effects occurred simultaneously in ventral occipitotemporal regions-in the vicinity of the putative visual word form area-around 160 msec, following task effects on orthographic typicality around 100 msec. Frequency and typicality also produced task-independent effects in anterior temporal lobe regions after 200 msec. The early task modulation for several specific psycholinguistic variables indicates that occipitotemporal areas integrate perceptual input with prior knowledge in a task-dependent manner. Still, later task-independent effects in anterior temporal lobes suggest that word recognition eventually leads to retrieval of semantic information irrespective of task demands. We conclude that even a highly overlearned visual task like word recognition should be described as flexible rather than automatic.

Concurrent brain responses to separate auditory and visual targets.

In the attentional blink, a target event (T1) strongly interferes with perception of a second target (T2) presented within a few hundred milliseconds. Concurrently, the brain's electromagnetic response to the second target is suppressed, especially a late negative-positive EEG complex including the traditional P3 wave. An influential theory proposes that conscious perception requires access to a distributed, frontoparietal global workspace, explaining the attentional blink by strong mutual inhibition between concurrent workspace representations. Often, however, the attentional blink is reduced or eliminated for targets in different sensory modalities, suggesting a limit to such global inhibition. Using functional magnetic resonance imaging, we confirm that visual and auditory targets produce similar, distributed patterns of frontoparietal activity. In an attentional blink EEG/MEG design, however, an auditory T1 and visual T2 are identified without mutual interference, with largely preserved electromagnetic responses to T2. The results suggest parallel brain responses to target events in different sensory modalities.

How are things adding up? Neural differences between arithmetic operations are due to general problem solving strategies.

A number of previous studies have interpreted differences in brain activation between arithmetic operation types (e.g. addition and multiplication) as evidence in favor of distinct cortical representations, processes or neural systems. It is still not clear how differences in general task complexity contribute to these neural differences. Here, we used a mental arithmetic paradigm to disentangle brain areas related to general problem solving from those involved in operation type specific processes (addition versus multiplication). We orthogonally varied operation type and complexity. Importantly, complexity was defined not only based on surface criteria (for example number size), but also on the basis of individual participants' strategy ratings, which were validated in a detailed behavioral analysis. We replicated previously reported operation type effects in our analyses based on surface criteria. However, these effects vanished when controlling for individual strategies. Instead, procedural strategies contrasted with memory retrieval reliably activated fronto-parietal and motor regions, while retrieval strategies activated parietal cortices. This challenges views that operation types rely on partially different neural systems, and suggests that previously reported differences between operation types may have emerged due to invalid measures of complexity. We conclude that mental arithmetic is a powerful paradigm to study brain networks of abstract problem solving, as long as individual participants' strategies are taken into account.

A framework for the design of flexible cross-talk functions for spatial filtering of EEG/MEG data: DeFleCT.

Brain activation estimated from EEG and MEG data is the basis for a number of time-series analyses. In these applications, it is essential to minimize "leakage" or "cross-talk" of the estimates among brain areas. Here, we present a novel framework that allows the design of flexible cross-talk functions (DeFleCT), combining three types of constraints: (1) full separation of multiple discrete brain sources, (2) minimization of contributions from other (distributed) brain sources, and (3) minimization of the contribution from measurement noise. Our framework allows the design of novel estimators by combining knowledge about discrete sources with constraints on distributed source activity and knowledge about noise covariance. These estimators will be useful in situations where assumptions about sources of interest need to be combined with uncertain information about additional sources that may contaminate the signal (e.g. distributed sources), and for which existing methods may not yield optimal solutions. We also show how existing estimators, such as maximum-likelihood dipole estimation, L2 minimum-norm estimation, and linearly-constrained minimum variance as well as null-beamformers, can be derived as special cases from this general formalism. The performance of the resulting estimators is demonstrated for the estimation of discrete sources and regions-of-interest in simulations of combined EEG/MEG data. Our framework will be useful for EEG/MEG studies applying time-series analysis in source space as well as for the evaluation and comparison of linear estimators.

Decoding Semantics from Dynamic Brain Activation Patterns: From Trials to Task in EEG/MEG Source Space.

The temporal dynamics within the semantic brain network and its dependence on stimulus and task parameters are still not well understood. Here, we addressed this by decoding task as well as stimulus information from source-estimated EEG/MEG human data. We presented the same visual word stimuli in a lexical decision (LD) and three semantic decision (SD) tasks. The meanings of the presented words varied across five semantic categories. Source space decoding was applied over time in five ROIs in the left hemisphere (anterior and posterior temporal lobe, inferior frontal gyrus, primary visual areas, and angular gyrus) and one in the right hemisphere (anterior temporal lobe). Task decoding produced sustained significant effects in all ROIs from 50 to 100 ms, both when categorizing tasks with different semantic demands (LD-SD) as well as for similar semantic tasks (SD-SD). In contrast, a semantic word category could only be decoded in lATL, rATL, PTC, and IFG, between 250 and 500 ms. Furthermore, we compared two approaches to source space decoding: conventional ROI-by-ROI decoding and combined-ROI decoding with back-projected activation patterns. The former produced more reliable results for word category decoding while the latter was more informative for task decoding. This indicates that task effects are distributed across the whole semantic network while stimulus effects are more focal. Our results demonstrate that the semantic network is widely distributed but that bilateral anterior temporal lobes together with control regions are particularly relevant for the processing of semantic information.

An eye on semantics: a study on the influence of concreteness and predictability on early fixation durations.

We used eye-tracking during natural reading to study how semantic control and representation mechanisms interact for the successful comprehension of sentences, by manipulating sentence context and single-word meaning. Specifically, we examined whether a word's semantic characteristic (concreteness) affects first fixation and gaze durations (FFDs and GDs) and whether it interacts with the predictability of a word. We used a linear mixed effects model including several possible psycholinguistic covariates. We found a small but reliable main effect of concreteness and replicated a predictability effect on FFDs, but we found no interaction between the two. The results parallel previous findings of additive effects of predictability (context) and frequency (lexical level) in fixation times. Our findings suggest that the semantics of a word and the context created by the preceding words additively influence early stages of word processing in natural sentence reading.

Linguistic and attentional factors - Not statistical regularities - Contribute to word-selective neural responses with FPVS-oddball paradigms.

Studies using frequency-tagging in electroencephalography (EEG) have dramatically increased in the past 10 years, in a variety of domains and populations. Here we used Fast Periodic Visual Stimulation (FPVS) combined with an oddball design to explore visual word recognition. Given the paradigm's high sensitivity, it is crucial for future basic research and clinical application to prove its robustness across variations of designs, stimulus types and tasks. This paradigm uses periodicity of brain responses to measure discrimination between two experimentally defined categories of stimuli presented periodically. EEG was recorded in 22 adults who viewed words inserted every 5 stimuli (at 2 Hz) within base stimuli presented at 10 Hz. Using two discrimination levels (deviant words among nonwords or pseudowords), we assessed the impact of relative frequency of item repetition (set size or item repetition controlled for deviant versus base stimuli), and of the orthogonal task (focused or deployed spatial attention). Word-selective occipito-temporal responses were robust at the individual level (significant in 95% of participants), left-lateralized, larger for the prelexical (nonwords) than lexical (pseudowords) contrast, and stronger with a deployed spatial attention task as compared to the typically used focused task. Importantly, amplitudes were not affected by item repetition. These results help understanding the factors influencing word-selective EEG responses and support the validity of FPVS-EEG oddball paradigms, as they confirm that word-selective responses are linguistic. Second, they show its robustness against design-related factors that could induce statistical (ir)regularities in item rate. They also confirm its high individual sensitivity and demonstrate how it can be optimized, using a deployed rather than focused attention task, to measure implicit word recognition processes in typical and atypical populations.

Minimum-norm cortical source estimation in layered head models is robust against skull conductivity error.

The conductivity profile of the head has a major effect on EEG signals, but unfortunately the conductivity for the most important compartment, skull, is only poorly known. In dipole modeling studies, errors in modeled skull conductivity have been considered to have a detrimental effect on EEG source estimation. However, as dipole models are very restrictive, those results cannot be generalized to other source estimation methods. In this work, we studied the sensitivity of EEG and combined MEG+EEG source estimation to errors in skull conductivity using a distributed source model and minimum-norm (MN) estimation. We used a MEG/EEG modeling set-up that reflected state-of-the-art practices of experimental research. Cortical surfaces were segmented and realistically-shaped three-layer anatomical head models were constructed, and forward models were built with Galerkin boundary element method while varying the skull conductivity. Lead-field topographies and MN spatial filter vectors were compared across conductivities, and the localization and spatial spread of the MN estimators were assessed using intuitive resolution metrics. The results showed that the MN estimator is robust against errors in skull conductivity: the conductivity had a moderate effect on amplitudes of lead fields and spatial filter vectors, but the effect on corresponding morphologies was small. The localization performance of the EEG or combined MEG+EEG MN estimator was only minimally affected by the conductivity error, while the spread of the estimate varied slightly. Thus, the uncertainty with respect to skull conductivity should not prevent researchers from applying minimum norm estimation to EEG or combined MEG+EEG data. Comparing our results to those obtained earlier with dipole models shows that general judgment on the performance of an imaging modality should not be based on analysis with one source estimation method only.

A role for the motor system in binding abstract emotional meaning.

Sensorimotor areas activate to action- and object-related words, but their role in abstract meaning processing is still debated. Abstract emotion words denoting body internal states are a critical test case because they lack referential links to objects. If actions expressing emotion are crucial for learning correspondences between word forms and emotions, emotion word-evoked activity should emerge in motor brain systems controlling the face and arms, which typically express emotions. To test this hypothesis, we recruited 18 native speakers and used event-related functional magnetic resonance imaging to compare brain activation evoked by abstract emotion words to that by face- and arm-related action words. In addition to limbic regions, emotion words indeed sparked precentral cortex, including body-part-specific areas activated somatotopically by face words or arm words. Control items, including hash mark strings and animal words, failed to activate precentral areas. We conclude that, similar to their role in action word processing, activation of frontocentral motor systems in the dorsal stream reflects the semantic binding of sign and meaning of abstract words denoting emotions and possibly other body internal states.