Brain Mapping in Verbal and Spatial Thinking

DOI 10.1007/s11055-015-0052-5 Neuroscience and Behavioral Physiology, Vol. 45, No. 2, February, 2015 Brain Mapping in Verbal and Spatial Thinking A. M. Ivanitskii,1 G. V. Portnova,1 O. V. Martynova,1 L. A. Maiorova,1,2 O. N. Fedina,2 and A. G. Petrushevskii2 UDC 612.821.6 Translated from Zhurnal Vysshei Nervnoi Deyatel’nosti imeni I. P. Pavlova, Vol. 63, No. 6, pp. 677–686, November–December, 2013. Original article submitted April 15, 2013. Accepted May 20, 2013. The aim of the present work was to describe the topography of active cortical zones and subcortical formations during verbal and spatial thought using functional magnetic resonance imaging (fMRI). A total of 18 right-handed subjects took part. Four types of task were presented – two study tasks, verbal (anagrams) and spatial (seeking figures to complement a square), as well as two types of control task (words and spatial tasks where all the figures are identical). Solution of verbal tasks was associated with a greater volume of activation of the left hemisphere, with involvement of Broca’s area, while solution of spatial tasks was linked with activation of the middle frontal gyrus on the right. In the occipital area, solution of spatial tasks was associated with greater activation of visual field 18, while solution of anagrams was linked with greater activation of field 19, which is associated with a higher level of visual information processing. The cerebellum was activated in both tasks, to a greater extent on solution of spatial tasks. These fMRI data provide evidence that the verbal and spatial types of thought are supported by the activity of a narrowly specific set of brain structures, while previous electrophysiological studies indicated distributed cerebral processes during thought. Combining these two approaches leads to the conclusion that cognitive functions are supported by systems-type cerebral processes with certain key structures. Keywords: brain mapping, fMRI, verbal and spatial thought, anagrams, mental rotation. The English philosopher Chalmers [10] suggested that the “mind and brain” problem consists of two subproblems, which can arbitrarily be defined as hard and easy. The first is to understand how subjective experiences arise from the movement of nerve spikes; the second (which in itself is also very complex) is to describe the concrete brain mechanisms determining mental functions. Several possible answers to the “hard” subproblem have been proposed, in the form of theories of consciousness (see [2]). A powerful thrust towards solving the “easy” problem came from the creation of methods of imaging the living brain, including positron emission tomography, functional magnetic resonance imaging, and multichannel recording of brain electrical activity. The present study was performed to compare the topography of the active cortical fields and subcortical formations in the two basic types of thought: verbal and spatial. There is a quite large body of literature on this question. However, most studies, according to the cognitive psychology paradigm, which separates mental acts into series of sequential actions, generally describe the cortical topography of individual mental operations. The aim of the present work was to follow that whole course of a mental process, from understanding the conditions of the task to taking a decision. We also wanted to minimize differences in second-degree mental operations required for finding the answer, leaving only their essence: operations with verbal or geometrical symbols. This led to selection of two types of task: solving anagrams and searching for figures which complement a base figure to form a square. These tasks are known to be to some extent similar, as finding solutions in both cases requires mental manipulation of image elements – letters or figures – though in the former case the elements form a word with an understood meaning, while in the latter they form a familiar geometrical figure in the form of a square. These studies were performed using functional magnetic resonance imaging, which has high spatial resolution. This provided precise iden- 1 Institute of Higher Nervous Activity and Neurophysiology, Russian Academy of Sciences; e-mail: [email protected]. 2 Speech Pathology and Neurorehabilitation Center, Moscow. 146 0097-0549/15/4502-0146 ©2015 Springer Science+Business Media New York Brain Mapping in Verbal and Spatial Thinking 147 Fig. 1. Examples of the main tasks presented to subjects (A – anagrams, B – figures) and control tasks (C – word, D – all figures identical). [Translator’s note: éääÄò is an anagram of äéòäÄ, which means cat. The control word ñàîêÄ means numeral.] tification of the topography of activated brain structures, including individual Brodmann fields in the cortex. Methods Experiments were performed at the Speech Pathology and Neurorehabilitation Center. A total of 18 right-handed subjects (11 men) took part in the study, mean age 30.7 ± 7.3 years. All participants in the study were Russian-speakers with no abnormalities of vision or hearing or any mental disorders. Subjects were given detailed instructions on solving the tasks and provided signed informed consent to take part in the studies, which were approved by the Ethics Committees of the Institute of Higher Nervous Activity and Neurophysiology, Russian Academy of Sciences, and the Speech Pathology and Neurorehabilitation Center. Experimental paradigm and stimuli. The study paradigm, as noted above, consisted of presenting tasks of two types: verbal and spatial. Verbal tasks consisted of solving anagrams, i.e., finding words from letters presented in random order (Fig. 1). The spatial task consisted of selecting, from three figures presented, the one which, after displacement and mental rotation, formed a square with the figure shown above them. A control task was selected for each study task. The control for anagrams consisted of presentation of a word; the control for the spatial task was presentation of four identical figures positioned as in the study task. The control tasks allowed subtraction of control maps from the maps obtained during solution of the study task to identify additional areas of the cortex activated during performance of mental actions of the verbal or spatial modalities, i.e., the actions intrinsically needed for finding solutions and not associated, for example, with examination of images. Tasks of different groups were presented in random order on an in vivo MRI-compatible monitor (USA). Stimuli presented on the screen were uniform in terms of intensity and brightness for all task types. After a 1-sec fixation spot, tasks of a given type were presented on the screen for 14 sec, after which this sequence (cross + task) was repeated for the new task type. Each type of task was presented 29 times, without repeating any individual task. The words for anagrams and control words were different, but were selected on the basis of encounter frequency and length and were specific words. Words were selected from a frequency dictionary of the Russian language [5], and each word consisted of five letters. All anagrams were normalized in terms of the number of letter movements required to solve the task. The number of such actions was 3–4 (mean 3.4). Tasks were adapted from studies of the classification of the type of mental operations using EEG rhythms [1, 4]; studies used only those spatial and verbal tasks whose solution times were not statistically significantly different in previous experiments using EEG recording. After taking part in the experiment, subjects were asked to comment on the correctness of task solutions, the percentage of correctly solved tasks, and the difficulty of finding solutions. Subjects were also asked to describe the task solution process itself, with allowance for leading questions, i.e., whether permutations of letters or rotations of figures were made or whether the solution came suddenly. Subjects could also spontaneously describe the solution process. Questioning results showed that spatial tasks were successfully solved by all subjects. Most anagrams were also solved successfully. Of the 18 subjects, 15 148 said they only had difficulty in solving 2–3 out of the 29 anagrams. Three subjects reported that they solved the anagrams very quickly, as they found the target word immediately, by insight, without sequential permutation of words. These features of solutions were considered during evaluation of fMRI maps. Structural and functional imaging. MRI scans were obtained on a Magnetomavanto 1.5 Tesla (Siemens, Germany) instrument. Anatomical images in the sagittal plane were obtained using T1MPRAGE sequences (TR = 1900 msec, TE = 3.4 msec, FA = 90°, 174 slices, slick thickness = 1 mm, slice separation distance = 0.5 mm, FoV = 250 mm, reconstruction matrix 256 × 256, voxel size 1 × 1 × 1 mm). BOLD (blood oxygenation level-dependent) effects, i.e., levels of oxygen consumption by given brain structures were recorded using sequences with the following characteristics: (TR = 3000 msec, TE = 50 msec, FA = 15°, 32 axially oriented slices, slick thickness = 3 mm, slice separation distance = 0.8 mm, matrix 64 × 64, voxel size 3 × 3 × 3 mm). The study included a set of 575 measurements (volumes) for each subject. Statistical data analysis methods. Data were analyzed using the statistical program suite SPM8 [24] running on the MATLAB 7.0.4 platform (MathWorks, Natick, MA, USA) at the individual and group levels. At the individual level, the first two functional volumes were excluded from the analysis (to ensure complete relaxation of protons after anatomical scanning); the remaining images were equalized relative to the first volume for correction of movements. Co-recording of mean functional images with structural images and normalization in MNI space were then performed, allowing subsequent analysis of activated areas at the group level. The final stage in data preprocessing consisted of smoothing images using a Gaussian filter with a filter nucleus of size 6 × 6 × 6 mm. Statistical parametric maps were constructed using a general linear model at a significance level of p < 0.05. Individual activation maps were constructed using two types of contrast: the first was “anagram solution – word reading” and the second was “spatial task solution – passive examination of figures.” At the group level, the activity of cerebral zones was assessed using the one-sided t test. The voxel activation threshold corresponded to pFWEcorr < 0.05 with a correction for multiple comparisons, z < 3.09. The activation cluster threshold was determined at the level puncorr < 0.05 (uncorrected for multiple comparisons) with a minimum cluster threshold (FWEc) of 100 voxels. The coordinates of the locations of activation areas were determined using Talairach Daemon [16]. Results The overall data on the topography (in MNI space) and volume of activated brain structures during solution of tasks of two types are shown in Tables 1 and 2. The topography of the activated brain zones is shown in Figs. 2 and 3. Solution of anagrams activated the left (Brodmann fields 44 Ivanitskii, Portnova, Martynova, et al. and 45, Broca’s area [11]) and right (Brodmann field 47) inferior frontal gyri, along with the left middle frontal gyrus (Brodmann field 6). The superior frontal gyrus was activated on both sides (fields 8 and 9). On the medial side of the frontal lobe, Brodmann field 32 was activated bilaterally. The total areas of activation in the frontal lobe on the left and right sides were 3443 and 395 voxels, respectively. In the occipital area, bilateral activation of the fusiform gyrus was seen (field 19; 1112 voxels on the left, 822 voxels on the right). The parietal lobe showed bilateral activation of the superior parietal lobule (fields 7 and 40, 1837 and 1240 voxels for the left and right hemispheres, respectively). Solution of spatial tasks was associated with activation of the left precentral gyrus (Brodmann field 6). On the right side, Brodmann fields 6, 9, 10, and 46 were activated in the middle frontal gyrus, along with field 44 in the inferior frontal gyrus. On the medial side of the frontal lobe, Brodmann field 32 was activated on both sides. The areas of activation of the frontal lobe on the left and right sides were 2225 and 2309 voxels, respectively. The occipital area showed bilateral activation of Brodmann field 18 (4898 and 4580 voxels for the left and right sides, respectively). The parietal area showed bilateral activation of Brodmann field 7 (4465 and 4808 voxels, respectively). In addition, activation of the right cerebellum (4808 voxels) was seen, while activation of the left cerebellum was significantly less marked (869 voxels), as were activation of the thalamus (397 and 401 voxels in the left and right hemispheres respectively) and left-sided activation of the insula (field 13, 111 voxels). Activation of the cerebellum was in both cases more marked on the right. Solution of anagrams was associated with the following individual differences. Three subjects showed no activation on solution of anagrams, as compared with the control word-reading task. Questionnaire results in all three cases indicated that anagram solution was very easy. The subjects noted that the task was solved immediately, after a very short period of time, and the proportion of correctly solved tasks was 98–100%. These subjects’ self-evaluations indicated that they did not permutate the letters or subselect words, as did other subjects; solutions came to them suddenly, as though by insight. On solution of spatial tasks, despite minor differences, the subjects solved the tasks identically successfully. The majority of subjects reported that the tasks were quite easy to solve. The proportion of successfully solved tasks ranged from 85% to 100%. fMRI data also showed that there were no significant individual differences in the subjects. In subjects whose fMRI maps showed differences compared with the other subjects during solution of anagrams, the activation maps during solution of spatial tasks were similar to those of other subjects. Discussion The results obtained here provide evidence that each of the two basic types of thought – verbal and spatial – is sup- Brain Mapping in Verbal and Spatial Thinking 149 TABLE 1. Active Zones of the Brain on Solution of Anagrams Notes. x, y, z are the coordinates of the maximal activation in the cluster, one-way t test (p < 0.05, FWEc = 100). TABLE 2. Active Zones of the Brain on Solution of Spatial Tasks Notes. x, y, z are the coordinates of the maximal activation in the cluster, one-way t test (p < 0.05, FWEc = 100). ported by the activity of a defined set of brain structures specific for the type of thought concerned. We will consider this in more detail after addressing the possible roles of each of these structures in supporting the thought process. For verbal thought, this set includes field 19 on the left and right sides, the superior parietal lobe on both sides, the inferior frontal gyrus mainly on the left, the middle frontal gyrus on the left, Broca’s area, which consists of fields 44 and 45, and the left lobe of the cerebellum. Field 19 is associated with detailed analysis of the shapes of visual stimuli, particularly letter symbols [21]. This field is adjacent to field 37 in the fusiform gyrus, which on the left side is often identified as the visual word shape analysis field. In our case, however, activation of field 19 was bilateral. It is important to bear in mind when evaluating these data that brain maps in control word reading were subtracted from maps obtained for solving anagrams. It can therefore be suggested that bilateral activation of field 19 was induced by the need for more detailed analysis of the shapes of unfamiliar combinations of letters. Visual information from the occipital cortex is known to be divided into two streams [14]. The ventral stream runs from the occipital to the temporal cortex for image recognition, while the dorsal runs to the temporal cortex for determination of the position of the stimulus in the subject’s visual space. Activation of these areas of the cortex is seen both on solution of anagrams and on displacing figures. In the former case, it may be associated with the need to permutate letters, in the latter with displacement of figures. The inferior frontal gyrus is linked with two important functions: the expression of insight [8] and (only the left field) the semantic categorization of words [23]. The function of Broca’s area is quite well known as the speech motor center function [12]. However, its role in verbal thought is more 150 Ivanitskii, Portnova, Martynova, et al. Fig. 2. Group activation maps in healthy volunteers for the contrast between “solving anagrams” and “reading words.” Dark areas are areas of activation (p < 0.01). A) Front view; B) right side view; C) bottom view; D) view from behind; E) left side view; F) view from above. complex and includes seeking the words required for expression of thoughts and lexical meanings [15]. Finally, activation of the left cerebellum plays an important role in mental control during the verbalization of words and is involved in mnemonic speech processes, as well as the process underlying the understanding of words [19]. Thus, structures supporting anagram solution are in any event linked with the processing of verbal information, starting with analysis of the shapes of letter stimuli and ending with insight and mental verbalization of words. These, apart from field 19 and the parietal cortex, which are activated at the early stages of solution, are located on the left side. As noted above, brain maps during solution of anagrams in three subjects who solved the task very easily and quickly showed very little difference from maps obtained as subjects read words. The task in this case was evidently too easy for the subjects, such that no additional influx of oxygen to the fields of the cerebral cortex was required. Solution of the spatial task was characterized by activation of the following structures: field 18, the parietal cor- tex (field 7 on the right and left sides), the precentral gyrus, the right middle frontal gyrus, and the left insula, as well as the following subcortical structures: the cerebellum and thalamus. Field 18 of the visual cortex is associated with analysis of image outlines and shape perception [7]. Its role is therefore similar to the function of field 19, which is activated during solution of anagrams and is linked with higher levels of visual information processing than field 18. It follows that analysis of verbal signals is more difficult than perception of geometrical figures. Activation of the superior parietal area during solution of tasks involving comparison and displacement of figures was more marked than during solution of anagrams, which should be expected for purely spatial tasks. Field 7, like the left precentral cortex, is linked with mental object rotation operations [22]. The right middle frontal gyrus includes fields 5, 9, and 10. The first of these is associated with topographic memory [9]. Fields 9 and 10 are activated during the performance of tasks requiring spatial memory [20]. Activation of the insula is associated with the Brain Mapping in Verbal and Spatial Thinking 151 Fig. 3. Group activation maps in healthy volunteers for the contrast between “solving spatial tasks” and “passive examination of figures.” For further details see caption to Fig. 2. occurrence of insight, as well as with emotional evaluation of actions [18]. The thalamus, apart from its role in transmitting information form the sensory organs to the cerebral cortex, is also a subcortical motor center [13]. Finally, the cerebellum is known to be involved in coordinating movements, and this function may be used in mental figure rotation operations. The cerebral basis of spatial thought is, thus, less lateralized than that of verbal thought, though there is some accent on activity in structures of the right hemisphere. This function also includes such subcortical formations as the thalamus. Spatial thought is evolutionarily older than verbal, and its functions evidently involve formations older than the cortex. Thus, the present study demonstrated the topography of cerebral structures supporting verbal and spatial thought, starting from the projection cortex and ending with the frontal cortex. Taking cognizance of published data, the roles of these structures in supporting the mental process as a whole has also been described. Attention is drawn to the fact that differences in mental actions during solution of cog- nitive tasks, as compared with the simple perception of images displayed on a screen, start at the earliest stages of analysis of visual information, i.e., fields 18 and 19. It is also important that only verbal thought involves activation of the intrinsic speech centers, such as Broca’s area. Spatial thought does not therefore obligately require verbalization of the mental process (to some extent, this also answers the old question: can thinking occur without words?). At the same time, fMRI methods, like other methods, have certain disadvantages. It answers the questions “where?” (the anatomy of a process) and, to some extent, “what” occurs in the brain (what operation is associated with the structure concerned). These data are undoubtedly important both for fundamental knowledge of brain functioning and for clinical aspects – neurology and neurosurgery. However, two important questions remain unanswered: “when” (the precise time sequence of events) and “how” information analysis occurs in the brain, its internal mechanism. These limitations are largely due to the fact that the activity of brain structures is identified indirectly by fMRI, via measurements of oxy- 152 gen consumption by brain tissues. Without answers to these two questions, there can be no general answer to the task identified by Chalmers [10] as “easy,” that of describing the cerebral basis of cognitive functioning. The answer to the “how” question is significant not only for brain science, but also for informatics and the creation of artificial intelligence. Analysis of brain electrical activity provides a partial answer to these questions. We note a number of reports [3, 6] of studies of cortical connections. Thought was shown to involve convergence of connections to particular regions of the cortex: the frontal for verbal and the parietal-temporal for imaginal thought. It was suggested that these cortical areas include the information synthesis processes required for finding a solution. Similar views were developed by Damasio [11], who held that active areas of the cortex identified by fMRI are neural connection centers. Studies reported by Ivanitskii et al. [1, 4] established that thought involves consistent rearrangements of electroencephalogram (EEG) rhythms. The EEG forms stable rhythmic patterns with combinations of frequency, amplitude, and topography of spectral components characteristic of a given type of thought, these forming a rhythmic code for type of thought concerned. This makes it possible to identify the type of mental operations occurring in the mind using EEG rhythms. Trained computer programs, in the form of artificial neural networks, are used for recognition. These have to be used because of the complexity of the combinations of individual spectral features forming during thought. It was also shown that differences between EEG patterns could also be seen in virtually all areas of the cortex. Thus, rhythmic patterns supporting a given type of thought are widely distributed across the cortex. Results from studies of brain electrical activity providing evidence of the distributed, systems, nature of the cerebral mechanisms supporting thought processes are to some extent contradicted by fMRI data on the exact binding of mental actions to particular structures. In fact, this argument is not new: it is found in the classical studies of Lashley and Pavlov and lasted a century [17]. Its continuation and development are found in the concept of the neural network principle of brain operation on the one hand, and the concept of the neuron as detector on the other. At our present level of knowledge, we can take it that the cerebral processes underlying the higher mental functions are systems processes and are distributed across the cortex. The signs of some integral function are represented in its various fields (except, perhaps, the projection zones). At the same time, some structure can be identified in this system which are characterized by greater levels of oxygen consumption and which appear to play leading roles in this complex cerebral integration. Conclusions 1. Verbal thought is associated with activation of Brodmann field 19, the superior parietal cortex, the inferior frontal gyrus mainly on the left, Broca’s area, and the cerebellum bilaterally, though more on the right. Ivanitskii, Portnova, Martynova, et al. 2. Solution of spatial tasks was characterized by activation of field 18 of the visual cortex, the superior parietal cortex, the precentral gyrus on the left, the right middle frontal gyrus, the left insula, and the cerebellum, more on the right, and the thalamus bilaterally. 3. In contrast to brain maps obtained by fMRI scans, studies of brain electrical activity show the systems and distributed nature of cerebral processes supporting thought processes. 4. Combining these two approaches suggests that cognitive functions are supported by systems processes in which the system has a number of key structures. This study was supported by the Russian Foundation for Basic Research (Project Nos. 11-04-01809a, 12-04-01254a, 13-04-01916), the Russian Humanities Scientific Foundation (Project Nos. 11-06-00959a, 12-06-00711a), the “Basic Sciences – Medicine” Program of the Presidium of the Russian Academy of Sciences, and the “Integrative Physiology” Program of the Department of Physiology and Basic Medicine, Russian Academy of Sciences. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. G. A. Ivanitskii, “Recognition of the type of mentally solved task using EEG traces lasting several seconds using a trained classifier,” Zh. Vyssh. Nerv. Deyat., 47, No. 4, 743–747 (1997). A. M. Ivanitskii and G. A. Ivanitskii, “Various aspects of the ‘consciousness and the brain’ problem,” Ros. Fiziol. Zh., 95, No. 10, 1108–1118 (2009). A. M. Ivanitskii and I. R. Il’yuchenok, “Mapping of brain biopotentials during solution of a verbal task,” Zh. Vyssh. Nerv. Deyat., 42, No. 4, 625–635 (1992). G. A. Ivanitskii, R. A. Naumov, and A. M. Ivanitskii, “Technology for identifying the type of mental operations in the mind from electroencephalogram patterns,” Tekhnol. Zhiv. Sistem., 4, No. 5–6, 20–28 (2007). O. N. Lyashevskaya and S. A. Sharov, “A new frequency dictionary of the Russian lexicon,” http://dict.ruslang.ru.freq/php. A. R. Nikolaev, A. P. Anokhin, G. A. Ivanitskii, et al., “Spectral rearrangements of the EEG and the organization of cortical connections in spatial and verbal thought,” Zh. Vyssh. Nerv. Deyat., 46, No. 5, 831–848 (1996). V. A. Fokin, Yu. E. Shelepin, A. K. Kharauzov, et al., “Locations of human cerebral cortical areas activated on perception of ordered and chaotic images,” Ros. Fiziol. Zh., 93, No. 10, 1089–1100 (2007). L. Aziz-Zadeh, J. T. Kaplan, and M. Iacoboni, “‘Aha!’: The neural correlates of verbal insight solutions,” Hum. Brain Mapp., 30, No. 3, 908–916 (2009). A. Berthoz, “Parietal and hippocampal contribution to topokinetic and topographic memory,” Phil. Trans. Roy. Soc. Lon. B. Biol. Sci., 352, No. 1360, 1437–1448 (1997). D. J. Chalmers, “Facing up to the problem of consciousness,” J. Consc. Stud., 2, 200–219 (1995). A. Damasio, Descartes’Error: Emotion, Reason and the Human Brain, Grosset/Putnam, New York (1994). N. F. Dronkers, O. Plaisant, M. T. Iba-Zizen, and E. A. Cabanis, “Paul Broca’s historic cases: high resolution MR imaging of the brains of Leborgne and Lelong,” Brain, 130, No. 5, 1432–1441 (2007). E. V. Evarts and E. T. Thach, “Motor mechanisms of the CNS: Cerebrocerebellar Interrelations,” Annu. Rev. Physiol., 31, 451–498 (1969). Brain Mapping in Verbal and Spatial Thinking 14. 15. 16. 17. 18. 19. M. A. Goodale and A. D. Milner, “Separate visual pathways for perception and action,” Trends. Neurosci., 15, No. 1, 20–25 (1992). S. Heim, K. Alter, A. K. Ischebeck, et al., “The role of the left Brodmann’s areas 44 and 45 in reading words and pseudowords,” Brain Res. Cogn. Brain Res. 25, No. 3, 982–993 (2005). J. L. Lancaster, M. G. Woldorff, L. M. Parsons, et al., “Automated Talairach Atlas labels for functional brain mapping,” Hum. Brain Map., 10, 120–131 (2000). I. P. Pavlov, “The reply of a physiologist to psychologists,” Psychol. Rev., 39, No. 2, 91–127 (1932). K. L. Khan, T. Wager, S. F. Taylor, and I. Liberzon, “Functional neuroanatomy of emotion: a meta-analysis of emotion activation studies in PET and fMRI,” Neuroimage, 1, No. 2, 331–348 (2002). M. C. Silveri, M. G. Leggio, and M. Molinari, “The cerebellum contributes to linguistic production: a case of agrammatic speech following a right cerebellar lesion,” Neurology, 44, 2047–2050 (1994). View publication stats 153 20. 21. 22. 23. 24. S. D. Slotnick and L. R. Moo, “Prefrontal cortex hemispheric specialization of categorical and coordinate visual spatial memory,” Neuropsychology, 44, No. 9, 1560–1568 (2006). B. Söderfeldt, M. Ingvar, J. Rönnberg, et al., “Signed and spoken language perception studied by positron emission tomography,” Neurology, 49, No. 1, 82–87 (1997). G. Vingerhoets, F. P. de Lange, P. Vandemaele, et al., “Motor imagery in mental rotation: an fMRI study,” Neuroimage, 17, No. 3, 1623–1633 (2002). S. Wang, Z. Zhu, J. X. Zhang, et al., “Broca’s area plays a role in syntactic processing during Chinese reading comprehension,” Neuropsychologia, 46, No. 5, 1371–1378 (2008). Welcome Trust Centre for Neuroimaging, www.fil.ion.ucl.ac.uk/spm.