Skip to main content
NCBI home page
Human Brain Mapping logoLink to Human Brain Mapping
. 2010 Aug 2;32(7):1081–1090. doi: 10.1002/hbm.21093

Two distinct neural networks support the mapping of meaning to a novel word

Zheng Ye 1,2, Anna Mestres‐Missé 3, Antoni Rodriguez‐Fornells 4,5, Thomas F Münte 2,6,✉
PMCID: PMC6870358  PMID: 20681014

Abstract

Children can learn the meaning of a new word from context during normal reading or listening, without any explicit instruction. It is unclear how such meaning acquisition is supported and achieved in human brain. In this functional magnetic resonance imaging (fMRI) study we investigated neural networks supporting word learning with a functional connectivity approach. Participants were exposed to a new word presented in two successive sentences and needed to derive the meaning of the new word. We observed two neural networks involved in mapping the meaning to the new word. One network connected the left inferior frontal gyrus (LIFG) with the middle frontal gyrus (MFG), medial superior frontal gyrus, caudate nucleus, thalamus, and inferior parietal lobule. The other network connected the left middle temporal gyrus (LMTG) with the MFG, anterior and posterior cingulate cortex. The LIFG network showed stronger interregional interactions for new than real words, whereas the LMTG network showed similar connectivity patterns for new and real words. We proposed that these two networks support different functions during word learning. The LIFG network appears to select the most appropriate meaning from competing candidates and to map the selected meaning onto the new word. The LMTG network may be recruited to integrate the word into sentential context, regardless of whether the word is real or new. The LIFG and the LMTG networks share a common node, the MFG, suggesting that these two networks communicate in working memory. Hum Brain Mapp, 2011. © 2010 Wiley‐Liss, Inc.

Keywords: meaning acquisition, frontal cortex, basal ganglia, thalamus, fMRI, functional connectivity

INTRODUCTION

To acquire the meaning of a novel word, learners need to establish an arbitrary correspondence between a concept and the particular sign or sound. Children can learn as many as 3,000 words every year between the age of 8 and 17 years but only a small proportion of this growth (about 200–300 words per year) is due to formal vocabulary instruction [Nagy and Anderson, 1984]. Most of these words must be learned from context during normal reading and listening [Nagy et al., 1987]. However, it is still unclear how this process is supported and achieved in the human brain.

Recent studies have used the human simulation paradigm [Gillette et al., 1999; Mestres‐Missé et al., 2007] to mimic contextual word learning processes. In this paradigm participants are exposed to a novel word that appears in successive sentences with increasing contextual constraint, and they can thus derive the meaning of the hidden word. For example,

  • Sentence 1: She likes people with nice and clean tankies

  • Sentence 2: After the meals you should brush your tankies.

Hidden word: teeth

A first fMRI study revealed a distributed set of brain areas as being modulated by the meaning acquisition process comprising the left inferior frontal gyrus (LIFG, BA 45), the left middle temporal gyrus (LMTG, BA 21), the parahippocampal gyrus, and subcortical structures (thalamus and striatum) [Mestres‐Missé et al., 2008]. Subsequent studies contrasted the acquisition of abstract and concrete nouns [Mestres‐Missé et al., 2009] and verbs and nouns [Mestres‐Missé et al., 2010] and again identified the LIFG and LMTG as major hubs in the process of mapping a meaning onto a novel word.

These activation patterns were obtained by contrasting sentences ending in a novel word with sentences ending in a real word as well as by contrasting the first and the second exposures to a novel word. The question arises how the different brain areas work together. We therefore subjected the data obtained by Mestres‐Missé et al. [ 2009, 2010] to a functional connectivity analysis using the two most prominently activated brain areas (LIFG and LMTG) as seed regions. Specifically, we applied the “β series correlation” method proposed by Rissman et al. [ 2004]. This method is implemented on the basis of a general linear model, using separate covariates to model hemodynamic responses of a particular stage in each single trial and giving rise to series of parameter estimates (β values) for the stage. If two regions are functionally connected within a network, their β series should be strongly correlated.

A brain network was considered to be involved in mapping a meaning onto a novel word, if the connectivity in this network was stronger for the presentation of new words than for the presentation of real words. Moreover, a stronger connectivity pattern for the first presentation of a novel word, i.e. when the sentential context might license more than one possible word, than for the second presentation would further support a role of the network in the meaning mapping process.

METHODS

fMRI data used for the functional connectivity analysis were acquired by Mestres‐Missé et al. [ 2010] in 21 healthy participants (11 women, mean age 24 ± 1.8 years). A second data set using a similar paradigm was also subjected to the connectivity analysis and is reported as Supporting Information (univariate analyses reported in Mestres‐Missé et al. [ 2009]).

Stimuli and Task

Participants read pairs of sentences in each functional run and completed a forced‐choice task and a recognition task after each run. There were two conditions, each comprising 80 pairs of sentences. In the experimental condition, the two sentences ended with the same new word. Participants had to discover the meaning of the hidden word (see example, given in the Introduction section). In the control condition, the two sentences ended with the same real word. Participants were asked to think about a semantically related word of the terminal word. Half of the new and the real words were nouns (Nn for new nouns and Rn for real nouns) and the other half were verbs. In the present study we focus on nouns, as the mapping of noun concepts and the mapping of verb concepts have been shown to recruit different brain areas [Mestres‐Missé et al., 2010]. The contextual constraint was increased from the first Nn sentence (low constraint) to the second Nn sentence (high constraint). Thus, it is more likely to derive an appropriate meaning after the second presentation of a novel word (for details of the stimulus construction, see the Supporting Information).

Each trial began with a fixation period of 500 ms. The stem of the first sentence (all words except the terminal word) was then displayed centrally for 2,000 ms. After a variable interval of 1,000 to 2,000 ms, the terminal word was displayed for 500 ms. The second sentence was presented in the same fashion after another variable interval of 1,000 to 6,000 ms. Following each functional run, a forced‐choice task was carried out to test whether participants assigned correct meanings to the new words [Mestres‐Missé et al., 2010].

Data Analysis

Data were collected on a 3T Siemens Trio system. Functional images were acquired using a T2*‐weighted echo planar imaging (EPI) sequence, with 2,000‐ms time repetition, 30‐ms time echo, and 80° flip angle, in eight runs. Each functional image consisted of 32 transversal slices, with 64 × 64 matrix, 224 × 224 mm field of view, 4‐mm thickness, no gap, and 3.5 × 3.5 mm inplane resolution.

Data were analyzed with SPM5 (available at: http://www.fil.ion.ucl.ac.uk/spm). The first four volumes were discarded owing to equilibration effects. Functional images were first phase‐shifted with reference to the first slice to correct differences in slice acquisition time. They were then realigned with a least squares approach and a rigid body spatial transformation to remove movement artifacts. Realigned images were normalized to the EPI‐derived MNI template (ICBM 152, Montreal Neurological Institute) and resampled to 4 × 4 × 4 mm3 voxel size. Normalized images were finally smoothed with a Gaussian kernel of 8 mm full‐width half‐maximum and filtered with a high‐pass filter of 128 seconds.

The functional connectivity analysis was performed to examine neural networks supporting the mapping of meaning onto a novel word. This approach is based on the hypothesis that if two regions interact within a network, their activity patterns should be strongly correlated [Rissman et al., 2004]. This analysis was implemented on the basis of a general linear model by using separate covariates to model hemodynamic responses of each terminal word in each single sentence. Estimated movement parameters (six per image: x, y, z, pitch, roll, and yaw) were included in the model to minimize signal‐correlated motion effect. For each participant, estimated β values were extracted to form a set of condition‐specific β series. Two seeds were defined as spheres with 4‐mm radius: one in the LIFG (coordinates in MNI: −44, 20, 20) and the other in the LMTG (−64, −40, −4). The coordinates were taken from the activation peaks observed in Mestres‐Missé et al. [ 2010]. For each seed, the β series were averaged across voxels within the critical region and correlated with beta series of every other voxel in the whole brain. We first calculated maps of correlation coefficients separately for Nn and Rn but across the word in the first sentence and that in the second sentence to examine what regions are functionally connected with the LIFG or the MTG during the mapping of meaning. The correlation maps were normalized by using an arc‐hyperbolic tangent transform on the subject level. To show interregional interactions for Nn and for Rn, we entered the correlation maps into one‐sample t tests on the group level and considered resulting maps at P < 0.0005 (family‐wise error correction, FWE‐corrected) with a minimum cluster size of 100 voxels. To show interregional interaction stronger for Nn than Rn, we entered the correlation maps into paired‐sample t tests (Nn > Rn) and considered differences at P < 0.05 (uncorrected) with a minimum cluster size of 5 voxels. Then we calculated correlation maps separately for the new noun in the first sentence (1Nn) and that in the second sentence (2Nn) to examine how the LIFG and the LMTG networks change during the exploration of new words. The correlation maps were normalized by using an arc‐hyperbolic tangent transform on the subject level. To show interregional interactions for 1Nn and for 2Nn, we entered the correlation maps into one‐sample t tests on the group level and considered resulting maps at P < 0.0005 (FWE‐corrected) with a minimum cluster size of 100 voxels. To show interregional interaction stronger for 1Nn than 2Nn, we entered the correlation maps into paired‐sample t tests (1Nn > 2Nn) and considered differences at P < 0.05 (uncorrected) with a minimum cluster size of 5 voxels.

RESULTS

Behavioral results indicated that participants assigned appropriate meanings to the new words and did pay attention to the real words in sentences (for details, see Mestres‐Missé et al. [ 2010]).

Figure 1 shows regions interacting with the LIFG and/or the LMTG during the mapping of meaning. Remarkably, the functional connectivity analysis revealed two distinct neural networks. The LIFG network showed significant differences between the new and the real nouns (Fig. 1A). For Nn, the LIFG strongly interacted with the medial superior frontal gyrus/supplementary motor area (mSFG/SMA), the bilateral middle frontal gyrus (MFG), the left inferior parietal lobule (IPL), the caudate nucleus, and the medial thalamus. Among these areas, only the caudate nucleus and the medial thalamus were connected with the LIFG for Rn. The LIFG connections were significantly stronger for Nn than Rn in the mSFG, the bilateral MFG, the left IPL, the caudate nucleus, and the medial thalamus (also see Table I). By contrast, the LMTG network showed similar patterns for the new and the real nouns (Fig. 1B). For both Nn and Rn, the LMTG interacted with the SMA, the anterior cingulate cortex (ACC), the middle/posterior cingulate cortex (MCC/PCC), the bilateral MFG, the bilateral IPL, the precuneus, and the left thalamus. The LMTG connection was not significantly different between Nn and Rn in these areas except the left precuneus and the right IPL (also see Table I). This pattern was replicated with data from another experiment [Mestres‐Missé et al., 2009] using a similar design but different materials and participants (Supp. Info., Fig. S2).

Figure 1.

Figure 1

Open in a new tab

Interregional interactions for the new noun (Nn) and for the real noun (Rn). A, Regions functionally connected with the LIFG for Rn and for Nn, and LIFG connections stronger for Nn than Rn (Nn > Rn). B, Regions functionally connected with the LMTG for Rn and for Nn, and LMTG connections stronger for Nn than Rn (Nn > Rn). mSFG, medial superior frontal gyrus; SMA, supplementary motor area; ACC, anterior cingulate cortex; MCC, middle cingulate cortex; MFG, middle frontal gyrus; IPL, inferior parietal lobule; Prec, precuneus; Caud, caudate nucleus; Thal, thalamus; L, left; R, right. Coordinates in MNI. Color scales indicate t values.

Table I.

LIFG connections and LMTG connections stronger for Nn than Rn

Region BA H x y z t Size
LIFG connections
 Medial superior frontal gyrus 9 L/R 8 44 56 4.53 399
 Middle frontal gyrus 9/8 L −36 28 56 2.99 81
R 24 28 56 2.85 70
 Inferior frontal gyrus 45 L −52 20 20 2.46 19
47 R 40 32 −12 3.43 88
 Inferior parietal lobule 40 L −52 −32 56 2.30 12
 Precuneus 30 R 4 −48 20 2.90 163
 Middle occipital gyrus/angular gyrus 19/39 L −44 −72 36 2.30 11
R 40 −76 32 3.07 121
 Inferior temporal gyrus 20 L −44 −4 −24 2.18 20
R 40 4 −36 2.63 26
 Fusiform gyrus 37 L −28 −44 −16 2.66 11
R 44 −16 −32 2.79 42
 Caudate nucleus L/R 8 8 −8 3.27 20
 Thalamus L/R 0 −4 4 2.89 8
LMTG connections
 Superior frontal gyrus 8 R 20 32 56 2.34 18
 Inferior frontal gyrus (orbital) 47 L −56 24 −4 3.71 382
R 44 16 −20 3.63 148
 Superior parietal lobule 7 R 28 −60 60 2.24 24
 Inferior parietal lobule 40 R 52 −56 52 2.33 10
 Precuneus 30 L −12 −56 16 3.32 103
 Angular gyrus 39 L −48 −72 24 2.68 26
 Inferior occipital gyrus 18 L −28 −84 −16 3.24 12
 Fusiform gyrus 37 L −28 −36 −16 2.42 20
Open in a new tab

BA, Brodmann Area; H, hemisphere; L, left; R, right; coordinates in MNI; t, statistic values; size in number of voxels.

Figure 2 shows the patterns of the LIFG and the LMTG networks changed from the first to the second presentation of a new word. For 1Nn, the LIFG strongly interacted with the mSFG/SMA, the bilateral MFG, the right IPL, the caudate nucleus, and the medial thalamus. Although the LIFG interacted with the same areas for 2Nn, the LIFG connections were significantly weakened from 1Nn to 2Nn in the SMA/MCC, the right MFG, and the caudate nucleus (Fig. 2A and Table II). For 1Nn, the LMTG interacted with the SMA, the MCC/PCC, the bilateral MFG, the bilateral IPL, the precuneus, the caudate nucleus, and the thalamus. Among these areas, only the left MFG and the precuneus were connected with the LMTG for 2Nn. The LMTG connections were significantly weakened from 1Nn to 2Nn in the SMA, the right MFG, and the caudate nucleus (Fig. 2B and Table II).

Figure 2.

Figure 2

Open in a new tab

Interregional interactions for the new noun in the first sentence (1Nn) and that in the second sentence (2Nn). A, Regions functionally connected with the LIFG for 1Nn and for 2Nn, and LIFG connections stronger for 1Nn than 2Nn (1Nn > 2Nn). B, Regions functionally connected with the LMTG for 1Nn and for 2Nn, and LMTG connections stronger for 1Nn than 2Nn (1Nn > 2Nn). SMA, supplementary motor area; MCC, middle cingulate cortex; IPL, inferior parietal lobule; Caud, caudate nucleus; L, left; R, right. Coordinates in MNI. Color scales indicate t values.

Table II.

LIFG connections and LMTG connections stronger for 1Nn than 2Nn

Region BA H x y z t Size
LIFG connections
 Supplementary motor area 6 L/R 4 0 48 3.01 57
 Middle frontal gyrus 46 R 36 32 48 2.95 93
 Rolandic operculum L −44 −24 24 3.23 354
 Precentral gyrus 6 L −60 4 24 3.06 111
 Postcentral gyrus 2 L −20 −40 56 2.75 59
R 20 −28 56 2.87 12
 Superior temporal pole 38 R 52 8 −16 2.32 5
 Superior parietal lobule 7 L −24 −72 60 2.93 5
R 24 −68 64 3.11 25
 Superior occipital gyrus 18 L −24 −92 20 3.88 40
R 28 −92 30 4.33 74
 Lingual gyrus 18 L −20 −80 −8 4.39 88
R 28 −80 −8 3.61 138
 Caudate nucleus L/R −8 16 −4 3.06 37
 Cerebellum L −24 −40 −48 3.39 43
LMTG connections
 Supplementary motor area 6 L/R −8 −12 64 3.09 99
 Superior frontal gyrus 8 L −16 8 56 2.30 22
R 16 24 56 3.58 133
 Middle frontal gyrus 46 R 44 28 12 4.60 305
 Inferior frontal gyrus (orbital) 47 L −48 40 −12 2.78 14
 Rolandic operculum L −36 −28 20 2.57 18
 Superior parietal lobule 7 L −20 −72 60 2.41 17
R 20 −32 52 3.14 185
 Precuneus 7 R 8 −64 40 2.18 12
 Superior temporal pole 38 L −32 8 −24 2.59 24
 Middle temporal pole 21 L −52 8 −28 2.66 19
 Middle occipital gyrus 19 R 44 −76 32 2.82 18
 Caudate nucleus L/R −8 8 −8 3.04 78
 Thalamus L −4 −24 12 4.39 42
 Cerebellum L −8 −64 −44 3.13 1,112
R 20 −18 −52 5.05
Open in a new tab

BA, Brodmann Area; H, hemisphere; L, left; R, right; coordinates in MNI; t, statistic values; size in number of voxels.

DISCUSSION

With the connectivity approach, we observed two neural networks during the mapping of meaning. One network connected the LIFG with the mSFG/SMA, MFG, IPL, caudate nucleus, and medial thalamus. The other network connected the LMTG with MFG, ACC, and PCC. The LIFG network showed stronger interregional interactions for the new than the real words, whereas the LMTG network showed similar connectivity patterns for the new and the real words. These two networks therefore appear to support different functions during our simulated word learning task and, by extension, in natural word learning. A new word in a sentence with low contextual constraint (as in the 1Nn condition) may activate a large set of candidate concepts. Further semantic constraints imposed by the second sentence allow to home in on the most appropriate meaning and to map the novel word to an existing concept. The connectivity dynamics of the LIFG network fit with this, as connectivity was much stronger for the first exposure of the new word than for the second exposure. Also, the LIFG network did not show up when connectivity was analyzed for real words. Here, no mapping of meaning has to take place because the concept‐word correspondence has been firmly established in semantic memory and it is unnecessary to select a meaning from competing alternatives. We therefore propose that the LIFG network is supporting the actual mapping process. By contrast, the LMTG network showed very different connectivity dynamics as it was similarly recruited for new and real words. Regardless of whether the sentence terminal stimulus was a word or a new word, it has to be integrated into the sentential context. We therefore suggest that the LMTG network is related to semantic integration during sentence reading. It is noteworthy that both, LIFG and LMTG networks share a common node, i.e. the MFG. This structure has been identified with working memory processes [e.g., McCarthy et al., 1994, 1996; Tsukiura et al., 2001] suggesting that LIFG and LMTG networks contribute information to a common workspace in working memory (see illustration in Fig. 3).

Figure 3.

Figure 3

Open in a new tab

Two neural networks underlying meaning acquisition. LIFG, left inferior frontal gyrus; LMTG, left middle temporal gyrus; MFG, middle frontal gyrus; mSFG/SMA, medial superior frontal gyrus/supplementary motor area; IPL, inferior parietal lobule; ACC, anterior cingulate cortex; PCC, posterior cingulate cortex.

LIFG Network

Within the LIFG network, the LIFG (especially pars triangularis, BA45) has been associated to control semantic retrieval and/or selection processes which operate on simultaneously activated representations to bias toward one particular lexical item during word comprehension and production [Badre and Wagner, 2007; Badre et al., 2005; Thompson‐Schill et al., 1997]. The dorsal mSFG is assumed to maintain task goals and to monitor and adjust cognitive processes in sensory, memory and motor systems [Botvinick et al., 2001; Carter et al., 1998; Dosenbach et al., 2007]. The IPL may play a role in binding a concept with a particular sign or sound during word learning [Breitenstein et al., 2005] or in binding a new label with a novel object during object naming [Cornelissen et al., 2004]. The current connectivity analysis demonstrated that these cortical regions are linked to subcortical structures, i.e. the basal ganglia and the thalamus. Animal work has shown that the caudate nucleus and the putamen receive cortical projections and project back to cortical regions via the thalamus, forming parallel [Middleton and Strick, 2002] and integrative [Bar‐Gad and Bergman, 2001] circuits in support of motor, cognitive, and emotional processes.

The loops passing through the caudate nucleus and the medial thalamus seem to play a role in the learning and usage of conceptual and lexical knowledge [Ullman, 2006]. On the one hand, the caudate nucleus has been proposed to mediate a form of learning in which the stimulus‐response association is incrementally acquired [Knowlton et al., 1996; Packard and Knowlton, 2002]. On the other hand, the thalamus has been stressed in the processing of lexical‐semantic and syntactic information [Nadeau and Crosson, 1997; Wahl et al., 2008]. Mestres‐Missé et al. [ 2008] further suggested that this loop can maintain a processing bias because the thalamus receives inhibitory projections from the basal ganglia. Once the most appropriate item is selected, the inhibition of the thalamus will be released by the excitation of the caudate nucleus via the inhibition of the globus pallidus, giving rise to the excitation of related cortical regions [Crosson et al., 2003].

LMTG Network

The involvement of the LMTG may reflect the activation of stored semantic information because the regions are proposed to be a supramodal structure involved in the long‐term storage of conceptual knowledge which is segregated from word‐form representations [Martin and Chao, 2001; Patterson et al., 2007; Vandenberghe et al., 1996]. A prominent node of the LMTG network is the ACC, which is called upon whenever cognitive control processes are needed and attentional resources need to be assigned [Botvinick et al., 2001; Gitelman et al., 1999]. Other nodes including the PCC and adjacent precuneus have been linked with the encoding and retrieval of episodic memory [Valenstein et al., 1987; Vogt et al., 1992] as well as the monitoring and evaluation of sensory events [Raichle et al., 2001]. In the language domain, the ACC and the LMTG have been shown to be more activated for nonwords than words and for irregular than regular words during word production [Binder et al., 2005]. The PCC and the angular gyrus have been found more activated for words than nonwords during word recognition [Binder et al., 2003]. These findings suggest that the ACC corresponds to the difficulty of lexical/semantic processes and the PCC corresponds to the access of lexical items.

Although the LMTG network did not show differences between the new and the real words, connectivity within this network was weaker for the second than the first presentation of a new word. The same pattern was obtained for real words: the interregional connectivity was stronger for the first than the second presentation of a real word (Supp. Info., Fig. S1). The changes in the connectivity pattern might be explained by strategic factors. Since participants knew that the second sentence of a pair always had the same terminal word as the first sentence, it was easier for them to access and integrate that word. First and second sentences of a pair also differed in contextual constraint. Since the first sentence had the weaker contextual constraint, the stronger connections between LMTG and related regions may reflect a more effortful integration process. Alternatively, it might be considered that the decrease in connectivity strength from the first to the second presentation might result from repetition suppression. Repetition suppression is a phenomenon in which the BOLD response in neuronal populations sensitive to a stimulus is reduced after repetition of that stimulus [Grill‐Spector et al., 2006; Krekelberg et al., 2006]. The distinction of these two possibilities, ease of contextual integration versus stimulus repetition, could be done by including a condition that repeats the novel word but in two sentential context that do not allow the extraction of a coherent meaning. If repetition was driving connectivity patterns, connectivity should still decrease from the first to the second presentation of a word. If ease of contextual integration was crucial, connectivity should not decrease. Indeed, we have used such a condition in prior word with the human simulation paradigm in conjunction with event‐related brain potentials [Mestres‐Missé et al., 2007] and fMRI [Mestres‐Missé et al., 2008]. Unfortunately, the timing of events in the latter study did not allow to apply connectivity analysis to this data set. Thus, a new study should be run, to clarify this point.

It is interesting to try to relate the current findings regarding the LIFG network to existing models of language processing. The functions that have been assigned to the LIFG are quite diverse. For example, Hagoort [ 2005] in his memory‐unification‐control (MUC) model proposed the LIFG to be an active workspace unifying word information retrieved from memory at the semantic, syntactic, and phonological levels. The LIFG, as the core of the unification component, is thought to interact with the memory component, situated in left temporal cortex, and the control component, located in dorsolateral prefrontal cortex and the anterior cingulate cortex. The MUC framework does not specify functions of subcortical structures. The current data suggest that the LIFG and MTG networks are dissociable and it would thus be interesting to apply connectivity analysis more formally to test the predictions of the MUC‐model.

A proposal put forward by Friederici [ 2002] specifies the functions of the frontal and temporal brain regions on the basis of fMRI and ERP data. In her view, frontal regions support the formation of relationships with a specialization of BA 44 and the frontal opercular cortex for syntactic relationships, and BA 45/47 for semantic relationships. By contrast, temporal regions support identification processes, again with a specialization for semantic (left MTG) and syntactic (left anterior STG) processes. In Friederici's [ 2002] model, building of relationships and identification are partially independent processes. It is interesting to speculate whether the dissociation of the LIFG and LMTG network in the current study corresponds to Friederici's model. Clearly, connectivity analyses might be used to test predictions of Friederici's [ 2002] and Hagoort's [ 2005] (model), as well as those made by other recent neuroanatomical models of language processing [Caplan, 2001, 2006; Hickok and Poeppel, 2000, 2007; Poeppel and Hickok, 2004].

In conclusion, we found two neural networks involved in a task requiring the mapping of meaning to a novel word. One network, involving the LIFG, MFG, mSFG/SMA, caudate nucleus, thalamus, and IPL, supports the selection of the most appropriate meaning over competing candidates and to map the selected meaning onto the new word. The second network, comprising MTG, MFG, ACC, and PCC, is recruited to integrate the incoming word into the sentential context, no matter whether the word is preexisting in semantic memory. A common node shared by both networks is the MFG which serves as a neural hub for working memory. Thus, the two networks may communicate in working memory to work together during word learning.

Supporting information

Additional Supporting Information may be found in the online version of this article.

Supporting Information

Click here for additional data file. (1.7MB, doc)

REFERENCES

  1. Badre D, Poldrack RA, Pare‐Blagoev EJ, Insler RZ, Wagner AD ( 2005): Dissociable controlled retrieval and generalized selection mechanisms in ventrolateral prefrontal cortex. Neuron 47: 907–918. [DOI] [PubMed] [Google Scholar]
  2. Badre D, Wagner AD ( 2007): Left ventrolateral prefrontal cortex and the cognitive control of memory. Neuropsychologia 45: 2883–2901. [DOI] [PubMed] [Google Scholar]
  3. Bar‐Gad I, Bergman H ( 2001): Stepping out of the box: Information processing in the neural networks of the basal ganglia. Curr Opin Neurobiol 11: 689–695. [DOI] [PubMed] [Google Scholar]
  4. 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 1: 372–393. [DOI] [PubMed] [Google Scholar]
  5. Binder JR, Medler DA, Desai R, Conant LL, Liebenthal E ( 2005): Some neurophysiological constraints on models of word naming. Neuroimage 27: 677–693. [DOI] [PubMed] [Google Scholar]
  6. Botvinick MM, Braver TS, Barch DM, Carter CS, Cohen JD ( 2001): Conflict monitoring and cognitive control. Psychol Rev 108: 624–652. [DOI] [PubMed] [Google Scholar]
  7. Breitenstein C, Jansen A, Deppe M, Foerster AF, Sommer J, Wolbers T, Knecht S ( 2005): Hippocampus activity differentiates good from poor learners of a novel lexicon. Neuroimage 25: 958–968. [DOI] [PubMed] [Google Scholar]
  8. Caplan D ( 2001): Functional neuroimaging studies of syntactic processing. J Psycholinguist Res 30: 297–320. [DOI] [PubMed] [Google Scholar]
  9. Caplan D ( 2006): Why is Broca's area involved in syntax? Cortex 42: 469–471. [DOI] [PubMed] [Google Scholar]
  10. Carter CS, Braver TS, Barch DM, Botvinick MM, Noll D, Cohen JD ( 1998): Anterior cingulate cortex, error detection, and the on‐line monitoring of performance. Science 280: 747–749. [DOI] [PubMed] [Google Scholar]
  11. Cornelissen K, Laine M, Renvall K, Saarinen T, Martin N, Salmelin R ( 2004): Learning new names for new objects: Cortical effects as measured by magnetoencephalography. Brain Lang 89: 617–622. [DOI] [PubMed] [Google Scholar]
  12. Crosson B, Benefield H, Cato MA, Sadek JR, Moore AB, Wierenga CE, Gopinath K, Soltysik D, Bauer RM, Auerbach EJ, Gökçay D, Leonard CM, Briggs RW ( 2003): Left and right basal ganglia and frontal activity during language generation: contributions to lexical, semantic, and phonological processes. J Int Neuropsychol Soc 9: 1061–1077. [DOI] [PubMed] [Google Scholar]
  13. Dosenbach NUF, Fair DA, Miezin FM, Cohen AL, Wenger KK, Dosenbach RAT, Fox MD, Snyder AZ, Vincent JL, Raichle ME, Schlaggar BL, Petersen SE ( 2007): Distinct brain networks for adaptive and stable task control in humans. Proc Natl Acad Sci USA 104: 11073–11078. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Friederici AD ( 2002): Towards a neural basis of auditory sentence processing. Trends Cogn Sci 6: 78–84. [DOI] [PubMed] [Google Scholar]
  15. Gillette J, Gleitman H, Gleitman L, Lederer A ( 1999): Human simulations of vocabulary learning. Cognition 73: 135–176. [DOI] [PubMed] [Google Scholar]
  16. Gitelman DR, Nobre AC, Parrish TB, LaBar KS, Kim Y‐H, Meyer JR, Mesulam M‐M ( 1999): A large‐scale distributed network for covert spatial attention: Further anatomical delineation based on stringent behavioural and cognitive controls. Brain 122: 1093–1106. [DOI] [PubMed] [Google Scholar]
  17. Grill‐Spector K, Henson R, Martin A (2006): Repetition and the brain: Neural models of stimulus‐specific effects. Trends in Cognitive Sciences 10:14–23. [DOI] [PubMed] [Google Scholar]
  18. Hagoort P ( 2005). On Broca, brain, and binding: A new framework. Trends Cogn Sci 9: 416–423. [DOI] [PubMed] [Google Scholar]
  19. Hickok G, Poeppel D ( 2000): Towards a functional neuroanatomy of speech perception. Trends Cogn Sci 4: 131–138. [DOI] [PubMed] [Google Scholar]
  20. Hickok G, Poeppel D ( 2007): The cortical organization of speech processing. Nat Rev Neurosci 8: 393–402. [DOI] [PubMed] [Google Scholar]
  21. Poeppel D, Hickok G ( 2004): Towards a new functional anatomy of language. Cognition 92: 1–12. [DOI] [PubMed] [Google Scholar]
  22. Knowlton BJ, Mangels JA, Squire LR ( 1996): A neostriatal habit learning system in humans. Science 273: 1399–1402. [DOI] [PubMed] [Google Scholar]
  23. Krekelberg B, Boynton GM, van Wezel RJ (2006): Adaptation: from single cells to BOLD signals. Trends Neurosci 29:250–256. [DOI] [PubMed] [Google Scholar]
  24. Martin A, Chao LL ( 2001): Semantic memory and the brain: structure and processes. Curr Opin Neurobiol 11: 194–201. [DOI] [PubMed] [Google Scholar]
  25. McCarthy G, Blamire AM, Puce A, Nobre AC, Bloch G, Hyder F, Goldman‐Rakic P, Shulman RG ( 1994): Functional magnetic resonance imaging of human prefrontal cortex activation during a spatial working memory task. Proc Natl Acad Sci USA 91: 8690–8694. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. McCarthy G, Puce A, Constable RT, Krystal JH, Gore JC, Goldman‐Rakic P ( 1996): Activation of human prefrontal cortex during spatial and nonspatial working memory tasks measured by functional MRI. Cereb Cortex 6: 600–611. [DOI] [PubMed] [Google Scholar]
  27. Mestres‐Missé A, Camara E, Rodriguez‐Fornells A, Rotte M, Münte TF ( 2008): Functional neuroanatomy of meaning acquisition from context. J Cogn Neurosci 20: 2153–2166. [DOI] [PubMed] [Google Scholar]
  28. Mestres‐Misse A, Münte TF, Rodriguez‐Fornells A ( 2009): Functional neuroanatomy of contextual acquisition of concrete and abstract words. J Cogn Neurosci 21: 2154–2171. [DOI] [PubMed] [Google Scholar]
  29. Mestres‐Missé A, Rodriguez‐Fornells A, Münte TF ( 2007): Watching the brain during meaning acquisition. Cereb Cortex 17: 1856–1966. [DOI] [PubMed] [Google Scholar]
  30. Mestres‐Missé A, Rodriguez‐Fornells A, Münte TF ( 2010): Neural differences in the mapping of verb and noun concepts onto novel words. Neuroimage 49: 2826–2835. [DOI] [PubMed] [Google Scholar]
  31. Middleton FA, Strick PL ( 2002): Basal‐ganglia ‘projections’ to the prefrontal cortex of the primate. Cereb Cortex 12: 926–935. [DOI] [PubMed] [Google Scholar]
  32. Nadeau SE, Crosson B ( 1997): Subcortical aphasia. Brain Lang 58: 355–402. [DOI] [PubMed] [Google Scholar]
  33. Nagy WE, Anderson RC ( 1984): How many words are there in printed school English? Read Res Q 19: 304–330. [Google Scholar]
  34. Nagy WE, Anderson RC, Herman PA ( 1987): Learning word meanings from context during normal reading. Am Educ Res J 24: 237–270. [Google Scholar]
  35. Packard MG, Knowlton BJ ( 2002) Learning and memory functions of the basal ganglia. Annu Rev Neurosci 25: 563–593. [DOI] [PubMed] [Google Scholar]
  36. Patterson K, Nestor PJ, Rogers TT ( 2007): Where do you know what you know? The representation of semantic knowledge in the human brain. Nat Rev Neurosci 8: 976–987. [DOI] [PubMed] [Google Scholar]
  37. Raichle ME, McLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL ( 2001): A default mode of brain function. Proc Natl Acad Sci USA 98: 676–682. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Rissman J, Gazzaley A, D'Esposito M ( 2004): Measuring functional connectivity during distinct stages of a cognitive task. Neuroimage 23: 752–763. [DOI] [PubMed] [Google Scholar]
  39. Thompson‐Schill SL, D'Esposito M, Aguirre GK, Farah MJ ( 1997): Role of left inferior prefrontal cortex in retrieval of semantic knowledge: A reevaluation. Proc Natl Acad Sci USA 94: 14792–14797. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Tsukiura T, Fujii T, Takahashi T, Xiao R, Inase M, Iijima T, Yamadori A, Okuda J ( 2001): Neuroanatomical discrimination between manipulating and maintaining processes involved in verbal working memory; a functional MRI study. Cogn Brain Res 11: 13–21. [DOI] [PubMed] [Google Scholar]
  41. Ullman MT ( 2006): Is Broca's area part of a basal ganglia thalamocortical circuit? Cortex 42: 480–485. [DOI] [PubMed] [Google Scholar]
  42. Valenstein E, Bowers D, Verfaellie M, Heilman KM, Day A, Watson RT ( 1987): Retrosplenial amnesia. Brain 110: 1631–1646. [DOI] [PubMed] [Google Scholar]
  43. Vandenberghe R, Price C, Wise R, Josephs O, Frackowiak RS ( 1996): Functional anatomy of a common semantic system for words and pictures. Nature 383: 254–256. [DOI] [PubMed] [Google Scholar]
  44. Vogt BA, Finch DM, Olson CR ( 1992): Functional heterogeneity in cingulate cortex: The anterior executive and posterior evaluative regions. Cereb Cortex 2: 435–443. [DOI] [PubMed] [Google Scholar]
  45. Wahl M, Marzinzik F, Friederici AD, Hahne A, Kupsch A, Schneider G‐H, Saddy D, Curio G, Klostermann F ( 2008): The human thalamus processes syntactic and semantic language violations. Neuron 59: 695–707. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Additional Supporting Information may be found in the online version of this article.

Supporting Information

Click here for additional data file. (1.7MB, doc)

Articles from Human Brain Mapping are provided here courtesy of Wiley

RESOURCES